(Raymond Cooper, John K. Snyder) Biology - Chemist (BookFi) PDF

The
Bio Iogy-C he mistry

Interface
ATribute to Koji Nakanishi
edited by
Raymond Cooper
Pharmanex, Inc.
San Francisco, California
John K. Snyder
Boston University
Boston, Massachusetts
MARCEL
MARCELDEKKER,
INC. NEWYORK BASEL
D E K K E R
ISBN: 0-8247-7116-8
This book is printed on acid-free paper.
Headquarters
Marcel Dekker, Inc.
270 Madison Avenue, New York, NY 10016
tel: 212-696-9000; fax: 212-685-4540
Eastern Hemisphere Distribution
Marcel Dekker AG
Hutgasse 4, Postfach 812, CH-4001 Basel, Switzerland
tel: 41-61-261-8482; fax: 41-61-261-8896
World Wide Web

http:/ /www.dekker.com
The publisher offers discounts on this book when ordered in bulk quantities. For more
information, write to Special Sales/Professional Marketing at the headquarters address
above.
Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

Neither this book nor any part may be reproduced or transmitted in any form or by any
means, electronic or mechanical, including photocopying, microfilming, and recording,
or by any information storage and retrieval system, without permission in writing from
the publisher.
Current printing (last digit):

10 9 8 7 6 5 4 3 2 1
PRINTED IN THE UNITED STATES OF AMERICA
TM
Copyright n 1999 by Marcel Dekker, Inc. All Rights Reserved.

Preface
Natural products science, a fascinating cornerstone of modern research, has long

bridged the traditional frontier between chemistry and biology. Humankind has
always been intrigued by the power and potential of plants and nature. Many old
texts reveal how the ancient cultures drew on the beneficial properties of plants.
They learned the wisdom of extracting the ingredients and using such potions as
foods, medicines, and mood enhancers long before anyone understood how these
worked. Slowly we have found the tools to explore the chemistry of these ingredi-
ents, and thus the systematic study of natural products began. Morphine was
isolated in 1805 and strychnine in 1819, although their structures remained mys-
teries for more than 100 years, and pure camphor has been an article of commerce
for centuries. Today the biosynthetic machinery of plants and other organisms
is purposefully manipulated to produce new ‘‘natural products’’ of biological
significance in medicine and agriculture.
In the nineteenth century, early progress in natural products research cen-
tered on the study of pigments from flowers as colored dyes. Originally, extrac-
tion of drug compounds, particularly alkaloids, from plants was achieved by using
simple isolation methods: a water steep or a solvent (generally alcohol) extrac-
tion. The impetus was set to explore this research area further and, not surpris-
ingly, more and more intellectual pursuit of natural sciences and our environment
encouraged universities and scientific centers throughout the world to study natu-
ral products, which then formed the nucleus of chemistry programs.
As source material to begin any research investigation, plants were abun-
dant and easily obtained. The first natural products to be studied in detail were
generally the major constituents of plants, since these often precipitated from
solution and could be purified through recrystallization. How well do we recall
TM

today that the purity of chemicals up to the latter half of the twentieth century
was determined solely by melting point? As increasing numbers of chemical
constituents with more complexity were found, structural analysis relied on chem-
ical transformations and degradation studies. Total synthesis was confirmatory.
The discovery of one or two compounds based on these research studies was
usually acceptable to earn a doctorate.
The second half of the twentieth century has witnessed incredible advances
in natural products research. These have been achieved through the discovery of
new chromatographic separation methods and remarkable advances in spectros-
copy. As new technologies for isolation and structure identification have evolved,
the isolation and detection of ever-diminishing amounts of natural products, cou-
pled with the determination of structures on a microscale, have become almost
routine.
In addition to identifying important targets for total synthesis, and thereby
spurring innumerable advances in fundamental organic chemistry, studies of nat-
ural products have led to significant research efforts in the related fields of bioor-
ganic chemistry and biosynthesis, as chemists, biologists, and biochemists have
striven to understand how these molecules are produced in nature and to establish
the molecular basis of the biological activity of these compounds. The structural
determination of natural products has impacted our basic understanding of nature.
One very important aspect of structure determination is the use of spectroscopy,
particularly nuclear magnetic resonance, mass spectrometry, circular dichroism,
and x-ray diffraction methods. Circular dichroism is particularly important in
establishing absolute stereochemistry, as chirality is correlated directly to biologi-
cal activity of the biomolecule. Thus, as we approach the end of this century,
we see the challenging questions in biology requiring answers at the molecular
level being met by increasingly sophisticated techniques and comprehension of
the chemistry of nature.
Professor Koji Nakanishi has been a pioneer and a towering figure in natural
products research. He has been a major contributor at the crossroads of bioorganic
chemistry over the past 40 years. His extraordinary and broad vision of natural
products chemistry and its close relationship to bioorganic studies is now univer-
sally accepted and was the inspiration for this book. He has constantly looked
at challenges in bioorganic chemistry and pushed ever closer the boundaries at
the interface between chemistry and biology. He has achieved this through his
lifelong studies in natural products, his investigations into the chemistry of vision,
his pursuit of new and ever more powerful analytical and spectroscopic micro-
techniques for solving complex structural problems, and his study of infrared and
circular dichroism and their applications to bioorganic science. Koji’s curiosity
and insights in applying the right solution to challenging problems are among
his legacies, to which we as students of his are deeply grateful.
Koji Nakanishi was born in 1925 in Hong Kong to parents of Japanese
TM

descent. As a result of his father’s business postings abroad, Koji’s early child-
hood was spent in various European capitals as well as in Alexandria, Egypt,
thereby giving birth to and nurturing his unique world vision. He returned to
Japan for his formal university training and received his B.Sc. degree at Nagoya
University in 1947. He first came to the United States in 1950–52 to study with
the legendary Professor Louis Fieser at Harvard University, and he returned to
Japan as an assistant professor to embark on his remarkable career in natural
products and bioorganic chemistry. He completed his Ph.D. in 1954 under the
mentorship of Professors Egami and Hirata, then took positions at Nagoya Uni-
versity (1955–58), Tokyo Kyoika University (1958–63), and Tohoku University
(1963–69). In 1969 he was invited to join the faculty at Columbia University,
New York, where he currently holds the chair of ‘‘Centennial Professor of Chem-
istry.’’
Indeed, it was at Columbia University that former and current students,
postdoctoral fellows, and esteemed colleagues of Professor Nakanishi gathered
to celebrate his 70th birthday and to honor him for his years of mentorship and
friendship, as well as for his considerable contributions to bioorganic science.
Two days of stimulating presentations on various topics in bioorganic chemistry
gave birth to the idea behind this book: to produce a volume with contributed
chapters from his former students in his honor.
This text reflects Koji’s own research interests in its scope and attempts to
bridge the gap between biology and chemistry: a gap that is rapidly diminishing
as investigators use the tools and vision that Koji has provided. Koji humbly
reminds us that he is ‘‘only a technician’’; we respectfully differ. He is a vision-
ary, and in essence the investigatory seeds planted by Koji are now in full bloom
in research gardens headed by those he taught.
We choose to highlight current research activities from former members of
his research groups from Asia, the United States, Europe, and Australia, thereby
illustrating Koji’s global scientific influence. As with Koji’s own research, one
goal of this text is to further dissolve the boundary that has kept chemistry and
biology apart; the contributions in this volume are by investigators for whom
this boundary has long since disappeared. It is hoped that readers will come to
understand the highly interactive nature of research in biological chemistry and
chemical biology, and find the transition between chemistry and biology far less
intimidating. Thus, this book reflects the ideals of Professor Nakanishi and his
impact.
Although the chapter titles may at first glance seem to suggest a relatively
large breadth of subjects, in fact they all fit snugly within the focus of the chemi-
cal basis of biological activity. Subjects range from hydrolytic enzymes to combi-
natorial chemistry, yet all the chapters strive to elucidate biological responses at
the molecular level. The contributing authors provide detailed accounts of their
current research rather than presenting formal reviews of disparate subjects. The
TM

rationale for this approach is to emphasize the interactive nature of the research
in bioorganic chemistry. The unifying theme throughout is the original skills that
developed in natural products research and chemistry of vision. The microanalyti-
cal techiques, the spectroscopic challenges, have now evolved into the application
of chemical minds to biological problems. Thus, the selection of authors reflects
a blend of investigations in academic and industrial research.
Koji’s pioneering contributions and world vision of science have inspired
several generations of chemists from around the globe, and demonstrations of
his mastery of the magical arts have left numerous audiences of the brightest
minds completely and delightfully mystified. We can identify the defining mo-
ment of our education and scientific growth as the time we spent with Koji, and
we offer our profoundest gratitude to him for his tireless leadership and support.
Raymond Cooper
John K. Snyder
TM

Contents
Preface
Contributors
Tribute Letters
1. Insect Antifeeding Limonoids from the Chinaberry Tree Melia azedarach

Linn. and Related Compounds
Munehiro Nakatani
2. Polygodial and Warburganal, Antifungal Sesquiterpene Dialdehydes and
Their Synergists
Isao Kubo
3. Marine Bromoperoxidases—Chemoenzymatic Applications
Chris A. Moore and Roy K. Okuda
4. LC-Hyphenated Techniques in the Search for New Bioactive Plant
Constituents
Kurt Hostett aryse Hostettmann, Sylvain Rodriguez, and
Jean-Luc Wolfender
5. Determination of the Absolute Configuration of Biologically Active
Compounds by the Modified Mosher’s Method
Takenori Kusumi and Ikuko I. Ohtani
6. Circular Dichroism Spectroscopy and the Absolute Stereochemistry of
Biologically Active Compounds
Nobuyuki Harada
TM

7. Recent Applications of Circular Dichroism to Carbohydrate
Conformational Analysis and Direct Determination of Drug Levels
Jesús Trujillo Vázquez
8. Furan-Terminated Cationic π-Cyclizations in the Synthesis of Natural
Products
Steven P. Tanis
9. Chemistry and Biology of Semisynthetic Avermectins
Timothy A. Blizzard
10. Chemical and Biological Approaches to Molecular Diversity:
Applications to Drug Discovery
Harold V. Meyers
11. Imidazoline Receptors and Their Endogenous Ligands
Colin J. Barrow and Ian F. Musgrave
12. Oxidoredox Suppression of Fungal Infections by Novel
Pharmacophores
Valeria Balogh-Nair
13. A Mechanistic Analysis of C—O Bond Cleavage Events with a
Comparison to 3,6-Dideoxysugar Formation
David A. Johnson and Hung-wen Liu
14. The Molecular Mechanism of Amyloidosis in Alzheimer’s Disease
Michael G. Zagorski
15. Bacteriorhodopsin Structure/Function Studies: Use of the Demethyl
Retinal Analogues for Probing of the Arg82Ala Mutant
Rosalie K. Crouch, Donald R. Menick, Yan Feng, Rajni ee, and
Thomas G. Ebrey
16. Autonomous Genomes
David G. Lynn
17. Stereochemical Considerations of Immunoglobulin Heavy Chain
Enhancer Activation
Barbara S. Nikolajczyk and Ranjan Sen
TM

Contributors
Valeria Balogh-Nair Department of Chemistry, The City College of the City

University, New York, New York
Colin J. Barrow School of Chemistry, The University of Melbourne, Parkville,

Victoria, Australia
Timothy A. Blizzard Medicinal Chemistry, Merck Research Laboratories,

Rahway, New Jersey
Rosalie K. Crouch Department of Ophthalmology, Medical University of

South Carolina, Charleston, South Carolina
Thomas G. Ebrey School of Cellular and Molecular Biology, University of

Illinois at Urbana-Champaign, Urbana, Illinois
Yan Feng Department of Ophthalmology, Medical University of South Caro-

lina, Charleston, South Carolina
Rajni Govindjee Center for Biophysics and Computational Biology and De-
partment of Cell and Structural Biology, University of Illinois at Urbana-Cham-
paign, Urbana, Illinois
TM

Contributors
Nobuyuki Harada Institute for Chemical Reaction Science, Tohoku Univer-

sity, Sendai, Japan
Kurt Hostettmann Institut de Pharmacognosie et Phytochimie, Université de

Lausanne, Lausanne, Switzerland
Maryse Hostettmann Institut de Pharmacognosie et Phytochimie, Université

de Lausanne, Lausanne, Switzerland
David A. Johnson Department of Chemistry, University of Minnesota, Minne-

apolis, Minnesota
Isao Kubo Department of Environmental Science, Policy, and Management,

University of California, Berkeley, California
Takenori Kusumi Faculty of Pharmaceutical Sciences, Tokushima University,

Tokushima, Japan
Hung-wen Liu Department of Chemistry, University of Minnesota, Minneapo-

lis, Minnesota
David G. Lynn Department of Chemistry, The University of Chicago, Chi-

cago, Illinois
Donald R. Menick Departments of Medicine, and Biochemistry and Molecular

Biology, Medical University of South Carolina, Charleston, South Carolina
Harold V. Meyers Chemistry and Drug Discovery Group, New Chemical Enti-
ties, Inc., Framingham, Massachusetts
Chris A. Moore Department of Chemistry, San José State University, San José,
California
Ian F. Musgrave Prince Henry’s Institute for Medical Research, Clayton, Vic-
toria, Australia
Munehiro Nakatani Department of Chemistry and Bioscience, Kagoshima

University, Kagoshima, Japan
TM

Barbara S. Nikolajczyk* Brandeis University, Waltham, Massachusetts
Ikuko I. Ohtani Department of Chemistry, Biology, and Marine Science, Uni-

versity of the Ryukyus, Okinawa, Japan
Roy K. Okuda Department of Chemistry, San José State University, San José,
California
Sylvain Rodriguez Institut de Pharmacognosie et Phytochimie, Université de

Lausanne, Lausanne, Switzerland
Ranjan Sen Rosenstiel Research Center and Department of Biology, Brandeis

University, Waltham, Massachusetts
Steven P. Tanis Medicinal Chemistry I, Pharmacia & Upjohn, Inc., Kalama-

zoo, Michigan
Jesús Trujillo Vázquez Instituto Universitario de Bio-Orgánica ‘‘Antonio

González,’’ Universidad de La Laguna, Tenerife, Spain
Jean-Luc Wolfender Institut de Pharmacognosie et Phytochimie, Université

de Lausanne, Lausanne, Switzerland
Michael G. Zagorski Department of Chemistry, Case Western Reserve Uni-

versity, Cleveland, Ohio
* Current affiliation: Immunobiology Unit, Departments of Medicine and Microbiology, Boston Uni-
versity Medical School, Boston, Massachusetts.
TM

Tribute Letters
March 16, 1998
Dear Koji,
On the occasion of your 70th birthday, many of your old friends and colleagues
came to Columbia for a wonderful celebration in 1995. Now we are assembling
a permanent record of our appreciation for your friendship and as a tribute to
your many elegant and important contributions to the chemistry and biology of
natural products.
I can remember our first meeting, 34 years ago, in Tokyo, at the Presympo-
sium to the IUPAC meeting in Kyoto. In the next two weeks you were our host
almost every evening and introduced us to Japanese food and customs. It was a
great awakening to the realization that Japanese chemistry was rapidly gaining
tremendous momentum and a turning point in my career. Since that meeting I
have had the pleasure and privilege of working with 26 Japanese colleagues (sev-
eral of whom came from your lab).
It was a pleasure to repay a little of your hospitality when you visited our
homes in Sussex, New Haven and College Station, and you know that you and
your wife are always welcome in Texas.
You are a true pioneer in solving different problems at the chemistry–
biology interface using every possible technique on vanishingly small amounts
of material and your work continues to be an inspiration to all of us. Most impor-
tantly your personal qualities have ensured a permanent legacy in your many
students who have done so well in our profession. It must make you feel very
TM

proud to have had so many loyal and dedicated coworkers. Above all (and almost
uniquely in our field) you have remained a gentleman, with the highest ethical
standards in dealing with your colleagues. I don’t know how you manage to work
so hard yet still find time for your magic and your friends.
I can guess the secret of your success in chemistry and life—that you are
fortunate like myself, to have such a long and happy marriage.
Betty joins with me in wishing you and your wife continued health, happi-
ness and success for many more birthdays to come.
As always!
Yours very sincerely,
A. I. Scott, F.R.S.
Davidson Professor of Science
Director of Center for Biological NMR
Texas A&M University, College Station, Texas
March 8, 1998
Dear Koji,
‘‘They’’ never stop celebrating you! ‘‘They,’’ of course, are those who have had
the privilege to obtain from you, as post-docs or as Ph.D. students, part of their
life baggage. They have been also kind enough to associate to them some of your
long time friends, and it was indeed a great pleasure to have the opportunity to
pay tribute to you in Columbia nearly half a century after we had first met in
the basement of Converse Laboratory, at Harvard, in Louis Fieser’s group.
When I was invited to contribute to this volume with a letter, I tried to call
back the oldest memories of our meetings I could muster. For some odd reason,
even though I am neither a gourmet nor a gourmand, they were nearly all memo-
ries of food. The experience of learning from you (and from Huang Wey Yuan)
to use chopsticks (a very useful lesson), the dinners of frog1 or lamb2 legs in
1
My wife was working at Harvard Medical School in Pharmacology with Fieser’s friend Prof. O.
Krayer, on the action of the Veratrum alkaloids on frog heart. A frog: one heart, two legs. We
always had a few dozen frozen legs aside for our friends. These legs, and frozen guinea pigs (one
heart, one guinea pig), helped us survive on our starvation scholarships.
2
On affluent months, for a change from frogs and guinea pigs, I was buying lamb by the half at the
TM

the stable-boy’s rooms of the mock-French castle I was living in with Paula in
Brookline.3
Food apart, another very old memory I can retrieve is that of the innumera-
ble small sealed tubes in which you were desperately heating pristimerin with
something (was it zinc or selenium?), to find its structure.
Food and pristimerin apart, I also revive with a little nostalgy our outing
to White Mountains, to a mountain the name of which escapes me, when we had
to climb very large oblique stone slabs and you lost grip at the top, to slide down
slowly at first, then quickly, on your fingers and stomach, to arrive at the bottom
half skinned.
But, food and pristimerin and outing apart, it is also then that I first became
one of your favorite stooges, always ready to serve on any available stage, many,
many times later, as a faire-valoir to your other profession. A simpleton glad to
oblige.
Koji, I have only one regret: that I could never find a pretext to share some
(serious) work with you, in Japan, in Nairobi or in New York, even though we
have both always held the same conviction that chemistry and biology share more
than one border and that to follow fashion is silly when there is so much else to
explore.
Merci, Koji, pour ton amitié
Professor Guy Ourisson

Vice President de l’Académie des Sciences
Strasbourg, France
Italian market. The subway passengers had to sit next to a very un-American young man carrying
half a lamb protruding from his rucksack.
3
We were living in the small rooms above the stables built by a former U.S. Ambassador to Italy
and to Japan, in the middle of the huge estate he had bequeathed to the township of Brookline,
Lars Andersen Park. The stables were in the form of the Château de Chambord, or nearly so, and
were the seat of the Veteran Motor Cars Association of America, the ‘‘vie de château,’’ which we
shared with some 100 old cars.
TM

1
Insect Antifeeding Limonoids from
the Chinaberry Tree Melia
azedarach Linn. and Related
Compounds
Munehiro Nakatani
Kagoshima University, Kagoshima, Japan
I. INTRODUCTION
Limonoids are tetranortriterpenoids derived from euphane (H-20β) or tirucallane

(H-20α) triterpenoids with a 4,4,8-trimethyl-17-furanylsteroidal skeleton [1].
Over 300 limonoids have been isolated to date, and they are the most distinctive
secondary metabolites of the plants in the order Rutales. Particularly, they charac-
terize members of the family Meliaceae, where they are abundant and varied [2–
4]. Almost every part of the trees of this family has been used in folkloric and
traditional systems of medicine [5,6]. Recent work has established a wide range
of biological activities for these compounds, including insect antifeedant and
growth-regulating properties, a variety of medicinal effects in animals and hu-
mans, and antifungal, bacteriocidal, and antiviral activities. The biological activi-
ties of limonoids from Rutales have been reviewed [7].
In particular, the limonoids from the neem tree Melia azadirachta indica
Juss and the Chinaberry tree Melia azedarach Linn. have attracted considerable
interest because of their marked insect antifeedant properties and intriguing struc-
tural variety. The most potent insect antifeedants are azadirachtin and related
highly oxidized C-seco limonoids from M. azadirachta. Their antifeedant activi-
ties and structure–activity relationships have been reviewed [8–10].
TM

Nakatani
Similarly, Melia azedarach is known to produce azadirachtin-type limo-

noids that are potent antifeedants such as the meliacarpinins, and C-19/C-29
bridged lactols and acyl acetals such as the azedarachins, and trichilins. We have
investigated the limonoid constituents of M. azedarach and some related plants
and evaluated the antifeedant properties of more than 30 limonoids in this work,
which includes all of the types of limonoids isolated from M. azedarach exclud-
ing glycosides. The structures and antifeedant activities of these compounds are
detailed here.
II. LIMONOIDS FROM Melia azedarach Linn.
Melia azedarach is a native of Persia, India, and China but is naturalized in a

number of continents including Africa, Australia, and the Americas. Thus, the
constituents of this tree from many regions have been studied. Limonoids may
be found in all tissues of the plant, but different organs within an individual may
produce different types of limonoids. Since the initial isolation of gedunin (29),
nimbolin A (9), and nimbolin B (32) from the trunk wood, reported in 1969 [11],
about 55 limonoids have been isolated from the fruits, stem bark, and root bark
to date.
A variety of oxidation and skeletal rearrangements of the basic limonoid
skeleton (meliacane) are found (Figs. 1–4). A characteristic oxidation of the skel-
eton that is found in M. azedarach produces C-19/C-29 bridged acyl acetals with
a 14,15-epoxide as in the azedarachins (10–16, Fig. 2) and trichilins (17–23, Fig.
2), or their D-ring keto compounds (24–28, Fig. 2). In addition, spirosendan (35,
Fig. 3), recently isolated by us, is a novel spiro limonoid possessing a C-12/C-
30 bridged system, the first such limonoid substructure found in nature, to the
best of our knowledge [12a]. In limonoids of other origins, the A- or D-rings
commonly is oxidized to lactones (A- or D-seco limonoids), in the latter case
presumably by introduction of a carbonyl function at C-16 followed by a Baeyer-
Villiger-type oxidation. But these derivatives are rarely found in M. azedarach,
with the sole exceptions to date being gedunin (29) and the related 3-glucoside
(30).
Many limonoids possess a 14,15-epoxide. Furthermore, the C-ring is fre-
TM

Figure 1 Limonoids from M. azedarach, Group 1: apo-euphol limonoids.
quently oxidatively cleaved to a carboxylic acid or lactone (C-seco limonoids).

The C-seco limonoids, which are major ring-seco limonoids of M. azedarach,
M. azadirachta, and M. toosendan in the family Meliaceae and are found only
in these three related species, may occur by one of several mechanisms, and there
is controversy about the mechanism by which the C-12/C-13 bond of the C-ring
is opened [1,13,14]. The series of sendanal (8), ohchinal (31), nimbolidins A
(43), and B (44), and nimbolin B (32) illustrates one possible process involved,
which includes introduction of an oxygen function at C-12, followed by oxidation
to a ketone. Cleavage of the C-12/C-13 bond is accompanied by the simultaneous
opening of a 14,15-epoxide to generate the CHO-12 and allylic 15-OH functions
TM

Nakatani
Figure 2 Limonoids from M. azedarach, Group 1: apo-euphol limonoids (continued ),

and Group 2: D-seco limonoids.
TM

Figure 3 Limonoids from M. azedarach, Group 3: C-seco limonoids.
TM

Nakatani
Figure 4 Limonoids from M. azedarach, Group 3: C-seco limonoids, highly oxidized

compounds.
Scheme 1
TM

(Scheme 1, Route 1) as found in (31). Rotation about the C-8/C-14 bond would
allow the 15-OH to recyclize with CHO-12 to form a lactol C-ring as in (32).
Alternatively, an electron transfer process may generate a radical cation that sub-
sequently opens to an allylic radical with a remote C-12 acylium-type ion inter-
mediate. Capture of the allyl radical by a neighboring C-7 oxygen substituent
would lead to C-seco derivatives with the C-7/C-15 oxo bridge such as (31)
(Scheme 1, Route 2). The C-12 keto limonoid nimbidinin found in the seeds of
Azadirachta indica [15] may serve as a precursor for this latter route.
Meliacarpinin A (46, Fig. 4), one of the most potent insect antifeedants,
was isolated from the root bark of Okinawan M. azedarach in 1993 [16]. This
limonoid and the related meliacarpinins (47–54) are more highly oxidized natural
products of the C-seco class similar to the azedarachins. 1-Cinnamoylmeliano-
lone (45), found in the fruits of the American variety, may be a precursor to the
meliacarpinins. It is of interest that none of the azadirachtin-type limonoids iso-
lated from M. azedarach possesses a 4β-carboxylate group like azadirachtin, but
rather a 4β-methyl group [17].
III. STUDIES ON THE LIMONOID ANTIFEEDANTS OF THE

CHINESE Melia azedarach
A. Isolation of Limonoids from the Root Bark
The ether and methanol extracts of the root bark contained a variety of limonoids,
which were detectable by the characteristic color upon treatment with Ehrlich’s
reagent on thin layer chromatography (TLC). Antifeedant limonoids are often
very sensitive to traces of acid and gradually decompose on a silica gel column. It
was, therefore, necessary to use droplet countercurrent chromatography (DCCC)
[18,19], flash chromatography, and high-performance liquid chromatography
(HPLC) separation techniques that eliminate or at least minimize contact with a
potentially acidic stationary phase. The isolation of the various limonoid conge-
ners was often a tedious process requiring careful use of HPLC. The isolations,
which were monitored by antifeedant assays described in the next section, were
accomplished as outlined in Fig. 5.
B. Bioassay of Antifeedants
The antifeedant assay results reported here were obtained with the Southern army
worm Spodoptera eridania (Boisduval) [20] as the test species. Spodoptera spe-
cies are distributed throughout the world and constitute a major agricultural
threat. The feeding bioassay was carried out by the conventional leaf disk method
[21], using 2-cm-diameter leaf disks cut from the Chinese cabbage Brassica
campestris L. var. chinensis (Cruciferae) with a cork-borer. Each disk was dipped
TM

Nakatani
Figure 5 Isolation scheme for the limonoids from the root bark of M. azedarach.
for 2 s in an acetone solution of the sample; 5 treated disks were arranged alter-
nately with another 5 control disks (immersed for 2 s in acetone alone), all
concentrically placed near the periphery in a Petri dish, as illustrated in Fig. 6.
Subsequently, 10 third-instar larvae were placed in the center of the dish, and
the treated and untreated leaves eaten by the larvae in 2-hour to 24-hour periods
were evaluated at appropriate intervals. The bioassay was terminated after the
TM

Figure 6 Strategy for the antifeedant bioassay.
larvae had eaten approximately 50% of the control disks, which usually took 6–
12 h. When the average eaten area of the treated disks was visually judged to
be less than 50% of that of the control disks, the test compound was judged to
be active. This bioassay was used to guide the isolations to the active limonoids,
as well as to assess their relative antifeedant activities. To determine the minimum
inhibitory concentration, this choice test was done at 50, 100, 150, 200, 300, 400,
500, and 1000 ppm, with 50 ppm corresponding to a concentration of ca. 1 µg/
leaf-cm2.
C. Structures of the Trichilins

The series of limonoids called the trichilins was first isolated from the root bark
of the East African medicinal plant Trichilia roka (Meliaceae) [22]. The struc-
tures of trichilin B (17) and its 12-epimer (trichilin A) were elucidated by exten-
sive 1H and 13C nuclear magnetic resonance imaging (NMR) studies and through
chemical correlation. Some pertinent points related to these structural studies are
listed as follows:
1. Irradiation of the 13- and 8-Me peaks induced 30% and 12% Nuclear
Overhauser Effect (NOE) enhancements of the 9-H and 19-H signals,
respectively (see conformational drawing, Fig. 7).
2. The assignment of the 12-OH stereochemistry as β in trichilin A and
α in trichilin B (17) was deduced from the finding that in their 12-p-
bromobenzoates, the aromatic protons of the benzoate and furan rings
were at higher field in trichilin B. Thus, the shifts of the p-bromoben-
zoate protons were (for trichilins A/B) o-H δ 7.99/7.65, m-H δ 7.51/
TM

Figure 7 1H NMR data for trichilin B (17) in CDCl3, 400 MHz, in δ (multiplicity and
J values), and NOE correlations (double-headed arrows). NMR, nuclear magnetic reso-
nance; NOE, nuclear overhauser effect.
7.59, and those of the furan were 21-H δ 7.20/7.02, 22-H δ 6.36/5.98,
and 23-H δ 7.37/7.10. The higher-field chemical shifts of these aro-
matic protons in trichilin B can be accounted for by the mutual
shielding induced by the ring currents of the two aromatic rings, which
are located on the same side of the molecule. More directly, however,
the 12-OH configurations were independently derived from a new addi-
TM

tivity relation found in the Cotton effects of multiple coupled chromo-
phores in the CD spectra [23]. For the trichilins A and B, the 7,12-
bis-p-bromobenzoate exciton couplings between the two benzoates as
well as with the furan chromophores distinguished the C12 stereo-
chemistry.
3. Oxidation of trichilins A and B with pyridinium chlorochromate (PCC)
in CH2Cl2 afforded the same α-diketone I (Scheme 2) as the 7,11,12-
trione; thus A and B are 12-epimers.
4. When trichilins A and B were treated with a catalytic amount of p-
TsOH, they were converted quantitatively into the isomeric trichilin C
(Scheme 2) and its 12-epimer, respectively.
5. Finally, treatment of trichilin B with Zn(BH4)2 led unexpectedly to a
Lewis acid–catalyzed acyl migration in ring A to give a mixture of
trichilin B and its 1,2-diacetyl analogue aphanastatin (23), the structure
of which had been determined by x-ray analysis [24]. This unexpected
conversion confirmed the structure of trichilin B.
Scheme 2
TM

D. The Structure of Meliacarpinin A
The meliacarpinins (46–50, and 52) are highly oxidized C-seco limonoids iso-
lated by us from the Chinaberry tree M. azedarach that are structurally very
similar to the azadirachtinins. The structure of meliacarpinin A (46) [25] was
mainly elucidated by comparing the NMR data with those of the azadirachtins
isolated from M. azadirachta. The 1H NMR spectrum was nearly superimposable
on that of 1-tigloyl-3-acetoxy-11-methoxyazadirachtinin [26] except for the
change of the 4β-methyl carboxylate to a 4β-methyl group. The NMR data and
selected NOE connectivities used to elucidate the structure and stereochemistry
are shown in Fig. 8.
Figure 8 1H NMR data for meliacarpinin A (46, cin ⫽ cinnamoyl) in CDCl3, 400 MHz,
in δ (multiplicity and J values), and NOE correlations (double-headed arrows). NMR,
nuclear magnetic resonance; NOE, nuclear overhauser effect.
TM

IV. ANTIFEEDANT ACTIVITY OF THE LIMONOIDS FROM
Melia azedarach
The limonoids isolated from M. azedarach and their effects on insect feeding
are summarized in Table 1. To develop a quantitative understanding of structure–
activity relationships among limonoids, the data are summarized primarily for
the activity against the third-instar larvae of a voracious pest insect, Spodoptera
eridania (Boisduval). It is important to note that differences in the response of
test insects when compared with different test species, or even different life cycle
stages of the same species, can mask any meaningful observations of structure–
activity relations. This is apparent from a comparison of the data reported against
different insects for compounds (11, 22, 39, and 40) in Table 1. Even where the
same bioassay species has been used, differences in the larval stage tested may
make comparisons invalid.
Some quantitative trends are apparent in the data in Table 1. Aside from
the highly oxidized C-seco meliacarpinins and azadirachtins, the most active
compounds appear to be intact apo-euphol limonoids with a 14,15-epoxide and
a C-19/C-29 acyl acetal bridged system. Structure–activity relationships in this
class will be discussed next.
Within the C-seco class, the highly oxidized meliacarpinins are the most
active of the limonoids from M. azedarach. Although their activities may be
weaker than that of the azadirachtins from M. azadirachta (effective dose of 50
ppm for 46–52 against S. eridania in comparison to 14 ppm for azadirachtin
against Epilachna varivestis), they are much more potent antifeedants than the
less oxidized class of C-seco limonoids exemplified by salannin (39), nimbolin
B (32), and ohchinolide B (37). The second most active class of limonoids from
M. azedarach is the C-19/C-29 bridged acetal class with the intact apo-euphol
skeleton (10–28) (effective dose 150–500 ppm). The two members of the intact
apo-euphol limonoid class without this C-19/C29 acetal bridge that were tested
against S. eridania, azadiron (1), and nimbolin A (9) showed little or no activity
(effective dose ⱖ 1000 ppm).
V. STRUCTURE–ACTIVITY RELATIONSHIPS IN C-19/C-29

BRIDGED ACETALS
During the course of our structural and chemical correlation studies, about 50
compounds belonging to the C-19/C-29 bridged acetal class of limonoids were
made available. Antifeedant assays with S. eridania (leaf disk method) showed
several interesting structure–activity correlations.
First, activity is insensitive to substituent variation in the A-ring except for
the nature of the C29 bridging position: A hemiacetal bridge (hydroxyl group at
C29) increases the activity in comparison to that of an acylated hemiacetal (com-
TM

Table 1 Antifeeding Activity of Limonoids from Melia azedarach
Effective References
concentration
Limonoid Test insect (ppm) Isolation Activity a
Group 1: Intact apo-euphol limonoids

Azadirone (1) Spodoptera eridania Inactive 27 12
Meldenin (2) 28,29
6-Acetoxy-7α-hydroxy-3-oxo-14β,15β-epoxymeliac-1,5-diene (3) 29 NA
6-Acetoxy-3β-hydroxy-7-oxo-14β,15β-epoxymeliac-1,5-diene 29 NA
3-O-β-d-glucopyranoside (4)
6-Acetoxy-3β-hydroxy-7-oxo-14β,15β-epoxymeliac-1,5-diene 30 NA
3-O-β-d-xylopyranoside (5)
6-Acetoxy-3β,11α-dihydroxy-7-oxo-14β,15β-epoxymeliac-1,5-diene 31 NA
3-O-α-l-rhamnopyranoside (6)
6,11α-Diacetoxy-3β-hydroxy-7-oxo-14β,15β-epoxymeliac-1,5-diene 32 NA
Sendanal (8) 33 NA
Nimbolin A (9) S. eridania 1000 11 12
C-19/C-29 bridged acyl acetals
Amoorastatin (10) S. eridania 150 34,35 12
Toosendanin (11) O. furnacalis 200 36,37 35
S. eridania 300 20
Azedarachin A (12) S. eridania 200 38 37
12-O-Acetylazedarachin A (13) S. eridania 400 38 37
Azedarachin C (14) S. eridania 400 39 37
Azedarachin B (15) S. eridania 200 12 12
12-O-Acetylazedarachin B (16) S. eridania 400 40 37
Trichilin B (17) S. eridania 200 22,40 40
TM

TM
C
12-O-Acetyltrichilin B (18) S. eridania 400 40 20
1,12-Di-O-acetyltrichilin B (19) S. eridania 400 40 40
Trichilin D (20) S. eridania 400 22,40 40
Trichilin H (21) S. eridania 400 40 40
Meliatoxin A 2 (22) S. litura 300 40,41 42
S. eridania 400 40
Aphanastatin (23) S. eridania 200 24 20
Amoorastatone (24) S. eridania 400 34,35 12
12-Hydroxyamoorastatone (25) S. eridania 300 35 12
iso-Chuanliansu (26) S. eridania 400 37 12
Meliatoxin B 1 (27) S. eridania 500 41 12
Meliatoxin B 2 (28) S. eridania 500 41 12
Group 2: D-seco limonoids
Gedunin (29) O. nubilalis 500 11 43
7α-Acetoxy-3β-hydroxy-14β,15β-epoxygedunan-1-ene 44 NA
Group 3: C-seco limonoids
Ohchinal (31) S. frugiperda Inactive 45 46
Nimbolin B (32) S. eridania 1000 11 12
Nimbolinin B (33) S. eridania 1000 47 48
1-Deacetylnimbolinin B (34) 47 NA
Spirosendan (35) 12 NA
Ohchinolide A (36) 47,49 NA
Ohchinolide B (37) S. eridania 700 47,49 12
Ohchinolal (38) S. eridania 1000 50 12
Salannin (39) S. eridania 1000 51 48
S. littoralis 100 52
Deacetylsalannin (40) S. eridania 1000 51 12
E. varivestis 30 53
Ohchinin (41) 50 NA
TM

TM
C
Table 1 Continued
Effective References
concentration
Limonoid Test insect (ppm) Isolation Activity a
Ohchinin acetate (42) S. eridania 1000 45 12

Nimbolidin A (43) S. eridania 500 47 20
Nimbolidin B (44) S. eridania 500 47 20
Highly oxidized compounds
1-Cinnamoylmelianolone (45) 54 NA
1-Cinnamoyl-3-acetyl-11-methoxymeliacarpinin (46, Meliacar- S. eridania 50 25 25
pinin A)
3-Deacetylmeliacapinin A (47) 55 NA
Meliacarpinin B (48) S. eridania 50 48 48
Meliacarpinin C (49) S. eridania 50 48 48
Meliacarpinin E (50) S. eridania 50 12 12
20-O-Acetylmeliacarpinin C (51) 55 NA
Meliacarpinin D (52) S. eridania 50 48 48
20-O-Acetylmeliacarpinin D (53) 55 NA
1-Deoxy-3-methacrylyl-11-methoxymeliacarpinin (54) 55 NA
Azadirachtin (55, Azadirachtin A) E. varivestis 14 56,57 53
Other trichilins from different species and reaction products
1: Natural products
Trichilin A (56) S. eridania 300 22 22
(C12-epimer of trichilin B)
7-O-Acetyltrichilin A (57) S. eridania 400 59 12
Trichilin C (58) S. eridania 500 22 12
(15-Keto isomer of trichilin A)
TM

TM
C
12α-Epimer of trichilin C (59) S. eridania 400 12 12
Trichilin E (60) S. eridania 300 22 12
(C12-Epimer of aphanastatin)
Trichilin F (61) S. eridania 300 60 60
(1-O-Acetyl-3-deacetyltrichilin A)
Trichilin G (62) S. eridania 300 60 60
(1-O-Acetyl-2,3-deacetyltrichilin A)
2: Acetylated compounds
12-O-Acetyltrichilin A (63) S. eridania 400 58 58
7,12-Di-O-acetyltrichilin A (64) S. eridania 500 58 58
1,7,12-Tri-O-acetyl-2-deacetyltrichilin A (65) S. eridania 500 61 61
1,7,12-Tri-O-acetyl-3-deacetyltrichilin A (66) S. eridania 500 61 61
7,12-Di-O-acetyltrichilin B (67) S. eridania 500 58 58
1-O-Acetyl-3-deacetyltrichilin B (68) S. eridania 200 61 61
12-O-Acetyltrichilin C (69) S. eridania ⬎500 12 12
3: Oxidized compounds
7-Oxotrichilin A (70) S. eridania Inactive 22 58
7-Oxotrichilin B (71) S. eridania Inactive 22 58
12-Oxotrichilin B (72) S. eridania 1000 22 12
1,12-Dioxotrichilin B (73) S. eridania Inactive 22 58
7,12-Dioxotrichilin B (74) S. eridania Inactive 22 58
a
NA ⫽ not available.
TM

TM
C
pare the activities of 10 and 11 with 12–28, Table 1). All natural C-19/C-29
bridged acetals possess the 11-keto group, and the configuration of the C-12
hydroxyl group has a pronounced effect on the activity (compare the activities
of (17) vs. 56, 23 vs. 60, and 58 vs. 59). The activity of compounds lacking the
12-OH (for example, 14 vs. 15, 20 vs. 17, and 24 vs. 25) or in which the C7 (57
vs. 56) and/or C12 positions (10 vs. 11, 12 vs. 13, 15 vs. 16, 17 vs. 18 vs. 67,
and 56 vs. 63 vs. 64, and 58 and 59 vs. 69) are acetylated or oxidized to ketones
(17 vs. 71, 72, and 74, and 56 vs. 70 and 72) are reduced. Replacement of the
14,15-epoxide with a C15 ketone also results in reduced activity (10 vs. 24, 11
vs. 25, 17 vs. 59, 20 vs. 27, and 56 vs. 58). These structure–activity relationships
in the trichilins are summarized in Fig. 9.
Figure 9 Structure–activity relations in the trichilins (F ⫽ 3-furyl). (Source: Modified

from Ref. 58.)
TM

Figure 10 Biogenesis and antifeedant activity of the limonoids from M. azedarach.
TM

VI. CONCLUSIONS
Research on biologically active limonoids has been motivated by the quest to

find useful compounds for specific agricultural or medicinal applications. Though
bioassay designs and choice of bioassay species have varied tremendously in the
literature, the common conclusion undoubtedly is that limonoids function primar-
ily as ‘‘antifeedants’’ in the host species. This suggests that the primary selective
advantage of the production of limonoids by the host plant is protection against
insect herbivory.
About 350 limonoids have been isolated to date. Although limonoids may
be found in all tissues of the plant, limonoid biosynthesis is characterized by an
evolutionary pattern of increasing oxidation and rearrangement of the original
limonoid skeleton [62]. Relations of biogenesis of the limonoids in M. azedarach
with their antifeedant activity are summarized in Fig. 10. From the scheme, it
can be seen that the evolutionary trend of increasing oxidation correlates with
increasing activity against insects.
REFERENCES
1. S. Siddiqui, B. S. Siddiqui, S. Faizi, and T. Mahmood, J. Nat. Prod., 51: 30 (1988).

2. J. D. Connolly, in Chemistry and Chemical Taxonomy of the Rutales (P. G. Water-
man and M. F. Grunden, eds.), Academic Press, New York, p. 175 (1983).
3. D. A. H. Taylor, in Chemotaxonomy: The Occurrence of Limonoids in the Melia-
ceae (T. D. Pennington and B. T. Styles, eds.), Academic Press, New York, p. 450
(1983).
4. D. A. H. Taylor, in Chemistry and Chemical Taxonomy of the Rutales (P. G. Water-
man and M. F. Grunden, eds.), Academic Press, New York, p. 353 (1983).
5. J. O. Kokwaro, Medicinal Plants of East Africa, East African Literature Bureau,
Nairobi, Kenya, (1976).
6. J. M. Watt and M. G. Breyer-Brandwick, The Medicinal and Poisonous Plants of
Southern and Eastern Africa, E. & S. Livingstone, London, (1962).
7. D. E. Champagne, O. Koul, M. B. Isman, G. G. E. Scudder, and G. H. N. Towers,
Phytochemistry, 31: 377 (1992).
8. H. Rembold, in Focus on Phytochemical Pesticides. Vol. 1. The Neem Tree (M.
Jacobson, ed.), CRC Press, Boca Raton, Fla., p. 47 (1988).
9. H. Rembold, in Insecticides of Plant Origin (J. T. Arnason, B. J. R. Philogene, and
P. Morand, eds.), ACS Symp. Series 387, Washington D.C., p. 150 (1989).
10. S. V. Ley, A. A. Denholm, and A. Wood, Nat. Prod. Rep., 109 (1993).
11. D. E. U. Ekong, C. O. Fakunle, A. K. Fasina, and J. I. Okogun, J. Chem. Soc. Chem.
Commun., 1166 (1969).
12. M. Nakatani, Unpublished data.
12a. M. Nakatani, J.-B. Zhou, K. Tadera, and H. Naoki, Chem. Lett., 1279 (1998).
TM

13. C. R. Mitra, H. S. Garg, and G. N. Pandy, Tetrahedron Lett., 2761 (1970).
14. D. A. H. Taylor, in Progress in the Chemistry of Organic Natural Products (W.
Herz, H. Grisebach, and G. W. Kirby, eds.), Springer, New York p. 1 (1984).
15. C. R. Mitra, H. S. Garg, and G. N. Pandy, Phytochemistry, 10: 857 (1971).
16. M. Nakatani, R. C. Huang, S. Arikawa, K. Yamauchi, H. Okamura, T. Iwagawa,
and H. Naoki, in 35th Symposium on the Chemistry of Natural Products, Kyoto,
Symposium papers, p. 385 (1993).
17. Azadirachtin has been reported from M. azedarach: E. D. Morgan and M. D. Thorn-
ton, Phytochemistry, 12: 391 (1973). But, no spectral nor analytical data were given,
nor has the isolation of azadirachtin from M. azedarach been reported elsewhere.
18. K. Hostettmann, Planta Med., 39: 1, (1980).
19. K. Hostettmann, C. Appolonia, B. Domon, and M. Hostettmann, J. Liquid Chro-
matogr., 7: 231 (1984).
20. R. C. Huang, J.-B. Zhou, H. Suenaga, K. Takezaki, K. Tadera, and M. Nakatani,
Biosci. Biotech. Biochem., 59: 1755 (1995).
21. K. Wada and K. Munakata, J. Agric. Food Chem., 16: 471 (1968).
22. M. Nakatani, J. C. James, and K. Nakanishi, J. Am. Chem. Soc., 103: 1228 (1981).
23. R. J. Stonard, D. A. Trainor, M. Nakatani, and K. Nakanishi, J. Am. Chem. Soc.,
105: 130 (1983).
24. J. Polonsky, Z. Varon, B. Arnoux, C. Poscard, G. R. Pettit, J. M. Schmidt, and
L. M. Lange, J. Am. Chem. Soc., 100: 2575 (1978).
25. M. Nakatani, S. Arikawa, H. Okamura, and T. Iwagawa, Heterocycles, 38: 327
(1994).
26. W. Kraus, M. Bokel, A. Bruhn, R. Cramer, I. Klaiber, A. Klenk, G. Nagle, H. Pöhnl,
H. Sadio, and B. Vogler, Tetrahedron, 43: 2817 (1987).
27. D. Lavie, E. C. Levy, and M. K. Jain, Tetrahedron, 27: 3927 (1971).
28. J. D. Connolly, K. L. Handa, and R. McCrindle, Tetrahedron Lett., 437 (1968).
29. S. K. Srivastava and H. O. Gupta, Indian J. Chem. Sect. B, 24B: 166 (1985).
30. K. Rusia and S. K. Srivastava, Proc. Natl. Acad. Sci., Ind., Sect. A, 58: 33 (1988).
31. S. D. Srivastava, J. Nat. Prod., 49: 56 (1986).
32. S. D. Srivastava, Planta Med., 53: 100 (1987).
33. M. Ochi, H. Kotsuki, and T. Tokoroyama, Chem. Lett., 621 (1978).
34. J. Polonsky, Z. Varon, C. Marazano, B. Arnoux, G. R. Pettit, J. M. Schmidt, M.
Ochi, and H. Kotzuki, Experientia, 35: 987 (1979).
35. J.-W. Ahn, S.-U. Choi, and C.-O. Lee, Phytochemistry, 36: 1493 (1994).
36. S.-F. Chiu, in Natural Pesticides from the Neem Tree (Azadirachta indica A. Juss)
and Other Tropical Plants. Proceedings of the 2nd International Neem Conference
(Rauischnolzhausen, 1983), (H. Schmutterer and K. R. S. Ascher, eds.), GTZ, Esch-
born, Germany, p. 315 (1984).
37. X. J-xi and Y. A-xing, Acta Pharm. Sin., 20: 188 (1985).
38. R. C. Huang, H. Okamura, T. Iwagawa, and M. Nakatani, Bull. Chem. Soc. Jpn.,
67: 2468 (1994).
39. R. C. Huang, H. Okamura, T. Iwagawa, K. Tadera, and M. Nakatani, Phytochemis-
try, 38: 593 (1995).
40. M. Nakatani, R. C. Huang, H. Okamura, H. Naoki, and T. Iwagawa, Phytochemistry,
36: 39 (1994).
TM

41. P. B. Oelrichs, M. W. Hill, P. J. Vallely, J. K. Macleod, and T. F. Molinski, Phyto-
chemistry, 22: 531 (1983).
42. J. K. McLeod, P. D. R. Moller, T. F. Molinski, and O. Koul, J. Chem. Ecol., 16:
2511 (1990).
43. J. T. Arnason, B. J. R. Philogene, N. Donskov, and I. Kubo, Entomol. Exp. Appl.,
43: 221 (1987).
44. M. Saxena and S. K. Srivastava, Ind. J. Chem. Sect. B, 25B: 1087 (1986).
45. M. Ochi, H. Kotsuki, T. Kataoka, T. Tada, and T. Tokoroyama, Chem. Lett., 331
(1978).
46. I. Kubo and J. A. Klocke, Colloques Inst. Nat. Recherches Agric., 7: 117 (1981).
47. W. Kraus and M. Bokel, Chem. Ber., 114: 267 (1981).
48. M. Nakatani, R. C. Huang, H. Okamura, T. Iwagawa, K. Tadera, and H. Naoki,
Tetrahedron, 51: 11731 (1995).
49. M. Ochi, H. Kotsuki, M. Ido, H. Nakai, M. Shiro, and T. Tokoroyama, Chem. Lett.,
1137 (1979).
50. Y. Fukuyama, I. Miura, and M. Ochi, Bull. Chem. Soc. Jpn., 56: 1139 (1983).
51. R. Henderson, R. McCrindle, A. Melera, and K. H. Overton, Tetrahedron, 24: 1525
(1968).
52. J. Meisner, K. R. S. Ascher, R. Aly, and J. D. Warthen, Jr., Phytoparasitica, 9: 27
(1981).
53. M. Schwinger, B. Ehhammer, and W. Kraus, in Natural Pesticides from the Neem
Tree (Azadirachta indica A. Juss) and Other Tropical Plants. Proceedings of the 2nd
International Neem Conference (Rauischnolzhausen, 1983), (H. Schmutterer and
K. R. S. Ascher, eds.), GTZ, Eschborn, Germany, p. 181 (1984).
54. S. M. Lee, J. A. Klocke, and M. F. Balandrin, Tetrahedron Lett., 28: 3543 (1987).
55. K. Takeya, Z.-S. Qiao, C. Hirobe, and H. Itokawa, Phytochemistry, 42: 709 (1996).
56. W. Kraus, M. Bokel, A. Klenk, and H. Pohnl, Tetrahedron Lett., 26: 6435 (1985).
57. H. V. Broughton, A. M. Z. Slawin, D. J. Williamson, and E. D. J. Morgan, J. Chem.
Soc., Chem. Comm., 46 (1986).
58. K. Nakanishi, R. Cooper, and M. Nakatani, Scientific Reports of the Institute of
Organic and Physical Chemistry of Wroclaw Technical University 22. Conferences
7. Proceedings of the International Conference of Regulation of Insect Development
and Behaviour Karpacz, Poland, 1981.
59. M. Nakatani, T. Iwashita, H. Naoki, and T. Hase, Phytochemistry, 24: 195 (1985).
60. M. Nakatani and K. Nakanishi, Heterocycles, 36: 725 (1993).
61. M. Nakatani and K. Nakanishi, Rep. Fac. Sci. Kagoshima Univ., 25: 59 (1992).
62. M. F. Das, G. F. D. Silva, and O. R. Gottlieb, Biochem. Syst. Ecol., 15: 85 (1987).
TM

2
Polygodial and Warburganal,
Antifungal Sesquiterpene
Dialdehydes and Their Synergists
Isao Kubo
University of California, Berkeley, California
I. INTRODUCTION
Many human pathogenic microorganisms can now be controlled with the antibiot-
ics that are presently available. However, with the increase in drug resistance
and prevalence of opportunistic infections, there is still a great need for new,
more effective agents. For example, systemic infections caused by filamentous
fungi, especially in patients with impaired immune defense mechanisms, have
become an increasingly serious, worldwide problem. Although various new anti-
fungal agents have been introduced, control of most fungal diseases has not yet
been achieved [1]. Hence, in our continuous search for antimicrobial agents from
tropical plants, a new emphasis has been placed on antifungal agents, particularly
against Candida albicans, considered one of the most devastating fungi responsi-
ble for human opportunistic systemic infections [2].
Tropical plants are exposed throughout the entire year to predation by vari-
ous parasites such as bacteria, fungi, and insects. In order to survive these de-
manding conditions, efficient built-in chemical defense mechanisms have
evolved, and thus tropical plants offer a rich and intriguing source of secondary
metabolites possessing attractive biological activities with potential medicinal
applications. These plants remain a good source of new antifungal agents [3].
However, there is a need to investigate them from a point of view different from
that of the traditional approach of mere isolation and structure determination of
biologically active principles. Today, understanding the mechanism, as well as
TM

enhancing the biological activity, have become important goals of natural product
chemists. This largely reflects the recent rapid developments in separation and
spectroscopic techniques that render isolation and structure determination more
routine.
The plants chosen for study were identified primarily on the basis of infor-
mation provided by medicine men in East Africa, mainly in Kenya and Tanzania
[4]. Botanically identified plants were then collected and extracted with methanol
at ambient temperatures. The extracts were first tested for antimicrobial activity
against four representative microorganisms, Bacillus subtilis, Escherichia coli,
Saccharomyces cerevisiae, and Penicillium chrysogenum, at a concentration of
100 µg/mL [5]. The active extracts were then tested against a larger number of
microorganisms to determine the scope of activity. Interestingly, the information
gathered from medicine men, known as Bwana Mganga in Swahili [4], proved
to be very useful in identifying species with significant activities. The plants
collected on the basis of their information had a much higher probability of pro-
ducing active extracts than those collected randomly [6]. From 79 extracts, which
included 72 species of plants distributed among 35 families, 40 extracts initially
gave positive results indicative of antimicrobial activity at the 100 µg/mL level
against one or more of the microorganisms. Noticeably few extracts showed ac-
tivity against fungi, especially C. albicans. This review focuses on a particular
group of compounds: the antifungal sesquiterpene dialdehydes isolated from trees
of the Warburgia genus. These sesquiterpene dialdehydes are among the rare
phytochemicals that exhibit potent antifungal activity against C. albicans.
II. INITIAL STUDIES
The genus Warburgia, endemic to East Africa, belongs to a small family, Canella-
ceae, and consists of only two species: W. ugandensis and W. stuhlmannii. We
first became interested in the tree W. ugandensis while I was on a tenured appoint-
ment at the International Centre of Insect Physiology and Ecology (ICIPE), lo-
cated in Nairobi, Kenya, because of its strong, hot taste. The methanolic extract
of the bark of W. ugendensis collected on the ICIPE grounds was originally found
to exhibit potent insect antifeedant activity against the African armyworm,
Spodoptera exempta [7]. Fractionation guided by the antifeedant assay led to
three active principles isolated from the bark, leaves, and fruit of this tree [8,9].
These active compounds, isolated after repeated chromatography and character-
ized by their unique hot taste, were the sesquiterpene dialdehydes [10] warburga-
nal (1) [11] and muzigadial (also known as canellal) (2) [12,13], in addition to
a known congener, polygodial (3). Most of the initial chemical study was per-
formed in Professor Koji Nakanishi’s labs together with his colleagues at Colum-
bia University in New York [11,12,14]. Subsequently, these antifeedant sesquiter-
pene dialdehydes received considerable attention from synthetic organic
TM

chemists, and various syntheses have been reported [15–17]. Included among
these efforts is the conversion of 1-abietic acid into (⫺)-warburganal, the first
reported synthesis of (1) with the natural absolute stereochemistry [18].
Since the Warburgia plants are widely used in folk medicine in East Africa
[19], their extracts were also submitted for testing in various available pharmaco-
logical assays together with other plant extracts. Among the extracts tested, the
two from the Warburgia species were found by Professor Makoto Taniguchi of
Osaka City University to exhibit a broad-spectrum antimicrobial activity, includ-
ing activity against Candida utilis [5]. This finding initiated an investigation to
determine the antifungal principles, which was achieved in collaboration with
Professor Taniguchi and his coworkers [20–24].
III. ISOLATION AND IDENTIFICATION OF

ANTIFUNGAL PRINCIPLES
A bioassay-directed fractionation of the n-hexane extract of the bark of W. ugan-

densis using Bacillus subtilis and Saccharomyces cerevisiae as test organisms
resulted in the isolation of three active sesquiterpene dialdehydes 1–3, previously
identified as the insect antifeedants from the same source [8,11,12,14]. Their
structures were identified by spectroscopic methods [11,12]. In the course of this
work it was noted that methanol, originally used to extract the Warburgia bark,
at least partly if not totally, inactivated the antifungal activity of these dialdehydes
through acetal formation. Therefore, the use of alcohols was avoided throughout
the isolation procedure.
Polygodial (3) was first isolated as a hot tasting substance from the sprouts
of Polygonum hydropiper (Polygonaceae) [25], which has been used in folk medi-
cine and food spices in several Asian countries including Japan and Vietnam.
For example, the sprout of P. hydropiper is a well-known relish for sashimi in
Japan [26,27]. Subsequently, warburganal (1) was also isolated from the same
source in minute amounts [28]. In addition to the antifungal sesquiterpene dialde-
hydes 1–3, a number of congeners such as mukaadial (4) [29], ugandensidial
(also known as cinnamodial) (5) [30], epipolygodial (6) [14], as well as conferti-
folin (7), 9α-hydroxycinnamolide (8) [31], cinnamosnolide (9) [14], colorata-4
[13], 8-dienolide (10), and bemadienolide (11) [14] were also isolated from
TM

Scheme 1
ous parts of W. ugandensis, but none exhibited antifungal activity at concentra-

tions of 100 µg/mL. Their structures were all identified by spectroscopic meth-
ods, particularly by nuclear magnetic resonance (NMR) spectroscopy. All the
sesquiterpenoids isolated (1–11) are considered oxidation products of the dri-
mane skeleton. The same sesquiterpenoids were also isolated from the bark of
W. stuhlmannii [11,12,32,34], as well as from several other plants [31–33]. It
was also discovered that each dialdehyde gave the corresponding lactone upon
treatment with 2N HCl or 1N NaOH. For example, alkaline treatment of (1) at
room temperature gave (8) through an intramolecular Cannizzaro reaction, and
acid treatment yielded futronolide (12) (Scheme 1). Similarly, alkaline treatment
of (5) yielded the lactone (13), which was converted to (9) by acetylation [29].
Nevertheless, these more stable lactones (in comparison to 3) were found to be
no longer active.
TM

IV. ANTIFUNGAL ACTIVITY
With the exception of (1)–(3), none of the other sesquiterpenoids (4)–(11)

showed any antifungal activity at concentrations up to 100 µg/mL. Bioassay re-
sults for these sesquiterpenoids against additional microorganisms are listed in
Table 1.
Table 1 Antimicrobial Activity of the Warburgia sesquiterpenoids a
MIC (µg/mL)
Microorganisms tested 1 2 3 4 6
Staphylococcus aureus 100 ⬎100 100 ⬎100 ⬎100

Bacillus subtilis ⬎100 ⬎100 ⬎100 ⬎100 ⬎100
Streptococcus mutans ⬎100 ⬎100 ⬎100 — —
Micrococcus luteus ⬎100 ⬎100 ⬎100 ⬎100 ⬎100
M. lysodeikticus ⬎100 ⬎100 ⬎100 ⬎100 ⬎100
Escherichia coli ⬎100 ⬎100 ⬎100 ⬎100 ⬎100
Enterobacter aerogenes ⬎100 — ⬎100 — —
Proteus vulgaris ⬎100 ⬎100 ⬎100 ⬎100 ⬎100
Pseudomonas aeruginosa ⬎100 ⬎100 ⬎100 ⬎100 ⬎100
Helicobacter pylori — — ⬎100 — —
Saccharomyces cerevisiae 3.13 1.56 0.78 ⬎100 ⬎100
Schizosaccharomyces pombe 12.5 25 6.25 ⬎100 ⬎100
Hansenula anomala 12.5 25 1.56 ⬎100 ⬎100
Candida utilis 3.13 3.13 1.56 ⬎100 ⬎100
C. albicans 6.25 — 3.13 — —
C. krusei — — 6.25 — —
Cryptococcus neoformans — — 3.13 — —
Pityrosporon ovale 25 — 50 — —
Sclerotinia libertiana 3.13 3.13 1.56 100 ⬎100
Mucor mucedo 25 25 6.25 ⬎100 ⬎100
Rhizopus chinensis 100 100 12.5 ⬎100 ⬎100
Aspergillus niger 50 50 25 ⬎100 ⬎100
A. flavus — — 50 — —
Trichophyton rubrum — — 0.78 — —
T. mentagrophytes 6.25 — 3.13 — —
Penicillium crustsum 50 50 25 ⬎100 ⬎100
P. marneffei — — 3.13 — —
a
The rate of activity varied slightly with the seed culture media, the physiological age of the culture
and the type of culture medium. The MIC was measured by twofold serial broth dilution. The MIC
was the lowest concentration of sample at which no growth of the test microorganism was visible.
—, not tested.
Source: Refs. 7, 35, and 36.
TM

Warburganal (1), muzigadial (2), and polygodial (3) exhibited activity
against all fungi tested [6]. In particular, high activity was shown against Candida
utilis, Saccharomyces cerevisiae, Pityrosporon ovale, Trichophyton rubrum,
Penicillium chrysogenum, Hansenula anomala, and Sclerotinia libertiana. The
activities of (1) and (3) against the pathogenic fungus Candida albicans were
also quite high: their minimum inhibitory concentrations (MICs) were 3.13 and
6.25 µg/mL, respectively [35]. Among the three antifungal sesquiterpene dialde-
hydes 1–3, the structurally simplest, (3), exhibited the most potent activity [20].
In addition, the activity of (3) against fungi was of broad scope. Polygodial (3)
was two to eight times more active than (1) and (2) against most species of fungi
tested, and its potency against these fungi was comparable to that of amphotericin
B, one of the most potent antibiotics known against filamentous fungi [22], al-
though its high toxicity limits extensive use. Therefore, (3) may be potent enough
to be considered for practical application, and further studies with (3) are ongoing.
The structural similarity of (2) with (3) suggests that similarly potent antifungal
activity should be found for this dialdehyde as well. Although the results in Table
1 indicate that this is the case, (2) could not be extensively tested because of its
limited availability.
Interestingly, in contrast to (3), the C-9 epimer, epipolygodial (6), did not
show antimicrobial activity at concentrations of 100 µg/mL. This difference in
antimicrobial activity between (3) and (6) may reflect differences in abilities to
cross the cell membranes of the target microorganisms since a marked difference
in their water solubilities was noted: (6) is much less soluble in water than (3).
The facile, reversible conversion of (3) to the cyclic dihemiacetal (14) in aqueous
media (Scheme 2) may assist its water solubilization [20]. Such hydration to a
cyclic dihemiacetal would be difficult in (6) with the pseudoaxial C-9α-aldehyde
orientation. In support of this suggestion, reduction of the C-9 aldehyde group
of (3) resulted in loss of activity. In addition to (6), mukaadial (4), possessing
two additional hydroxyl groups at C-6 and C-9, did not exhibit any activity at
concentrations up to 100 µg/mL.
Numerous phytochemicals have been isolated as potential antifungal agents
[3,36]. As natural products, these phytochemicals are typically biodegradable
and, more importantly, renewable. The efficient utilization of such renewable
Scheme 2
TM

natural products is becoming increasingly important worldwide to conserve natu-
ral resources. Unfortunately, although these antifungal phytochemicals may play
an important role in the defense of the host plants against fungal predation, their
biological activities are usually not sufficiently potent for practical application
in medicine. This is a common dilemma when the biological activities of phyto-
chemicals are considered; hence, studies to enhance activities are needed for effi-
cient utilization of renewable natural products.
An attempt to enhance the antimicrobial activity of some purified antimi-
crobial agents through combination with other agents was made. In a preliminary
experiment, 3 was first tested with several antibiotics such as actinomycin D and
rifampicin. In these experiments, one agent is maintained at a concentration be-
low the MIC (for example, 1/2MIC) while the concentration of the second agent
is varied to determine its MIC in the presence of the first agent. The roles of the
two agents are then reversed. Polygodial (3) significantly enhanced the antifungal
activity of these antibiotics against C. utilis and S. cerevisiae, but not vice versa
[20,22,23,37]. Polygodial (3) also synergized the antifungal activity of maesanin
(15), isolated from the fruit of the East African medicinal plant Maesa lanceolata
(Myrsinaceae) [38], against C. utilis [39].
The reason for these combination effects seems to be an increase in the

permeability of the plasma membrane toward the antibiotics induced by 3 [40,41].
Thus, when cells of S. cerevisiae were treated with 3, ultrastructural changes in
the cell membrane were observed [22], as shown in Fig. 1. These morphological
alterations suggested that the primary site of action of 3 is the plasma membrane
with simultaneous organelle disorganization followed by the fatal loss of cellular
constituents such as proteins and polysaccharides, as illustrated in Fig. 2 [37].
The addition, of excess Ca 2⫹ was found to protect against the polygodial-induced
cell membrane damage in S. cerevisiae, and this protection was abolished by the
addition of EDTA [24]. It has been suggested that Ca 2⫹ stabilizes the structure
of membranes by forming ion bridges between phosphate groups of phospholip-
ids and the carboxyl groups of membrane proteins, similar to bacterial membranes
[42]. Interestingly, the protection seems to be specific for Ca 2⫹ since Mg 2⫹, a
similar divalent cation, did not give this protection, though both Ca 2⫹ and Mg 2⫹
suppressed miconazole-induced leakage and slightly suppressed amphotericin B–
induced leakage. This result indicates that 3, miconazole, and amphotericin B
differ in their modes of action on fungal cell membranes [24]. The Ca 2⫹ cation
has many properties that make it better suited for chelation than the more abun-
TM

(a)
(b)
Figure 1 (a) Section of untreated control cell of Saccharomyces cerevisiae. (b) Section
of S. cerevisiae cell treated with 50 µg/mL of polygodial (3) for 10 min. CW, cell wall;
PM, plasma membrane (cell membrane); N, nucleus; M, mitochondrion; V, vacuole.
TM

Figure 2 Leakage of (a) folin-reagent and (b) phenol-H2 SO4-positive substances from
Saccharomyces cerevisiae cells during incubation with polygodial (3). , None; ,
1 µg/mL of 3; , 10 µg/mL of 3.
dant Mg 2⫹, such as its larger radius, its lower energy of hydration, and the pres-
ence of d-orbitals, allowing it to chelate more readily. In addition, both 3 and 1
have been reported to exhibit the membrane leakage activity in human neuro-
blastoma cells [43]. The binding site of 3 on the cell membrane and the mecha-
nism of the suppressive effects of Ca 2⫹ on the polygodial-induced leakage remain
to be established.
TM

V. COMBINATION EFFECTS
In our continuing search for antimicrobial agents from tropical plants, several
phenylpropanoids, such as anethole (16), isolated from the seeds of Pimpinella
anisum (Umbelliferae) [44], and eugenol (17), methyleugenol (18), and safrole
(19), in addition to anethole from the seeds of Licaria puchuri-major (Lauraceae)
[45], were isolated as antimicrobial principles in rather large quantities. All exhib-
ited moderate but broad-spectrum activity. Their MICs, ranging from 100 to 800
µg/mL, were not potent enough to warrant further studies on the individual com-
pounds. However, it was still considered worthwhile to investigate the possibility
of their use as antifungal agents in combination with other antimicrobial natural
products, since they were all isolated from various food spices that have long
been consumed by people from many different cultures. Hence, they were first
examined in combination with 3 in an attempt to enhance their antifungal activity
against several fungi such as C. albicans and S. cerevisiae as well as the dermato-
mycotic fungus Pityrosporon ovale. Unexpectedly, 3 did not synergize the anti-
fungal activity of any of these phenylpropanoids, though its antifungal activity
was significantly increased when combined with one of the phenylpropanoid
compounds. Notably, a dramatic increase in the antifungal activity of 3 occurred
when it was combined with a sublethal amount (1/2MIC) of 16: the activity of
3 against C. albicans and S. cerevisiae was increased 32- and 64-fold, respec-
tively. Thus, the MIC of 3 against C. albicans was lowered from 3.13 to 0.098
µg/mL, and against S. cerevisiae, from 1.56 to 0.024 µg/mL, when 3 was com-
bined with 100 µg/mL of 16 (1/2MIC for both C. albicans and S. cerevisiae)
[44].
VI. FUNGICIDAL ACTIVITY
These combination effects, based on the MICs obtained after 48-h incubations,
do not fully characterize the antifungal activity. For example, it was not clear
whether the combination of 3 and 16 was fungicidal or fungistatic. Hence, the
viable count method [46] to analyze the growth curve of C. albicans was used
TM

in order to obtain the minimum fungicidal concentration (MFC). The growth
curves of C. albicans in the presence of 3 and 16 are illustrated in Fig. 3 [47].
Polygodial (3) and anethole (16) exhibited fungicidal activity against C. albicans
at 3.13 and 200 µg/mL, respectively; however, they were not fungicidal at 1.56
and 100 µg/mL, respectively. Thus, C. albicans was not viable after 18 and 42
h at 3.13 µg/mL of 3 and 200 µg/mL of 16. The MFCs of both 3 and 16 against
C. albicans were therefore the same as their MICs. The combination of 3 and
16, however, was found to possess fungicidal rather than fungistatic activity at
concentrations far below their MICs. Fungicidal activity against C. albicans with
the combination of 3 and 16 is shown in Fig. 3. When 100 µg/mL of 16 (1/
2MIC for C. albicans) was combined with more than 0.098 µg/mL of 3, C.
albicans was not viable after 6 or 12 h. Thus, the fungicidal activity of 3 against
C. albicans was increased 32-fold by 16. In contrast, the fungicidal activity of
16 was enhanced only 4-fold by 3.
After this discovery, warburganal (1) was also examined in combination
with 16 to see whether 16 had the same enhancing activity [35]. As expected,
16 also significantly increased the activity of 1 against both C. albicans and S.
cerevisiae. In this combination, the activity of 1 against C. albicans and S. cere-
visiae was enhanced 32- and 256-fold, respectively, when combined with 100
µg/mL of 16. The MIC of 1 against C. albicans was lowered from 6.25 to 0.20
µg/mL, and against S. cerevisiae from 6.25 to 0.024 µg/mL. Anethole (16) also
enhanced the activity of 3 and 1 against P. ovale but not as much as against C.
albicans and S. cerevisiae. The MICs were reduced only from 50 to 6.25 µg/
mL for 3, and from 25 to 3.13 µg/mL for 1 in combination with 50 µg/mL 16
Figure 3 Growth curves of Candida albicans in the presence of anethole (16), polygod-
ial (3), and their combination.
TM

(1/2MIC for P. ovale). Similarly, the antifungal activity of 3 was significantly
increased when combined with subinhibitory concentrations of methyleugenol
(18) and safrole (19), whereas eugenol (17) did not induce any meaningful en-
hancement of activity, as shown in Table 2.
As shown by the results presented in Table 2, 18 induced a 128-fold en-
hancement of the activity of 3 against P. ovale; the MIC was lowered from 50
to 0.39 µg/mL. A preliminary analysis suggests that those phenylpropanoids that
do not possess a free phenolic group, such as 16, 18, and 19, enhance the activity
of 3 more than 17, which has a free phenolic group, though the number of samples
tested to date is rather limited. Anethole (16) was very effective in enhancing
the antifungal activity of 3 and 1 against C. albicans, C. utilis, and S. cerevisiae,
and the combinations with 18 were the most effective against P. ovale [44].
In addition, 16 was also combined with other antifungal agents, such as
amphotericin B. This combination was investigated since amphotericin B is also
known to damage the plasma membrane by interacting with sterols [41] in fungal
cells [40]. However, 16 did not enhance the antifungal activity of amphotericin B
against C. albicans and S. cerevisiae; in fact, its antifungal activity was somewhat
antagonized by 16, as shown in Table 3.
Increasing amounts of 16 decreased the antifungal activity of amphotericin
B significantly. Specifically, the activity of amphotericin B against C. albicans
and S. cerevisiae decreased 16-fold when it was combined with 25 and 50 µg/
mL of 16, respectively. In contrast, 16 did increase the antifungal activity of
amphotericin B against P. ovale 8-fold. Thus, the MIC was lowered from 3.13
to 0.39 µg/mL when amphotericin B was combined with 50 µg/mL of 16. These
results indicate that a chemical reaction between 16 and amphotericin B did not
occur prior to the assay. The effect of 16 on the activity of amphotericin B and
Table 2 Antifungal Activity of Polygodial (3) in Combination with 1/2MIC of

Phenylpropanoids 16–19
MIC (µg/mL)
Candida Saccharomyces Pityrosporon
Compounds combined albicans cerevisiae ovale
Polygodial (3) alone 6.25 3.13 50

⫹ Anethole (16) 0.20 0.049 6.25
⫹ Eugenol (17) 3.13 1.56 25
⫹ Methyleugenol (18) 1.56 0.78 0.39
⫹ Safrole (19) 0.78 0.39 6.25
Source: Refs. 35, 44, 45, and 47.
TM

Table 3 Antifungal Activity of Amphotericin B in Combination with Anethole (16)
MIC of amphotericin B (µg/mL)
Concentration of Candida Saccharomyces Pityrosporon

16 (µg/mL) albicans cerevisiae ovale
0 0.78 0.78 3.13

6.25 1.56 1.56 3.13
12.5 3.13 3.13 0.78
25 12.5 6.25 0.78
50 6.25 12.5 0.39
100 0.20 3.13 —
Source: Ref. 44.
other compounds therefore depends on the species of fungi being tested and the
antifungal agents being combined. Although accumulation of this kind of knowl-
edge may provide new insight into the molecular basis of fungicidal activity, the
mechanism of the combination treatment remains to be established.
VII. MODES OF FUNGICIDAL ACTION
A deeper probe into the mechanism of the antifungal activity of polygodial 3

was also undertaken. It is known that the addition of glucose to an unbuffered
suspension of S. cerevisiae cells results in the extrusion of acid. The change in
external pH upon the addition of glucose is characteristic of yeast cells and is
accepted to be due to the action of the plasma membrane H⫹ –adenosine triphos-
phatase (H⫹-ATPase) [48]. The activation of H⫹-ATPase by glucose at the molec-
ular level is not yet fully understood, but the maintenance of internal pH homeo-
stasis is essential for the cell to survive since intracellular pH is important for
the activity of a number of enzymes with pH optima [49,50]. So presumably,
the added glucose results in an increase in intracellular acidity that must be elimi-
nated. This glucose-induced medium acidification was inhibited by 3, as illus-
trated in Fig. 4. The inhibition was presumably caused by inhibition of H⫹-AT-
Pase. In support of this conclusion, 3 was also found to inhibit the isolated
H⫹-ATPase of S. cerevisiae. Therefore, it is possible that the potent antifungal
activity of 3 is, at least in part, due to its inhibition of the plasma membrane
H⫹-ATPase. Interestingly, the inhibitory action of 3 on the glucose-induced acid-
ification of the medium was strongly suppressed by Ca 2⫹, but only weakly by
Mg2⫹.
TM

Figure 4 Inhibitory effect of polygodial (3) on medium acidification by the plasma
membrane H ⫹-ATPase of Saccharomyces cerevisiae. The medium acidification was in-
duced by adding the glucose solution (final concentration 2%) and was evaluated with the
mole concentration of protons calculated with external medium pH. The ratio of inhibition
(percentage) was calculated as follows: (1 ⫺ [H ⫹]inhibitor /[H ⫹]inhibitor free) ⫻ 100. H ⫹-ATPase,
H ⫹ –adenosine triphosphatase.
It should also be noted that the activity of 3 is enhanced under acidic condi-
tions [22]. It is known that yeast cells are able to maintain a normal internal pH
when suspended in an acidic medium with relatively little change in the intracel-
lular pH. The acidic conditions appear to stimulate the plasma membrane H⫹-
ATPase activity and ‘‘excess’’ protons are pumped out to the external medium,
maintaining constant internal pH during growth [48]. As a result of the inhibition
of the plasma membrane H⫹-ATPase by 3, the intracellular pH may drop into
the range where phosphofructokinase is sensitive [51]. The subsequent inhibition
of glycolysis caused by this inactivation of phosphofructokinase results in a drop
in adenosine triphosphate (ATP) levels and thus restricts growth [49]. This ratio-
nale may explain why 3 is more potent in the acidic conditions.
In contrast to the potent antifungal activity of 3, its congener 4 did not
exhibit any activity up to 800 µg/mL; 1 was active, though to a lesser extent
than 3 [33]. Thus, the activity decreased for each additional hydroxyl group
‘‘added’’ to the molecular framework of 3. The fungicidal activity of 3 was ex-
plained as the result of the structural disruption of the cell membrane [21,52,53],
as illustrated in Fig. 1. Moreover, in previous work using human neuroblastoma
cells [43], the increase in membrane permeability was demonstrated to depend
TM

on the accumulation of the unsaturated dialdehydes in the membranes with the
reactive aldehyde groups oriented toward the membrane surface. Considering the
results so far reported, the antifungal mechanism of 3 may result, at least in part,
from its ability to function as a nonionic detergent, similar to long-chain alcohols
[54]. The greater activity of the dialdehydes 1 and 3 could be due primarily to
a balance between the hydrophilicity of the unsaturated aldehyde subunits and
the hydrophobicity of the decalin portion of the molecules; 4 does not possess
this balance because of its increased hydrophilicity and, hence, is inactive.
As a nonionic detergent, 3 would likely approach the binding site with the
electronegativity of the aldehyde oxygen atom. The aldehyde oxygens are potent
hydrogen bond acceptors that will disrupt existing hydrogen bonds. For example,
in the lipid bilayer the hydroxyl group of ergosterol, a major component of the
plasma membrane of S. cerevisiae, resides near the membrane–water interface
and is likely hydrogen bonded with the carbonyl group of phospholipids [55,56].
Since ergosterol owes its membrane-fixing properties to its rigid, longitudinal
orientation in the membrane and has profound influence on membrane structure
and function, if these hydrogen bonds are disrupted, the cell will die.
Interestingly, the polygodial-induced membrane damage was prevented by
Ca 2⫹, but this protection is eliminated by adding EDTA [24]. The role of Ca 2⫹
is still unclear and many mechanisms to explain it seem possible. For example,
the possibility of Ca 2⫹ binding to the negatively charged phosphate oxygen atoms
on the membrane, similar to bacterial membranes [42], cannot be entirely ruled
out. If this is so, it would result in formation of a cross-linked membrane structure
that may impede the approach of 3 to the binding site on the cell membrane.
Needless to say, further study is needed to clarify the Ca 2⫹ protection mechanism.
Given the detergentlike properties of 3, it is possible to suggest that 3 also
acts at the lipid–protein interface of H ⫹-ATPase, denaturing its functioning con-
formation. In a system containing both lipids and proteins, it is difficult to deter-
mine whether a conformational change of a protein is the result of a direct H ⫹-
ATPase interaction or of motional or conformational modification of the lipids
themselves that exist at the lipid–protein interface. Nevertheless, the binding of
3 as a nonionic detergent can only involve relatively weak head group interac-
tions, such as hydrogen bonding. It is suggested that the intrinsic proteins of
membranes are held in position by hydrogen bonding, as well as by hydrophobic
and electrostatic forces, and that hydrogen bonding also mediates the penetration
of membranes by proteins. As proposed, hydrogen bonds may be disrupted by
3 and redirected. Thereby the conformation of the protein may be changed, and
consequently the H ⫹-ATPase in particular may lose its functioning conformation.
Although H ⫹-ATPase is the most abundant plasma membrane protein, constitut-
ing over 20% of the total membrane protein in S. cerevisiae, other plasma mem-
brane proteins may also be disrupted by 3. All of this is consistent with the
previous report that the primary active site of 3 is at the membrane [22].
TM

All the data so far obtained can be explained by the detergent concept. It
should be remembered that 3 showed potent fungicidal activity, particularly
against yeasts, but the detergent mechanism mentioned previously should not be
specific. The specificity of 3 against yeasts is likely based on its hydrophobic
decalin moiety. The activity of polygodial may be increased by modifying this
hydrophobic decalin moiety, but experimentally this may not be practical.
In addition, α,β-unsaturated aldehydes are highly reactive substances, and
they readily react with biologically important nucleophiles, such as sulfhydryl,
amino, and hydroxyl groups. The main reaction appears to be 1,4-addition under
physiological conditions, although the formation of Schiff bases is also possible
[10,57]. An earlier report demonstrated a good correlation between the antifungal
activity and the papain inhibitory activity of 3. Both activities appear to result
from highly specific reactivity of sulfhydryl groups with the enal group [21].
Yeast plasma-membrane H ⫹-ATPase is reported to contain nine cysteines [58].
Polygodial (3) may bind directly to the plasma-membrane H ⫹-ATPase, possibly
with sulfhydryl groups of the three cysteines in the presumed transmembrane
segments (C148A, C312S, C867A). However, Petrov and Slayman [59] reported
that no single cysteine is required for the enzyme activity, on the basis of their
site-directed mutagenesis study. This does not exclude, however, the possibility
that 3 breaks a hydrogen bond as a detergent and then reacts with the freed
sulfhydryl group of the H ⫹-ATPase. This is supported by the previous report by
Monk and his colleagues [60] that covalent modification of the conserved C148
in the transmembrane segment 2 may be important for inhibition of H ⫹-ATPase
activity and cell growth. Nevertheless, the involvement of this kind of biochemi-
cal reaction is still unclear.
VIII. CONCLUSIONS
The data so far obtained indicate that polygodial (3) initially acts as a nonionic
detergent. More importantly, 3 inhibits the plasma-membrane H ⫹-ATPase by dis-
rupting and disorganizing the hydrogen bonds at the lipid bilayer–protein inter-
face. It seems that 3 targets the extracytoplasmic region and thus does not need
to enter the cell, thereby avoiding most cellular pump–based resistance mecha-
nisms.
For the otherwise healthy person, fungal infections are more of a nuisance
than health-threatening and are normally kept in check by a strong immune sys-
tem and by otherwise innocuous bacteria of the throat and gut. However, when
outside forces such as cancer chemotherapy or heavy doses of antibiotics disrupt
the body’s natural defenses, fungal populations can sharply increase and cause
serious health problems. Synergistic substances that enable physicians to use
lower, safer antifungal dosages would be a useful addition to the therapeutic
TM

arsenal. Further studies of increased potency of currently used antifungal agents
due to combination with 3, as well as 1 and 2, may provide a more rational
and scientific approach for the development of new and extraordinarily powerful
antifungal treatments, especially against the deadly C. albicans.
ACKNOWLEDGMENT
This work has involved a number of scientists. I am greatly indebted to my col-

leagues cited in the references, especially Professor K. Nakanishi, Professor M.
Taniguchi, Dr. M. Himejima, and Dr. S. H. Lee.
REFERENCES
1. S. Sternberg, Science, 266: 1632 (1994).

2. Y. Fukuzawa and K. Kagawa, in Filamentous Microorganisms (T. Arai, ed.) Japan
Scientific Societies, Tokyo, pp. 247–253. (1985).
3. L. A. Mitscher, S. Drake, S. R. Gollapudi, and S. K. Okwute, J. Nat. Prod., 50:
1025 (1987).
4. I. Kubo, in Science Year 1982, World Book-Childcraft International, Chicago, pp.
126–137. (1981).
5. M. Taniguchi, A. Chapya, I. Kubo, and K. Nakanishi, Chem. Pharm. Bull., 26: 2910
(1978).
6. M. Taniguchi and I. Kubo, J. Nat. Prod., 56: 153, (1993).
7. I. Kubo, in Methods in Plant Biochemistry, Vol. 6 (K. Hosttetmann, ed.) Academic
Press, London, pp. 179–193. (1991).
8. I. Kubo and K. Nakanishi, in The Chemical Basis for Host Plant Resistance to Pest,
ACS Symposium Series 62, (P. A. Hedin, ed.), American Chemical Society, Wash-
ington, D.C., pp. 165–178. (1977).
9. I. Kubo, in Recent Advances in Phytochemistry: Phytochemical Potential of Tropical
Plants, Vol. 27, (K. R. Downum, J. T. Romeo, and H. A. Stafford, eds.) Plenum,
New York, pp. 133–151. (1993).
10. I. Kubo and I. Ganjian, Experientia, 37: 1063 (1981).
11. I. Kubo, Y. W. Lee, M. Pettei, F. Pilkiewicz, and K. Nakanishi, J. Chem. Soc. Chem.
Commun., 1013 (1976).
12. I. Kubo, I. Miura, M. Pettei, Y. W. Lee, F. Pilkiewicz, and K. Nakanishi, Tetrahe-
dron Lett., 4553 (1977).
13. F. S. El-Feraly, T. McPhail, and K. D. Onan, J. Chem. Soc. Chem. Commun., 75
(1978).
14. K. Nakanishi and I. Kubo, Isr. J. Chem., 16: 28 (1978).
15. S. P. Tanis and K. Nakanishi, J. Am. Chem. Soc., 101: 4399 (1979).
16. M. Jalani-Naini, D. Guillerm, and J. Y. Lallemand, Tetrahedron, 39: 749 (1983).
17. K. Mori and H. Watanabe, Tetrahedron, 42: 273 (1986).
TM

18. H. Okawara, H. Nakai, and M. Ohno, Tetrahedron Lett., 23: 1087 (1982).
19. J. O. Kokwaro, in Medicinal Plants of East Africa, East African Literature Bureau,
Nairobi, p. 45 (1976).
20. M. Taniguchi, T. Adachi, S. Oi, A. Kimura, S. Katsumura, S. Isoe, and I. Kubo,
Agric. Biol. Chem., 48: 73 (1984).
21. M. Taniguchi, T. Adachi, H. Haraguchi, S. Oi, and I. Kubo, J. Biochem., 94: 149
(1983).
22. M. Taniguchi, Y. Yano, E. Tada, K. Ikenishi, S. Oi, H. Haraguchi, K. Hashimoto,
and I. Kubo, Agric. Biol. Chem., 52: 1409 (1988).
23. M. Taniguchi, Y. Yano, K. Motoba, S. Oi, H. Haraguchi, K. Hashimoto, and I. Kubo,
Agric. Biol. Chem., 52: 1881 (1988).
24. Y. Yano, M. Taniguchi, T. Tanaka, S. Oi, and I. Kubo, Agric. Biol. Chem., 55: 603
(1991).
25. C. S. Barnes and J. W. Loder, Aust. J. Chem., 15: 322 (1962).
26. A. Ohsuka, Nippon Kagaku Zasshi, 84: 748 (1963).
27. I. Kubo, Drug News Persp., 2: 292 (1989).
28. Y. Fukuyama, T. Sato, I. Miura, and Y. Asaka, Phytochemistry, 24: 1521 (1985).
29. I. Kubo, T. Matsumoto, A. B. Kakooko, and N. K. Mubiru, Chem. Lett., 979 (1983).
30. C. J. W. Brooks and G. H. Draffan, Tetrahedron, 25: 2887 (1969).
31. D. Kioy, A. I. Gray, and P. G. Waterman, J. Nat. Prod., 52: 174 (1989).
32. R. F. McCallion, A. L. J. Cole, J. R. L. Walker, J. W. Blunt, and M. H. G. Munro,
Planta Med., 44: 134 (1982).
33. M. S. Al-Said, S. M. El-Khawaja, F. S. El-Feraly, and C. H. Hufford, Phytochemis-
try, 29: 975 (1990).
34. D. Kioy, A. I. Gray, and P. G. Waterman, Phytochemistry, 29: 3535 (1990).
35. I. Kubo and M. Himejima, Experientia, 48: 1162 (1992).
36. L. A. Mitscher, R. P. Leu, M. S. Bathala, W. N. Wu, J. L. Beal, and R. White,
Lloydia, 35: 157 (1972).
37. I. Kubo and M. Taniguchi, J. Nat. Prod., 51: 22 (1988).
38. I. Kubo, T. Kamikawa, and I. Miura, Tetrahedron Lett., 24: 3825 (1983).
39. M. Taniguchi, Y. Yano, E. Tada, T. Tanaka, S. Oi, H. Haraguchi, K. Hashimoto,
and I. Kubo, Agric. Biol. Chem., 53: 1525 (1989).
40. S. C. Kinsky, Annu. Rev. Pharmacol., 10: 119, (1970).
41. J. M. T. Hamilton-Miller, Adv. Appl. Microbiol., 17: 109 (1974).
42. M. A. Asbell and R. G. Eagon, J. Bacteriol., 92: 380 (1966).
43. A. Forsby, E. Walum, and O. Sterner, Chem. Biol. Interact., 84: 85 (1992).
44. I. Kubo and M. Himejima, J. Agric. Food Chem., 39: 2290 (1991).
45. M. Himejima and I. Kubo, J. Nat. Prod., 55: 620 (1992).
46. C. W. Norden, H. Wentzel, and E. Keleti, J. Infec. Dis., 140: 629 (1979).
47. M. Himejima and I. Kubo, J. Agric. Food Chem., 41: 1776 (1993).
48. P. Eraso and C. Gancedo, FEBS Lett., 224: 187 (1987).
49. W. B. Busa and R. Nuccitelli, Am. J. Physiol., 246: 409 (1984).
50. S. Ramos, M. Balbin, M. Raposo, E. Valle, and L. A. Pardo, J. Gen. Microbiol.,
135: 2413 (1989).
51. H. A. Krebs, D. Wiggins, and M. Stubbs, Biochem. J., 214: 657 (1983).
TM

52. M. N. Jones and D. Chapman, Micelles, Monolayers, and Biomembranes, Wiley-
Liss, New York (1995).
53. A. Kocková-Kratochvílová, Yeasts and Yeast-like Organisms, VCH, Weinheim
(1990).
54. I. Kubo, H. Muroi, and A. Kubo, Bioorg. Med. Chem. Lett., 3: 873 (1995).
55. H. Brockerhoff, Lipids, 9: 645 (1970).
56. V. P. S. Chauhan, L. S. Ramsammy, and H. Brockerhoff, Biochim. Biophys. Acta.,
772: 239 (1984).
57. E. Schauenstein, H. Esterbauer, and H. Zollner, in Aldehydes in Biological Systems,
Pion, London, pp. 172–200 (1977).
58. R. Serrano, M. C. Kielland-Brandt, and G. R. Fink, Nature, 319: 689 (1986).
59. V. V. Petrov and C. W. Slayman, J. Biol. Chem., 270: 28535 (1995).
60. B. C. Monk, A. B. Mason, G. Abramochkin, J. E. Haber, D. Seto-Young, and
D. S. Perlin, Biochim. Biophys. Acta, 1239: 81 (1995).
TM

3
Marine Bromoperoxidases—
Chemoenzymatic Applications
Chris A. Moore and Roy K. Okuda

San José State University, San José, California
I. INTRODUCTION
A distinctive feature of many marine natural products is the presence of halogen

atoms, particularly chlorine, bromine, and iodine, as structural components (Fig.
1). There is little doubt that the organically bound halogens are derived from
halide salts that occur naturally in seawater and are incorporated during the bio-
synthetic process. Approximately 2400 marine natural products that are known
contain one or more halogen atoms, of which chlorine and bromine are the most
abundant [1,2]. Considering that the concentration of bromide in seawater is
1/300 that of chloride (65 mg/L bromide vs. 19,000 mg/L chloride) [3], the
frequent occurrence of bromine in marine natural products suggests that it may
be selected for incorporation into secondary products over chlorine. In contrast
to marine secondary products, a much smaller number of terrestrially derived
natural compounds contains halogen. Indeed, only a handful of brominated natu-
ral products from terrestrial sources has been reported [2].
To date, the only enzymes that are known to incorporate halogen into or-
ganic substrates are known collectively as the haloperoxidases. Chloroperoxi-
dases are described primarily from the fungi and some bacteria [4]. These en-
zymes in the presence of both hydrogen peroxide, a ‘‘suitable’’ organic substrate
(defined later), and either chloride, bromide, or iodide catalyzes the halogenation
of the substrate [5]. Bromoperoxidases are able to utilize only bromide and iodide
but are much more widely distributed in nature, having been reported from nu-
merous species of bacteria, marine algae, and some marine invertebrates (e.g.,
TM

worms and mollusks). Iodoperoxidases are able to utilize only iodide as the halide
and are found in marine algae and in the thyroid gland in humans [6].
Although haloperoxidases are believed to be responsible for the incorpora-
tion of halogen during the biosynthetic process, this activity has only been defini-
tively demonstrated in a small number of cases [7,8]. However, the existence of
these enzymes in organisms that also produce significant quantities of haloge-
nated secondary metabolites is unlikely to be coincidental. It should be noted
that some haloperoxidases are also found in organisms that are not known to
produce halogenated organics [9]. This may be due to either as yet undetected
halogenated organic products (e.g., proteins or nucleic acids) or a possible alter-
native role for the enzyme in the source organism.
This chapter will focus only on those marine-derived bromoperoxidases,
which compose the largest group of haloperoxidases reported and on which much
recent research has focused. The distribution of these enzymes among diverse
taxonomic sources leads to an excellent potential for finding variants that have
distinct chemoenzymatic applications.
II. BACKGROUND
A. The Enzymes
The first haloperoxidase to be characterized was named chloroperoxidase and
was identified from the fungus Caldariomyces fumago [10]. This enzyme con-
tains heme-bound iron as the cofactor and has been extensively investigated [11].
Several bacteria are reported to contain a nonheme chloroperoxidase, which is
involved in the chlorination step of chlorinated metabolites [4]. Except for a small
number of cases of enzymes from a marine sponge [12] and a marine worm
[13], all chloroperoxidases reported to date have been isolated from nonmarine
fungi [6].
Bromoperoxidases are so named because of their requirement for hydrogen
peroxide and bromide (or iodide) for catalytic activity. If an organic substrate
containing a functional group that is susceptible to electrophilic attack (e.g. al-
kene, alkyne, aromatic, β-diketone) is also present, an electrophilic substitution
or addition occurs. In the mechanistic sense, the bromide is oxidized to a bromon-
ium ion equivalent and reacts as an electrophilic species. In some cases, the bro-
mide becomes incorporated into the substrate; in others, it may be part of a tran-
sient intermediate and may not be found in the product [6].
In addition to being classified according to the halides used by an enzyme,
haloperoxidases are also categorized by the metal cofactor present. Two general
classes are recognized—those that contain heme (the ‘‘H’’ haloperoxidases, such
as chloroperoxidase), and the nonheme type (the ‘‘NH’’ enzymes, which include
TM

most marine bromoperoxidases). Both H and NH bromoperoxidases have been
found in marine organisms. Interestingly, a significant number of marine algal
bromoperoxidases has been found to have vanadium species as the metal cofactor,
with or without nonheme iron present [14]. The bioinorganic chemical processes
of this system have been of considerable interest. Indeed, chemical mimics of
the vanadium bromoperoxidases that have been prepared and exhibit similar gross
catalytic activity [15,16].
B. Distribution in Marine Organisms

Among marine organisms, bromoperoxidases have been most widely reported
from algae. In particular, these enzymes appear to be common in many species
of red algae. These enzymes are also found in a number of species of brown
algae but are reported only from several green and blue–green algae. Two com-
prehensive surveys of marine algae and bromoperoxidases have appeared [9,17].
Table 1 lists some of the marine algae from which a significant level of bromoper-
oxidase activity has been reported.
1). Bromoperoxidases have also been reported from a small number of
marine invertebrates (Table 1). One report indicates their presence in a marine
annelid and in a marine hemichordate, or ‘‘acorn worm’’ [22], which is known
to produce copious quantities of halogenated aromatic metabolites [28].
That brominating enzymes have been found in marine algae and inverte-
brates is perhaps not surprising, since many are known to produce significant
quantities of halogenated natural products (Fig. 1). Hewson and Hager [17] found
a strong correlation between the presence of bromoperoxidase and the presence
of halogenated lipids in marine algae collected from the Gulf of California. Pre-
sumably, the enzymes are involved in the halogenation step(s) of the biosynthesis
of these compounds. In our own work, we have found bromoperoxidases in many
species of algae that are not known to produce halogenated compounds. This
suggests that these enzymes may serve another catalytic role in the algae, other
than biological halogenation [9].
It is interesting to note that although chlorinated, as well as brominated
and iodinated, metabolites appear in marine natural products, few chlorinating
enzymes have been reported from marine sources. An ‘‘abnormal’’ bromoperoxi-
dase with chlorinating ability was reported from the sponge Iotrochoa birotulata
[12]. The haloperoxidase from the brown alga Ascophyllum nodosum can be in-
duced to incorporate chloride into organic substrates, but at a rate 500 times less
that if bromide were used [29]. Although nearly all marine-derived haloperoxi-
dases have been bromoperoxidases, recently an unusual flavin-containing chloro-
peroxidase has been reported from a marine polychaete worm [13].
TM

Table 1 Representative Marine Organisms That Contain Bromoperoxidase
Marine algae
Reference
Red algae (phylum Rhodophyta)
Asparagopsis taxiformis United States (Hawaii) 9
Bonnemaissonia hamifera Mexico (Gulf of California); 17
United States (New Hampshire) 9
Ceramiun rubrum Netherlands 18
Corallina officinalis Great Britain; 14
United States (Massachusetts) 19
C. pilulifera Japan 20
C. vancouveriensis United States (California) 9
Cystoclonium purpureum 21
Gracilaria sjoestedtii United States (California) 9
Laurencia nipponica Japan 8
Plocamium cartilageneum United States (California) 9
Portieria hornemanni United States (Hawaii) 9
Rhodomela larix United States (Washington) 22
Brown algae (phylum Phaeophyta)
Ascophyllum nodosum Netherlands 23
Fucus distachia Netherlands 24
Laminaria digitata France 23
Macrocystis pyrifera United States (California) 24
Green algae (phylum Chlorophyta)
Codium cylindricum Japan (Okinawa) 9
Halimeda incrassata Mexico (Gulf of California) 17
Penicillus capitatus United States (Florida) 25
P. lamourouxii United States (Florida) 26
Rhipocephalus phoenix United States (Florida) 26
Udotea flabellum Mexico (Gulf of California) 17
Invertebrates
Sponge
Iotrochoa birotulata United States (Florida) 12
Worms
Ptychodera flava (acorn worm) United States (Hawaii) 22
Thelepus setosus (annelid) United States (Hawaii) 22
Gastropod (phylum Mollusca)
Murex trunculus Not given in ref. 27
TM

Figure 1 Examples of halogenated marine natural products.
C. Mechanism of Action
Several articles have reviewed the mechanism of action of marine bromoperoxi-
dases [5,30], and thus only a brief discussion will be given here.
Heme bromoperoxidases react via oxidation of the heme iron by hydrogen
peroxide to yield an activated species, called Compound 1, which then oxidizes
the halide to a halonium ion equivalent. It is unclear whether the halonium ion
is enzyme bound or exists as a hypohalous acid (Fig. 2) [6]. Nonheme vanadium
bromoperoxidases from marine algae are reported to yield an ‘‘enzyme-trapped’’
bromonium species; if an organic substrate is present, it is bound to the enzyme,
whereupon reaction occurs [31].
Bromoperoxidases react primarily with substrates that are susceptible to
electrophilic attack. In biosynthesis, the bromoperoxidases obviously exert their
halogenating capabilities in a regio- and stereoselective fashion, which gives rise
to many natural products containing chiral halogens. Since chemical routes to
chiral halogen moieties starting from achiral substrates are still rather circuitous,
if bromoperoxidases can be induced to perform enantioselective halogenation in
a chemoenzymatic mode, these enzymes could be extremely useful for synthetic
applications. To date, enantioselective halogenation has not been achieved by
using bromoperoxidases. However, work in this area is still in its preliminary
development, and the enzymes may require selection of proper experimental con-
ditions before progress is made.
TM

Figure 2 Possible mechanism of iron-heme bromoperoxidase reactions, ‘‘EOX’’ vs.
HOBr. (Source : Ref. 10.)
III. CHEMOENZYMATIC APPLICATIONS OF MARINE

BROMOPEROXIDASES
A. Key Parameters for Enzymatic Activity
As with all chemoenzymatic applications, a number of reaction parameters must
be carefully manipulated to give optimal yield of the product. With bromoperoxi-
dases, the most important conditions are related to the pH of the buffer, the con-
centration and rate of addition of hydrogen peroxide, and, to a lesser extent, the
concentration of bromide in solution. These reactions are typically conducted in
aqueous buffer solutions, but if a substrate is insoluble in water, organic cosol-
vents may be used. In general, the bromoperoxidases are reported to be relatively
‘‘robust’’ and are fairly stable to storage, in contrast to other enzymes [14].
It is rather difficult to establish one common set of reaction conditions that
TM

will work best with all bromoperoxidases, since each investigator uses a unique
set of parameters to perform reactions with these enzymes. Furthermore, enzymes
from different biological sources are also likely to have different reaction charac-
teristics, thus adding to the complexity. However, for most of the marine algal
bromoperoxidases, which by far are the largest group under study, the optimal
reaction parameters fall within a definable range.
In nearly all cases, the optimal ranges of enzyme activity have been deter-
mined by using the standard monochlorodimedone (MCD) assay for haloperoxi-
dases (Fig. 3) [10]. Although this may not necessarily reflect the reactivity of
the bromoperoxidase with other substrates, the general use of this assay at least
gives benchmark values across which all of these enzymes may be compared.
The major reaction parameters determined by the MCD assay will be briefly
discussed for representative bromoperoxidases.
1. pH
Nearly all marine bromoperoxidases have a pH optimum for reactivity that falls
in the range of 5.0 and 7.0. This is in marked contrast to fungal chloroperoxidase,
which has a pH optimum near pH 3.3 [10]. The pH optimum for most reported
marine bromoperoxidases is a fairly sharp peak, and generally activity drops to
less than 50% of the maximum if the solution is ⫾2 pH units from the optimal
value [9]. Fig. 4 shows examples of pH optima for some marine algal bromoper-
oxidases.
2. Hydrogen Peroxide (H 2 O 2) Concentration and Rate

of Addition
Hydrogen peroxide provides the oxidative impulse necessary for the bromoperox-
idase to activate bromide to an electrophilic species. In too-high titers, however,
H 2O 2 has been shown to inactivate bromoperoxidase, probably by irreversible
Figure 3 Monochlorodimedone (MCD) assay for haloperoxidase activity.
TM

Figure 4 pH Optima of some algal bromoperoxidases. (Source : Ref. 9.)
oxidation of amino acid residues or of the metal cofactor. Hydrogen peroxide is

reported to be a noncompetitive inhibitor of A. nodosum bromoperoxidase, with
greater inhibition at higher pH [32]. This sensitivity may reflect that in the natural
system, the enzyme is subjected to only very low levels of hydrogen peroxide
at all times. Product formation occurs at much higher yield if peroxide is added
in small aliquots over an extended period [33]. Examples of enzyme activity vs.
added hydrogen peroxide are shown in Fig. 5.
TM

Figure 5 Hydrogen peroxide optima of some algal bromoperoxidases. (Source: Ref.
9.)
3. Bromide Concentration
Optimal bromide concentrations for bromoperoxidase activity vary rather widely
among different sources of the enzyme. In some cases, the enzyme activ-
ity diminishes significantly upon addition of additional bromide. Interestingly,
some bromoperoxidases are still active at very high levels of bromide (e.g., 1M
NaBr) [9].
4. Buffer and Organic Cosolvents

One problem commonly associated with chemoenzymatic reactions is that these
reactions are often carried out in aqueous buffered solutions. Hence, the organic
substrates must be water soluble in order to react. We have found that many
nonpolar organic compounds do eventually dissolve in water over extended pe-
riods and show a high level of reactivity with these enzymes. However, bromo-
peroxidases are known to be quite robust enzymes [14] and maintain a good level
of catalytic activity if used in mixtures of aqueous buffer and an organic solvent
(e.g. methanol, dimethyl sulfoxide) to solubilize the substrate. Bromoperoxidase
immobilized on agarose also maintains good catalytic activity in some organic
solvents [34].
TM

Figure 6 Bromide optima of some algal bromoperoxidases. (Source: Ref. 9.)
B. Functional Group Modifications

The following section summarizes information on the reactions of organic sub-
strates with marine bromoperoxidases. Examples are taken from published litera-
ture and from the authors’ laboratory. It is important to note that, in most cases,
a single set of reaction conditions was used, so any yields shown are not opti-
mized for the particular substrate. It is assumed that all appropriate controls have
been performed with each substrate, to ensure that the product is not spontane-
ously formed.
1. Alkenes and Alkynes (Table 2)

The primary products obtained from reaction of alkenes with bromoperoxidases
are bromohydrins. The mechanism is probably via electrophilic addition of the
bromonium ion (or equivalent) to the alkene, followed by the addition of a nucleo-
phile. However, although the Markovnikov regiochemistry is typically observed,
in several cases, anti-Markovnikov products are also found [35,36]. This suggests
that the enzyme is able to orient the substrate to some extent before the addition
occurs, and thus impart some degree of regioselectivity.
TM

Table 2 Alkene and Alkyne Substrates
Substrate Product(s) Enzyme source Reference
Propene Laurencia pacifica 36

Corallina offinalis
9 : 1
Allyl alcohol Corallina sp. 37
Cinnamyl C. pilulifera 38
alcohol [α] D ⫽ 0
Cinnamic C. pilulifera 38
[α] D ⫽ 0
acid
Cyclohexene C. pilulifera 38
[α]D ⫽ 0
Styrene C. pilulifera 38
[α]D ⫽ 0
cis-Propenyl C. pilulifera 38
[α]D ⫽ 0
phosphonic
acid
Geraniol Corallina 35
vancouveriensis
X ⫽ Br, Y ⫽ OH
X ⫽ OH, Y ⫽ Br
TM

Table 2 Continued
Linalool Corallina 35
X ⫽ Y ⫽ Br vancouveriensis
X ⫽ OH, Y ⫽ B
X ⫽ BR, Y ⫽ OH
α-Pinene a Corallina 35
vancouveriensis
a
See Fig. 7.
Dibromides and epoxides are also found in alkene product mixtures. By

using a mixture of chloride and bromide in a ratio of 500: 1 (mimicking the
concentration in seawater), Geigert et al. [36] were able to produce mixed dihal-
ides (Table 2). Epoxides are also seen as products of alkenes, such as geraniol,
which are more likely derived from intramolecular cyclization of the bromohy-
drins. α-Pinene yielded several ring-opened products when reacted with
Corallina vancouveriensis bromoperoxidase (Fig. 7) [35].
Itoh et al. measured the optical activity of four halohydrin products that
contained chiral centers and found that the [α] D was 0° in all cases [38]. When
reacted with the bromoperoxidase from C. vancouveriensis, α-pinene and linalool
both result in a surprising array of products. Some of the identified products are
shown in Table 2; a scheme explaining the origin of the pinene products is also
shown (Fig. 7). Enzymatic products of linalool also possess optical rotation mea-
surements of 0 [35].
Alkynes appear to be poor substrates for bromoperoxidases. Only phenyl
acetylene was found to react with C. officinalis enzymes, although the product
was not structurally characterized and may involve reaction with the aromatic
ring, and not the alkyne itself [38].
2. β-Diketones and Enolizable Carbonyl Compounds (Table 3)

The principal assay used by all researchers to quantify the activity of a haloperox-
idase is based on the α-halogenation of monochlorodimedone (2-chloro-3,3-
dimethyl-1,5-cyclohexanedione, or MCD), shown in Figure 3. In aqueous buffer
TM

Figure 7 Proposed mechanism of formation of α-pinene products.
TM

Table 3 β-Diketones and Enolizable Ketones
CHBr3 Penicillus capitatus 39
3-oxo-octanoic acid
R ⫽ CH 2 Br a
CHBr2
CBr3
3-oxo-octanoic acid CHBr3 Bonnemassonia 7

CH 2 Br2 hamifera
Monochlorodimedone Dihalodimedone Standard assay; see 10

(MCD) Fig. 3
a
The mono- and dibromoheptanones were also converted to the tribromoheptanone, but the formation
of CHBr 3 is nonenzymatic.
solutions, the enol form, which has an ⑀ value of 20,000/M-cm at 290 nm, pre-
dominates. If a haloperoxidase is added to a solution of MCD, hydrogen peroxide,
and bromide, the dihalo product is formed, which has an ⑀ of 200/M-cm. One
unit (U) of bromoperoxidase activity is defined as the amount of enzyme neces-
sary to convert one micromole of MCD to dihalo product at 25°C in 1 min [10].
The role of bromoperoxidases in the biosynthesis of brominated carbonyl
compounds was investigated by Theiler et al. [7]. When reacted with 3-oxoocta-
noic acid, the enzyme from Bonnemaissonia hamifera yielded a mixture of dibro-
momethane, bromoform, and 1-bromopentane. In subsequent studies, the bro-
moperoxidase from Penicillus capitatus produced brominated heptanones and
bromoform from the same substrate, which were similar or identical to com-
pounds found naturally in B. hamifera. The authors suggest that the difference
in product formation may be due to either distinctive catalytic abilities of each
or two simultaneous reaction pathways [39].
The ready formation of bromoform by action of bromoperoxidases has led
to the suggestion that these enzymes, through their natural catalytic activity with
simple carbonyl compounds, are responsible for a significant percentage of atmo-
spheric halomethanes, which are contributing factors to the degradation of the
earth’s ozone layer. Indeed, bromoperoxidases from marine algae are suggested
to be ultimately responsible for the formation of significant quantities of volatile
TM

organohalogen compounds that are found in the atmosphere [40,41], and it has
been suggested that these compounds are responsible for some damage to the
earth’s ozone layer [42,43].
3. Aromatic Compounds (Table 4)

Benzene rings containing substitutents that activate the ring toward electrophilic
aromatic substitution (e.g., phenols) react readily with bromoperoxidase, yielding
brominated products in good to high yields. Compounds with strongly deactivat-
ing substituents alone are not reactive, but if present with an activating substituent
(e.g., OH), may yield a brominated product. In all cases, the regiochemistry ap-
pears to follow the expected patterns for electrophilic aromatic substitution. Mul-
tiple brominated products are possible, but the yields and ratio of single vs. multi-
ple halogenated products may vary widely with reaction conditions.
4. Heterocycles (Table 5)
A variety of nitrogen heterocycles and one sulfur heterocycle have yielded prod-
ucts with bromoperoxidases. Itoh et al. have compared the relative halogenating
abilities of fungal chloroperoxidase and the bromoperoxidase from the alga
Corallina pilulifera and report some significant differences between them. Using
these enzymes, they were able to produce halogenated derivatives of nucleoside
and nucleotide bases, which are known to have activity as potential anticancer
agents. In this study, the reactions of the same substrates with fungal chloroperox-
idase were compared. With the exception of the chlorinated products, most of
the brominated and iodinated nucleic acid base products were similar. However,
the report indicates several practical differences that favor the use of
bromoperoxidase over the fungal enzyme [44]. Barbituric acid and derivatives
were brominated by the enzyme from Ascophyllum nodosum to only mono- or
dibromo products [45].
5. Amino Acids and Proteins (Table 5)

Using the bromoperoxidase from the brown alga Ascophyllum nodosum, a protein
was successfully radiolabeled with 77 Br and the conditions were optimized [46].
This represents a novel technique for labeling of therapeutically useful proteins.
The amino acid derivative methoxytyrosine yielded interesting decarboxylated
nitrile and aldehyde products [47].
6. Role of Bromoperoxidase in the Cyclization of a Complex

Natural Product
Elegant studies by Fukuzawa et al. [8] have demonstrated that the bromoperoxi-
dase from the red alga Laurencia nipponica is involved in the final stages of the
TM

Table 4 Aromatic Compounds
Phenol 2,4,6-Tribromophenol Corallina pilulifera 20
C. vancouveriensis 33
Anisole o-Bromoanisole C. pilulifera 38
p-Bromoanisole
o-Hydroxybenzyl alcohol 2,4,6-Tribromophenol Amphiroa ephedraea 20
Corallina pilulifera
C. vancouveriensis 33
m-Hydroxybenzyl alcohol 3-Hydroxy-2,4,6-tribromobenzyl alcohol C. vancouveriensis 33
4-Bromo-3-hydroxybenzyl alcohol
p-Hydroxybenzyl alcohol 2,6-Dibromo-4-hydroxy benzyl alcohol Rhodomela larix 22
Ptychodera flava
Methyl salicylate C. vancouveriensis 33
(91% yield)
Thymol C. vancouveriensis 33
56 : 31
1-Methoxynaphthalene C. pilulifera 38
Phenol red Bromphenol blue Most bromoperoxidases 9

(may be used as a
general indicator)
TM

TM
C
Table 5 Heterocyclic Compounds
Thiophene C. pilulifera 38
Pyrazole C. pilulifera 44
R ⫽ Br, I
Indoles ‘‘V-bromoperoxidase’’ 31
R ⫽ Me, Phenyl (from marine alga)
Barbituric acid and A. nodosum 45

derivatives
R1 ⫽ R2 ⫽ R3 ⫽ H R1 ⫽ R2 ⫽ H
Nucleic acid bases and nucleosides
Cytosine 5-Bromocytosine C. pilulifera 44
Uracil 5-Bromouracil, 5-iodouracil
Cytidine 5-Bromocytidine
Thymine (Decomposes)
(No reaction with adenine, adenosine, guanine, and 2-deoxyuridine)
Amino acid
P. capitatus 47
R ⫽ CH 2-CN
CH2-CHO
TM

TM
C
biosynthesis of cyclic metabolites that occur naturally in the organic extract. Us-
ing laurediols (also found as trace metabolites in L. nipponica) as substrate for
both lactoperoxidase and later the actual bromoperoxidase from the Laurencia,
it was demonstrated that each laurediol isomer yielded a cyclized product that was
identical to a natural product found in the alga (Fig. 8). This is the first example
in which a direct bromonium-induced cyclization is clearly demonstrated to be
catalyzed by a bromoperoxidase. Many halogenated marine-derived products
may also be envisioned to be derived from halonium ion cyclization [48].
IV. CONCLUSIONS
This review indicates bromoperoxidases from marine organisms are capable of

halogenating a wide variety of organic substrates, often in high yields. The gen-
eral mechanism appears to be via electrophilic substitution or addition routes, and
this is supported by most of the regiochemistry that is observed in the products.
However, in the reaction with certain alkenes, anti-Markovnikov products are
observed along with Markovnikov products, suggesting some difference from the
‘‘normal’’ chemical route to this functionality. Although stereoselectivity has not
yet been observed through the use of bromoperoxidases, it is also clear that the
development of these enzymes as chemoenzymatic reagents still requires a great
deal of effort. Most studies (including those from our laboratory) report the use
of these reactions under only one set of reaction conditions. Relatively little work
has been done in the optimization of the reaction conditions under which bromo-
peroxidases might show enhanced selectivity. As an example of the critical role
of these parameters, Allain et al. [49] have shown that chloroperoxidase may be
induced to produce chiral epoxides in high enantiomeric excess from disubsti-
tuted alkenes, but only, if the rate of hydrogen peroxide is precisely controlled.
Subsequent work with chloroperoxidase has demonstrated that chiral epoxides
with a high level of enantioselectivity may be formed with certain alkene sub-
strates [50,51]. The catalytic activity of the bromoperoxidases involves numerous
experimental parameters and will require a significant amount of effort to deter-
mine whether enantioselectivity can be shown in the laboratory.
Bromoperoxidases are widespread in the marine environment, with the ma-
rine algae probably being the largest natural resource of these enzymes. The
bromoperoxidase from Corallina officinalis is now offered commercially from
Sigma Chemical Co., and recent reports on the immobilization of this enzyme
on agarose have appeared [34]. Numerous other sources of bromoperoxidases
are known but have not been investigated in any detail (or reported in the litera-
ture) [9,17]. No reports of bromoperoxidases from marine fungi or marine bacte-
ria have appeared, although halogenated metabolites have been reported from
both groups. This is probably due to lack of investigation, rather than absence
TM

Figure 8 Cyclization of laurediols by Laurencia nipponica bromoperoxidase. (Source :
Ref. 8.)
TM

of the enzyme since bromoperoxidases are known from Pseudomonas sp., Strep-
tomyces sp., and other bacteria [4,52], and fungi are the traditional source of
chloroperoxidases. Genes encoding for bromoperoxidases have been cloned [53],
and thus genetic manipulation to produce quantities of these enzymes is feasible.
Bromoperoxidases have a significant advantage over chemical halogenation
methods in organic synthesis. In contrast to classical halogenation methods (e.g.,
Br 2 in CCl 4), the bromoperoxidases are extraordinarily simple to use and require
no unusual precautions. The most hazardous component of the reaction system is
hydrogen peroxide (usually diluted). The workup of the reaction merely involves
extraction of the product by an organic solvent. The resulting aqueous solution
may then be relatively easily disposed. Because the bromoperoxidase reaction
involves reagents that are relatively innocuous, it may be considered an ‘‘environ-
mentally benign’’ process, especially when compared to typical chemical meth-
ods to achieve the same synthetic goals. Nearly all of the conventional chemical
halogenation reactions require either a significant level of caution and/or special
disposal arrangements to minimize environmental contamination. Especially for
industrial applications, the use of enzymes may offer a beneficial alternative to
chemical methods. One potential application of bromoperoxidase is the radiola-
beling of proteins for diagnostic application using 77 Br. [54]
One major benefit of the use of bromoperoxidases will be realized if stereo-
or regiochemistry can be controlled for specific substrates, reflecting the natural
ability of these enzymes. It is clear that many experimental attributes of bromo-
peroxidases are not yet fully understood, yet these enzymes are undoubtedly in-
volved in the biosynthesis of chiral halogens in natural products. By analogy
with widely used enzymes such as lipases and esterases, bromoperoxidases from
distinct biological sources likely have similar gross catalytic properties but may
differ in the exact nature of their interaction with a specific substrate. This unique-
ness has yet to be taken advantage of by those investigating these potentially
highly useful enzymes.
The role of enzymes in organic synthesis is rapidly growing in both research
and industrial applications, as the versatility of enzymes and methodology for
their use become apparent. Monographs dedicated to the practical use of enzymes
in organic synthesis are published with regular updates, which include ‘‘recipes’’
and protocols for their use. One group of enzymes that are certain to gain greater
use and interest in the near future are the bromoperoxidases.
ACKNOWLEDGMENTS
The authors wish to gratefully acknowledge the financial support of the National
Institutes of Health (GM-38040), the CSU Awards for Research, and the Tropical
Technology Center (Okinawa, Japan).
TM

REFERENCES
1. G. W. Gribble, J. Nat. Prod., 55: 1353 (1992).

2. G. W. Gribble, Progress in the Chemistry of Organic Natural Products, Vol. 68
(W. Herz, et al., eds.), Springer, New York (1996).
3. E. D. Goldberg, The Sea, Vol. 2 (M. N. Hill, ed.), Wiley-Interscience, New York
(1963).
4. K.-H. van Pee, Annu. Rev. Microbiology, 50: 375 (1996).
5. S. L. Neidleman and J. Geigert, Endeavour, 11: 1 (1987).
6. (a) S. L. Neidleman and J. Geigert, Biohalogenation: Principles, Basic Roles and
Applications, Ellis Horwood Press (Wiley), New York (1986). (b) S. L. Neidleman
and J. Geigert, Biochem. Soc. Symp., 48: 39 (1983).
7. R. F. Theiler, J. S. Siuda, and L. P. Hager, in Food-Drugs from the Sea (P. N. Kaul
and C. J. Sindermann, eds.) Marine Technology Society, Norman, Okla. (1978).
8. A. Fukuzawa, M. Aye, Y. Takasugi, M. Nakamura, M. Tamura, and A. Murai, Chem.
Lett., 2307 (1994).
9. C. A. Moore and R. K. Okuda, J. Nat. Toxins, 5: 295 (1996).
10. D. R. Morriss and L. P. Hager, J. Biol. Chem., 421: 1763 (1966).
11. A. Zaks and D. R. Dodds, J. Am. Chem. Soc., 117: 10419 (1995).
12. D. G. Baden and M. D. Corbett, Comp. Biochem. Physiol., 64B: 279 (1979).
13. Y. P. Chen, D. E. Lincoln, S. A. Woodin, and C. R. Lovell, J. Biol. Chem., 266:
23909 (1991).
14. D. J. Sheffield, T. Harry, A. J. Smith, and L. J. Rogers, Phytochem., 32: 21 (1993).
15. M. J. Clague and A. Butler, Adv. Inorg. Chem., 9: 219 (1994).
16. M. Andersson, V. Conte, F. Di Furia, and S. Moro, Tetrahedron Lett., 36: 2675
(1995).
17. W. D. Hewson and L. P. Hager, J. Phycol., 16: 340 (1980).
18. B. E. Krenn, H. Plat, and R. Wever, Biochim. Biophys. Acta, 912: 287 (1987).
19. H. Yu and J. W. Whittaker, Biochem. Biophys. Res. Comm., 160: 87 (1989).
20. H. Yamada, N. Itoh, S. Murakami, and Y. Izumi, Agric. Biol. Chem., 49: 2961 (1985).
21. M. Pedersen, Physiol. Planta, 37: 6 (1976).
22. T. J. Ahern, G. G. Allan, and D. G. Medcalf, Biochim. Biophys. Acta, 616: 329
(1980).
23. H. Vilter, Phytochem., 23: 1387 (1984).
24. H. S. Soedjak and A. Butler, Biochim. Biophys. Acta, 1079: 1 (1991).
25. J. A. Manthey and L. P. Hager, J. Biol. Chem., 260: 9654 (1985).
26. D. G. Baden and M. D. Corbett, Biochem. J., 187: 205 (1980).
27. R. Jannun and E. L. Coe, Comp. Biochem. Physiol., 88B: 917 (1987).
28. T. Higa, in Marine Natural Products: Chemical and Biological Perspectives, Vol.
5 (P. J. Scheuer, ed.), Academic Press, New York (1981).
29. H. S. Soedjak and A. Butler, Biochem., 29: 7974 (1990).
30. A. Butler and J. V. Walker, Chem. Rev., 93: 1937 (1993).
31. R. A. Tschirret-Guth and A. Butler, J. Am. Chem. Soc., 116: 411 (1994).
32. H. S. Soedjak, J. V. Walker, and A. Butler, Biochem., 34: 12689 (1995).
33. M. Shang, R. K. Okuda, and D. R. Worthen, Phytochem., 37: 307 (1994).
TM

34. D. J. Sheffield, T. R. Harry, A. J. Smith, and L. J. Rogers, Phytochem., 38: 1103
(1995).
35. C. A. Moore and R. K. Okuda, unpublished results.
36. J. Geigert, S. L. Neidleman, S. K. DeWitt, and D. J. Dalietos, Phytochem., 23: 287
(1984).
37. S. L. Neidleman, W. F. Jr. Amon, and J. Geigert, U.S. Patent 4,247,641, Cetus Corp.
(1981).
38. N. Itoh, A. K. M. Q. Hasan, Y. Izumi, and H. Yamada, Eur. J. Biochem., 172: 477
(1988).
39. R. S. Beissner, W. J. Guilford, R. M. Coates, and L. P. Hager, Biochem., 20: 3724
(1981).
40. R. Theiler, J. C. Cook, L. P. Hager, and J. F. Siuda, Science, 202: 1094 (1978).
41. B. Walter and K. Ballschmiter, Chemosphere, 22: 557 (1991).
42. N. Itoh and M. Shinya, Marine Chem., 45: 95 (1994).
43. R. M. Moore, M. Webb, R. Tokarcyzk, and R. Wever, J. Geophys. Res., 101(C9):
20,899 (1996).
44. N. Itoh, Y. Izumi, and H. Yamada, Biochem., 26: 282 (1987).
45. M. C. R. Franssen, J. D. Jansma, H. C. van der Plan, E. de Boer, and R. Wever,
Bioorg. Chem., 16: 352 (1988).
46. F. Lambert and G. Slegers, Appl. Radiat. Isot., 45: 11 (1994).
47. M. Nieder and L. P. Hager, Arch. Biochem. Biophys., 240: 121 (1985).
48. W. Fenical, J. Phycology, 11: 245 (1975).
49. E. J. Allain, L. P. Hager, L. Deng, and E. N. Jacobsen, J. Am. Chem. Soc., 115:
4415 (1993).
50. A. F. Dexter and L. P. Hager, J. Am. Chem. Soc., 117: 817 (1995).
51. F. J. Lakner, K. P. Cain, and L. P. Hager, J. Am. Chem. Soc., 119: 443 (1997).
52. N. Itoh, N. Morinaga, and A. Nomura, Biochim. Biophys. Acta, 1122: 189 (1992).
53. S. J. Facey, F. Grob, L. C. Vining, K. Yang, and K. H. van Pee, Microbiology, 142:
657 (1996).
54. F. Lambert and G. Slegers, Appl. Radiat. Isot., 45: 11 (1994).
TM

4
LC-Hyphenated Techniques in the
Search for New Bioactive
Plant Constituents
Kurt Hostettmann, Maryse Hostettmann, Sylvain Rodriguez,

and Jean-Luc Wolfender
Université de Lausanne, Lausanne, Switzerland
I. INTRODUCTION
The use of medicinal plants for curing illnesses can be traced back over five
millennia to written documents of the early civilizations in China, India, and the
Near East, but it is doubtless an art as old as mankind. Even today, plants are a
major and significant source of drugs for the majority of the world’s population.
Although in industrialized countries medicinal plant research has faced cyclic
fortune during the last few decades [1], nonetheless, substances derived from
higher plants still constitute ca. 25% of prescribed medicines [2,3].
The plant kingdom represents an extraordinary reservoir of novel mole-
cules. Of the estimated 400,000–500,000 plant species around the globe, only a
small percentage has been investigated phytochemically and the fraction submit-
ted to biological or pharmacological screening is even lower. Since plants may
contain hundreds, or even thousands, of metabolites, there is currently a resur-
gence of interest in the vegetable kingdom as a possible source of new lead
compounds for introduction into therapeutic screening programs. Furthermore,
the rapid disappearance of tropical forests and other important areas of vegetation
has added renewed pressure for the rapid isolation and identification of bioactive
natural products.
In general, the approach adopted to obtain an exploitable pure plant constit-
TM

uent involves interdisciplinary work in botany, pharmacognosy, pharmacology,
chemistry, and toxicology and can be formulated as follows [1] (Fig. 1):
Selection, collection, botanical identification, and preparation of plant ma-
terial
Extraction with suitable solvents and preliminary analysis
Biological and pharmacological screening of crude extracts
Chromatographic separation of pure bioactive constituents guided by bio-
assay
Structure determination
Analysis and pharmacological profile of pure compounds
Toxicological testing
Partial or total synthesis
Preparation of derivatives for structure–activity relationships
By following this approach alone, there is a very high risk of unnecessarily
isolating bioactive, but known plant constituents. Furthermore, interesting lead
compounds that do not exhibit the tested activity will simply be missed. In order
to eliminate the time-consuming isolation of known constituents, hyphenated
techniques such as liquid chromatography–mass spectrometry (LC/MS), LC–
ultraviolet spectroscopy (LC/UV), and recently even LC–nuclear magnetic reso-
nance (LC/NMR) imaging are being introduced at the earliest stage of separation
to screen the crude extracts spectroscopically (Fig. 1) [4]. Indeed, efficient detec-
tion and rapid characterization of natural products play important roles as analyti-
Figure 1 Procedure for obtaining active principles from plants and use of LC hyphen-
ated techniques as strategic analytical screening tools during the isolation process of a
plant extract. LC, liquid chromatography.
TM

cal support in the work of phytochemists. This type of analysis is valuable for
detecting compounds with interesting structural features and targeting their isola-
tion. In this chapter, the role of the LC-hyphenated screening of the crude extracts
used as a complement to the biological screening will be particularly emphasized.
II. HYPHENATION IN HIGH-PERFORMANCE LIQUID

CHROMATOGRAPHY
High-performance liquid chromatography (HPLC) is used routinely in phyto-

chemistry to ‘‘pilot’’ the preparative isolation of natural products (optimization
of the experimental conditions, checking of the different fractions throughout the
separation) and to control the final purity of the isolated compounds [5–8]. For
chemotaxonomic purposes, the botanical relationships among different species
can be shown by chromatographic comparison of their chemical compositions.
Comparison of chromatograms, used as fingerprints, of authentic samples and of
unknowns permits identification of drugs and/or their adulteration. The selective
detection of a given product in a complex mixture allows good quantitative mea-
surement, as well as precise chemotaxonomic comparisons. Furthermore, in many
applications, it may be necessary not only to detect but also to identify compounds
in extracts. With conventional detection methods such as monochromatic UV
detection, the identity of peaks can be confirmed only from their retention times
by comparison with those of authentic samples. In order to get more information
on the metabolites of interest, there is a need for a multiple detection system that
takes advantage of chromatography as a separation method and spectroscopic
techniques as detection and identification methods. At present, a number of im-
portant hyphenated techniques are available, the application of which depends
upon the types of problems that have to be solved (Fig. 2).
In our laboratory, for the screening of crude plant extracts, HPLC coupled
with UV photodiode array detection (LC/DAD-UV) and LC/MS have been
mainly used [4,9,10]. Recently, several studies have also shown that HPLC cou-
pled with nuclear magnetic resonance (LC/NMR) is a very powerful complemen-
tary technique for on-line plant metabolite identification [11,12]. This new tool
has been configured into our LC/UV/MS systems [13].
The use of several complementary detection techniques is necessary to ob-
tain a complete picture of the extract composition. Indeed, a crude plant extract is
a very complex mixture containing sometimes hundreds or thousands of different
metabolites [1]. The chemical nature of these constituents varies considerably
within a given extract, and the variability of the physicochemical and spectro-
scopic properties of these compounds causes numerous detection problems. Al-
though different types of LC detectors such as UV, RI, fluorescence, electrochem-
ical, and evaporative light scattering exist, none of these detectors permits within
TM

Figure 2 Summary of different coupling possibilities for chromatographic separation
methods with spectroscopic techniques. GC, gas chromatography; SFC, supercritical fluid
chromatography; HPLC, high-performance liquid chromatography; CE, capillary electro-
phoresis; TLC, thin layer chromatography; UV/VIS, ultraviolet/visible spectroscopy;
FT-IR, Fourier transform infrared spectroscopy; MS, mass spectrometry; NMR, nuclear
magnetic resonance.
the same analysis the detection of all the secondary metabolites encountered in
a plant extract; each method has its own selectivity.
A. Liquid Chromatography/Ultraviolet Photodiode Array

For more than a decade HPLC coupled with UV photodiode array detection has
been used by phytochemists for screening extracts [14–16]. The UV spectra of
natural products often give useful information on the type of constituents and
TM

also, as in the case of polyphenols, on the oxidation pattern. New instruments
allow storage of UV spectra of reference compounds in databases and computer
matching can be realized automatically when screening for known constituents.
The LC/DAD-UV technique has been used extensively for the screening
of polyphenols [17,18], in combination with the postcolumn addition of classical
UV shift reagents, allowing the determination of the positions of free hydroxyl
groups on the polyphenol nucleus by comparison of the spectra obtained with
each of the specific reagents [14,19]. The equipment used for this purpose is
shown schematically in Fig. 3. Reagents are added to the effluent, leaving the
HPLC column by means of a low-volume mixing tee. Diode array detection is
then performed on the modified mobile phase. A weak base (sodium acetate)
deprotonates only the more acidic phenolic groups, whereas a strong base (so-
dium methoxide, potassium hydroxide) reacts with all phenolic groups. The shift
reagents AlCl 3 and H 3 BO 3 can also be employed and are compatible with the
acidic acetonitrile–water eluent system. As disodium hydrogen phosphate is un-
suitable as a weak base because of solubility problems in aqueous organic systems
(for example, with an acetonitrile–water eluent), it can be replaced by sodium
acetate (0.5 M ). However, this reagent is insufficient to deprotonate acidic pheno-
lic groups in an aqueous solvent system, and NaOH (0.01 M ) may be added via
a second HPLC pump to obtain a pH of 8 for the solvent exiting the HPLC
column (Fig. 3). With the combination NaOAc-H 3 BO 3 it is not always possible
to locate ortho-dihydroxyl groups in the acetonitrile–water system. This problem
Figure 3 Schematic representation of the experimental setup used for the postcolumn
addition of UV shift reagents. UV, ultraviolet.
TM

can be solved by using AlCl3 in the acidified mobile phase (0.1% trifluoroacetic
acid [TFA], pH 3). Similar shifts are obtained with AlCl 3-HCl in methanol, the
reagent typically used in the characterization of flavonoids. By comparing on-
line UV spectra with the addition of aluminum chloride at pH 7 and pH 3, it is
possible to detect labile ortho-dihydroxyl complexes [14,18].
However, LC/DAD-UV is limited to the detection of UV-active constit-
uents. The information recorded on-line is generally not sufficient for a complete
identification, and other on-line complementary spectroscopic information is
needed.
B. Liquid Chromatography/Mass Spectrometry

Although LC/UV has often been used for the analysis of crude plant extracts,
LC/MS has been introduced more recently; its use is still not widespread in the
phytochemical community [10]. At present, MS is one of the most sensitive meth-
ods of molecular analysis. Moreover, it has the potential to yield information on
the molecular weight as well as on the structure of the analytes. Because of its
high power of mass separation, very good selectivities can be obtained. For vola-
tile, nonpolar compounds from essential oils, MS detection used in combination
with gas chromatography (GC) separation has been applied successfully [20].
Indeed, GC/MS has become essential for applications in this field and has already
been used routinely for two decades as the ideal GC detector [21].
The coupling between LC and MS has not been straightforward since nor-
mal operating conditions of a mass spectrometer (high vacuum, high tempera-
tures, gas-phase operation, and low flow rates) are diametrically opposed to those
used in HPLC, namely, liquid-phase operation, high pressures, high flow rates,
and relatively low temperatures [22]. Because of the basic incompatibilities be-
tween HPLC and mass spectrometry (MS), on-line coupling of these instrumental
techniques has been difficult to achieve.
To cope with these different problems, a considerable number of LC/MS
interfaces have been developed. These LC/MS interfaces must accomplish nebu-
lization and vaporization of the liquid, ionization of the sample, removal of excess
solvent vapor, and extraction of the ions into the mass analyzer. Thus, different
techniques may involve variations in the methods by which these steps are accom-
plished. The most commonly used LC/MS interfaces are particle beam (PB) [23],
thermospray (TSP) [24], atmospheric pressure chemical ionization (APCI) [25],
and electrospray (ES) [26]. Each of these interfaces has its own characteristics
and range of applications. With the appropriate LC/MS configuration, the analy-
sis of small nonpolar molecules to very large polar molecules such as proteins
is now possible.
A number of LC/MS interfaces are suitable for the analysis of plant second-
ary metabolites. In our approach to LC/MS, mainly used for the HPLC screening
TM

of crude plant extracts, three interfaces, thermospray (TSP) [24], continuous flow
fast-atom bombardment (FAB) (CF-FAB) [27], and electrospray (ES) [26], have
been investigated [28]. They cover the ionization of relatively small nonpolar
products (aglycones, 200 µ) to more polar molecules (glycosides, 2000 µ)
(Table 1). In our laboratory, thermospray is the most widely used interface.
1. Thermospray Liquid Chromatography/Mass Spectroscopy

The TSP interface [2] has most often been used for natural product analysis in
crude plant extracts [10,29], and it has emerged as one of the most popular LC/
MS interfaces, although ES is also gaining more acceptance. Thermospray allows
introduction of the liquid aqueous phase into the MS at a flow rate (about 1–2
mL/min) compatible with that usually employed for the standard HPLC condi-
tions (organic solvents and aqueous mixtures in either isocratic or gradient mode).
Thermospray operation provides mass spectra that are closely related to chemical
ionization (CI solvent mediated) and is thus well suited for the analysis of moder-
ately polar molecules in a mass range from 200 to 1000 µ or more in certain
cases.
In the TSP system, the LC effluent is partially vaporized and nebulized in
the heated vaporizer probe to produce a supersonic jet of vapor containing a mist
of fine droplets or particles. As the droplets travel at high velocity through the
heated ion source, they continue to vaporize because of rapid heat input from
the surrounding hot vapor. A portion of the vapor and ions produced in the ion
source escapes into the vacuum system through a sampling cone, and the remain-
der of the excess vapor is pumped away by a mechanical vacuum pump [30] (Fig.
Table 1 Characteristics of the Liquid Chromatography/Mass Spectroscopy (LC/MS)

interfaces: Thermospray (TSP), Continuous-Flow Fast-Atom Bombardment (CF-FAB),
and Electrospray (ES)
Interface TSP CF-FAB ES
Flow 1–2 mL/min 5–10 µL/min 1–1000 µL/min

Molecular weight 200–800 µ 500–2000 µ 200–500,000 µ
Fragments Moderately polar Polar Moderately polar
and polar
Liquid chromato- Good Poor Excellent
graphy resolution
Stability Medium Poor Good
Operation Many parameters Difficult to handle Easy, great influ-
to optimize ence of the mod-
ifiers
TM

Figure 4 Schematic representation of a TSP–LC/MS interface. TSP, thermospray; LC,
liquid chromatography; MS, mass spectrometry.
4). Four ionization modes exist with this interface. Besides chemical ionization
initiated by conventional electron bombardment using a heated filament (‘‘fila-
ment-on’’ mode) or a discharge electrode (‘‘discharge-on’’ mode), a third ioniza-
tion method, in which ammonium acetate or some other volatile buffer in the
mobile phase plays a major role (‘‘thermospray buffer’’ ionization) is available;
the fourth ionization mode is based on an ion evaporation mechanism. This latter
mechanism may be operative in TSP buffer ionization mode as well [30].
The chemical process of ionization plays an important role and should al-
ways be taken into account when analyzing samples with LC/TSP-MS. Usually,
as for CI, the values of the proton affinity of the analytes, the solvent, and the
buffer should be considered. In our experience, for the analysis of plant metabo-
lites (150–1000 µ, aglycones, mono- and diglycosides), the TSP operated mainly
in the positive-ion mode in ‘‘thermospray buffer’’ ionization generally afforded
MS spectra similar to those produced by (desorption chemical ionization (DCI)
with NH 3 as reagent gas, positive-ion mode).
In order to perform satisfactory LC/TSP-MS investigations of plant metab-
olites, or any other sample, different parameters should be optimized. Many inter-
dependent experimental parameters are important in this respect: the mobile
phase composition (amount and type of buffer and organic modifier), the solvent
flow rate, the temperature of the vaporizer and the ion source, together with the
geometry, the position, and the potential of the repeller electrode. The tuning of
TM

these parameters is of utmost importance in order to get high sensitivity and
selectivity for a given type of analyte. Different articles have already emphasized
the problems of thermospray optimization [31] and this particular aspect will not
be further developed in detail here.
C. Liquid Chromatography/Nuclear Magnetic Resonance

Nuclear magnetic resonance (NMR) spectroscopy is by far the most powerful
spectroscopic technique for obtaining detailed structural information about or-
ganic compounds in solution. Given a molecular mass, NMR spectroscopy can
usually provide sufficient additional information to identify a completely un-
known compound unambiguously [32]. Thus LC/NMR represents a potentially
important complementary hyphenated technique to LC/UV and LC/MS.
Proton-detected high-performance liquid chromatography/nuclear mag-
netic resonance (LC/ 1 H-NMR), despite being known for over 15 years [33], has
not yet become a widely accepted technique, mainly because of its lack of sensi-
tivity. However, the recent progress in pulse field gradients and solvent suppres-
sion, the improvement in probe technology, and the construction of high-field
magnets have given a new impulse to this technique.
Although LC coupling itself was rather straightforward compared to LC/
MS [30], one of the main problems of LC/NMR was the difficulty of observing
analyte resonances in the presence of the much larger resonances of the mobile
phase. This problem was compounded under LC conditions when more than one
protonated solvent was used, and when solvent composition changed during gra-
dient analysis. Furthermore, the continuous flow of sample in the detector coil
complicated solvent suppression. These problems have now been overcome in
part because of the development of fast, reliable, and powerful solvent suppres-
sion techniques such as the ‘‘water suppression enhanced through T 1 effect’’
technique (WET) [34], which produces high-quality spectra in both on-flow and
stop-flow mode. These techniques consist of a combination of pulsed field gradi-
ents, shaped rf pulses, shifted laminar pulses, and selective 13 C decoupling and are
much faster than classical presaturation techniques previously used [34]. Thus, in
reversed-phase HPLC conditions, nondeuterated solvents such as MeOH or
MeCN can be used, and water is replaced by D 2O.
In contrast to conventional NMR probes, with LC/NMR the sample to be
analyzed is not placed in a rotating 5-mm internal diameter (i.d.) tube. The mobile
phase is flowing continuously through a nonrotating glass tube connected at both
ends with HPLC tubing. The detection is made in most LC/NMR systems in
flow cells of 60–180 µL with 2- to 4-mm i.d. placed in the active NMR coil
region of a conventional probe body [32]. The choice of the volume of the flow
cell is dependent on the sensitivity and/or the LC resolution required for each
specific application.
TM

D. Setup for Liquid Chromatography/Ultraviolet/Mass
Spectroscopy and Liquid Chromatography/Nuclear
Magnetic Resonance Analyses
In order to perform LC/UV/MS and LC/NMR analyses of crude plant extracts,
a setup allowing the use of standard HPLC conditions (1 mL/min, gradient with
aqueous solvent systems) with all the detectors has been used. An HPLC pump
equipped with a gradient controller provides the eluent for the HPLC separation
of the plant extracts (Fig. 5). A reversed-phase column (i.d. about 4 mm) is most
often used. At the column outlet, the eluent passes through a photodiode array
detector equipped with a high-pressure cell.
For LC/MS, and with the aim of using the same columns without changing
the chromatographic conditions, the buffer needed for the various LC/MS inter-
faces is added post column after the UV detector. For LC/TSP-MS, all the eluent
enters the spectrometer, whereas for CF-FAB and ES, the eluent is split and the
excess sample discarded (Fig. 5). Thus for LC/TSP-MS operation, postcolumn
addition of the buffer needed for ‘‘TSP buffer’’ ionization is provided by an
additional HPLC pump (usually ammonium acetate 0.5 M, 0.2 mL/min). The
total eluent (1.2 mL/min) containing the buffer passes through the TSP interface,
and the exhaust eluent is pumped away by a mechanical pump and trapped in a
cold trap.
For LC/CF-FAB-MS operation, the additional HPLC pump allows the
postcolumn addition of the glycerol matrix needed for FAB ionization (usually
Figure 5 Schematic representation of the experimental setup used for LC/UV/MS (1)
and LC/UV/NMR (2) analyses. LC, liquid chromatography; UV, ultraviolet; MS, mass
spectroscopy; NMR, nuclear magnetic resonance.
TM

glycerol 50%, 0.2 mL/min). The viscous matrix is efficiently mixed into the
eluent with a visco mixer. The mixture is then split with an accurate splitter (1/
100), and only 10 µL/min of the total eluent enters the CF-FAB interface through
a fused silica capillary. For LC/ES-MS operation, trifluoroacetic acid (TFA) or
ammonium acetate is added into the eluent and the flow is split to leave ca. 200
µL/min flow rate entering the interface.
For LC/NMR, no postcolumn addition is necessary and the water is re-
placed by D 2O. As with LC/MS, the separation is first monitored by a UV detec-
tor. In the on-line mode, all the eluent passes through the UV detector and then
through the LC/NMR probe without splitting. In the stop-flow mode, the HPLC
pump is stopped as soon as a peak of interest reaches the center of the NMR
detection cell. A computer controlling the LC pump, the UV detector, and the
NMR is used to synchronize this operation. The UV monitor is used to detect
the peak and the HPLC is stopped after a delay corresponding to the time needed
by this peak to reach the NMR cell.
With such a setup, standard HPLC conditions for crude extract analysis (1
mL/min, 4 mm i.d. column) can be maintained without alteration. Furthermore,
LC/DAD-UV detection is not affected by the buffer or the matrix used.
III. ROLE OF LIQUID CHROMATOGRAPHY/ULTRAVIOLET,

LIQUID CHROMATOGRAPHY/MASS SPECTROSCOPY,
AND LIQUID CHROMATOGRAPHY/NUCLEAR
MAGNETIC RESONANCE AS SCREENING METHODS
IN PHYTOCHEMICAL ANALYSIS
When searching plant extracts for compounds with interesting properties, a multi-
dimensional approach to their chromatographic analysis is of great significance.
By combining HPLC with on-line UV, MS, and NMR, a large amount of prelimi-
nary information can be obtained about the constituents of an extract before isola-
tion of the compounds of interest. In the case of polyphenols, for example, UV
spectra recorded on-line give useful information (type of chromophore or pattern
of substitution) complementary to that obtained with LC/MS and provide an
initial assignment of the peak of interest at an early stage. When this information
is not sufficient for a precise structural assignment, LC/NMR provides a useful
complement for a full structural identification on-line. To illustrate this approach,
examples of on-line LC/UV/MS, LC/MS/MS, and LC/NMR analyses of xan-
thones, flavones, and secoiridoids in crude extracts of plant species belonging to
the Gentianaceae family will be discussed. Gentianaceae plants have indeed been
extensively screened in our laboratory, especially in the search for novel antide-
pressive agents.
TM

A. Search for Antidepressive Compounds
from Gentianaceae
It is generally accepted that depression is characterized by low neurotransmitter
concentration in the synapses. From a therapeutic viewpoint, there are two ways
to increase the neurotransmitter concentration: slow down their recapture or in-
hibit their deamination. In this latter approach, a search for selective and revers-
ible inhibitors of monoamine oxidase (MAO) has become of major importance
[35,36].
Disturbances in monoamine oxidase levels have been reported in a series
of disorders such as Parkinson’s disease, Huntington’s chorea, depression, anxi-
ety, and senile dementias [37]. The enzyme monoamine oxidase (MAO) plays a
key role in the regulation of certain physiological amines in the human body. It
causes deactivation by deamination of neurotransmitters such as catecholamines
and serotonin, and of endogenous amines such as tyramine. This enzyme exists
as two isoenzymes, MAO-A and MAO-B, which exhibit different substrate speci-
ficities. Inhibitors of MAO, in particular of the A type isoenzyme, have potential
as antidepressive drugs since they increase both noradrenaline and serotonin lev-
els in the brain. There is current interest in the discovery of new reversible inhibi-
tors of MAO [37] because existing drugs such as clorgyline (selectively inhibits
MAO-A) and (⫺)-deprenyl (selectively inhibits MAO-B) are irreversible inhibi-
tors and exhibit serious side effects.
It is well known that plants contain inhibitors of MAO [38]. For example,
the bark of the tropical liana Banisteriopsis caapi (Malpighiaceae) contains β-
carboline alkaloids such as harmaline, a potent reversible inhibitor of MAO-A
[39]. Infusions of St. John’s wort, Hypericum perforatum (Guttiferae), have been
used for the treatment of melancholia in European folk medicine, and numerous
preparations of this plant have now been commercialized for the management of
different depressive states. The antidepressive activity of H. perforatum has been
demonstrated in vivo [40], and it was originally thought that the anthrone dimer
hypericin, which has shown MAO-A and MAO-B inhibitory activity in vitro
[41], was responsible for this activity. However, more recent investigations point
toward xanthones and flavonoids as the major source of the antidepressive activ-
ity since these have potent MAO inhibitory properties [42].
Investigations with extracts of Canscora decussata (Gentianaceae) [43] in
vivo showed that the fractions containing xanthones had a stimulating effect on
the central nervous system in mice, rats, and dogs [44]. A strong inhibition of
monoamine oxidase (IMAO) by trisubstituted xanthones was observed by Suzuki
and coworkers [45]. In fact, a number of xanthones that inhibit MAO have been
identified from plant sources [46–48]. Of those that have been tested, the agly-
cones are, in general, more active than the corresponding glycosides. Xanthones
with the substitution patterns 1-hydroxy-3,8-dimethoxy- and 1,3-dihydroxy-7,8-
TM

dimethoxy- showed pronounced activity. They were also selective inhibitors, be-
ing more effective against MAO-A than MAO-B. All 1,3,5,8-tetrasubstituted xan-
thones isolated from the aerial parts of Gentiana lactea (Gentianaceae) have been
tested against monoamine oxidases from rat brain mitochondria. Two glycosides
and the aglycone swerchirin (1,8-dihydroxy-3,5-dimethoxyxanthone) were inac-
tive. Bellidifolin (1,5,8-trihydroxy-3-methoxyxanthone), on the other hand, in-
hibited MAO-A significantly. The activities of these xanthones were compared
to those of specific synthetic inhibitors of MAO-A (clorgyline) and MAO-B (par-
gyline). The MAO-A activity of bellidifolin was of the same order of magnitude
as that of pargyline, and 40 times less than that of clorgyline. It was also selective
with a marked reduction in activity against MAO-B (IC 50 MAO-A/IC 50 MAO-
B ⫽ 0.001). Desmethylbellidifolin (1,3,5,8-tetrahydroxyxanthone) gave an inhib-
itory concentration (IC 50) against MAO-A that was approximately 10 times higher
than that obtained with bellidifolin [48].
In a search for more active xanthones, other plants (especially of the fami-
lies Gentianaceae, Guttiferae, and Polygalaceae) have been investigated. A com-
parison of the IMAO activity of some representative extracts showed, for exam-
ple, that the dichloromethane extract of Chironia krebsii, a member of the
Gentianaceae from Malawi, was much more active (100% IMAO-A at 15 µg/
mL) than that of Hypericum perforatum (18% IMAO-A at 15 µg/mL), which
was of the same order of magnitude as that of the yellow gentian, Gentiana lutea
(30% IMAO-A at 15 µg/mL) [49]. In order to find new xanthones, numerous
extracts of Gentianaceae have been screened by LC-hyphenated techniques, and
several examples of this research are described in the next section.
B. Liquid Chromatography/Ultraviolet/Mass Spectroscopy

Analysis of the Aperitif Suze
One of the first gentian extracts to be analyzed by LC/UV/MS was not a crude
plant extract but a beverage made with the roots of the yellow gentian G. lutea.
Gentians contain bitter compounds belonging to the monoterpene glycosides (sec-
oiridoids) and have been known since ancient time to stimulate appetite [50]. As
it is known that yellow gentian also contains xanthones such as gentisin (1) that
showed IMAO activity, it was of interest to check for the presence of 1 in the
French aperitif called Suze. Direct injection of an aliquot of this beverage resulted
in an LC/UV chromatogram with many different peaks, but only one having the
characteristic xanthone UV spectrum (Fig. 6) [51]. As xanthones usually appear
as aglycones or relatively small glycosides (mono- or diglycosides), the TSP
interface was chosen for their ionization. After careful tuning of the TSP interface
(especially the source and vaporizer temperatures), the peaks recorded by UV
detection (254 nm) gave a clearly discernible MS response in the total ion current
trace (TIC). The LC/TSP-MS spectrum of this peak exhibited a protonated mo-
TM

Figure 6 LC/UV analysis of the aperitif Suze  and UV and TSP-MS spectra of the
only xanthone found in this beverage: gentisin (1). UV traces were recorded at 254 nm
and UV spectra from 200 to 500 nm. HPLC: column, RP-18 NovaPak (4 µm, 150 ⫻ 3.9
mm i.d.); gradient, CH 3 CN-H 2 O (0.1% TFA) 5 :95 → 65: 35 in 50 min (1 mL/min),
NH 4 OAc 0.5 M (0.2 mL/min post column). TSP: positive-ion mode; filament off; vapor-
izer 100°C; source 280°C. LC/UV, liquid chromatography/ultraviolet; TSP-MS, ther-
mospray–mass spectrometry; HPLC, high-performance liquid chromatography; TFA, tri-
fluoroacetic acid.
lecular [M ⫹ H] ⫹ ion at m/z 259. The molecular weight of this xanthone was
thus 258 µ. As the mass of a bare xanthone nucleus is 196 µ, this indicated that
the polyphenol was substituted by two hydroxyls and one methoxyl group. A
comparison of the on-line UV spectra with those of trisubstituted xanthones from
an in-house UV database confirmed that this compound was indeed gentisin, the
main xanthone of the yellow gentian [52]. This example shows that for simple
known compounds like xanthones, a precise identification can be performed on-
line with the help of LC/UV and LC/MS, provided some information about the
TM

type of the constituents and their occurrence in given plant families is already
available.
C. Liquid Chromatography/Ultraviolet/Mass Spectroscopy

and Liquid Chromatography/Ultraviolet Analyses of
Chironia krebsii Extracts with Postcolumn Addition of
Ultraviolet Shift Reagents
The extracts of the roots of Chironia krebsii, a species from Africa, showed
significant inhibition of MAO-A in vitro (CH 2Cl 2 ext. 100%; MeOH ext. 75%
IMAO-A (15 µg/mL)). As mentioned, these activities were several orders of
magnitude greater than those measured for H. perforatum extracts (CH 2Cl 2 ext.
18%; MeOH ext. 14% IMAO-A, 15 µg/mL) [49]. The LC/UV analysis of C.
krebsii indicated that this plant was very rich in xanthones (see the structures in
Fig. 7). The number of bands and the general aspects of the UV spectra of xan-
thones (UV spectra 4–15 in Fig. 8) allowed a first attribution of the type of
oxygenation patterns encountered [51,53]. The constituents having a weaker
chromophore (UV max. ca. 250 nm) were attributable to the secoiridoids swertia-
marin (2) and sweroside (3), widespread bitter principles of the Gentianaceae.
In order to get more information on the molecular weight, the molecular formula,
and the sugar sequence (glycosides) of the xanthones, LC/TSP-MS analysis of
the extract was carried out. The same LC conditions as used for the LC/UV/
MS analysis of the aperitif Suze were employed; LC/TSP-MS was performed in
positive-ion mode with the filament off mode using ammonium acetate as buffer.
The source was set at 280°C and the vaporizer at 100°C [18]. Under these condi-
tions, the total ion current trace recorded with LC/TSP-MS showed ionization
of all the peaks recorded in the UV trace of the extract (254 nm) (Fig. 8 and 9).
Nevertheless, the total ion current responses for highly hydroxylated xanthones
with high melting point temperatures (i.e., 11, Fig. 8) were found to be very weak
in comparison to the UV trace of the corresponding peak, leading to a difficult
MS detection [10].
The TSP mass spectra of the xanthone aglycones recorded on-line after
HPLC separation exhibited only [M ⫹ H] ⫹ ions as the main peak (see TSP spectra
of 15, Fig. 8, or 14 and 16, Fig. 9). Corresponding mass spectra of the xanthone
glycosides usually showed two weak ions due to [M ⫹ H] ⫹ and [M ⫹ Na] ⫹
adducts, and a main peak for the protonated aglycone moiety [A ⫹ H] ⫹ (see TSP
spectra of 9, Fig. 8, or 4 and 8, Fig. 9). In the case of diglycosides, successive
losses of the monosaccharide units were marked by corresponding peaks in the
spectrum. Thus, for the diglycosides of Chironia species, a loss of 132 µ was
first observed, corresponding to a pentosyl residue, followed by a loss of 162 µ
corresponding to a hexosyl moiety, leading to the aglycone ion [A ⫹ H] ⫹ (see
TM

Figure 7 Structure of the xanthones, flavones, and secoiridoids found in Chironia kreb-
sii, Gentiana rhodantha, Halenia corniculata, and Swertia calycina.
TM

Figure 8 LC/UV and LC/TSP-MS analysis of the root methanolic extract of Chironia
krebsii (Gentianaceae). UV spectra 2–3 are characteristic of secoiridoids. Peaks 4–11 are
xanthone-O-glycosides; 12–15 are xanthone aglycones. LC/UV/MS: same conditions as
for Fig. 6. LC/UV, liquid chromatography/ultraviolet; TSP, thermospray; MS, mass spec-
troscopy.
TSP spectra of 9, Fig. 8, or 4 and 8, Fig. 9). After subsequent isolation of these
glycosides, the disaccharide was shown to be primeverose, a β-d-xylopyranosyl-
(1 → 6)-β-d-glucopyranoside disaccharide unit, often encountered in the Gen-
tianaceae family [54].
Since MS detection allows the selective recording of each ion trace, it was
possible to reconstruct the chromatogram of the ion m/z 657 and to obtain a
precise assignment of the peak corresponding to compound 9 in the extract. Simi-
larly, the specific ion trace at m/z 363 gave responses for the HPLC peaks of 9
TM

Figure 9 Summary of all the on-line spectral data obtained by LC/TSP-MS, LC/UV,
and LC/UV with postcolumn addition of shift reagents for xanthones 4, 8, 14, and 16
from the root methanolic extract of Chironia krebsii (Gentianaceae). LC/UV/MS: same
conditions as for Fig. 6. LC/UV/MS, liquid chromatography/ultraviolet/mass spectros-
copy.
TM

with AlCl 3 reagent. In the case of 8, an important bathochromic shift was ob-
served, characteristic of the chelated hydroxyl group at position C-1, whereas
for 3 no such shift was recorded (Fig. 9: UV of 4 and 8). The primeverosyl
moiety was thus attached at position C-1 in the case of 4 and at C-5 for 8. The
corresponding aglycone of these two glycosides, 1,5-dihydroxy-3-methoxyxan-
thone (12) was also found as a slower eluting peak in the extract (Fig. 8), con-
firming the assignments of 4 and 8. Thus, the structures were established as
5-hydroxy-3-methoxy-1-O-primeverosylxanthone (4) and 1-hydroxy-3-methoxy-
5-O-primeverosylxanthone (8). In similar fashion, the structures of compounds
4–16 were also deduced (Fig. 7).
Xanthones 4–16 were isolated [56] in order to test their IMAO activity.
Several of them showed important selective and reversible inhibition of MAO-
A. The most active xanthone was 1,5-dihydroxy-3-methoxyxanthone (12) with
an inhibitory activity of 0.04 µM (IC 50 MAO-A) [49]. The UV spectra of the
pure products, fully identified, were registered in a UV database in order to permit
a computer matching search against xanthones from other extracts and thus accel-
erate on-line identification of these compounds in other Gentianaceae species
[57].
D. Liquid Chromatography/Thermospray–Mass
Spectroscopy and Liquid Chromatography/Continuous
Flow–Fast-Atom Bombardment–Mass Spectroscopy of
Gentiana rhodantha
Among the other Gentianaceae species screened by LC/UV/MS, Gentiana rho-
dantha from China presented a completely different extract composition profile
from that found for C. krebsii. The LC/UV analysis of G. rhodantha showed
the presence of only one predominant xanthone 17 and different secoiridoids.
Compound 17 was easily identified as the widespread xanthone C-glycoside man-
giferin (see structure in Fig. 7). The TSP-MS spectrum of 17 exhibited a pseudo-
molecular ion at m/z 423, characteristic fragments for C-glycosides at [M ⫹ H-
90] ⫹ and [M ⫹ H-120] ⫹, and a weak aglycone ion at m/z 261 (xanthone with
four OH groups). The computer fit for the UV spectrum of 17 with our in-house
UV spectral library allowed the definitive characterization of 17.
With the help of the LC/TSP-MS spectra recorded on-line, chemotaxonom-
ical considerations, and comparison with pure standards, the secoiridoids with
retention times less than 10 min were also easily identified [58]; among them a
very minor secoiridoid, sweroside (3)(MW 358), was found to be present. The
slower eluting peaks 18 and 19 also exhibited the same characteristic UV spectra
of secoiridoids (one band around 240 nm, Fig. 10). These compounds, which
were less polar than common secoiridoids, were studied in more detail. The LC/
TSP-MS analysis of 18 and 19 in each case gave a spectrum identical to that
TM

(this ion is the main fragment of 9) and also of compound 15, a slower-running
component of the extract (see ion traces in Fig. 8). According to the UV and
TSP mass spectra of 15, this compound was readily identified as the correspond-
ing free aglycone of 9. This information allowed the rapid identification of both
aglycones and their corresponding glycosides simultaneously in a crude plant
extract.
In order to obtain more structural information on the position of the free
hydroxyl groups on the xanthone nucleus, LC/UV with postcolumn addition of
shift reagents was performed, using the setup described previously (see Fig. 3).
The structural information obtained on-line by the combination of LC/UV, LC/
TSP-MS, and LC/UV with postcolumn addition of shift reagents is illustrated
for four xanthones (4, 8, 14, and 16, Fig. 9).
The two xanthone aglycones (14 and 16) exhibited almost identical UV
spectra (Fig. 9). According to Kaldas [55] the UV spectra of 14 and 16 with four
absorption maxima and a higher intensity for band II are characteristic for 1,3,7,8-
tetraoxygenated xanthones with a free hydroxyl at position C-1. The TSP-MS
spectra recorded on-line permitted the molecular weight (MW) determination and
the assignment of the number and type of substituents of 14 (MW 274: three
OH, one OMe) and 16 (MW 302: one OH, three OMe, Fig. 9). The shifted UV
spectra recorded on-line for 14 confirmed this compound to be 1,7,8-trihydroxy-
3-methoxyxanthone. Indeed, the presence of hydroxyl groups at positions C-1
and C-8 was characterized by the important bathochromic shift recorded with
AlCl 3, and the ortho-dihydroxyl group at C-7 and C-8 was confirmed by the shift
due to the complexation of boric acid. The position of the methoxyl at C-3 was
confirmed by the absence of shift registered with the weak base NaOAc. The
spectra of 16 with added KOH showed a significant decrease in band intensity
and only a very small shift, indicating no free hydroxyl groups, with the exception
of a chelated one. This was confirmed by the shift measured with AlCl 3. The
spectra with added NaOAc and H 3BO 3 remained unchanged, confirming the
structure of 16 to be 1-hydroxy-3,7,8-trimethoxyxanthone (Fig. 9).
The two isomeric xanthone glycosides 4 and 8 also exhibited nearly identi-
cal UV spectra indicating the same oxygenation pattern (probably 1,3,5) [55].
The TSP-MS spectra of 4 and 8 were comparable: they both exhibited an [M ⫹
H] ⫹ pseudomolecular ion at m/z 553 with consecutive losses of 132 and 162 µ,
leading to the same aglycone ion [A ⫹ H] ⫹ at m/z 259 (Fig. 9: TSP-MS of 4
and 8). The mass of the aglycone ion was characteristic of a xanthone with one
methoxyl and two hydroxyl groups, and the presence of a disaccharide moiety
consisting of a pentosyl and hexosyl subunits. The disaccharide was assumed to
be primeverosyl since it is the only disaccharide of this mass that is known in
the Gentianaceae family [54]. No shift was observed for either compound in the
presence of weak base, indicating no free hydroxyl group at position C-3. The
only difference between these two isomers was shown by the spectra recorded
TM

Figure 10 Combined TSP (a) and CF-FAB (b) LC/MS of the enriched n-BuOH fraction
of the methanolic extract of Gentiana rhodantha (Gentianaceae). HPLC: column, RP-18
Novapak (4 µm, 150 ⫻ 3.9 mm i.d.); gradient, CH 3 CN-H 2 O (0.05% TFA) 5 :95 → 50:
50 in 30 min (0.9 mL/min). TSP (a): positive-ion mode; filament off; vaporizer, 90°C;
source, 230°C; AcONH 4 0.5M (0.2 mL/min post column). CF-FAB (b): negative-ion
mode; FAB tip 50°C; source 100°C; glycerol 50% (v/v) (0.15 mL/min post column); LC
flow post column split 1:100. TSP, thermospray; CF-FAB, LC/MS, liquid chromatogra-
phy; HPLC, high-performance liquid chromatography; TFA, trifluoroacetic acid.
obtained for 3; all exhibited an intense ion at 359 u and no ion at higher mass
(Fig. 10). However, the chromatographic behavior was quite different for 18, 19,
and 3. In order to obtain complementary information about these constituents, a
second LC/MS analysis with CF-FAB was undertaken, using the same HPLC
conditions. The total ion current recorded for the whole chromatogram showed
a very important MS response for compounds 18 and 19, while the more polar
metabolites were only weakly ionized (Fig. 10). The CF-FAB spectrum of 18
recorded on-line in negative-ion mode exhibited a very intense pseudomolecular
TM

ion [M ⫺ H] ⫺ at m/z 913 together with a weak ion at m/z 555 corresponding
to the loss of a ‘‘swerosidelike’’ unit [M ⫺ H-358] ⫺ (CF-FAB spectrum of 18,
Fig. 10b). This complementary information indicated clearly that the molecular
weight of 18 was 914 u. For the same compound, only a fragment corresponding
to a ‘‘sweroside-like’’ unit m/z 359 was recorded during the LC/TSP-MS analy-
sis (TSP spectrum of 18, Fig. 10a). According to the different results obtained
on-line for 18 in the HPLC screening of the extract of G. rhodantha, it was
concluded that 18 was probably a type of moderately polar, large secoiridoid
containing at least one unit very similar to 3.
The CF-FAB spectrum of 19 exhibited an intense [M ⫺ H] ⫺ pseudomolecu-
lar ion at m/z 1629, indicating a molecular weight of 1630 for this compound.
Fragments at m/z 1271, 913, and 555, respectively, showed the consecutive losses
of ‘‘swerosidelike’’ units (⫺358 u). These on-line LC/MS results suggested that
19 was probably similar to 18 with two more ‘‘swerosidelike’’ units attached to
it.
After the LC/MS screening results, a targeted isolation of 18 and 19 was
undertaken. A full structure determination of 18 with the help of one-dimensional
(1D) and two-dimensional (2D) NMR experiments as well as with different
chemical transformations showed that 18 consisted of two secoiridoid units linked
together with a monoterpene unit through two ester groups (structure in Fig. 7)
[58]. The full structure determination of 19 showed that this compound possessed
two more secoiridoid units than 18, which were esterified on the free carboxylic
functions of 18 [59]. These two compounds were found to be natural products
of a new type. The structure determination of other closely related compounds
from this extract is still in progress.
This example illustrates the use of different LC/MS ionization techniques
for targeting unknowns; LC/TSP-MS indicated that compounds 18 and 19 had
common subunits (identical fragments m/z 359), and CF-FAB allowed the on-
line molecular weight determination of all of these oligomeric compounds. The
combination of both types of information has thus allowed the early recognition
of this type of large secoiridoid glycoside.
E. Liquid Chromatography/Thermospray–Mass
Spectroscopy and Liquid Chromatography/
Electrospray–Mass Spectroscopy of
Halenia corniculata
As shown in the previous example, the use of different complementary MS ion-
ization techniques is often important for unambiguous on-line determination of
molecular weights. The analysis of another of the Gentianaceae species collected
in Mongolia, Halenia corniculata, by TSP and LC/ES-MS also demonstrated
the importance of an LC/MS analysis using two independent ionization methods
for screening unknowns.
TM

The whole-plant MeOH extract of H. corniculata, which contains xan-
thones, flavones, and secoiridoid glycosides [57], was analyzed by both LC/TSP-
MS and LC/ES-MS. In the latter case, the analysis was carried out with TFA as
buffer. Ionization of all the glycosides (20–30) found in the extract was observed
in the negative ion (N.I.) LC/ES-MS (Fig. 11), but the response for the aglycones
was very selective and most of them were not ionized [28]. The LC/ES-MS base
peak trace showed a better chromatographic resolution than the corresponding
LC/TSP-MS trace. The ES spectra of flavonoids 22, 23, and 30; xanthones 25–
29; as well as secoiridoid glycosides 20, 21, and 24 exhibited clearly discernible
TFA anion adducts [M ⫹ CF 3 CO 2] ⫺, together with deprotonated ions [M ⫺ H] ⫺.
No aglycone ions or other fragments were observed (see 22, Fig. 11). For poly-
phenols 22, 23, and 25–30, the positive ion (P.I.) TSP spectra recorded on-line
exhibited [M ⫹ H] ⫹ ions together with base peak [A ⫹ H] ⫹ aglycone ions. In
the case of diglycosides such as 22 (Fig. 11), the loss of the terminal sugar unit
was also observed. This demonstrated that the spectra of a given compound such
as 22 will differ considerably according to the ionization mode chosen. A good
knowledge of the ionization process is thus necessary for a correct peak assign-
ment [28].
In this example it was also shown that compound 24, a minor secoiridoid
glycoside ester of the extract, was not ionized in TSP but in LC/ES-MS exhibited
a deprotonated molecular ion [M-H] ⫺ at m/z 697 and a strong TFA adduct
[M ⫹ CF 3 CO 2] ⫺ at m/z 811, indicating a molecular weight of 698 u (Fig. 12).
The UV spectrum of 24 (λ max 219, 244, 297, 321 nm) differed from those usually
encountered in Gentianaceae species extracts. The two UV bands, at 219 and
244 nm, together with the retention time (ca. 16 min), suggested that 24 was an
iridoid-type compound. The two other maxima at 297 and 321 nm indicated the
presence of an aromatic unit. No literature reference matched such a compound
corresponding to this molecular weight. Thus, as in the case of G. rhodantha, a
targeted isolation of 24 was undertaken [60]. Different hydrolyses of 24 followed
by the HPLC of the degradation products showed the presence of a caffeoyl, a
secoiridoid, and two hexosyl moieties. The structure was fully established by
NMR experiments as 7-β-[4′-O-(β-d-glucopyranosyl)-trans-caffeoyloxy] swero-
side, and this new natural product was named corniculoside [60].
F. Liquid Chromatography/Ultraviolet/Mass Spectroscopy

and Liquid Chromatography/Mass Spectroscopy/Mass
Spectroscopy Analysis of the Methanolic Extract of
Swertia calycina
Besides xanthones and secoiridoids, the LC/UV analysis of the methanolic ex-
tract of Swertia calycina, a Gentianaceae native of Rwanda in central Africa,
revealed the presence of four flavonic constituents, 31–34 (Fig. 13) [61]. The
TSP-MS spectra of these four flavonoids exhibited protonated molecular
TM

Figure 11 Combined LC/TSP-MS and LC/ES-MS of the MeOH extract of Halenia
corniculata (Gentianaceae). HPLC: C18 NovaPak (4 µm, 150 ⫻ 3.9 mm i.d.); gradient,
CH 3 CN-H 2O 5 :95 → 65:35 in 50 min (1 mL/min). Injection 200 µg. LC/TSP-MS (P.I.)
and LC/ES-MS (N.I.) allowed the ionization of the different glycosides 20–30. Only the
secoiridoid glycoside 24 was not ionized by TSP. As shown for the spectra of the flavonoid
diglycoside 22, [M⫹H]⫹ ion and fragments (aglycone ion, sugar losses) were observed
in LC/TSP-MS (P.I.), only [M⫺H] ⫺ and [M⫹CF 3 COO] ⫺ were visible in LC/ES-MS
(N.I.). LC/TSP-MS, liquid chromatography/thermospray–mass spectroscopy; ES, elec-
trospray; HPLC, high-performance liquid chromatography; P.I., positive ion; N.I., nega-
tive ion.
TM

Figure 12 LC/ES-MS analysis of the MeOH extract of Halenia corniculata (Gentiana-
ceae) with the ES-MS and UV spectra of corniculoside (24). This compound was not
recorded by LC/TSP-MS. LC: same conditions as for Fig. 6. LC, liquid chromatography;
ES, electrospray; MS, mass spectroscopy; UV, ultraviolet; TSP, thermospray.
TM

Figure 13 (a) LC/TSP-MS and (b) LC/TSP-MS/MS spectra obtained on-line for isovi-
texin (33) in the methanolic extract of Swertia calycina. The LC/TSP-MS/MS spectrum
of the ion m/z 313 (c) exhibits characteristic fragments for the substitution of the B-ring
and for the glycosidation at C-6. LC/UV/MS: same conditions as for Fig. 6. LC/TSP-
MS, liquid chromatography/thermospray–mass spectroscopy; UV, ultraviolet.
TM

[M ⫹ H] ⫹ at m/z 449 (31), 463 (32), 433 (33), and 447 (34). Each spectrum
also showed fragments [M ⫹ H ⫺ 90] ⫹ and [M ⫹ H ⫺ 120] ⫹ characteristic of
the cleavage of C-glycoside moieties [38] (see TSP-MS spectrum of 33, Fig.
13A). Thus, 31–34 were all flavone C-glycosides, and according to their molecu-
lar weights, it was deduced that 31 bore four OH groups; 32, three OH groups
and one OMe group; 33, three OH groups; and 34, two OH groups and one
OMe group. The LC/TSP-MS/MS [62] analysis of the characteristic ions of these
flavonoids provided more precise information about the fragmentation of the C-
glycoside moieties and the localization of the substituents on the A- and B-rings.
The LC/TSP-MS/MS was performed on the extract of S. calycina under the same
LC and ionization conditions established for LC/TSP-MS. In the collision cell,
argon was used for collisionally induced decomposition (CID). In a first analysis,
the pseudomolecular ions m/z 449, 463, 433, and 447, corresponding to 31, 32,
33, and 34, respectively, were selected as parent ions and the daughter ions were
recorded. The TSP-MS/MS spectra obtained on-line in the extract clearly showed
for each flavonoid the characteristic [M ⫹ H ⫺ 120] ⫹ and [M ⫹ H ⫺ 150] ⫹
ions, due to the cleavage of the C-glycoside moieties [38], but the aglycone ion
[A ⫹ H] ⫹ was not visible and further fragmentation of the aglycone moiety was
not observed (see LC/TSP-MS/MS spectrum of 33, Fig. 13B). Thus, the LC/
TSP-MS/MS analysis of the flavonoid C-glycosides using the protonated pseudo-
molecular ions as parent ions did not provide information on the substitution sites
on the A- and B-rings of the aglycones of these compounds. In contrast to O-
glycosides that exhibit intense aglycone ions [A ⫹ H] ⫹ [62], the [A ⫹ H] ⫹ ions
of C-glycosides are usually undetectable in the LC/TSP-MS spectra of flavonoids
(see LC/TSP-MS spectrum of 33, Fig. 13A) and cannot be used for fragmentation
studies. Thus, a second LC/TSP-MS/MS analysis of these constituents in the
extract was performed by choosing the intense [M ⫹ H ⫺ 120] ⫹ ions as parent
ions. In this case, a classic retro-Diels-Alder (RDA) fragmentation of the agly-
cone moiety of the flavonoids was observed [63]. For 31 and 32, a clearly discern-
ible ion at m/z 137 was characteristic for a fragment of the B-ring bearing two
hydroxyl groups, for 33 and 34, this fragment was recorded at m/z 121, indicating
only one hydroxyl (see LC/TSP-MS/MS spectrum of 33, Fig. 13C).
With this analysis it was also possible to ascertain the position of C glyco-
sylation. As shown in Fig. 13, the LC/TSP-MS/MS spectrum of 33, obtained by
choosing the [M ⫹ H ⫺ 120] ⫹ as parent ion, exhibited fragments at m/z 149
and 177 characteristic of 6-C-glycosylated flavones. Indeed, it is known that TSP-
MS/MS spectra of the [M ⫹ H ⫺ 120] ⫹ fragments of isomeric C-glycosylfla-
vones show different specific daughter ions [64]. The A-C-ring fragments issued
from RDA cleavages are only observable for C-6 positional isomers as in the
case of 33. With these data, 33 was identified as isovitexin [65]. Similarly com-
pounds 31, 32, and 34 were established as 6-C-glycosylflavones, and with addi-
tional computerized UV comparisons and retention time measurements were
TM

Figure 14 LC/UV and LC/TSP-MS analysis of the crude CH 2 Cl 2 extract of Swertia
calycina. For each major peak, the single-ion LC/MS traces of the protonated molecular
ions [M⫹H] ⫹ were displayed, together with the UV spectra obtained on-line. LC/
liquid chromatography/ultraviolet; TSP, thermospray; MS, mass spectroscopy.
TM

identified as isoorientin, swertiajaponin, and swertisin, respectively. All these
flavonoids are well known in Swertia species and further isolation was not per-
formed [61].
G. Liquid Chromatography/Ultraviolet/Mass Spectroscopy

and Liquid Chromatography–Nuclear Magnetic
Resonance Analysis of the Dichloromethane Extract
of Swertia calycina
The dichloromethane extract of Swertia calycina was strongly antifungal against
Cladosporium cucumerinum and Candida albicans. Using the TLC bioautogra-
phy assay [66], this activity was linked to a strong UV visible spot (R f ⫽ 0.38;
petroleum ether/ethylacetate 1: 1). As some xanthones are known antifungal
agents [67], one of these was suspected.
The LC/UV chromatogram of the dichloromethane extract of S. calycina
was simpler than the methanolic extract, and three main peaks (3, 35, and 16)
were detected (Fig. 14). Compound 16 presented a UV spectrum with four ab-
Figure 15 Bidimensional LC/ 1H-NMR chromatogram of the crude CH 2Cl2 extract of

Swertia calycina. Methoxyl groups and aromatic proton signals of 35 and 16 are clearly
visible together with all the resonances of the monoterpene glycoside 3. The signal of
HOD is negative and was continually shifted during the LC gradient. LC, liquid chroma-
tography; NMR, nuclear magnetic resonance.
TM

sorption bands characteristic of a xanthone. The TSP-MS spectrum exhibited a
strong protonated ion at 303 u, indicating a xanthone with a molecular weight
of 302, and thus substituted with one hydroxyl and three methoxyl groups (Fig.
14). This information, together with the comparison with a UV spectral library,
permitted the identification of 16 as decussatin, a xanthone widespread in the
Gentianaceae family, also previously found in C. krebsii [56]. The on-line data
obtained for 3 indicated the presence of a secoiridoid-type molecule with a molec-
ular weight of 358 µ. The loss of 162 µ observable in the TSP spectrum was
characteristic of the presence of a hexosyl moiety. These data suggested strongly
that 3 was most probably sweroside, a well-known secoiridoid of the family [68].
The UV spectrum of 35 was not attributable to a common polyphenol of the
Gentianaceae such as a flavone or xanthone. It was very weakly ionized in TSP
(see the important background ions of the TSP spectrum in Fig. 16, for example,
at m/z 231 and 200), but a protonated molecular ion was nevertheless found at
m/z 189. This low molecular weight (188 u) and the UV spectrum suggested
Figure 16 Summary of all the spectroscopic data obtained on-line for naphthoquinone
35 in the CH 2 Cl 2 extract of Swertia calycina.
TM

that 35 could be a quinonic compound, but as no metabolite of this type was
previously found in Gentianaceae, it was not possible to identify it on-line.
In order to confirm these attributions and to obtain more structural informa-
tion on-line, the same extract of S. calycina was submitted to an on-line LC/ 1 H-
NMR analysis on a 500-MHz instrument [13]. The same LC conditions as in the
LC/UV/MS analysis were used except that the water of the LC gradient system
was replaced by D 2O. However, the quantity of extract injected onto the column
was increased to 1 mg to collect at least 20 µg for each peak of interest.
For the suppression of the solvent signal of MeCN and its two 13 C satellites,
as well as the residual HOD peak, WET solvent suppression [34] was run before
each acquisition. In the gradient LC run, the solvent peaks change frequency
during the course of the experiment. As a result, the solvent suppression must
be continuously adjusted for optimal performance. To do this, a one-transient-
one-pulse experiment is used to find the solvent peaks prior to WET suppression
(Scout Scan) [34]. As a result of this sequence, the transmitter is automatically
adjusted to keep the biggest solvent peak at a constant frequency (MeCN), while
the spectrometer is locked on D 2 O [13].
The on-line LC/NMR analysis of S. calycina provided 1 H-NMR spectra
for all the major constituents. A plot of the retention time ( y axis) versus the
NMR shifts (x axis) permitted the localization of the resonances of compounds
3, 35, and 16 (Fig. 15). On this unusual two-dimensional chromatogram, strong
signals of aromatic methoxyl groups were observed around 4 ppm for 35 and
16. Xanthone 16 exhibited two pairs of aromatic protons, and the quinonic com-
pound 35 presented five other low-field protons. The more polar secoiridoid 3
showed different signals between 3 and 6 ppm. The important trace starting from
4.8 ppm (at 0 min) and ending to 4 ppm (at 30 min) was due to the change of
the chemical shift of the residual negative water (HOD) signal during the LC
gradient. The traces between 1 and 2.6 ppm were due to residual MeCN signal
and solvent impurities.
The cross sections of this bidimensional plot along the x-axis produced
single on-line LC/ 1 H-NMR spectra for each constituent, allowing a precise as-
signment of their specific resonances. Xanthone 16 exhibited three methoxyl reso-
nances at 3.92, 3.93, and 3.95 ppm; a pair of meta-coupled aromatic protons (δ
6.55, d, J ⫽ 2.4 Hz, H-4; 6.41, d, J ⫽ 2.4 Hz, H-2) was indicative of a 1,3-
disubstituted A-ring. The B-ring protons exhibited a pair of ortho-coupled protons
(δ 7.36, d, J ⫽ 9.2 Hz, H-5; 7.62, d, J ⫽ 9.2 Hz, H-6), suggesting a 1,3,5,6- or
1,3,7,8-substitution pattern for 16. These NMR data together with UV and MS
information led to the conclusion that 16 was 1-hydroxy-3,7,8-trimethoxyxan-
thone (decussatin), as already suggested by the LC/UV/MS data.
In the LC/ 1 H NMR spectrum of 35 (Fig. 16), two signals (δ 8.11, m, 2H:
H-5,8; 7.89, m, 2H: H-6,7) were characteristic of four adjacent protons of an
aromatic ring with two equivalent substituents. The low-field shift of the H-5,8
TM

signal indicated that these two protons were in peri positions of carbonyl func-
tions, suggesting the presence of a naphthoquinone nucleus [69]. The strong
bands recorded in the UV spectrum at 243, 248, 277, and 330 nm confirmed this
deduction. The singlet at 6.35 ppm was attributed to H-3 and the remaining me-
thoxyl group was thus at position C-2. With these on-line data and the molecular
weight deduced from the LC/TSP-MS spectrum (MW 188), 35 was finally identi-
fied as 2-methoxy-1,4-naphthoquinone. As this was the first naphthoquinone to
Figure 17 Stop-flow WET-COSY spectrum of sweroside (3) in the CH 2 Cl 2 extract of

Swertia calycina; 3 was kept 30 min in the cell for the total acquisition (8 scans ⫻ 220
increments). WET-COSY, water suppression enhanced through J, effect correlated spec-
troscopy.
TM

be reported in Gentianaceae, it was isolated and was found to be the compound
responsible for the strong antifungal activity of the extract of S. calycina [61].
As indicated earlier, UV and MS data suggested that 3 was the secoiridoid
sweroside. The on-line LC/ 1 H-NMR data indicated different resonances than
those corresponding to this secoiridoid, but a rather poor signal-to-noise-ratio
was obtained for this compound. In order to obtain clearer resonances, the LC/
1
H-NMR spectrum of 3 was also recorded in the stop-flow mode. To perform
this experiment, the LC flow was stopped when 3 was eluting into the LC/NMR
flow cell. This was possible by following the separation of the extract by UV
and synchronizing the LC pump to stop when the peak of interest entered the
NMR detection cell (see setup in Fig. 5). In this mode, the analyte was maintained
in the NMR for a greater number of transients and a better signal-to-noise ratio
was obtained. All the resonances for 3 [70] were then easily identified. In order
to prove the 1H-1 H connectivities for this secoiridoid, an 1 H, 1 H-correlated spec-
troscopy (COSY) spectrum was also acquired during the same stop-flow experi-
ment (Fig. 17). This stop-flow WET-COSY experiment performed on the LC
peak of 3 allowed assignment of all the 1 H-1 H correlations within the molecule.
Only 35 min was needed to acquire and process this 2D spectrum. With these
data, it was indeed possible to link all the protons of the glucosyl moiety of 3,
starting from the anomeric proton (δ 4.84, d, J ⫽ 8.6 Hz, H-1′) and ending at
the methylene H-6′a (δ 3.92) and H-6′b (δ 3.73) signals. Also, all the connectivit-
ies from the terminal methylene H-10 to H-7 progressing through the monoter-
pene skeleton were clearly assigned. These 1D and 2D LC/ 1H-NMR data, to-
gether with UV and MS information, allowed the identification of 3 as sweroside.
This example showed that the LC/UV/MS and LC/NMR information obtained
for S. calycina permitted a full structure identification of its main constituents.
IV. CONCLUSIONS
The LC-hyphenated techniques are playing an increasingly important role as stra-

tegic tools to support phytochemical investigations. Indeed, these techniques pro-
vide a great deal of preliminary information about the content and nature of con-
stituents of crude plant extracts. In certain cases, combination with a spectral
library and precolumn or postcolumn derivatization allow structure determination
on-line. This is very useful when large numbers of samples have to be processed
because unnecessary isolation of compounds is avoided. Once the novelty or
utility of a given constituent is established, it is then important to process the
plant extracts in the usual manner, to obtain samples for full structure elucidation
and biological or pharmacological testing.
In this chapter, it has been shown that in the search for xanthones with
potential monoamine oxidase–inhibitory activity, LC/MS and LC/UV analyses
TM

have been of great value for the preliminary screening of numerous species of
Gentianaceae. These methods have been used not only to determine the presence
of xanthones but also to provide an indication of their structures.
The use of LC/UV and LC/MS has also aided the localization of new types
of constituents such as secoiridoids of unusual molecular weight and helped to
target their isolation. The introduction of complementary techniques such as post-
column addition of UV shift reagents or MS/MS have extended the possibilities
of on-line structural determination with LC/UV/MS.
Analysis of crude plant extracts using LC/MS is, however, not straightfor-
ward because of the great variety of constituents. As has been shown, no interface
allows an optimal ionization of all the metabolites within a single crude plant
extract. Often, different ionization modes or different interfaces are necessary to
obtain a complete picture of the extract composition.
The recent introduction of LC/NMR for crude plant extract screening will
probably allow another breakthrough in the on-line structural determination of
natural products. This hyphenated method allows the recording of precious com-
plementary on-line structure information when LC/UV/MS data are insufficient
for an unambiguous peak identification. Indeed, LC/NMR has proved to be very
effective in obtaining 1D spectra of both flowing and nonflowing samples, as
well as stop-flow 2D spectra. However, compared with UV or MS, NMR remains
a rather insensitive detection method. Thus, in the on-line mode, this technique
is at present limited to the characterization of the major constituents of crude
plant extracts. In order to obtain 1 H-NMR spectra of minor metabolites or to
perform bidimensional experiments, the use of the stop-flow mode is necessary.
With the full set of spectroscopic information obtained by LC/UV, LC/
MS, and LC/NMR, the phytochemist will be able to characterize the main constit-
uents of a given plant rapidly and choose which metabolites are to be isolated
for in-depth structural or pharmacological study. The chemical screening of ex-
tracts with such elaborate hyphenated techniques generates a huge amount of
information. In order to rationalize this approach and to use it efficiently with
high sample throughput, the next challenge will be to find a way to centralize
all these data for rapid pattern recognition by reference to standard compounds.
With such an analytical system, phytochemists will then be able to concentrate
their efforts on finding new biological targets. This aspect remains the more dif-
ficult problem to solve when searching for new leads.
Both biological and chemical screenings provide important information on
the plant constituents, but these results will not be sufficient for the discovery of
new potent drugs if no suitable pharmacological model exists. Thus, an important
condition for success in the discovery of new bioactive plant constituents is the
establishment of effective collaborations among botanists, phytochemists, and
pharmacologists in order to realize all the steps involved, starting with plant mate-
rial in the field and concluding with viable pharmacologically active compounds.
TM

Only in this manner can medicinal plants be investigated efficiently and new
leads rapidly discovered.
ACKNOWLEDGMENT
Financial support was provided by the Swiss National Science Foundation.

Thanks are due to Dr. Wolf Hiller, Varian AG, Darmstadt, Germany, for the
LC/NMR measurements and to Dr. Winfried Wagner-Redeker, Spectronex AG,
Basel, Switzerland, for the LC/ES-MS experiments.
REFERENCES
1. M. Hamburger and K. Hostettmann, Phytochemistry, 30: 3864 (1991).

2. N. R. Farnsworth and A. S. Bingel, in New Natural Products and Plant Drugs with
Pharmacological, Biological or Therapeutical Activity (H. Wagner and P. Wolff,
eds.), Springer, New York, pp. 61 (1977).
3. P. Principe, in Economic and Medicinal Plant Research, (H. Wagner, H. Hikino,
and N. R. Farnsworth, eds.), Academic Press, London, p. 1 (1989).
4. J.-L. Wolfender and K. Hostettmann, in Phytochemistry of Medicinal Plants (J. T.
Arnason, R. Mata and J. Romeo eds.), Plenum Press, New York, pp. 189 (1995).
5. F. Gränicher, P. Christen, and P. Vuagnat, Phytochem. Anal., 5: 297 (1994).
6. N. Nykolov, T. Iossifova, E. Vassileva, I. Kostova, and G. Stoev, Phytochem. Anal.,
4: 86 (1993).
7. R. L. Arslanian and F. R. Stermitz, Phytochem. Anal., 2: 35 (1991).
8. D. G. I. Kingston, J. Nat. Prod., 42: 237 (1979).
9. K. Hostettmann, A. Marston, and J.-L. Wolfender, in Phytochemistry of Plants Used
in Traditional Medicine (K. Hostettmann, M. Marston, M. Maillard, and M. Ham-
burger, eds.), Oxford Science Publications, Oxford, p. 17 (1995).
10. J.-L. Wolfender, M. Maillard, and K. Hostettmann, Phytochem. Anal., 5: 153 (1994).
11. A. Höltzel, G. Schlotterbeck, K. Albert, and E. Bayer, Chromatographia, 42: 499
(1996).
12. O. Spring, H. Buschmann, B. Vogler, E. Schilling, M. Spraul, and M. Hoffmann,
13. J.-L. Wolfender, S. Rodriguez, K. Hostettmann, and W. Hiller, Phytochem. Anal.,
8: 97 (1997).
14. K. Hostettmann, B. Domon, D. Schaufelberger, and M. Hostettmann, J. Chro-
matogr., 283: 137 (1984).
15. I. Yoshimura, K. Yasuhiro, Y. Yoshikazu, S. Huneck, and Y. Yasuyuki, Phytochem.
Anal., 5: 197 (1994).
16. P. M. Bramley, Phytochem. Anal., 3: 97 (1992).
17. B. Ducrey, J.-L. Wolfender, A. Marston, and K. Hostettmann, Phytochemistry, 38:
129 (1995).
TM

18. J.-L. Wolfender and K. Hostettmann, J. Chromatogr., 647: 191 (1993).
19. I. Mueller-Harvey and P. M. S. Blackwell, Phytochem. Anal., 2: 38 (1991).
20. Y. Masada, Analysis of Essential Oils by Gas Chromatography and Mass Spectrome-
try, John Wiley & Sons, New York (1976).
21. F. David and P. Sandra, Phytochem. Anal., 3: 145 (1992).
22. D. A. Garteiz and M. L. Vestal, LC Mag., 3: 334 (1985).
23. R. C. Wiloughby and R. F. Browner, Anal. Chem., 56: 2625 (1984).
24. C. R. Blakley and M. L. Vestal, Anal. Chem., 55: 750 (1983).
25. A. P. Bruins, T. R. Covey, and J. T. Henion, Anal. Chem., 59: 2642 (1987).
26. R. C. Whitehouse, R. N. Dreyer, M. Yamashita, and J. B. Fenn, Anal. Chem., 57:
675 (1985).
27. R. M. Caprioli, F. Tan, and J. S. Cotrell, Anal. Chem., 58: 2949 (1986).
28. J.-L. Wolfender, S. Rodriguez, K. Hostettmann, and W. Wagner-Redeker, J. Mass
Spectrom., S35 (1995).
29. R. Verpoorte and W. M. A. Niessen, Phytochem. Anal., 5: 217 (1994).
30. W. M. A. Niessen and J. van der Greef, Liquid Chromatography Mass Spectrometry,
Marcel Dekker, New York (1992).
31. C. E. M. Heeremans, R. A. M. Van der Hoeven, W. M. A. Niessen, U. R. Tjaden,
and J. van der Greef, J. Chromatogr., 474: 149 (1989).
32. K. Albert, J. Chromatogr. A, 703: 123 (1995).
33. N. Watanabe, E. Niki, and S. Shimizu, JEOL News, 15 A: 2 (1979).
34. S. H. Smallcombe, S. L. Patt, and P. A. Keiffer, J. Magn. Reson. A 117: 295 (1995).
35. P. Kielholtz, Schweiz. Apot. Ztg., 120: 507 (1982).
36. M. Delini-Stula, The Origin of Depression: Current Concept and Approaches, Dah-
lem Konferenzen, Springer, Heidelberg, pp. 351 (1983).
37. M. Stroelin-Benedetti and P. Dostert, in Advances in Drug Research (B. Testa, ed.),
Academic Press, London, pp. 65 (1992).
38. A. Bakhtiar, J. Gleye, C. Moulis, and I. Fourasté, Phytochem. Anal., 5: 86 (1994).
39. V. Deulofeu, in Ethnopharmacologic Search for Psychoactive Drugs (D. H. Efron,
B. Holmstedt and N. S. Kline eds.), US Public Health Service publ. No 1645, Wash-
ington D.C., pp. 393 (1967).
40. S. N. Okpanyi and M. L. Weischer, Arzneimittelforschung, 37: 10 (1987).
41. O. Suzuki, Y. Katsumata, M. Oya, S. Bladt, and H. Wagner, Planta Med., 50: 272
(1984).
42. B. Sparenberg, MAO-inhibierende Eigenschaften von Hypericum-inhaltsstoffen und
Untersuchungen zur Analytik und Isolierung von Xanthonen aus Hypericum perfora-
tum, PhD Thesis, University of Marburg (1993).
43. S. K. Bhattacharya, S. Ghosal, R. K. Chaudhuri, and A. K. Sanyal, J. Pharm. Sci.,
61: 1838 (1972).
44. S. Ghosal, D. K. Jaiswal, S. K. Singh, and R. Srivastava, Phytochemistry, 24: 831
(1985).
45. O. Suzuki, Y. Katsumata, M. Oya, V. M. Chari, R. Klapfenberger, H. Wagner, and
K. Hostettmann, Biochem. Pharmacol., 27: 2075 (1978).
K. Hostettmann, Planta Med., 42: 17 (1981).
TM

K. Hostettmann, Planta Med., 39: 19 (1980).
48. D. Schaufelberger and K. Hostettmann, Planta Med., 54: 219 (1988).
49. U. Thull, Monoamine oxidase inhibitors of natural and synthethic origin: Biological
assay and 3D-OSAR, Thesis, Université de Lausanne (1995).
50. H. Wagner and K. Muenzig-Vasirian, Dtsch. Apoth. Z., 115: 1233 (1975).
51. K. Hostettmann and M. Hostettmann, in Methods in Plant Biochemistry, (J. B. Har-
borne ed.), Academic Press, London, pp. 493 (1989).
52. G. G. Nikolaeva, V. I. Glyzin, M. S. Mladentseva, V. I. Sheichenko, and A. V.
Patudin, Khim. Prir. Soedin., 1: 107 (1983).
53. A. A. Lins Mesquita, D. De Barros Correa, O. R. Gottlieb, and M. Taveira Magal-
haes, Anal. Chim. Acta, 42: 311 (1968).
54. K. Hostettmann and H. Wagner, Phytochemistry, 16: 821 (1977).
55. M. Kaldas, Identification des composés polyphénoliques dans Gentiana campestris
L., Gentiana germanica Willd. et Gentiana ramosa Hegetschw, Thesis, Université
de Neuchâtel, Switzerland (1977).
56. J.-L. Wolfender, M. Hamburger, J. D. Msonthi and K. Hostettmann, Phytochemistry,
30: 3625 (1991).
57. S. Rodriguez, J.-L. Wolfender, G. Odontuya, O. Purev, and K. Hostettmann, Phyto-
chemistry, 40: 1265 (1995).
58. W.-G. Ma, N. Fuzzati, J.-L. Wolfender, K. Hostettmann, and C. Yang, Helv. Chim.
Acta, 77: 1660 (1994).
59. W. Ma, N. Fuzzati, J.-L. Wolfender, C.-R. Yang and K. Hostettmann, Phytochemis-
try, 43: 805 (1996).
60. S. Rodriguez, J.-L. Wolfender, G. Odontuya, O. Purev and K. Hostettmann, Helv.
Chim. Acta, 79: 363 (1996).
61. S. Rodriguez, J.-L. Wolfender, E. Hakizamungu, and K. Hostettmann, Planta Med.,
61: 362 (1995).
62. Y. Y. Lin, K. J. Ng, and S. Yang, J. Chromatogr., 629: 389 (1993).
63. T. J. Mabry and K. R. Markham, in The Flavonoids (J. B. Harborne, T. J. Mabry,
and H. Mabry, eds.), Academic Press, New York, pp. 78 (1975).
64. G. Rath, A. Touré, J.-L. Wolfender, and K. Hostettmann, Chromatographia, 41:
332 (1995).
65. K. Hostettmann, G. Bellmann, R. Tabacchi, and A. Jacot-Guillarmod, Helv. Chim.
Acta, 56: 3050 (1973).
66. L. Rahalison, M. Hamburger, M. Monod, E. Frenk, and K. Hostettmann, Planta
Med., 60: 41 (1994).
67. A. Marston, M. Hamburger, I. Sordat-Diserens, J. D. Msonthi, and K. Hostettmann,
68. H. Inouye, S. Ueda, and Y. Nakamura, Chem. Pharm. Bull., 18: 1856 (1970).
69. R. H. Thomson, Naturally Occurring Quinones, University Press, Cambridge
(1987).
70. T. A. van Beek, P. P. Lankhorst, R. Verpoorte, and A. Baerheim Svendsen, Planta
Med., 44: 30 (1982).
TM

5
Determination of the Absolute
Configuration of Biologically Active
Compounds by the Modified
Mosher’s Method
Takenori Kusumi
Tokushima University, Tokushima, Japan
Ikuko I. Ohtani
University of the Ryukyus, Okinawa, Japan
I. INTRODUCTION
Most biologically active natural products have stereogenic centers and are ob-
tained from their natural sources enantiomerically pure. In pharmaceutical drug
development of naturally occurring and synthetic compounds, identification of
the absolute configuration of the drug is crucial. In many cases, when a drug is
chiral, only one enantiomer is effective, and sometimes the other enantiomer
may be seriously hazardous [1]. Because of the continuous development of new
spectroscopic techniques, unambiguous structure determination of natural and
synthetic products has become relatively routine even in cases in which the mo-
lecular weight of a compound is greater than 1000 daltons.
Few methods exist, however, to elucidate the absolute configuration of or-
ganic compounds; the most commonly employed nonchemical procedures are x-
ray crystallography and the exciton chirality method [2]. The latter circular di-
chroic methodology has been most prominently developed by Nakanishi [3] and
Harada [4] (see Chapter 5) over the past decade.
We have been developing methodology to elucidate the absolute configu-
ration of organic compounds using nuclear magnetic resonance (NMR) spectros-
TM

copy, generally considered the most important instrumental method in organic
chemistry laboratories. Our methodology, which we have called the ‘‘modified
Mosher’s method’’ [5], evolved from Mosher’s method [6] and uses Mosher’s
reagent, 1-methoxy-1-phenyl-1-trifluoromethylacetic acid (MTPA), as a chiral
derivatizing agent. A number of chemists have now applied this method to both
natural and synthetic products, establishing it as highly reliable and effective
for determining absolute configuration. This chapter introduces the concept and
application of the modified Mosher’s method and lists numerous compounds
whose absolute configurations were determined by this method.
Since our initial reports on the modified Mosher’s method, several new,
structurally similar chiral derivatizing reagents for NMR analysis have also been
reported as tools to elucidate absolute configuration. All of these reagents consist
of a chiral auxiliary with an aromatic group such as phenyl, 1- or 2-naphthyl
[7,8,] binaphthyl [9], 2-anthryl [10], or 9-anthryl systems, which cause shifts of
the protons by the anisotropic effect of the aromatic rings. Thus, for these com-
pounds, we have proposed the term chiral anisotropic reagents [11]. Most of the
chiral anisotropic reagents are applicable to organic compounds having a second-
ary alcohol moiety. Although most of these new chiral anisotropic reagents will
be summarized elsewhere, N,N-dimethylphenylglycinamide (phenylglycine di-
methylamide [PGDA]) and methyl phenylglycinate (phenylglycine methyl ester
[PGME]), developed for the absolute configuration of carboxylic acids, are de-
scribed in this chapter.
II. MODIFIED MOSHER’S METHOD

A. Basic Principles of Mosher’s Method
By careful examination of the chemical shift differences observed between the
diastereomeric pairs of (R)- and (S)-MTPA esters of chiral secondary alcohols,
Mosher proposed [6] that in solution, the carbinyl proton, ester carbonyl, and
trifluoromethyl groups of the MTPA moiety are oriented on the same plane (Fig.
1). PCILO calculations [12] and x-ray analysis [13] of the MTPA esters supported
Figure 1 Conformational model proposed by Mosher for MTPA esters. MTPA, 1-me-
thoxy-1-phenyl-1-trifluoromethylacetic acid. (Source : Ref. 6.)
TM

Modified
this concept in a qualitative sense. As Mosher pointed out, when the MTPA ester
exists in this conformation, resonances of ligand L 3 in the (S)-MTPA ester are
upfield in the 1 H NMR spectrum relative to those of the (R)-MTPA ester as a
result of the diamagnetic shielding effect of the benzene ring (Fig. 1). The reverse
sense of nonequivalence is true for the 1 H resonances of ligand L 2, with those
of the (R)-MTPA ester being upfield relative to those of the (S)-MTPA ester.
Therefore, comparison of the chemical shifts of the proton resonances of L 2 and
L 3 of the diastereomeric esters allows assignment of the absolute configuration
of the secondary alcohol.
When Mosher first put forward his analysis in the early 1970s however,
the NMR instruments available were mostly 60–100 MHz in proton frequency,
and the resolution and assignment of protons of complex organic molecules were
practically impossible with such limited field strengths (14.1–23.6 kG). Conse-
quently, the use of 19 F NMR [6,14] or a lanthanide-shift reagent to resolve the
easily identified OMe singlets of the MTPA-esters [15] was often preferred to the
original 1 H NMR method, which required identification of the usually overlapped
protons bonded to the original alcohol (on L 2 and L 3). Since Mosher’s conforma-
tional analysis did not address the relative shielding of these resonances, unam-
biguous assignment of absolute stereochemistry was difficult (if not impossible!),
and early applications of the MTPA ester method were primarily limited to deter-
mination of diastereomeric excess (enantiomeric excess of original alcohol) of
enriched synthetic mixtures.
The intrinsic drawback in these early applications of Mosher’s method for
absolute stereochemistry assignment was that it usually depended on only two
data points—the chemical shift difference of the two CF 3’s (19 F) or OMe’s (1 H)
of the (R)- and (S)-MTPA esters—and not on the more difficulty assigned reso-
nances of the substrate alcohol. The modified Mosher’s method, on the contrary,
uses the chemical shifts of many protons as Mosher originally intended and thus
is more reliable.
B. Basic Concept of the Modified Mosher’s Method

The basic concept of the modified Mosher’s method is essentially the same as
Mosher proposed: the idealized conformation depicted in Fig. 2a accounts for
the observed chemical shift differences in the MTPA esters of secondary alcohols.
(For convenience, this conformation will be called the ideal conformation, and
the plane defined by the carbonyl and the trifluoromethyl groups, and the carbinol
proton H α in this conformation will be called the MTPA plane.) Because of the
diamagnetic effect of the benzene ring, the NMR signals of H X ,Y,Z . . . of the (S)-
MTPA ester should appear upfield relative to those of the (R)-MTPA ester. The
reverse should hold true for H A ,B ,C. . . Therefore, when ∆δ ⫽ δ S ⫺ δ R , protons
on the right side of the MTPA plane must have positive values (∆δ ⬎ 0) and
TM

Figure 2 (a) MTPA plane of an MTPA ester; H A,B,C and H X,Y,Z are on the right and left
sides of the plane, respectively, when viewed down the plane from the MTPA subunit.
(b) Model A to determine the absolute configuration of secondary alcohols; Model A is
the view of the MTPA ester drawn in (a) from the direction specified by the outlined
arrow. MTPA, 1-methoxy-1-phenyl-1-trifluoromethylacetic acid. (Source: Ref. 5.)
protons on the left side of the plane should have negative values (∆δ ⬍ 0, Fig.
2B).
Mosher’s original method can be extended as follows: (1) assign as many
proton signals as possible of the (R)- and (S)-MTPA esters; (2) obtain ∆δ values
for the protons; (3) place the protons with positive ∆δ on the right side, and those
with negative ∆δ on the left side of Model A (Fig. 2b); (4) construct a molecular
model of the compound in question and confirm that all the assigned protons
with positive and negative ∆δ values are actually found on the right and left sides
of the MTPA plane, respectively. The absolute values of ∆δ should be propor-
tional to the distance from the MTPA moiety. When these conditions are all
satisfied, Model A will indicate the correct absolute configuration of the com-
pound.
This modified Mosher’s method allows examination of chemical shift dif-
ferences of as many protons as may be assigned by means of up-to-date NMR
techniques including two-dimensional (2D) spectra such as H,H-Correlation
Spectroscopy (H,H-COSY) and Homonuclear Hartmann-Hahn spectroscopy
(HOHAHA) spectra. We feel that as this method relies on a larger number of
TM

Modified
data points (i.e., all the protons assignable by 2D NMR techniques), it is far more
reliable than the previously mentioned variations of Mosher’s method using 19 F
NMR or a lanthanide shift reagent.
1. Preparation of 1-Methoxy-1-Phenyl-1-Trifluoromethylacetic
Acid Esters
The MTPA esters of a secondary alcohol are easily prepared [5] by treatment of
the alcohol (1) with (R)- or (S)-MTPA acid and 1-ethyl-3-(3-dimethylaminopro-
pyl)-carbodiimide (EDC) and 4-dimethylaminopyridine (DMAP) in dichloro-
methane, or (2) with (R)- or (S)-MTPA chloride and DMAP/Et 3N in dichloro-
methane or pyridine. Usually the reaction is completed within 15 h at room
temperature. In a few cases, the reaction rate of (R)- and (S)-MTPA acids or
acid chlorides with the chiral alcohol differs significantly (kinetic resolution),
and either diastereomer may be slow, or even difficult to obtain.
Note the following: because of the nomenclature priority rules, the chloride
of (R)-MTPA acid is (S)-MTPA chloride and the chloride of (S)-MTPA acid is
(R)-MTPA chloride. The stereochemical correlation of MTPA acids and their
chlorides is correctly shown in the scheme below.
When the amount of sample is limited, less than 1 mg of the sample per
esterification is usually sufficient for assignment of the absolute configuration
thanks to the high sensitivity of superconductive NMR instruments. One advan-
tage of MTPA esters is that the MTPA moiety possesses ultraviolet (UV) ab-
sorbing phenyl chromophore. Combinatory use of preparative thin layer chroma-
tography (TLC) (fluorescent under UV 254 nm) and HPLC successfully achieves
purification of the MTPA esters.
2. Choice of Nuclear Magnetic Resonance Solvents for

Recording the Spectra
It is advisable to use nonaromatic solvents such as deuterochloroform or deutero-
methanol for the modified Mosher’s method. Use of deuterobenzene (and proba-
bly other aromatic solvents such as deuteropyridine) often gives confusing results
[5]. Interaction between the aromatic solvent and the phenyl group of MTPA
may be the cause of this difficulty.
TM

3. Application to Natural Products or Their Simple Derivatives
Having Secondary Hydroxyl Groups
To confirm the validity of the present method, several compounds, including
cholesterol, ergosterol, (⫺)-borneol, and (⫺)-menthol (the absolute configuration
of which are known), were subjected to the method [5]. The assignments of the
proton chemical shifts (all in CDCl 3) were accomplished using 1D and 2D
(COSY, HOHAHA, NOESY [Nuclear Overhauser and Exchange Spectroscopy])
NMR techniques. As illustrated with (⫺)-menthol-(1), the ∆δ values (in parts per
million [ppm]) were all systematically arranged so that the positive and negative
nonequivalences were oriented on the right and left sides of the MTPA plane,
respectively (Fig. 3). The absolute configuration predicted by this method was
the same as known for the natural product. The results for the other three natural
products were all consistent with the known absolute configurations.
The modified Mosher’s method was subsequently applied [5] to several
marine natural products, including sanadaol (2) [16] hydroxyacetyldictyolal (fu-
kurinolal) (3) [17], cembranolide (4) [18], denticulatolide (5) [19], and crenula-
cetal B (6) [20], with unknown absolute configurations. These results are shown
in Fig. 4. Without exception, systematic arrangement of positive and negative
∆δ values was observed for these compounds, and their absolute configurations
were deduced as shown in the structures. For compounds 2 and 5, the absolute
configurations determined by the present method have been confirmed by total
synthesis [21] and X-ray crystallography [22], respectively.
The modified Mosher’s method has now become an indispensable means
to elucidate the absolute configuration of natural products, and additional exam-
ples using this method are continually being reported. In the following section
are examples of compounds whose absolute configuration was determined by the
modified Mosher’s method. As these examples demonstrate, the absolute config-
Figure 3 ∆δ Values (ppm) recorded for the MTPA esters of (⫺)-menthol (1). MTPA,
1-methoxy-1-phenyl-1-trifluoromethylacetic acid. (Source : Ref. 5.)
TM

Modified
Figure 4 ∆δ Values (ppm) recorded for the MTPA esters of marine terpenoids 2–6.
MTPA, 1-methoxy-1-phenyl-1-trifluoromethylacetic acid. (Source: Ref. 5.)
uration of compounds that are not secondary alcohols yet possess functionality
routinely converted into one with high stereoselectivity are also amenable to this
method. For example, reduction of the aragupetrosine A (59) ketone group to
the secondary alcohol with cyanoborohydride allowed absolute stereochemistry
assignment by the modified Mosher’s method [23].
4. Application to Natural Products Without a Secondary

Hydroxyl Group
When the compound in question has no secondary hydroxyl group, chemical
introduction of the hydroxyl group provides the handle to use the modified Mosh-
TM

er’s method for the absolute configuration assignment. As illustrated in the previ-
ous section (Fig. 5), simple deacylation (e.g., 8, 31, 34, 44, 52, and 55) or reduc-
tion of a ketone (e.g., 59) may expose a latent secondary alcohol for subsequent
formation of the MTPA esters. In addition, reduction of peroxides (e.g., 14, 15
and 47), epoxide or larger cyclic ether opening (e.g., 24, 41, and 63a,b), and
an ozonation/reduction sequence (e.g., 10) have also been utilized to obtain the
requisite secondary alcohol. Two less routine examples of such functionalizations
are presented in this section.
Figure 5 Examples of natural and synthetic secondary alcohols whose absolute config-
urations were determined by the modified Mosher’s method.
TM

Modified
TM

Figure 5 Continued
TM

Modified
TM

Figure 5 Continued
TM

Modified
TM

Figure 5 Continued
TM

Modified
Lobatriene (64) is a marine diterpene originally isolated from a soft coral

[71]. The relative configuration of the substituents on ring A had been determined,
but the stereochemical correlation of the substituents between rings A and B by
NMR spectroscopy was not possible because of rapid rotation along the single
bond connecting the rings. The absolute configurations of all the asymmetric
centers were, therefore, determined by chemical transformations with subsequent
application of the modified Mosher’s method, as shown in Fig. 6 [72]. Reductive
cleavage of the allylic ether bond afforded 65, which possesses a secondary alco-
hol at C-17. The R-configuration at this carbon was deduced by the results de-
tailed on the structure of the MTPA ester 65a. Ozonolysis of 65 gave methyl
ketone 66 (1 mg), which was subjected to a Baeyer-Villiger reaction that proceeds
with retention. Basic hydrolysis of the resulting acetate yielded the secondary
alcohol 67 (ca. 500 µg). This was divided into two portions and treated with (R)-
and (S)-MTPA chloride, respectively. The ∆δ values obtained are detailed on
Figure 6 Chemical transformation of lobatriene (64) into secondary alcohols 65 and

67, and the ∆δ values (ppm) recorded for their MTPA esters 65a and 67a, respectively.
TM

structure 67a. Beautifully systematic arrangement of positive and negative values
was observed, and the S-configuration of the hydroxy group was easily estab-
lished. Thus, the configuration of four asymmetric centers of 64 was determined
as shown in Fig. 6.
It should be noted that the sign of the ∆δ values of the methyl, ethyl, and
isopropyl groups on the A-ring is negative despite its apparent location on the
right side of the MTPA plane in 65a. One plausible explanation of this apparent
inversion of the sign is that the long-side chain may exist in a twisted conforma-
tion, so that the alkyl substituents are actually oriented on the left side of the
MTPA plane. Thus, application of the modified Mosher’s method to acyclic com-
pounds must be interpreted with caution when nonequivalence of protons quite
remote from the acylation site is observed.
Another nonroutine example is isoclavukerin A (68), isolated from an Oki-
nawan soft coral as a very volatile liquid [73]. The relative stereochemistry of
68 was deduced by comparison of the NMR data with those of clavukerin A, the
C-2 epimer of 68 [74]. Fortunately isoclavukerin A autooxidized during storage
in a refrigerator to give 69, which had a secondary hydroxyl group. Application of
the modified Mosher’s method led to the absolute configuration of the secondary
alcohol 69a, and thus of isoclavukerin A (Fig. 7). It is noteworthy that the benzo-
ate 69b showed no split circular dichroism (CD), although it is an allyl benzoate
[2,75]. Molecular models indicated that the dihedral angle between the ben-
zoyloxy and the olefinic groups was 180°, which does not in principle give a
Davydov-split Cotton effect [2,76]. Thus in such cases in which the exciton
Figure 7 Auto oxidation product 69 of isoclavukerin A (68) and ∆δ values (ppm) re-
corded for the MTPA esters of 69. MTPA, 1-methoxy-1-phenyl-1-trifluoromethylacetic
acid. (Source : Ref. 73.)
TM

Modified
chirality method is not applicable, the modified Mosher’s method may be suc-
cessfully applied.
5. Exceptions to the Rule of the Modified Mosher’s Method

and Countermeasures
a. Axial Alcohols. In our attempt to elucidate the absolute configuration of
the marine triterpene sipholenol-A (70) [77] using the modified Mosher’s method,
we noticed that ∆δ values of the protons on the A- and B-rings of the MTPA
derivatives were irregularly arranged, as seen in Fig. 8 [78]. These data, of course,
could not be used to determine the absolute configuration. However, this case
illustrates another advantage of the present method: a self-examination mecha-
nism based on systematic sense of nonequivalences to establish whether the ob-
served data can be used or must be abandoned.
Molecular models of 70 suggested that the OMTPA group exists in an
axiallike orientation and was sterically compressed by the two axial protons on
C-2 and C-7 (70′), which may compel the MTPA group to take a different confor-
mation from the ideal one. Therefore, 70 was converted into the C4-epimer 71,
in which the OH group exists in a sterically less crowded equatorial position, by
oxidation of 70 (PDC [pyridinium dichromate] in CH 2 Cl 2) followed by reduction
Figure 8 The ∆δ values (ppm) recorded for the MTPA esters of sipholenol-A (70).
TM

with sodium borohydride. (The Mitsunobu reaction did not work.) To our delight,
the positive and negative ∆δ values of 71 were found to be systematically ar-
ranged, and that enabled us to determine the absolute configurations of 71, and
thus of 70, as shown in the respective structures. The absolute configuration was
subsequently verified by x-ray crystallography on the (S)-MTPA ester of 70 [79].
In order to confirm that steric interference may prevent adoption of the
ideal conformation of the MTPA group in such cases with axial alcohols, and
thereby cause irregular arrangement of ∆δ values, friedelan-3β-ol [72, R ⫽ H]
was examined [5]. Because of the rigidity of the rings, the C3 hydroxyl group
of 72 is axially oriented. The ∆δ values of the MTPA derivative of 72 (R ⫽
MTPA) are, as anticipated, irregularly distributed, in contrast to the results ob-
tained for the C3 epimer 73 (Fig. 9), wherein the ∆δ values are completely in
accord with the rule.
These results suggest a general device to overcome the problem of the
present method: if the ∆δ values of the MTPA derivatives of a secondary alcohol
are irregularly arranged and cannot be used, epimerization of the hydroxyl group
and subsequent application of the MTPA method may succeed.
b. Anomalies Caused by the Anisotropy of Other π-Systems. As described, the
modified Mosher’s method utilizes the differences in the corresponding proton
chemical shifts between the diastereomers of (R)- and (S)-MTPA esters. If the
conformation of the MTPA group of one diastereomer greatly differs from that
of the other, as seen in the axial alcohol examples cited, the ∆δ values will not
be systematically arranged about the MTPA plane. The question also arises, What
Figure 9 The ∆δ values (ppm) recorded for the MTPA esters of friedelan-3β-ol (72)
with the axially oriented C3 alcohol, and those of the MTPA esters of its C3 epimer 73.
MTPA, 1-methoxy-1-phenyl-1-trifluoromethylacetic acid. (Source : Ref. 5.)
TM

Modified
will be the consequences if the conformations of other functional groups differ

in the diastereomers?
During synthetic studies on tautomycin, the modified Mosher’s method was
applied to intermediates 74a and 74b [80]. Ambiguous ∆δ values were observed
for one of the α-protons of these two derivatives (∆δ ⫽ ⫺0.04 and 0, Fig. 10);
however, other ∆δ values were systematically arranged [81]. To determine
whether this irregularity of ∆δ values was common in compounds bearing a β-
aromatic ring, the modified Mosher’s method was applied to synthetic β-aromatic
secondary alcohols, 75a–80a and 75b–80b.
All the protons on the left and right sides of the MTPA plane possessed
negative and positive ∆δ values, respectively, except these α-protons (H A and
Figure 10 Synthetic compounds 74a′ and 74b′ with irregular ∆δ values for one of the
α-protons (shown in boldface) and other model compounds 75–80. (Source : Ref. 81.)
TM

Figure 11 Conformations of 74a′–80a′ and 74b′ and 80b′ deduced from NMR studies.
NMR, nuclear magnetic resonance. (Source: Ref. 81.)
H B). Although these methylene protons should have positive ∆δ values, in many
compounds they were negative or zero. The conformations of these esters de-
duced from their NMR properties (Fig. 11) were supported by molecular model-
ing studies of 74b′ (Fig. 12). Since the α-protons are located near the β-aromatic
ring, these were anisotropically affected by both this β-aromatic ring as well as
the benzene ring of the MTPA moiety. Even though a difference of the dihedral
angles (H c-C-C-C1) in the predicted most stable conformations for 74b′ was not
observed between the (S)- and (R)-MTPA esters (352° and 351°, respectively)
by molecular mechanics calculation, subtle differences in the conformation
around the β-carbon between (S)- and (R)-MTPA esters might cause these irregu-
larities of the ∆δ values.
A similar anomaly was observed in tanabalin [81], a natural product iso-
lated from the dried flowers of the Brazilian medicinal plant Tanacetum bal-
samita, which has a secondary alcohol adjacent to a furan ring [82]. One of the
C-11 methylene protons showed an anomalous positive ∆δ value thought to be
caused by the furan ring anisotropy. The anomaly was ‘‘verified’’ by using model
Figure 12 Stable conformations predicted for 74b′ by molecular modeling. (Source:

Ref. 81.)
TM

Modified
Figure 13 The ∆δ values (ppm) obtained for the MTPA esters of tanabalin 81a and
model compound 82a. MTPA, 1-methoxy-1-phenyl-1-trifluoromethylacetic acid. (Source:
Ref. 82.)
compound 82, which also showed the irregular ∆δ value of the α-proton denoted
by a thick arrow (Fig. 13).
Minale et al. [40] encountered similar anomalies in the structure determina-
tion of superstolide A that can be interpreted in the same manner: an irregular
∆δ value (denoted by the arrow) in both 28a and 28b (Fig. 14). This anomaly
was apparently caused by the N-acetyl group anisotropy, the conformation of
which may be slightly different between the (R)- and (S)-MTPA diastereomers,
affecting the chemical shifts of the protons close to the carbonyl group. Since
other ∆δ values were systematically arranged, their absolute configurations were
safely assigned.
In conclusion, the modified Mosher’s method can be applied to secondary
alcohols possessing other π-systems, such as the aromatic rings and carbonyl
groups that can fully or partially rotate along a single bond, in the vicinity of
the hydroxy group. In such cases, however, one must be aware of the possible
impact of this second anisotropy in generating irregular ∆δ values.
6. Applications in the Enantioselective Synthesis of

Natural Products
Biologically active natural products are primary targets of synthetic chemists.
Most natural products exist as single enantiomers, and thus enantioselective syn-
thesis is particularly important to organic chemists. When a chemical reaction
yields only one enantiomer, the absolute configuration of the product can often
TM

Figure 14 The ∆δ values (ppm) reported by Minale for the MTPA esters 28a and 28b.
MTPA, 1-methoxy-1-phenyl-1-trifluoromethylacetic acid. (Source : Ref. 40.)
be determined by using the modified Mosher’s method. Products obtained by

‘‘enantioselective’’ reactions, however, are commonly accompanied by minor
amounts of the second enantiomer. In such cases, the absolute configuration of
the major and minor products can be assigned by preparing only (S)- [or (R)-]
MTPA esters of both enantiomers [83]. Fig. 15 shows why this is possible. [Bear
in mind that the NMR spectrum of the (S)-MTPA ester of one product is identical
with that of the (R)-MTPA ester of its enantiomer.]
Esterification of the hypothetical product mixture consisting of the major
enantiomer (R′) and its minor antipode (S′) with (S)-MTPA acid will give the
S ⫺ R′ (major diastereomer) and S ⫺ S′ (minor diastereomer) esters, which are
separated by chromatography. Because the 1 H NMR spectrum of S ⫺ S′ is identi-
cal with that of its enantiomer R ⫺ R′, subtraction of the chemical shifts of S ⫺
R′ from those of S ⫺ S′ (⫽ R ⫺ R′) will give ∆δ values for assigning the absolute
configuration (R′) of the major product: [∆δ ⫽ δ (S)-MTPA ester ⫺ δ (R)-MTPA ester ⫽
δ S⫺R′ ⫺ δ R⫺R′ (⫽ δ S⫺S′)].
TM

Modified
Figure 15 Strategy for the determination of the absolute configuration of synthetic prod-
ucts obtained from enantioselective synthesis using the modified Mosher’s method.
In the total synthesis of (⫺)-dendrobine, Mori successfully used this idea

[28]. (⫹)-Carveol [83] was converted to benzylamine 84, whose enantiomeric
purity (ee) was found to be 90%, although the absolute configuration of the major
enantiomer was unknown since either an S N2 or S N2′ mechanism could dominate
in the Mitsunobu reaction (Fig. 16). The enantiomeric mixture of 84 and 84′
was transformed into allyl amine 85, and then into a tricyclic ketone, sodium
borohydride reduction of which yielded 86 (plus its antipode). Conversion into
the (S)-MTPA esters gave the two diastereomers, and ∆δ values (shown in 11)
were calculated by subtraction of the chemical shifts (1H) of the major one from
those of the minor. Because the (S)-MTPA esters were used, the absolute config-
uration of the major enantiomer of 86 was determined as shown in the structure.
A similar application of the modified Mosher’s method has recently been reported
by Fujimoto [38].
7. Applications to Primary Amines Bonded to Methine Carbons

The modified Mosher’s method is also applicable to the amines attached to ter-
tiary carbons (methines). The MTPA amides may exist in the same conformation
TM

Figure 16 Synthetic pathway to the (⫺)-dendrobine intermediate 86, and ∆δ values
(ppm) recorded for a mixture of the (S)-MTPA esters of 86 (ester 11) and its enantiomer.
(S)-MTPA, (S)-1-methoxy-1-phenyl-1-trifluoroacetic acid. (Source: Ref. 28.)
as the MTPA esters; therefore, ∆δ ⫽ δ S ⫺ δ R can lead to the absolute configura-

tion of the amines. Some examples using amino acid esters and amino alcohol
derivatives are shown in Fig. 17 [84].
Other examples of the assignment of the absolute stereochemistry of amine-
containing natural and synthetic compounds [98–100] using the modified Mosh-
er’s method are listed in Fig. 18 [85–87]. The modified Mosher’s method was
also successful in assigning the absolute configuration of the primary amine-
containing subunit of fumonisin B 1 (32), as illustrated in Fig. 5 [44]. As these
TM

Modified
Figure 17 The ∆δ values (ppm) recorded for amino compounds 87–97. Proton assign-
ments that could be interchanged lead to two possible values, as indicated by the second
value in parentheses. Values for the methoxyl group of the MTPA subunit (all positive)
are also given. MTPA, 1-methoxy-1-phenyl-1-trifluoroacetic acid. (Source: Ref. 84.)
TM

Figure 18 Examples of primary amines whose absolute configurations were determined
by the modified Mosher’s method.
examples demonstrate, a large number of research groups have now found this
procedure to be tremendously useful for the assignment of absolute stereochemis-
try for primary amines as well as secondary alcohols.
III. PHENYLGLYCINE DIMETHYLAMIDE AND

PHENYLGLYCINE METHYL ESTER, CHIRAL
ANISOTROPIC REAGENTS FOR THE ABSOLUTE
CONFIGURATION DETERMINATION OF
CARBOXYLIC ACIDS
A. Molecular Design for the Chiral Anisotropic Reagents
Applications of the modified Mosher’s method and other NMR reagents (chiral
anisotropic reagents) are primarily restricted in their usage to secondary alcohols,
and more recently to primary amines. By extending Mosher’s concept that com-
parison of the chemical shifts of diastereomeric derivatives of a chiral compound
obtained from (R)- and (S)-derivatizing agents can lead to absolute configuration
TM

Modified
assignment, new chiral anisotropic reagents may be developed for the compounds
possessing functional groups other than a secondary alcohol. In this context, we
have elaborated new chiral anisotropic reagents, which allow determination of
the absolute configuration of carboxylic acids, in which the carboxylic group is
attached to a methine carbon [88].
The general molecular design for such chiral anisotropic reagents is shown
in Fig. 19. A chiral carboxylic acid of type I is converted to acyl derivative II,
where X can be any heteroatom. In order for the phenyl ring to cast its diamag-
netic field effectively, and selectively, onto R 1 or R 2, coplanarity of C1 to Y5 in
the dominant conformation is necessary. When X is NH, coplanarity from C1 to
C4 may be achieved because of the planar structure as well as the preferred s-
trans conformation of the amide group. If Y is also a polar group that would
interact repulsively with the polar carbonyl group at C2 in a syn relationship,
the C2 through Y5 orientation about the X3-C4 bond may also adopt a trans
conformation. We chose the dimethylaminocarbonyl and methoxycarbonyl
groups for Y, the former being preferable because of the enhanced dipole caused
by the greater participation of the lone-pair electrons of the nitrogen atom with
carbonyl resonance. Furthermore, hydrogen bonding between N 3H and C 5 ⫽ O,
Figure 19 Rationale to elucidate the absolute configuration of carboxylic acids by using

PGDA and PGME. PGDA, phenylglycine dimethylamide; PGME, phenylglycine methyl
ester. (Source : Ref. 88.)
TM

which will reinforce conformation IV, can be expected. It is equally important
that the proton signals of the Z group (NMe 2 and OMe in these examples) do
not obscure other 1 H NMR signals. The chiral anisotropic reagents thus designed
are phenylglycine dimethylamide (PGDA) and phenylglycine methyl ester
(PGME).
If PGDA and PGME derivatives of a chiral carboxylic acid predominantly
populate conformation IV, H X,Y,Z would be more shielded by the phenyl group
of the (S)-PGDA (or PGME) moiety than those of the (R)-derivative. The reverse
will be true for H A,B,C, which would be more shielded in the (R)-derivative. Like
the modified Mosher’s method for secondary alcohols and amines, Model B, in
which ∆δ ⫽ δ S ⫺ δ R , will show the correct absolute configuration of the carbox-
ylic acid.
B. Preparation of Phenylglycine Dimethylamide and

Phenylglycine Methyl Ester
Although both enantiomers of PGME are commercially available, (R)-(⫺)-phe-
nylglycine was used as starting material for both PGME and PGDA. Enantiomer-
ically pure PGME {(R)-PGME hydrochloride; [α] D ⫺135°, (S)-PGME hydro-
chloride [α] D ⫹ 135°} was obtained by treating phenylglycine with thionyl
chloride in methanol. The preparation of (R)-(⫺)-PGDA was also routine: N-
Boc-phenylglycine was condensed (PyBOP [benzotriazol-1-yloxytris(dimethyl-
amino)phosphonium hexafluorophosphate], HOBT [1-hydroxybenzotriazole])
with dimethylamine to give the dimethylamide. The protecting group was re-
moved under acidic condition (4 M HCl) to give PGDA hydrochloride {(R)-
PGDA hydrochloride; [α] D ⫺ 111°}. Partial epimerization of the phenylglycinate
was observed in several attempts to prepare PGDA hydrochloride, so caution
must be exercised to use only enantiomerically pure material for assignment of
absolute stereochemistry, and especially for the determination of e.e. (enantio-
meric excess).
C. Applications of Phenylglycine Dimethylamide and

Phenylglycine Methyl Ester to the Determination of
Carboxylic Acid Absolute Stereochemistry
Prior to applying PGDA and PGME to carboxylic acids of unknown absolute
stereochemistry, it was important to clarify whether derivatized amides actually
had the expected conformation IV (Fig. 19). To this end, PGDA was condensed
(PyBOP, HOBT) with 4-methylcyclohexanecarboxylic acid (101) to give crystal-
line 101a. An x-ray structure of 101a revealed that the diamide moiety actually
exists in the expected conformation as shown in IV: C1 to Y-5 (NMe 2 in 101a)
TM

Modified
Figure 20 The ∆δ values (ppm) of PGDA derivatives of 4-methylcyclohexanecarbox-

ylic acid 101a and (S)-2-methylbutanoic acid 102a. PGDA, phenylglycine dimethylamide.
(Source: Ref. 88.)
were nearly coplanar, and, because the phenyl plane faces the methylcyclohexane
residue, the anisotropy of the phenyl group was effectively cast over the methyl-
cyclohexyl protons. The ∆δ values for this amide were calculated (Fig. 20): they
are positive and negative on the right and left sides of the PGDA plane, respec-
tively, as shown in 101a. Analogous results were also obtained with the PGDA
derivative of (S)-2-methylbutanoic acid 102a. The absolute configuration de-
Figure 21 The ∆δ values (ppm) and NOEs supporting the proposed conformation of
the PGME derivative of 4-methylcyclohexanecarboxylic acid 101c and the ∆δ values of
the PGME derivative of (S)-2-methylbutanoic acid 102c. NOE [nuclear Overhauser ef-
fect], PGME, phenylglycine methyl ester. (Source : Ref. 88.)
TM

duced from Model B (Fig. 19), the conformation in agreement with that predicted
by molecular mechanics calculations, was identical to the known absolute stereo-
chemistry.
With the promising results using PGDA as a chiral anisotropic reagent for
the assignment of absolute configurations of chiral carboxylic acids obtained from
the model studies discussed, PGME derivatives were subsequently examined for
comparison. Condensation of (R)-(⫺)-PGME with 4-methylcyclohexanecarbox-
ylic acid (101) gave amide 101c as an oil. Various NOE (nuclear Overhauser
effect) experiments on this amide, particularly the NOE, which crossed the amide
bond, suggest the conformation as shown (Fig. 21). The ∆δ values calculated for
101c are shown on the structure. Values having opposite signs are nicely arranged
on the right and left sides of the PGME plane, respectively, in accord with the
proposed conformation.
To complete the comparison with PGDA, (R)- and (S)-PGME were con-
densed (PyBOP, HOBT) with (S)-2-methylbutanoic acid (102) and the ∆δ values
recorded for the PGME derivative 102c (Fig. 21). The (S)-configuration assigned
using Model B was identical to the known absolute configuration as also deter-
mined by using PGDA (Fig. 20, 102a). Although the ∆δ values for the PGME
Figure 22 The ∆δ values (ppm) recorded for PGME derivatives 103–106. PGME, phe-
nylglycine methyl ester. (Source: Ref. 88.)
TM

Modified
amide 102c are smaller than those for the PGDA amide 102a as expected, they
are still of practical magnitude for absolute stereochemistry assignment. Consid-
ering the ease of the preparation, the authors prefer PGME to PGDA, and thus
for further derivatizations, PGME was subsequently applied to several carboxylic
acids with known absolute configurations [88]. The results are summarized in
103–106 (Fig. 22).
In all cases, the protons of ∆δ ⬎ 0 and ∆δ ⬍ 0 are on the right and left
sides of the PGME plane, respectively, without exception, confirming the validity
of the method. The absolute configurations predicted by the present method are
identical with the known ones. It should be noted that PGME was applicable to
rather complex compounds such as gibberellic acid (106). It was also found that
the magnitude of the absolute values of ∆δ was proportional to the distance from
the PGME moiety.
The present method should be appropriate for absolute stereochemical as-
signment of natural products possessing a carboxyl group of type (I) as well as for
primary alcohols 107 that can be routinely oxidized into carboxylic acid 108 (Eq. 1).
IV. CONCLUSIONS
The number of applications appearing in the literature of the modified Mosher’s

method to assign absolute stereochemistry is increasing dramatically, demonstrat-
ing the usefulness and reliability of the method. In principle, Professor Mosher’s
original idea of using the magnetic anisotropy of an aromatic ring on a chiral
reagent can be extended to a variety of functional groups. The use of PGME
and PGDA derivatives is an example that shows that intense organic chemical
considerations directed toward new chemical methodology for absolute stereo-
chemistry determination can result in new chiral anisotropic reagents applicable
to particular functional groups.
REFERENCES
1. (a) D. Parker, Chem. Rev., 91: 1441 (1991). (b) S. Ahuja, in Chiral Separations,
Applications and Technology (S. Ahuja, ed.), American Chemical Society, Washing-
ton, D. C., chapters 1 and 6 (1997).
TM

2. N. Harada and K. Nakanishi, Circular Dichroic Spectroscopy: Exciton Coupling in
Organic Stereochemistry, University Science Books, Mill Valley, Calif., and Oxford
University Press, Oxford (1983).
3. (a) K. Nakanishi and N. Berova, in Circular Dichroism, Principles and Applications
(K. Nakanishi, N. Berova, and R. W. Woody, eds.), VCH Publishers, New York,
chapter 13 (1994). (b) J.-G. Dong, A. Wada, T. Takakuwa, K. Nakanishi, and N.
Berova, J. Am. Chem. Soc., 119: 12024 (1997).
4. N. Harada, in Circular Dichroism, Principles and Applications, (K. Nakanishi, N.
Berova, and R. W. Woody, eds), VCH Publishers, New York, chapter 12 (1994).
5. I. Ohtani, T. Kusumi, Y. Kashman, and H. Kakisawa, J. Am. Chem. Soc., 113: 4092
(1991).
6. (a) J. A. Dale and H. S. Mosher, J. Am. Chem. Soc., 95: 512 (1973). (b) G. R.
Sullivan, J. A. Dale, and H. S. Mosher, J. Org. Chem., 38: 2143 (1973).
7. T. Kusumi, H. Takahashi, P. Xu, T. Fukushima, Y. Asakawa, T. Hashimoto, Y. Kan,
and Y. Inouye, Tetrahedron Lett., 35: 4397 (1994).
8. (a) J. M. Seco, S. K. Latypov, E. Quinoa, and R. Riguera, Tetrahedron Lett., 35:
2921 (1994). (b) S. K. Latypov, J. M. Seco, E. Quinoa, and R. Riguera, J. Org.
Chem., 60: 504 (1995).
9. Y. Fukushi, C. Yajima, and J. Mizutani, Tetrahedron Lett., 35: 599 (1994).
10. K. Kouda, T. Kusumi, X. Ping, Y. Kan, T. Hashimoto, and Y. Asakawa, Tetrahedron
Lett., 37: 4541 (1996).
11. T. Kusumi, H. Takahashi, T. Hashimoto, Y. Kan, and Y. Asakawa, Chem. Lett.,
1093 (1994).
12. E. M. Merckx, L. Vanhoeck, J. A. Lepoivre, F. C. Alderweireldt, B. J. Van Der
Veken, J. P. Tollenaere, and L. A. Raymaekers, Spectros Int. J., 2: 30 (1983).
13. H. M. Doesburg, G. H. Petit, and E. M. Merckx, Acta Crystallogr., B38: 1181
(1983).
14. (a) P. L. Rinaldi, Prog. NMR Spec., 15: 291 (1982). (b) S. Yamaguchi, in Asymmetric
Synthesis (J. D. Morrison, ed.), Academic Press, New York, Vol. 1, chapter 7 (1983).
15. (a) S. Yamaguchi, F. Yasuhara, and K. Kabuto, Tetrahedron, 32: 1363 (1976). (b)
F. Yasuhara and S. Yamaguchi, Tetrahedron Lett., 4085 (1977).
16. M. Ishitsuka, T. Kusumi, and H. Kakisawa, Tetrahedron Lett., 23: 3179 (1982).
17. (a) N. Enoki, R. Ishida, and T. Matsumoto, Chem. Lett., 1749 (1982). (b) M. Ochi,
N. Masui, H. Kotsuki, I. Miura, and T. Tokoroyama, Chem. Lett., 1927 (1982).
18. Y. Yamada, S. Suzuki, K. Iguchi, H. Kikuchi, Y. Tsukitani, and H. Horiai, Chem.
Pharm. Bull., 28: 2035 (1980).
19. Y. Uchio, S. Eguchi, J. Kuramoto, M. Nakayama, and H. Hase, Tetrahedron Lett.,
26: 4487 (1985).
20. T. Kusumi, D. M. Nkongolo, M. Goya, M. Ishitsuka, T. Iwashita, and H. Kakisawa,
J. Org. Chem., 51: 384 (1986).
21. H. Nagaoka, K. Kobayashi, and Y. Yamada, Tetrahedron Lett., 29: 5945 (1988).
22. T. Kusumi, I. Ohtani, Y. Inouye, and H. Kakisawa, Tetrahedron Lett., 29: 4731
(1988).
23. M. Kobayashi, K. Kawazoe, and I. Kitagawa, Tetrahedron Lett., 30: 4149 (1989).
24. I. Kitagawa, M. Kobayashi, T. Katori, M. Yamashita, J. Tanaka, M. Doi, and T.
Ishida, J. Am. Chem. Soc., 112: 3710 (1990).
TM

Modified
25. H. Itokawa, E. Kishi, H. Morita, K. Takeya, and Y. Iitaka, Tetrahedron Lett., 32:
1803 (1991).
26. M. Natsume, K. Yasui, S. Kondo, and S. Marumo, Tetrahedron Lett., 32: 3087
(1991).
27. M. Nishizawa, M. Emura, Y. Kan, H. Yamada, K. Ogawa, and N. Hamanaka, Tetra-
hedron Lett., 21: 2983 (1992).
28. M. Mori, F. Saitoh, N. Uesaka, and M. Shibasaki, Chem. Lett., 213 (1993).
29. I. Kubo, and T. Kusumi, unpublished results.
30. J. Tanaka, T. Higa, K. Suwanborirux, V. Kokpol, G. Bernardinelli, and C. W. Jef-
ford, J. Org. Chem., 58: 2999 (1993).
31. K. Shin-ya, M. Tanaka, K. Furihata, Y. Hayakawa, and H. Seto, Tetrahedron Lett.,
34: 4943 (1993).
32. K. Iguchi, M. Fujita, H. Nagaoka, H. Mitome, and Y. Yamada, Tetrahedron Lett.,
34: 6277 (1993).
33. M. Ochi, S. Ariki, A. Tatsukawa, H. Kotsuki, Y. Fukuyama, and K. Shibata, Chem.
Lett., 89 (1994).
34. Y. Inouye, Y. Sugo, T. Kusumi, and N. Fusetani, Chem. Lett., 419 (1994).
35. M. Kobayashi, S. Aoki, and I. Kitagawa, Tetrahedron Lett., 35: 1243 (1994).
36. M. Sugano, A. Sato, Y. Iijima, K. Furuya, H. Haruyama, K. Yoda, and T. Hata, J.
Org. Chem., 59: 564 (1994).
37. G. Trimurtulu, I. Ohtani, G. M. L. Patterson, R. E. Moore, T. H. Corbett, F. A.
Valeriote, and L. Demchik, J. Am. Chem. Soc., 116: 4729 (1994).
38. H. Shimada, S. Nishida, S. Singh, M. Sahai, and Y. Fujimoto, Tetrahedron Lett.,
35: 3961 (1994).
39. H. Yada, H. Sato, S. Kaneko, and A. Ichihara, Tetrahedron Lett., 35: 4393 (1994).
40. M. V. D’Auria, C. Debitus, L. G. Paloma, L. Minale, and A. Zampella, J. Am. Chem.
Soc., 116: 6658 (1994).
41. Z. Gu, X. Fang, L. Zeng, and J. L. McLaughlin, Tetrahedron Lett., 35: 5367 (1994).
42. T. Hashimoto, M. Morie, M. Toyota, Z. Taira, R. Takeda, M. Tori, and Y. Asakawa,
Tetrahedron Lett., 35: 5457 (1994).
43. S. Paik, S. Carmeli, J. Cullingham, R. E. Moore, G. M. L. Patterson, and M. A.
Tius, J. Am. Chem. Soc., 116: 8116 (1994).
44. T. R. Hoye, J. I. Jiménez, and W. T. Shier, J. Am. Chem. Soc., 116: 9409 (1994).
45. M. L. Ciavatta, E. Trivellone, G. Cimino, and M. J. Uriz, Tetrahedron Lett., 35:
7871 (1994).
46. K. Stratmann, D. L. Burgoyne, R. E. Moore, G. M. L. Patterson, and C. D. Smith,
J. Org. Chem., 59: 7219 (1994).
47. G. J. Hopper and M. T. Davies-Coleman, Tetrahedron Lett., 36: 3265 (1995).
48. F. Kong and R. J. Andersen, Tetrahedron, 51: 2895 (1995).
49. S. Fukuzawa, S. Matsunaga, and N. Fusetani, J. Org. Chem., 60: 608 (1995).
50. W. Y. Yoshida, P. J. Bryan, B. J. Baker, and J. B. McClintock, J. Org. Chem., 60:
780 (1995).
51. Y. Matsuo, M. Suzuki, and M. Masuda, Chem. Lett., 1043 (1995).
52. J. Tanaka, T. Higa, G. Bernardinelli, and C. W. Jefford, Chem. Lett., 255 (1996).
53. G. Shi, D. Alfonso, M. O. Fatope, L. Zeng, Z. Gu, G. Zhao, K. He, J. M. MacDougal,
and J. L. McLaughlin, J. Am. Chem. Soc., 117: 10409 (1995).
TM

54. Y. Jiao, T. Yoshihara, S. Ishikuri, H. Uchino, and A. Ichihara, Tetrahedron Lett.,
37: 1039 (1996).
55. J. Kobayashi, K. Yuasa, T. Kobayashi, T. Sasaki, and M. Tsuda, Tetrahedron, 52:
5745 (1996).
56. J. Shin, Y. Seo, J.-R. Rho, E. Baek, B.-M. Kwon, T.-S. Feong, and S.-H. Bok, J.
Org. Chem., 60: 7582 (1995).
57. T. Ichiba, P. J. Scheuer, and M. Kelly-Borges, Tetrahedron, 51: 12195 (1995).
58. C. M. Cerda-Garcia-Rojas and D. J. Faulkner, Tetrahedron, 51: 1087 (1995).
59. T. Shibata, S. Kurihara, K. Yoda, and H. Haruyama, Tetrahedron, 51: 11999 (1995).
60. L. Zeng, Z. Gu, X. Fang, P. E. Fanwick, C. Chang, D. L. Smith, and J. L. McLaugh-
lin, Tetrahedron, 51: 2477 (1995).
61. E. Palagiano, S. De Marino, L. Minale, R. Riccio, F. Zollo, M. Iorizzi, J. B. Carré,
C. Debitus, J. Provost, and L. Lucarain, Tetrahedron, 51: 3675 (1995).
62. R. M. Rzasa, D. Romo, D. J. Stirling, J. W. Blunt, and M. H. G. Munro, Tetrahedron
Lett., 36: 5307 (1995).
63. M. Ojika, T. Nagoya, and K. Yamada, Tetrahedron Lett., 36: 7491 (1995).
64. A. F. Barrero, E. Alvarez-Manzaneda, and A. Lara, Tetrahedron Lett., 36: 6347
(1995).
65. P. A. Searle, R. K. Richter, and T. F. Molinski, J. Org. Chem., 61: 4073 (1996).
66. L. Zeng, F.-E. Wu, Z. Gu, and J. L. McLaughlin, Tetrahedron Lett., 36: 5291 (1995).
67. E. W. Schmidt and D. J. Faulkner, Tetrahedron Lett., 37: 3951 (1996).
68. N. Sitachitta, M. Gadepalli, and B. S. Davidson, Tetrahedron, 52: 8073 (1996).
69. Y. Guo, A. Madaio, E. Trivellone, G. Scognamiglio, and G. Cimino, Tetrahedron,
52: 8341 (1996).
70. J.-L. Abad, J. Casas, F. Sánchez-Baeza, and A. Messeguer, J. Org. Chem., 60: 3648
(1995).
71. R. W. Dunlop and R. J. Wells, Aust. J. Chem., 32: 1345 (1979).
72. T. Kusumi, T. Hamada, M. O. Ishitsuka, I. Ohtani, and H. Kakisawa, J. Org. Chem.,
57: 1033 (1992).
73. T. Kusumi, T. Hamada, M. O. Ishitsuka, H. Ginda, and H. Kakisawa, Tetrahedron
Lett., 33: 2019 (1992).
74. M. Kobayashi, B. W. Son, M. Kido, Y. Kyogoku, and I. Kitagawa, Chem. Pharm.
Bull., 31: 2160 (1983).
75. N. Gonnella, K. Nakanishi, V. S. Martin, and K. B. Sharpless, J. Am. Chem. Soc.,
104: 3775 (1982).
76. N. Harada, S. L. Chen, and K. Nakanishi, J. Am. Chem. Soc., 97: 5345 (1975).
77. (a) U. Shmueli, S. Carmely, A. Groweiss, and Y. Kashman, Tetrahedron Lett., 22:
709 (1981). (b) S. Carmely and Y. Kashman, J. Org. Chem., 48: 3517 (1983).
78. I. Ohtani, T. Kusumi, Y. Kashman, and H. Kakisawa, J. Org. Chem., 56: 1296
(1991).
79. Y. Inouye, I. Ohtani, T. Kusumi, Y. Kashman, and H. Kakisawa, Chem. Lett., 2073
(1990).
80. A. Naganawa, Y. Ichikawa, and M. Isobe, Tetrahedron, 50: 8969 (1994).
81. I. I. Ohtani, K. Hotta, Y. Ichikawa, and M. Isobe, Chem. Lett., 513 (1995).
82. I. Kubo, V. Jamalamadaka, T. Kamikawa, K. Takahashi, K. Tabata, and T. Kusumi,
Chem. Lett., 441 (1996).
TM

Modified
83. T. Kusumi, J. Syn. Org. Chem. Jpn., 51: 462 (1993).

84. T. Kusumi, T. Fukushima, I. Ohtani, and H. Kakisawa, Tetrahderon Lett., 32: 2939
(1991).
85. Y. Ichikawa, K. Tsuboi, and M. Isobe, J. Chem. Soc. Perkin Trans. 1, 2791 (1994).
86. T. Ohtsuka, Y. Itezono, N. Nakayama, A. Sakai, N. Shimma, M. Yanagisawa, K.
Yokose, and H. Seto, Symposium Paper of the 35th Symposium on the Chemistry
of Natural Products, Kyoto, Japan, pp. 306–313 (1993).
87. K. Iida, S. Fukuda, T. Tanaka, M. Hirama, S. Imajo, M. Ishiguro, K. Yoshida, and
T. Otani, Tetrahedron Lett., 3: 4997 (1996).
88. Y. Nagai and T. Kusumi, Tetrahedron Lett., 36: 1853 (1995).
TM

6
Circular Dichroism Spectroscopy
and the Absolute Stereochemistry
of Biologically Active Compounds
Nobuyuki Harada
Tohoku University, Sendai, Japan
I. INTRODUCTION
The theoretical calculation of circular dichroism (CD) spectra by the π-electron

self-consistent field/configuration interaction/dipole velocity molecular orbital
(SCF-CI-DV MO) method has become an important tool for the study of the
absolute configuration of a variety of twisted π-electron systems including exci-
ton-coupled systems [1–4]. We have assigned the absolute stereochemistry of
(⫹)-1,8a-dihydro-3,8-dimethylazulene [(⫹)-1] isolated from a liverwort, a labile
intermediate on the biosynthetic pathway to 1,4-dimethylazulene (2), as (8aS) by
the application of this method to the calculation of the theoretical CD spectrum
of the twisted tetraene system for comparison with that recorded for the natural
product (Section III, Chart 1) [5,6]. In this case, we also succeeded in the experi-
mental verification of the absolute configuration theoretically determined, by
comparison of the CD spectrum of the natural product with those of synthetic
chiral model compounds.
We have also established the power of the π-electron SCF-CI-DV MO
method for nonempirical determination of the absolute configuration of the more
complicated, novel marine natural products of the halenaquinol family 18–27
with a new pentacyclic skeleton, isolated from tropical marine sponges (Section
IV, Charts 2 and 3) [7,8]. In addition, we have confirmed that the absolute stereo-
chemistry of the halenaquinol family, theoretically predicted, is correct by the
first total synthesis of the chiral halenaquinols and halenaquinones [9–12]. The
TM

absolute stereochemistry of several related natural products was subsequently
determined by the total synthesis of their natural enantiomers [12].
In addition to these natural products with twisted π-electron systems, the
π-electron SCF-CI-DV MO method was successfully applied to the determination
of another unusual class of natural products, the biflavones. Optically active
biflavones are unique in the sense that they are chiral molecules devoid of ste-
reogenic centers. One such chiral biflavone, (⫺)-4′,4⵮,7,7″-tetra-O-methylcu-
pressuflavone [(⫺)-61], was isolated from the tropical plant Garcinia man-
gostana (Section V, Chart 4) [13]. This biflavone is composed of two flavone
monomers linked at the 8- and 8′-positions; the monomeric subunits are almost
perpendicular to each other with hindered rotation about the 8/8′-bond due to
steric hindrance. Such molecules can exist as optically active compounds and
belong to the class of chiral molecules known as atropisomers. Although the
structure of 61 was reported in 1968 [14], the absolute stereochemistry remained
undetermined for more than 20 years. Finally in 1992, we reported the determina-
tion of the absolute stereochemistry of the natural atropisomer as (aR) by theoreti-
cal calculation of the CD spectrum [13].
There are only two nonempirical methods for determining the absolute con-
figuration of chiral organic compounds. One is the x-ray crystallographic Bijvoet
method using the anomalous dispersion effect of heavy atoms within the mole-
cule. The other is the theoretical CD method described in this chapter, which
includes applications to the CD exciton chirality method. Although these two
methods are based on totally different physical phenomena, they necessarily must
give the same assignment of absolute configuration. In 1995, however, it was
claimed that the absolute configuration of (⫺)-61 determined by x-ray crystallog-
raphy disagreed with that obtained by the theoretical CD method [15]. This was
a very interesting incident; however, the claim based on x-ray was later retracted
[16,17]. Since then, we have succeeded in the total synthesis of (⫺)-61, confirm-
ing that the absolute configuration determined by the theoretical CD method was
correct [18,19].
In this chapter, the theoretical calculation of CD spectra by the π-electron
SCF-CI-DV MO method and its applications to the determination of the absolute
stereochemistry of the biologically active compounds mentioned are described.
The method has also been successfully applied to various chiral synthetic com-
pounds with twisted π-electron systems for determining their absolute configura-
tions [20–27].
II. COMPUTATIONAL METHODS

A. Molecular Structure
To calculate the CD spectra of chiral molecules with twisted π-electron systems,
the molecular geometry and atomic coordinates of the most stable conformation
TM

are needed. The molecular geometries of chiral compounds with π-electron sys-
tems were optimized by using the molecular mechanics (MMP2) and/or MOPAC
93, AM1 programs [28,29]. The atomic coordinates obtained were then used for
the next step: calculation of CD and ultraviolet (UV) spectra.
B. ␲-Electron Self-Consistent Field/Configuration

Interaction/Dipole Velocity Molecular Orbital Method:
Numerical Calculation of Circular Dichroism and
Ultraviolet Spectra
The CD and UV spectra of twisted, conjugated π-electron systems, including
exciton-coupled systems, can be calculated by the π-electron SCF-CI-dipole ve-
locity (DV) molecular orbital method (the computer program CD/NH-Sendai)
[1–4,30]. In the dipole velocity method, the rotational strength Rba and dipole
strength Dba are expressed as follows:
Rba ⫽ 2(ϕa | ∇ |ϕb ) (ϕa | r ⫻ ∇ | ϕb )βM2 /(πσba ) (1)
Dba ⫽ 2(ϕa | ∇ | ϕb ) 2 βM2 /(πσba ) 2 (2)
where ∇ is the del operator, r is a distance vector, βM is the Bohr magneton, and
σba is the excitation wave number of the transition a → b. The z axis components
of the electric and magnetic transition moments are expressed, respectively, as
(ϕa | ∇| ϕb ) z ⫽ 冱 (C C
bonds
ra sb ⫺ Csa Crb ) 〈∇rs 〉 cos Zrs (3)
(ϕa | r ⫻ ∇ | ϕb ) z ⫽ 冱 (C
bonds
ra Csb ⫺ Csa Crb )
(4)
〈∇rs〉 (Xrs cos Yrs ⫺ Yrs cos Xrs )
cos Zrs ⫽ (Zr ⫺ Zs )/Rrs (5)
Xrs ⫽ (Xr ⫹ Xs )/2 (6)
where Cra is the coefficient of atomic orbital r in the wave function ϕa ; 〈∇rs 〉 is
the expectation value of a dipole velocity vector ∇rs that is directed along the
bond rs in the direction r → s; Xr, Yr, and Zr are the x, y, and z coordinates of
an atom r, respectively; and Rrs is the interatomic distance between atoms r and
s. In a similar way, the x and y components of the electric and magnetic transition
moments were calculated.
In the π-electron SCF-CI-DV MO calculation, the configuration interac-
tions (CIs) among all singly excited states were included, and the following
standard values of atomic orbital parameters were used: for aromatic carbon,
W(C) ⫽ ⫺11.16 eV, (rr | rr) (C) ⫽ 11.13 eV, β(C ⫺ C, 1.388 Å) ⫽ ⫺2.32 eV,
〈∇〉 (C ⫺ C, 1.388 Å) ⫽ 4.70 ⫻ 107 cm⫺1; for carbonyl oxygen, W(O)
TM

⫺17.28 eV, (rr | rr) (O) ⫽ 14.58 eV, β(C ⫽ O) ⫽ ⫺2.54 eV, 〈∇〉 (C ⫽ O) ⫽
5.0 ⫻ 107 cm⫺1; for nitrogen, Z(N) ⫽ 2.0, W(N) ⫽ ⫺27.70 eV, (rr| rr)(N) ⫽
17.44 eV, β(C ⫺ N) ⫽ ⫺1.899 eV, 〈∇〉(C ⫺ N) ⫽ 5.50 ⫻ 107 cm⫺1 [3,4]. The
electric repulsion integral (rr | ss) was estimated by the Nishimoto–Mataga equa-
tion [31]. The resonance integral and del values were calculated by employing
the following equations, respectively;
β ⫽ [S/S (C ⫺ C, 1.388 Å)] β(C ⫺ C, 1.388 Å) (7)
〈∇〉 ⫽ [〈∇〉 (empir., 1.388 Å)/〈∇〉 (theor., 1.388 Å)] 〈∇〉 (theor.) (8)
where S is the overlap integral.
In the case of Method A, the component CD and UV bands were approxi-
mated by the Gaussian distribution [3].
∆ε(σ) ⫽ ∑ ∆εk exp[⫺((σ ⫺ σk )/∆σ) 2 ] (9)
ε(σ) ⫽ ∑ εk exp[⫺((σ ⫺ σk )/∆σ) ] 2
(10)
where 2∆σ is the 1/e width of the bands.
When the shape of the component CD and UV bands is much better approx-
imated by the observed shape of the UV bands of model compounds than by the
Gaussian distribution, the following equations (Method B) were adopted [3]:
∆ε(σ) ⫽ ∑ ∆εk f (σ ⫹ σo ⫺ σk ) (11)
ε(σ) ⫽ ∑ εk f (σ ⫹ σo ⫺ σk ) (12)
where f (σ) is the function describing the shape of component CD and UV bands
and is taken from the observed UV spectrum of a model compound. For example,
in the case of compounds containing naphthalene chromophores, the shape of
the UV spectrum of naphthalene was adopted. On the other hand, when the ob-
served CD and UV spectra are composed of many component bands, as in the
case of the natural products discussed in this chapter, Method A was employed.
The numerical calculations were carried out on a Sun S-4/10 Work Station
and/or the NEC ACOS-3900 computer at the Computer Center of Tohoku Uni-
versity.
III. ABSOLUTE STEREOCHEMISTRY OF (ⴙ)-1,8a-

DIHYDRO-3,8-DIMETHYLAZULENE [(ⴙ)-1], A
LABILE BIOSYNTHETIC PRECURSOR OF
1,4-DIMETHYLAZULENE [5,6]
Considerable attention has focused on the chemistry of trinorsesquiterpenes with

a 1,4-dimethylazulene skeleton isolated from marine soft corals and liverworts
TM

[32–36]. Takeda first isolated chiroptically active 1,8a-dihydro-3,8-dimethylazu-
lene [(⫹)-1] (Chart 1), from the cell culture of the liverwort Calypogeia granulata
Inoue, as a labile trinorsesquiterpenoid biosynthetic precursor of 1,4-dimethyla-
zulene (2) [32]. The absolute stereochemistry of this intermediate is quite interest-
ing from the viewpoint of the biosynthesis of trinorsesquiterpenes with 1,4-di-
methylazulene skeleton. In fact, trinoranastreptene 3 [inflatene [35], also known
as clavekerin B [36]], a probable biosynthetic precursor of (⫹)-1, was isolated
from the same liverwort, and also from marine soft corals. Furthermore, Kitagawa
and coworkers had established that the absolute structure of clavukerin A [(⫺)-
4], a related trinorsesquiterpene isolated from a marine soft coral, was (8S,8aS)-
(⫺)-3,8-dimethyl-1,2,6,7,8,8a-hexahydroazulene [33]. Since the π-electron sys-
tem of 1 consists of a twisted, conjugated polyene chromophore, the problem
of the absolute configuration is also interesting from the theoretical viewpoint.
However, the absolute configuration of (⫹)-1 has remained undetermined be-
cause of its extreme instability and limited amounts of sample available. In the
following, we discuss the absolute stereochemistry of (⫹)-1, as determined by
theoretical calculation of the CD spectrum. In addition, we describe the experi-
mental verification of the absolute configuration of (⫹)-1 by the synthesis of
model compounds.
The labile intermediate (⫹)-1 with a unique 1,8a-dihydroazulene skeleton
Chart 1 1,8a-Dihydro-3,8-dimethylazulene [(8aS)-(⫹)-1] and related compounds.
TM

showed very intense chiroptical activity, [α]D ⫹1165°, and intense CD Cotton
effects (Fig. 1), suggesting a strongly twisted conjugated tetraene system. There-
fore, it was reasonable to consider that the chiroptical activity of 1 is mainly due
to the distortion of the π-electron chromophore. In order to predict the absolute
configuration of (⫹)-1 theoretically, we performed the calculation of the CD
curve of a model compound, 1,8a-dihydroazulene [(8aR)-9], on the basis of the
π-electron framework approximation, using the SCF-CI-DV MO method [1–4].
The absolute configuration of 9 was arbitrarily chosen to be (8aR) for the calcula-
tion. The molecule has a very rigid skeleton in which the triene part in the seven-
membered ring is almost symmetric when reflected through the yz plane (Fig.
2). Therefore, it is also reasonable to consider that conjugation of the triene sub-
unit with the additional double bound in the five-membered ring generates the
chiroptical activity.
A. Molecular Structure
Model compound (8aR)-9 has the following stereochemical features: The 1,8a-
dihydroazulene ring skeleton is conformationally very rigid. The triene subunit
Figure 1 Circular dichroism and ultraviolet spectra of naturally occurring (8aS)-(⫹)-

1,8a-dihydro-3,8-dimethylazulene [(⫹)-1] in hexane. (Source: Ref. 5.)
TM

Figure 2 Stereochemistry of model compound (8aR)-1,8a-dihydroazulene [(8aR)-9]
calculated by the MOPAC 93, AM1 program.
(3a-C/4-C/5-C/6-C/7-C/8-C) of the seven-membered ring is almost symmetric

when reflected through the plane defined by 8a-C, 8a-H, and the midpoint be-
tween 5-C and 6-C. The additional double bond (2-C/3-C) in the five-membered
ring is coplanar with the 3a-C/4-C double bond.
B. Theoretical Determination of Absolute Stereochemistry

by Calculation of Circular Dichroism Spectra
The calculated CD and UV curves of (8aR)-9 are illustrated in Fig. 3. The UV
spectrum exhibited two allowed π → π* absorption bands: λmax 313 nm (ε 9900)
and 219 nm (ε 27,300). The calculated values agree closely with the observed
UV data of (⫹)-1: λmax 308.5 nm (ε 5400) and 227.5 nm (ε 25,600). Analysis of
the calculations clarified that the two absorption bands at 313 and 219 nm consist
of single π → π* transitions, respectively, and also revealed that, to a first approx-
imation, the transition at 313 nm is polarized along the short axis of the molecule,
whereas the transition at 219 nm is along the long axis. Therefore, the absorption
band at 219 nm is stronger than that at 313 nm because the π-electron region
along the long axis is larger than that along the short axis. The calculated values
of the dipole strength are D ⫽ 12.6 ⫻ 10⫺36 and 24.4 ⫻ 10⫺36 cgs unit for the
313- and 219-nm transitions, respectively.
In the calculated CD spectrum, these two transitions yielded negative and
positive CD Cotton effects at longer and shorter wavelengths, respectively (Fig.
3): λext 313 nm (∆ε ⫺ 13.9) and 219 nm (∆ε ⫹ 46.2). These calculated values
are in a good agreement with the observed data of (⫹)-1, though with reversed
sign of the ∆ε values (Fig. 1): λext 314.0 nm (∆ε ⫹ 19.7) and 235.2 nm (∆ε ⫺
47.4). The observed CD curve of (⫹)-1 showed a pattern characteristic of the
1,8a-dihydroazulene chromophore: the CD Cotton effect at longer wavelength is
weaker than that at shorter wavelength, and the two Cotton effects are opposite
in sign. It is evident that the observed CD spectral pattern characteristic of the
TM

Figure 3 Circular dichroism and ultraviolet curves of (8aR)-1,8a-dihydroazulene
[(8aR)-9] calculated by the SCF-CI-DV MO method. SCF-CI-DV MO, self-consistent
field/configuration interaction/dipole velocity molecular orbital. (Source: Ref. 5.)
1,8a-dihydroazulene system was well reproduced by the calculation when Figs.

1 and 3 are compared. The rotational strengths were calculated to be R ⫽
⫺44.3 ⫻ 10⫺40 and ⫹103.3 ⫻ 10⫺40 cgs units for the 313- and 219-nm transitions
of (8aR)-9, respectively.
The observed CD curve of (⫹)-1 is almost a mirror image of that calculated
for the model compound (8aR)-9, although the ∆ε values are slightly different.
Accordingly, the absolute stereochemistry of the labile biosynthetic intermediate
(⫹)-1 was theoretically predicted to be (8aS), as shown in structure 1. This con-
clusion was experimentally proved by the synthesis of model compounds, as
described in the following section. The present results imply that the presumed
precursor, trinoranastreptene 3, has the same absolute configuration at the bridge-
head position as (⫹)-1. It should be noted that naturally occurring compounds
(⫹)-1 and clavukerin A[(⫺)-4], partially hydrogenated azulene derivatives, have
the same absolute configuration at the 8a position, regardless of the different
sources.
TM

C. Experimental Verification by the Synthesis of
Model Compounds
In order to verify the absolute configuration of (⫹)-1 theoretically predicted by
the SCF-CI-DV MO method, we synthesized model compounds, (1S,8aS)-(⫹)-
1,8a-dihydro-1-methoxy-8a-methylazulene [(⫹)-7] and (1S,8aS)-(⫹)-1,8a-dihy-
dro-1-methoxy-6,8a-dimethylazulene [(⫹)-8], starting from the optically active
Wieland-Miescher ketone (8aS)-(⫹)-10 (Scheme 1) [37–39]. These compounds
were chosen as models since they resist oxidation to azulene, and the angular
position 8a is substituted with a methyl group and thus not prone to epimerization.
In addition, the methoxyl group makes the compounds less volatile, and hence
the compounds can be easily handled.
Scheme 1 Synthesis of model compounds (⫹)-7 and (⫹)-8. (Source: Ref. 5.)
TM

Starting from the optically pure Wieland-Miescher ketone (8aS)-(⫹)-10
[39], the model compounds (1S,8aS)-(⫹)-7 and (1S,8aS)-(⫹)-8 were synthesized
[5]. Thus, according to the procedure reported by Heathcock et al., the Wieland-
Miescher ketone (8aS)-(⫹)-10 was converted to perhydroazulenone 11 (Scheme
1). The mixture of trans-fused ketone 11 and its cis-fused isomer was reduced
to the corresponding glycols, then methylated to dimethyl ethers, from which a
single stereoisomer (⫹)-12 was obtained as the major product by recrystallization
from diethyl ether. Bromination of the ketal (⫹)-12 with PyHBr3 occurred exclu-
sively at the C-7 position, yielding bromide (1S,3aR,4S,7R,8aS)-(⫹)-13. The po-
sition of the bromine atom in (⫹)-13 was determined by 1H NMR spectroscopy:
H-7 appeared as a doublet of a doublet. Dehydrobromination of (⫹)-13 with
t-BuOK and deketalization afforded enone 14. Successive eliminations of metha-
nol and dehydrogenation with DDQ in the presence of a catalytic amount of
p-TsOH yielded the desired trienone 15.
Reduction of ketone 15 with LiAlH4 gave an epimeric mixture of alcohols
16; since these alcohols were extremely unstable, they were immediately used
in the next reaction. For the purpose of dehydration of 16, various reaction condi-
tions were examined, and iodine in benzene was finally found to be best. When
a mixture of 16 and a catalytic amount of iodine in benzene was heated, the
desired tetraene model compound (1S,8aS)-(⫹)-7 was obtained in good yield.
Although (⫹)-7 was relatively unstable, it could be distilled in vacuo to give a
faint yellow liquid: bp 35–45°C (0.067 kPa); [α]D ⫹ 393.3°. The other model
compound (⫹)-8 was similarly synthesized: bp 60–70°C (0.029 kPa); [α]D ⫹
323.8°. Since both of the dihydroazulenes 7 and 8 were unstable in pure form,
they were stored in a freezer as dilute hexane solutions. The structure of (⫹)-7
was determined by 1H NMR studies including two-dimensional shift correlation
and difference NOE spectra; all hydrogen atoms were fully assigned.
The absolute configurations of these model compounds were also estab-
lished by x-ray crystallographic analysis of synthetic intermediate (⫹)-13. The
anomalous dispersion effect of the bromine atom led to the (1S,3aR,4S,7R,8aS)
absolute configuration assignment as shown in Fig. 4.
The CD and UV spectra of (1S,8aS)-(⫹)-7 are shown in Fig. 5; the UV
spectrum exhibited a π → π* band of medium intensity at 324.3 nm (ε 6000)
and an intense band at 223.2 nm (ε 23,700), which are characteristic of the 1,8a-
dihydroazulene chromophore. In the region of these transitions, the CD spectrum
showed a weak positive Cotton effect at λext 321.1 nm (∆ε ⫹5.7), and an intense
negative Cotton effect at λext 221.3 nm (∆ε ⫺24.5). The other model compound
(1S,8aS)-(⫹)-8 showed similar CD and UV spectra.
The model compounds (⫹)-7 and (⫹)-8 have extra chirality due to the
methoxyl group at the C-1 position. To exclude this extraneous stereogenic cen-
ter, we synthesized the more ideal model compounds (8aS)-(⫹)-5 and (8aS)-(⫹)-
TM

Figure 4 ORTEP drawing of the absolute stereostructures of the two crystallographi-
cally independent molecules of (1S,3aR,4S,7R,8aS)-(⫹)-13 contained in an asymmetric
unit. (Source: Ref. 6.)
6 starting from 11 following routine procedures (Scheme 2) [6]. The CD

spectrum of (8aS)-(⫹)-5 showed more intense Cotton effects than that of
(1S,8aS)-(⫹)-7 (Fig. 6); the CD spectrum of (8aS)-(⫹)-6 was similar to that of
(8aS)-(⫹)-5.
The CD spectra of model compounds (1S,8aS)-(⫹)-7, (1S,8aS)-(⫹)-8,
(8aS)-(⫹)-5, and (8aS)-(⫹)-6 were similar in both sign and shape of the Cotton
effects to that of dihydroazulene (⫹)-1. In the case of (8aS)-(⫹)-5, the CD inten-
sity of the Cotton effects was also comparable to that of the natural product (⫹)-
1. Therefore, it was experimentally established that the natural dextrorotatory
1,8a-dihydro-3,8-dimethylazulene [(⫹)-1] had the (8aS) absolute configuration.
The present results thus verified the theoretical determination of the absolute
configuration of (⫹)-1 discussed previously.
There were differences in the ∆ε values in the CD spectra of the natural
product (⫹)-1 and model compounds (1S,8aS)-(⫹)-7 and (1S,8aS)-(⫹)-8, which
may be due to the added chirality of the C-1 methoxyl substitution and/or to the
difference in the positioning of the methyl groups. In fact, the methyl group at
the C-6 position in (⫹)-6 diminished the ∆ε value, compared with that of
(⫹)-5. Furthermore, it seems likely that the replacement of a hydrogen in the
angular position 8a by a methyl group changes the molecular geometry to some
extent and hence affects the CD activity. Nevertheless, the consistent pattern of
TM

Figure 5 Circular dichroism and ultraviolet curves of (1S,8aS)-(⫹)-1,8a-dihydro-1-
methoxy-8a-methylazulene [(⫹)-7] in EtOH. (Source: Ref. 5.)
a weak, long wavelength transition centered around 315–325 nm, with a stronger,
oppositely signed band around 225 nm for these models, and the natural dihy-
droazulenes, along with the agreement of this pattern with the theoretical CD
spectrum of (8aR)-9, indicated that the absolute stereoconfiguration of (⫹)-1 and
related dihydroazulenes may be assigned on the basis of the theoretical calcula-
tions.
Scheme 2 Synthesis of model compounds (⫹)-5 and (⫹)-6. (Source: Ref. 6.)
TM

Figure 6 Circular dichroism and ultraviolet curves of (8aS)-(⫹)-1,8a-dihydro-8a-
methylazulene [(⫹)-5] in EtOH. (Source: Ref. 6.)
IV. ABSOLUTE STEREOCHEMISTRY OF THE

HALENAQUINOL FAMILY, MARINE NATURAL
PRODUCTS WITH A NOVEL PENTACYCLIC
SKELETON, AS DETERMINED BY THE THEORETICAL
CALCULATION OF CIRCULAR DICHROISM SPECTRA
[7–12]
In recent years, there have been many reports concerning isolation, structure de-
termination, and biological activity studies of marine natural products. Many
novel, biologically active compounds have been isolated from marine sponges.
For example, Scheuer and coworkers isolated halenaquinone [(⫹)-18], an antibi-
otic with a novel pentacyclic skeleton, from the tropical marine sponge Xestos-
pongia exigua collected off the Western Caroline Islands (Chart 2) [40]. The
structure of (⫹)-18 was determined by x-ray crystallographic analysis. The abso-
lute configuration, however, remained undetermined since the x-ray Bijvoet
TM

Chart 2 Halenaquinone [(12bS)-(⫹)-18] and its family of related marine natural prod-
ucts.
method for determination of the absolute configuration on the basis of the anoma-
lous dispersion effect of heavy atoms could not be applied.
Subsequent to Scheuer’s work, Kitagawa and coworkers isolated halenaqui-
nol [(⫹)-19], the hydroquinone form of (⫹)-18, from the Okinawan sponge Xes-
tospongia sapra, together with halenaquinol sulfate [(⫹)-20] [41]. Halenaquinol
[(⫹)-19] was easily oxidized either at UV irradiation or through heating in air
to give (⫹)-18. Furthermore, Nakamura and coworkers isolated xestoquinone
[(⫹)-21], a powerful cardiotonic, from the same Okinawan sponge, X. sapra [42].
More recently, Schmitz and Bloor isolated a series of similar natural products,
tetrahydroxestoquinol (23), the related dihydrofuran compound 24, adociaqui-
none A [(⫹)-25], adociaquinone B [(⫹)-26], and 3-ketoadociaquinone A (27),
from a marine sponge, Adocia sp. collected in Truk Lagoon, in addition to (⫹)-
18 and (⫹)-21 (Chart 3) [43]. They also revealed that halenaquinone [(⫹)-18]
and adociaquinone B [(⫹)-26] showed mild cytotoxicity. Harada and coworkers
also reported (⫺)-prehalenaquinone [(⫺)-22], a putative biosynthetic precursor
to (⫹)-18 and (⫹)-21 (Chart 2) [11]. Considering the interest in the halenaquinol
family of marine natural products due to the physiological activity of these novel
TM

Chart 3 Other members of the halenaquinone family of marine compounds.
compounds, the determination of their absolute stereochemistries became an im-

portant problem.
In this section, the application of the π-electron SCF-CI-DV MO method
to calculate the CD spectra of the complex chromophore of the natural products
of the halenaquinol family is described [7,8]. In the course of this work, we
have also achieved the first total syntheses of (⫹)-halenaquinol [(⫹)-19], (⫹)-
halenaquinone [(⫹)-18], xestoquinol, (⫹)-xestoquinone [(⫹)-21], (⫺)-prehalen-
aquinone [(⫺)-22], (⫹)-adociaquinone A [(⫹)-25], and (⫹)-adociaquinone B
[(⫹)-26] [9–12]. By these total syntheses, we experimentally confirmed that the
absolute stereochemistry theoretically determined for these compounds with their
unique, twisted π-electron system was correct.
A. An Attempt to Apply the Circular Dichroism Exciton

Chirality Method for Determining the Absolute
Configuration of the Halenaquinols
The CD exciton chirality method has been extensively applied to degenerate sys-
tems consisting of two identical chromophores such as dibenzoates, binaphthyls,
TM

bianthryls, and others [3]. In addition to such degenerate systems, the CD exciton
chirality method is useful for determination of the absolute stereochemistry of
nondegenerate systems that contain two different chromophores that experience
exciton coupling. For example, Harada and Nakanishi determined the absolute
configuration of chromomycin A3 by interpretation of the exciton coupling be-
tween the long-axis-polarized 1Bb transition of the naphthalene chromophore and
the long-axis-polarized transition of the p-methoxybenzoate chromophore
[3,44,45]. We considered that this method would be analogously applicable to
the halenaquinols and related compounds. Specifically, we planned to synthesize
benzoate derivative 32, then apply the CD exciton chirality method to the ex-
pected interaction between the naphthalene and benzoate chromophores (Scheme
3) [8].
Halenaquinol [(⫹)-19] was methylated in refluxing acetone with iodometh-
ane in the presence of potassium carbonate with the exclusion of light to yield
Scheme 3 Strategy to apply the circular dichroism exciton chirality method.
TM

dimethyl ether (⫹)-28 as yellow needles (Scheme 3). To differentiate the two
carbonyl groups at the 3- and 6-positions, dimethyl ether (⫹)-28 was selectively
reduced with NaBH4 in the presence of CeCl3 ⋅ 7H2O, which catalyzes the regio-
selective 1,2-reduction of conjugated enones [46]. Keto-alcohol (⫹)-29 was ob-
tained, and assignment of the C-3 stereocenter was secured by 1H NMR coupling
constant data: J2β-H,3α-H ⫽ 8.0 Hz, J2α-H,3α-H ⫽ 8.0 Hz. The two large coupling
constants indicate that the D-ring adopts a twisted half-chair conformation. This
stereochemical assignment was subsequently confirmed by the NOE data of de-
rivative 34 shown in Fig. 8 (Section IV.B). The alcohol was then protected as
the t-butyldimethylsilyl ether (⫹)-30, followed by treatment with NaBH4 /CeCl3 ⋅
7H2O in MeOH/CH2Cl2 to reduce the C-6 carbonyl group. However, the expected
product, which was formulated as alcohol 31, was extremely unstable and could
not be isolated, nor could it be trapped by benzoylation. Therefore, the prepara-
tion of naphthalene–benzoate compound 32, and hence the application of the CD
exciton chirality method, was unsuccessful.
B. Attempted Application of the ␲-Electron Self-

Consistent Field/Configuration Interaction/Dipole
Velocity Molecular Orbital Method to Halenaquinol
Dimethyl Ether [(ⴙ)-28]
We next attempted to apply the π-electron SCF-CI-DV MO method to halenaqui-
nol dimethyl ether [(⫹)-28] to determine the absolute configuration of halenaqui-
nol [(⫹)-19] since (⫹)-28 has a conjugated π-electron system composed of a
naphthalene–ketone–furan–ketone chromophore that is twisted as a result of chi-
ral center bearing the angular methyl group at the 12b-position [47]. Halenaquinol
itself was not employed in this case because of its photochemical and thermal
instability (even at 40°C). Furthermore, as a protecting group, a methyl ether
group is better than others such as an acetate group, because the π-electron system
of dimethyl ether (⫹)-28 containing only the lone pair of electrons of the ether
oxygens is simpler than that of a diacetate. In the case of a diacetate, the π-
electron system becomes complex as a result of the contribution of the ester
carbonyls and their rotational conformation variability.
Although we expected relatively intense CD Cotton effects for (⫹)-28, the
CD spectrum showed only weak Cotton effects (Fig. 7). The weak intensity of
the CD Cotton effects may be due to the existence of two carbonyl groups of
strong electron-withdrawing nature, which makes the total π-electron system of
(⫹)-28 less symmetric, and hence the electronic transitions more complex and
weaker. Therefore, from the viewpoint of reliability of the determination, (⫹)-
28 was deemed unsuitable for the theoretical determination of the absolute con-
figuration since it is rather difficult to discriminate small positive and negative
∆ε values. Although we actually carried out the calculation of the CD spectrum
TM

Figure 7 Circular dichroism spectrum of halenaquinol dimethyl ether [(12bS)-(⫹)-28]
in EtOH derived from the natural sample of halenaquinol [(⫹)-19]. (Source: Ref. 47.)
of (⫹)-28, and the results suggested the 12bS absolute configuration for (⫹)-28
(as later shown to be correct), we don’t consider these results a convincing and
unambiguous determination of the absolute configuration because of the small
∆ε values of CD Cotton effects.
C. Circular Dichroism Spectra of Naphthalene–Diene

Derivatives with Twisted ␲-Electron Systems
As described, we could not isolate alcohol 31 because of its instability (Scheme
3) and therefore failed to determine the absolute configuration of the halenaqui-
nols by application of the nondegenerate CD exciton chirality method. However,
we were very happy to find that the reduction of ketone (⫹)-30 with NaBH4 /
CeCl3 gave the elimination/addition products (⫺)-33 and (⫹)-34 instead of 31
(Scheme 4) [7,8]. It was presumed that (⫺)-33 and (⫹)-34 are derived from 31
by elimination of the hydroxyl group and subsequent addition of methanol at the
4-position. Furthermore, treatment of the reaction mixture with a catalytic amount
of aqueous HCl following the borohydride reduction of (⫹)-30 gave trans-me-
thoxy diene (⫺)-33 and cis-methoxy diene (⫹)-34 in moderate yields, 44% and
20%, respectively. The structures of acetal epimers (⫺)-33 and (⫹)-34 were de-
termined on the basis of the spectroscopic data, with the relative stereochemistries
unambiguously determined by the 1H NMR coupling constant data and NOE
enhancement data (Fig. 8).
It was very interesting to note that naphthalene–diene compounds (⫺)-33
and (⫹)-34 exhibited much stronger CD Cotton effects than other halenaquinol
derivatives. For example, the UV spectrum of trans-methoxysilyl ether (⫺)-33
TM

Scheme 4 Synthesis of naphthalene–diene derivatives (⫺)-33 and (⫹)-34.
Figure 8 NOE and coupling constant data and conformation of naphthalene–diene de-
rivative (⫹)-34. (Source: Ref. 8.)
TM

Figure 9 Observed circular dichroism and ultraviolet spectra of halenaquinol trans-
methoxydiene derivative (3R,4R,12bS)-(⫺)-33 in MeOH. (Source: Refs. 7 and 8.)
showed two intense π → π* bands (Fig. 9), a broad band at 324 nm (ε 27,000)
with complex vibrational structure, and a sharp band at 218 nm (ε 42,000). In
these corresponding regions, the CD spectrum of (⫺)-33 exhibited three major,
intense Cotton effects: λext 338 nm (∆ε ⫹6.4), 301 nm (∆ε ⫺23.3), and 229 nm
(∆ε ⫹40.9). The other naphthalene–diene compound (⫹)-34 also exhibited three
major CD Cotton effects of similar intensity and of the same sign as those of
(⫺)-33. These results clearly indicated that the CD Cotton effects originated
mainly from the π-electron chromophore composed of the naphthalene–diene
moiety, which is twisted by the angular methyl group at the 12b-position. Further-
more, the additional chiralities due to the silyloxy group at the 3-position and
the methoxy group at the 4-position are only minor contributors to the CD Cotton
TM

effects. These observations were critical to our choice of a model compound for
the theoretical study discussed in the next section. Thus, these naphthalene–diene
compounds are ideal systems for the theoretical determination of the absolute
stereochemistry by application of the π-electron SCF-CI-DV MO method. It is
interesting to note that the orientation of the methoxyl group apparently dictated
the sign of [α]D: negative for 33 and positive for 34.
D. Application of the ␲-Electron Self-Consistent Field/

Configuration Interaction/Dipole Velocity Molecular
Orbital Method to Naphthalene–Diene Derivatives
As a model compound for the theoretical calculation of CD spectra, we adopted
the molecule (12bS)-35, retaining the essential part of the π-electron system con-
tained in the naphthalene-diene compounds 33 and 34 (Fig. 10) [7,8]. Of specific
importance, in addition to the naphthalene and conjugated diene chromophores,
the lone pair of electrons of the two methyl ether and furan ring oxygens were
also included. The absolute configuration of 35 was arbitrarily chosen to be 12bS
for the calculation. The molecular geometry of (12bS)-35 was calculated using
the MOPAC 93, AM1 programs and is illustrated in Fig. 10. The molecular
framework of this model compound is relatively rigid, and the terminal cyclohex-
ene ring adopts a half-chair conformation. This conformation was supported by
the 1H NMR coupling constant and NOE enhancement data of compounds (⫺)-
33 and (⫹)-34 (Fig. 8): the coupling constant J1α-H,2β-H ⫽ 12.0 Hz indicated the
trans-diaxial relation between these protons. By contrast, J2β-H,3α-H ⫽ 8.0 Hz was
smaller than J1α-H,2β-H and is equal to J2α-H,3α-H, indicating the half-chair conforma-
tion of the terminal cyclohexene ring. The C-6/C-5a double bond and naphthalene
Figure 10 Model compound (12bS)-35 and its stable conformation calculated by the
MOPAC 93, AM1 programs.
TM

chromophore of (12bS)-35 constitute a clockwise helicity (dihedral angle of 5a-
6-6a-7: ⫹170°), whereas the conjugated diene moiety by itself constitutes a coun-
terclockwise helicity (dihedral angle of 3a-12c-5a-6: ⫺167°). The helical sense
of these two moieties is not changed even if the terminal cyclohexene ring adopts
the higher-energy boat conformation. Thus, the sense of the twist of the conju-
gated π-electron system is governed solely by the chirality of the angular methyl
group at the 12b position.
Calculation of the theoretical CD and UV spectra of (12bS)-35 by the π-
electron SCF-CI-DV MO method gave the curves illustrated in Fig. 11. The UV
spectrum exhibited two intense π → π* bands: a broad band at 349 nm (ε 29,900)
Figure 11 Circular dichroism and ultraviolet curves of the model compound (12bS)-
35 calculated by the π-electron SCF-CI-DV MO, self-consistent field/configuration
interaction/dipole velocity molecular orbital method. (Source: Refs. 7 and 8.)
TM

and a sharp band at 219 nm (ε 40,300). These calculated values agreed closely
with the observed UV data of (⫺)-33: λmax 324 nm (ε 27,000) and 218 nm (ε
42,000) (Fig. 9). In the corresponding region, the calculation predicted three prin-
cipal CD Cotton effects: a weak positive band at 378 nm (∆ε ⫹3.3), a negative
band of medium intensity at 322 nm (∆ε ⫺22.4), and a positive intense one
at 223 nm (∆ε ⫹35.5). These theoretically obtained CD bands were in a good
agreement with the observed data of (⫺)-33: λext 338 nm (∆ε ⫹6.4), 301 nm
(∆ε ⫺23.3), and 229 nm (∆ε ⫹40.9) (Fig. 9). The calculation thus reproduced
the basic pattern of the observed CD and UV spectral curves, including the sign,
position, intensity, and shape of the bands. Since the absolute configuration of
the model compound 35 was fixed to be (12bS), comparison of the calculated
and observed CD data leads to the conclusion that (⫺)-33 and other related naph-
thalene–diene compounds have the (12bS) absolute configuration. Accordingly,
the absolute stereochemistry of halenaquinol [(⫹)-19] was theoretically deter-
mined to be (12bS). Since UV irradiation of (⫹)-19 in the presence of air gave
halenaquinone [(⫹)-18,] and solvolysis of halenaquinol sulfate [(⫹)-20] quantita-
tively yielded halenaquinol [(⫹)-19], the absolute stereostructures of (⫹)-18 and
(⫹)-20 were also established as (12bS), respectively.
E. Circular Dichroic Power of a Twisted Naphthalene–

Diene System
In the case of (8aS)-(⫹)-1,8a-dihydro-3,8-dimethylazulene [(⫹)-1], the composi-
tion of the apparent CD and UV bands was rather simple, because each of the
apparent bands was composed of a single electronic transition. In contrast, the
π-electron chromophores of the twisted naphthalene–diene systems 33–35 are
complex and have no symmetric character [8]. Therefore, to confirm the applica-
bility of the π-electron SCF-CI-DV MO method to such complex systems, it is
important to analyze the composition of the apparent CD and UV bands theoreti-
cally obtained. As illustrated in Fig. 12, there are nine major electronic transitions
that contribute to the CD and UV bands. The first and second electronic transi-
tions, with weak positive rotational strengths at 374.5 and 351.6 nm, respectively,
generate the weak positive Cotton effect at 378 nm (Fig. 12). The third electronic
transition, with an intense negative rotational strength at 324.4 nm, results in the
negative Cotton effect at 322 nm, and the sixth electronic transition, with a strong
positive rotational strength, dominates the intense positive Cotton effect at 223
nm. The correspondence between the calculated CD rotational strengths and the
observed CD Cotton effects in the spectra of (⫺)-33 and (⫹)-34 is quite evident.
Therefore, this analysis using hypothetical model (12bS)-35 for comparison with
(⫺)-33 and (⫹)-34 makes the theoretical determination of the absolute configu-
rations of halenaquinol [(⫹)-19] and related compounds more reliable.
TM

Figure 12 Calculated rotational and dipole strengths of the transitions of the model
compound (12bS)-35. (Source: Ref. 8.)
F. The Synthetic Strategy for the First Total Synthesis of

(ⴙ)-Halenaquinol [(ⴙ)-19] and (ⴙ)-Halenaquinone
[(ⴙ)-18]
As discussed, we succeeded in the theoretical determination of the absolute ste-
reochemistry of novel marine natural products of the halenaquinol family. It is
quite natural that chemists, as the next step, want to prove experimentally the
absolute configurations theoretically predicted. So, we undertook the synthesis
of halenaquinol [(⫹)-19] and halenaquinone [(⫹)-18] in their natural enantio-
meric forms in order to corroborate their structures by comparison of the CD
spectra of the synthetic samples with that of the natural ones [9].
Our retrosynthetic strategy centered around the cycloaddition of a transient
o-quinodimethane generated from benzocyclobutene 39 with enone 41 as the key,
convergent step, forming 38 with the carbocyclic halenaquinol skeleton (Scheme
TM

Scheme 5 Retrosynthetic analysis for halenaquinol [(12bS)-(⫹)-19].
5). The naphthoquinone moiety of (12bS)-(⫹)-18, which can be reduced to halen-

aquinol [(⫹)-19], can be obtained by oxidative cleavage of the hydroquinone
dimethyl ether (12bS)-(⫹)-28; the furan ring of (12bS)-(⫹)-28 was considered
accessible by oxidation of triol 36. The diosphenol moiety of 36, would be ob-
tained by the air oxidation of ketone 37 [48,49]. The tetracyclic skeleton of 37
would be constructed by the key Diels-Alder reaction between the transient o-
quinodimethane 40 (generated from benzocyclobutane 39) and enone 41.
Further analysis for dienophile 41 suggested that this critical synthon could
be derived from the Wieland-Miescher ketone 43 (Scheme 6), in which the one-
carbon unit ultimately to become C-4 in 19 is introduced by application of Stork’s
reductive hydroxymethylation procedure to 43 [50]. In this synthetic route, the
absolute configuration of the bridgehead methyl group of optically active 43 is
retained as that of the corresponding methyl group in the final product. Since the
absolute configurations of halenaquinol and halenaquinone were predicted to be
TM

Scheme 6 Retrosynthesis of halenaquinol [(12bS)-(⫹)-19]; preparation of enone 41.
12bS as discussed, the synthesis of the natural enantiomeric forms of halenaquinol

and halenaquinone should start from the (8aR)-(⫺) enantiomer of 43. Among
several synthetic methods to prepare optically active 43, the most practical is the
asymmetric cyclization of triketone 44 with a catalytic amount of optically active
proline, which was independently discovered by Hajos’s and by Eder’s groups
[39].
We achieved the first total synthesis of (⫹)-halenaquinol [(⫹)-19] and (⫹)-
halenaquinone [(⫹)-18] by following this strategy [9]. The carbonyl group at
the 1-position of optically pure (8aR)-(⫺)-43 was selectively protected to give
monoacetal (⫺)-45, which was then reductively hydroxymethylated according to
the procedure of Stork [50] (Scheme 7): enone (⫺)-45 was reduced with lithium
in liquid ammonia, and the resultant enolate was trapped as the trimethylsilyl
ether 46. Regeneration of the enolate anion by treatment of 46 with methyllithium
and then addition of gaseous formaldehyde gave keto alcohol (⫹)-42 as the sole
stereoisomer in 82% overall yield from (⫺)-45. Keto alcohol (⫹)-42 was reduced
with lithium tri-sec-butylborohydride (L-Selectride), producing cis-diol (⫹)-47
in 92% yield, which was then converted to ketodiol (⫹)-48 in 98% yield by
treatment with aqueous p-toluenesulfonic acid ( p-TsOH). The relative stereo-
chemistry of the 6(ax)-hydroxyl and 5(eq)-hydroxymethyl groups of (⫹)-48 was
secured by the 1H NMR coupling constant data of its acetonide (⫹)-51 (Fig. 13).
Formation of the tosylhydrazone of (⫹)-48, followed by treatment with
methyllithium, gave olefin (⫺)-49 in quantitative yield. Next, the 1,3-diol of (⫺)-
49 was protected as the acetonide to give olefin (⫺)-50. Finally, allylic oxidation
of acetonide olefin (⫺)-50 with CrO3 /3,5-dimethylpyrazole [51] gave conjugated
TM

Scheme 7 Synthesis of chiral dienophile 41.
Figure 13 1H nuclear magnetic resonance (NMR) coupling constant data of keto-aceto-

nide (⫹)-51. (Source: Ref. 47.)
TM

enone (⫹)-41 in 63% yield, which was used as the dienophile in the Diels-Alder
reaction with 40 generated in situ from 3,6-dimethoxybenzocyclobutene (39).
Although dimethoxybenzocyclobutene 39 had been previously synthesized
by photocycloaddition chemistry [52], we prepared it by pyrolysis of sulfone 55,
which itself was prepared from 2,3-dimethyl-1,4-dimethoxybenzene [52]
(Scheme 8) [9]. Bromination of 52 with N-bromosuccinimide (NBS), followed
by treatment of the resulting dibromide 53 with sodium sulfide in aqueous etha-
nol, gave 54 in 70% yield. Oxidation of 54 with m-chloroperbenzoic acid
(MCPBA) in dichloromethane afforded sulfone 55 in 89% yield. Various reaction
conditions were examined for the thermal elimination of sulfur dioxide, and fi-
nally it was found that direct heating of solid sulfone 55 without solvent afforded
the desired 3,6-dimethoxybenzocyclobutene (39) in moderate yield. Thus, crys-
tals of 55 were pyrolyzed at 305–310°C in a muffle furnace under a stream of
nitrogen to give 39 in 48% yield.
The Diels-Alder reaction of o-quinodimethane 40 derived from 39 and (⫹)-
41 was achieved by heating a benzene solution of the two reactants in a sealed
tube at 210–215°C for 20 h, giving tetrahydronaphthalene derivative (⫹)-38 in
33% yield (Scheme 9). Numerous reaction conditions were examined to increase
the yield, but no further improvement was achieved. The 13C NMR spectrum of
(⫹)-38 indicated that the product was composed of a single stereoisomer. How-
ever, the relative stereochemistry of the newly formed chiral centers was not
investigated further since they were eliminated in the subsequent dehydrogena-
tion. To dehydrogenate the tetrahydronaphthalene moiety of (⫹)-38, a benzene
solution was refluxed with 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ) to
Scheme 8 Synthesis of 3,6-dimethoxybenzocyclobutene (39).
TM

Scheme 9 Synthesis of halenaquinol derivative 56.
afford naphthalene derivative (⫺)-37 in 89% yield, which was then subjected to
air oxidation in the presence of base [48]. Thus, oxygen was bubbled through a
solution of (⫺)-37 and potassium t-butoxide in t-butyl alcohol for 5 h, and the
mixture was worked up with aqueous ammonium chloride to give diosphenol 56
in 90% yield. The structure of 56 was secured by the 1H NMR (a sharp singlet
at δ 7.60 ppm disappeared upon addition of D2O), UV spectra (a red shift and
a hyperchromic effect of the longer-wavelength UV absorption band upon adding
aqueous NaOH), along with high-resolution MS data. Deprotection of the aceto-
nide group of 56 upon treatment with 60% aqueous acetic acid yielded triol 36,
which was subjected to the next reaction without purification (Scheme 10). The
oxidation of the primary and secondary hydroxyl groups of 36 and subsequent
cyclization to form the furan ring were accomplished by treatment with dimethyl
sulfoxide (DMSO) and 1,3-dicyclohexylcarbodiimide (DCC) in benzene in the
presence of trifluoroacetic acid and pyridine (PTFA), giving the desired halena-
quinol dimethyl ether [(12bS)-(⫹)-28] with the furan–diketone system, in 44%
overall yield from (⫺)-37. All of the spectroscopic data of the synthetic sample
of (⫹)-28 were identical with those of an authentic sample of (⫹)-28 prepared
from natural halenaquinol.
The dimethyl ether groups of (12bS)-(⫹)-28 were next oxidatively cleaved
with cerium (IV) ammonium nitrate (CAN) in aqueous methanol, affording halen-
aquinone [(12bS)-(⫹)-18] in 45% yield (Scheme 10). The 1H NMR and UV spec-
tra of the synthetic sample agreed with those of natural (⫹)-18. Finally, (12bS)-
TM

Scheme 10 Synthesis of halenaquinone [(12bS)-(⫹)-18] and halenaquinol [(12bS)-
(⫹)-19].
(⫹)-18 was reduced with aqueous sodium hydrosulfite in acetone to give halena-
quinol [(12bS)-(⫹)-19] in near-quantitative yield. The 1H NMR spectrum of
(12bS)-(⫹)-19 in DMSO-d6 exhibited two broad singlets at δ 9.6 and 9.8 that
were due to the phenolic hydroxyl groups, which disappeared when D2O was
added. The remaining 1H NMR peaks and UV spectrum curve were in good
agreement with those of the natural sample. The first total synthesis of halenaqui-
nol [(12bS)-(⫹)-19] and halenaquinone [(12bS)-(⫹)-18] with a novel polyketide
skeleton had been thus accomplished.
G. Circular Dichroism Spectrum of Halenaquinol Dimethyl

Ether [(12bS)-(ⴙ)-28] and Experimental Proof of the
Absolute Stereochemistry of the Halenaquinol Family
In the total synthesis of halenaquinol [(12bS)-(⫹)-19] and of halenaquinone
[(12bS)-(⫹)-18], we started with the Wieland-Miescher ketone (8aR)-(⫺)-43, as
presented earlier, and thus the synthetic sample of (⫹)-28 had the (12bS) absolute
configuration. If the theoretical determination of the absolute stereochemistry of
the halenaquinol family based on the π-electron SCF-CI DV MO method dis-
cussed in Section IV.D was correct, the CD spectrum of the synthetic sample
must be identical to those of the authentic sample of (⫹)-28 derived from natural
TM

Figure 14 Circular dichroism spectrum of the synthetic sample of halenaquinol di-
methyl ether [(12bS)-(⫹)-28] in EtOH. (Source: Ref. 47.)
halenaquinol. This was verified as shown in Figs. 7 and 14: the CD spectra of
the synthetic and natural samples are identical. These results lead to the experi-
mentally confirmed, unambiguous determination that the absolute stereochemis-
try of (⫹)-halenaquinol and (⫹)-halenaquinone is 12bS. In addition, these syn-
theses also proved that the theoretically predicted absolute configurations of
halenaquinol compounds were correct.
H. The First Total Synthesis of (ⴙ)-Xestoquinone [(ⴙ)-21]

and Xestoquinol, and Their Absolute Stereochemistries
In similar fashion, we also achieved the first total synthesis of (⫹)-xestoquinone
[(⫹)-21] and xestoquinol [10,11]. The CD spectrum of the synthetic sample of
(12bS)-21 was identical to that of natural xestoquinone [(⫹)-21] [10]. Thus, be-
ginning with Wieland-Miescher ketone (8aR)-(⫺)-43, synthetic (12bS)-xestoqui-
none [(⫹)-21] was prepared that was identical to natural xestoquinone [(⫹)-21],
which must therefore also have the 12bS absolute configuration. Xestoquinone
[(12bS)-(⫹)-21] was subsequently reduced to xestoquinol, so this hydroquinone
also has the 12bS absolute configuration. The absolute stereochemistry of (⫹)-
xestoquinone and of xestoquinol was thus unambiguously determined.
The synthetic sample of xestoquinone [(12bS)-(⫹)-21] was also converted
into the adociaquinones A [(⫹)-25] and B [(⫹)-26] [12]. Therefore, the absolute
configurations of these members of the halenaquinol family were also deter-
mined, as shown in Chart 3.
TM

I. Efficient Synthesis of (ⴙ)-Halenaquinone [(ⴙ)-18] and
(ⴙ)-Xestoquinone [(ⴙ)-21] via Prehalenaquinol Dimethyl
Ether [(ⴚ)-57]
During the synthetic studies described, we found a new procedure to convert triol
36 into dihydrofuran (3S,3aS,12bS)-(⫺)-57 (Scheme 11). When 36 was treated
with a limited amount of DMSO/DCC (5.0 equiv.)/PTFA (1.3 equiv.) in benzene
at room temperature for 2.5 h, the formation of dihydrofuran-alcohol (⫺)-57
instead of furan-diketone (⫹)-28 resulted (75%). When (⫺)-57 was resubjected
to the DMSO/DCC (8.6 equiv.)/PTFA (6.0 equiv.) oxidation at room temperature
for 18 h, the final product (⫹)-28 was obtained. Therefore, dihydrofuran-alcohol
(⫺)-57 is a synthetic intermediate of the Pfitzner-Moffatt oxidation of triol 36.
When we found this unexpected product, we immediately considered that
dihydrofuran alcohol (⫺)-57 could be a pivotal synthetic intermediate common
to the halenaquinone and xestoquinone natural products. If dihydrofuran alcohol
(⫺)-57 was dehydrated, the newly formed double bond should migrate into the
five-membered ring to give xestoquinol dimethyl ether [(⫹)-58] (Scheme 12).
Since the 3α-hydroxyl group and 3aβ-hydrogen of (⫺)-57 are in a trans-diaxial
relationship, the configuration was ideal for dehydration. A solution of dihydro-
furan alcohol (⫺)-57 in benzene was refluxed for 24 h in the presence of p-TsOH
to yield xestoquinol dimethyl ether [(⫹)-58] as expected in 77% yield. Compound
(⫹)-58 had previously been converted to xestoquinone [(⫹)-21] and xestoquinol
Scheme 11 Synthesis of dimethylhalenaquinol [(12bS)-(⫹)-28] from prehalenaqui-

none derivative (3S,3aS,12bS)-(⫺)-57.
TM

Scheme 12 Synthesis of xestoquinone [(12bS)-(⫹)-21] and xestoquinol [(12bS)-(⫹)-
59] from prehalenaquinone derivative (57).
[11]. This synthetic route is more efficient than our previous total synthesis of
xestoquinone and xestoquinol. We have thus achieved the total syntheses of xes-
toquinol, halenaquinol, and related compounds from the common synthetic inter-
mediate (⫺)-57.
J. Synthesis and Absolute Stereochemistry of

Prehalenaquinone and Prehalenaquinol, Putative
Biosynthetic Precursors Common to Halenaquinone,
Xestoquinone, and Related Natural Products
The chemical conversion of dihydrofuran alcohol (⫺)-57 into xestoquinol, halen-
aquinol, and related compounds described poses a very attractive explanation for
the biosynthesis of these natural products in marine sponges. That is, dihydro-
furan–alcohol–quinone 22 and dihydrofuran–alcohol–hydroquinone 60 are puta-
tive biosynthetic precursors of both the halenaquinols and xestoquinols (Scheme
13). As in the case of the chemical conversion, dehydration of 22 or 60 would lead
to xestoquinone [(⫹)-21] or xestoquinol [(⫹)-59], respectively, and oxidation of
22 or 60 would lead to halenaquinone [(⫹)-18] or halenaquinol [(⫹)-19]. The
related compound, dihydrofuran–alcohol 24 (Chart 3), isolated by Schmitz [43],
may be an earlier intermediate than precursors 22 and 60. Oxidation of the sec-
ondary alcohol group in the A-ring of 24 would give a diketone, which would
be expected to tautomerize to 60; hydroquinone 60 could be easily oxidized to
TM

Scheme 13 Synthesis of prehalenaquinone [(3S,3aS,12bS)-(⫺)-22] and prehalenaqui-
nol [(3S,3aS,12bS)-60].
quinone 22. Therefore, if the biosynthesis of halenaquinone, xestoquinone, and

related compounds proceeded as postulated here, we thought biosynthetic precur-
sors 22 and/or 60 may be found in marine sponges. To this end, samples of
putative biosynthetic precursors (⫺)-22 and 60 were synthesized (Scheme 13).
Treatment of dimethyl ether (⫺)-57 with ammonium cerium(IV) nitrate
(CAN) afforded quinone (⫺)-22 in 79% yield, which was then quantitatively
converted to hydroquinone 60 by treatment with sodium hydrosulfite. The struc-
tures of these compounds were confirmed by spectroscopic and physical data.
From the synthetic route described, the absolute stereochemistry of (⫺)-22 and 60
must be (3S,3aS,12bS). The putative biosynthetic precursors to the halenaquinol
family of natural products, (⫺)-22 and 60, were named prehalenaquinone and
prehalenaquinol, respectively.
We next sought to determine whether (⫺)-22 and 60 were actually present
in the marine sponges from which the halenaquinols and xestoquinols were iso-
lated. Using the synthetic samples of (⫺)-22 and 60 as standards, we analyzed
by high-performance liquid chromatography (HPLC) the components of an ex-
tract of the Okinawan marine sponge Xestospongia sapra from which halenaqui-
none [(⫹)-18] and xestoquinone [(⫹)-21] had been isolated (HPLC conditions:
silica gel, EtOA/hexane/MeOH 100 :100 :5, or ODS C18 column, 50% aqueous
MeOH). Although no peak corresponding to prehalenaquinol (60) was detected,
a peak corresponding to prehalenaquinone [(⫺)-22] was observed, and we subse-
quently succeeded in the isolation of (⫺)-22 from this source, as follows: The
TM

wet sponge (305 g) was treated with acetone, and the crude extract thus obtained
upon removal of the solvent (6.5 g) was successively separated and purified by
HPLC (silica gel, EtOAc, hexane/EtOAc 1: 3, and hexane/EtOAc/MeOH 100:
100 :2), giving (⫺)-22 (8.3 mg, 0.13% from the acetone extract). The spectral
and physical data, including the CD data, of natural (⫺)-22 thus isolated were
identical with those of synthetic (⫺)-22. These results imply that dihydrofuran–
alcohol (⫺)-22 may be a biosynthetic precursor of the halenaquinone and xesto-
quinone families of compounds.
V. ABSOLUTE STEREOCHEMISTRY OF A NATURAL

ATROPISOMER, THE BIFLAVONE (ⴚ)-4′,4ⵯ,7,7″-TETRA-
O-METHYLCUPRESSUFLAVONE [(ⴚ)-61] [13,18,19]
The title biflavone was isolated from Garcinia mangostana L. [13] and was iden-
tified as optically active cupressuflavone tetramethyl ether [(⫺)-61] by a direct
comparison of spectral data with those of the authentic sample isolated earlier
from Araucaria cunninghamii and A. cookii (Chart 4) [14]. Compound 61 is a
dimeric flavone as indicated by the MS spectral data ([M]⫹, m/z 594) and has a
chiral C2-symmetric structure that exhibits an optical rotation of [α]D27 ⫺25.3°,
with only seven signals in the 1H NMR spectrum, and 15 peaks in the 13C NMR
spectrum. We established the dimeric structure with an 8-8″-linkage of 61 on the
basis of detailed 1H and 13C NMR studies, including 1H NOESY, heteronuclear
gated decoupled 13C NMR, 1H-13C COSY, and long-range 1H-13C COSY
(COLOC) methods, and have achieved full assignment of all proton and carbon
Chart 4 Natural biflavone atropisomer (aR)-4′,4⵮,7,7″-tetra-O-methylcupressuflavone

[(⫺)-(aR)-61], and monomer 62.
TM

signals, leading to the structure as shown in Chart 4. The absolute stereochemistry
of (⫺)-61, however, had never been determined.
A. Ultraviolet and Circular Dichroism Spectra of the

Natural Atropisomer of Biflavone (ⴚ)-61
The UV spectrum of biflavone (⫺)-61 exhibited three intense π → π* bands at
324.2 nm, 273.0 nm, and 225.8 nm. The two UV bands at 324.2 nm and 273.0
nm were accompanied by bisignate CD Cotton effects, respectively (Fig. 15).
For the first UV band at 324.2 nm, positive and negative CD Cotton effects were
observed at 362.0 and 326.2 nm, respectively; for the second UV band at 273.0
nm, a negative CD shoulder around 300 nm and a positive Cotton effect at 267.5
Figure 15 Circular dichroism and ultraviolet spectra of biflavone (⫺)-4′,4⵮,7,7″-tetra-

O-methylcupressuflavone [(aR)-(⫺)-61] in EtOH. (Source: Ref. 13.)
TM

nm were seen. Therefore, these Cotton effects seemed to originate from exciton
coupling between the two flavone chromophores.
To determine absolute stereochemistry by application of the CD exciton
chirality method, the direction and position of the transition moments in the mole-
cule must be determined. From a qualitative viewpoint, the UV bands at 324.2
and 273.0 nm may be assigned to the intrachromophoric charge transfer, and the
long-axis-polarized-transitions of the p-methoxycinnamoyl and p-methoxyben-
zoyl chromophores, respectively. On the basis of such a simple assignment of
the transition moments, the S configuration had once been deduced for biflavone
(⫺)-61 [53].
However, the actual flavone chromophore is a composite of these two moie-
ties. Furthermore, it is difficult to predict the exciton chirality between the long
axes of the two p-methoxycinnamoyl subunits, even if the transition at 324.2 nm
may be ascribed to the p-methoxycinnamoyl moiety, because the two transition
moments incline toward the inside of the molecule. It is thus not simple to deduce
the absolute stereochemistry of such a complex system by application of the CD
exciton chirality method. In such a case, the theoretical calculation of CD and
UV spectra by means of the π-electron SCF-CI-DV MO method is extremely
useful for determination of the absolute stereochemistry.
B. Theoretical Calculation of the Ultraviolet and Circular

Dichroism Spectra of the Natural Atropisomer of
Biflavone (ⴚ)-61 and Nonempirical Determination of
Absolute Stereochemistry
We adopted the biflavone 61 with the (aR) absolute configuration as the model
compound for the calculation of the CD and UV spectra. The atomic coordinates
of (aR)-61 were obtained by the molecular mechanics calculation using the
MMP2 program (Fig. 16), and the CD and UV spectral curves of (aR)-61 were
calculated by use of the π-electron SCF-CI-DV MO method [1–4]. The calculated
UV spectrum exhibited three π → π* bands: a broad and intense band at λmax
322.6 nm (ε 66,200), a weak band appearing as a shoulder around 270 nm, and
an intense band at 226.8 nm (ε 78,300) (Fig. 17). These calculated values agreed
well with the observed UV data of (⫺)-61: λmax 324.2 nm (ε 40,900), 273.0 nm
(ε 41,400), and 225.8 nm (ε 51,800) (Fig. 15 and 17). The basic pattern of the
UV spectral curve of biflavone 61 was thus well reproduced by the theoretical
calculation, with the only differences being in the relatively intensities of these
bands.
The CD spectral curve for (aR)-61 was similarly calculated to give CD
Cotton effects as shown in Fig. 17. In the region of the first UV band, a positive
Cotton effect at λext 359.7 nm (∆ε ⫹28.6) and a negative one at λext 317.5 nm
(∆ε ⫺45.0) were predicted. For the second UV band, the calculation gave a nega-
TM

Figure 16 Stereoscopic view of the stable conformation of biflavone, (aR)-4′,4⵮,7,7″-
tetra-O-methylcupressuflavone [(aR)-(⫺)-61] calculated by molecular mechanics.
(Source: Ref. 13.)
tive Cotton effect that appeared as a shoulder around 290 nm and a positive one
at λext 263.2 nm (∆ε ⫹21.7). These calculated CD bands were in excellent agree-
ment with those observed in the CD spectrum of (⫺)-61, including sign, intensity,
and position of the Cotton effects (compare Figs. 15 and 17): λext 362.0 nm (∆ε
⫹25.6), 326.2 (∆ε ⫺54.4), a negative shoulder around 300 nm, and 267.5 (∆ε
⫹21.3). Since the calculation was performed for the enantiomer with the aR con-
figuration, the absolute stereochemistry of natural biflavone, (⫺)-4′,4⵮,7,7″-tetra-
O-methylcupressuflavone [(⫺)-61], was thus determined to be aR (or M helicity)
in a nonempirical manner.
C. Circular Dichroic Power Due to the Atropisomerism of

Biflavone (aR)-61
The calculated CD and UV spectral data of (aR)-61 were analyzed in detail to
clarify the mechanism of CD and UV activity. As shown in Fig. 18, there are
10 π → π* transitions above 250 nm. The electric transitions 1, 4, 7, and 10 are
polarized along the C2-symmetrical axis, and those of transitions 2, 3, 8, and 9
are polarized perpendicular to the axis. In the region of 400–300 nm, electric
TM

Figure 17 Circular dichroism and ultraviolet spectral curves of biflavone, (aR)-
4′,4⵮,7,7″-tetra-O-methylcupressuflavone [(aR)-(⫺)-61] calculated by the π-electron SCF-
CI-DV MO method. SCF-CI-DV MO, self-consistent field/configuration interaction/
dipole velocity molecular orbital. (Source: Ref. 13.)
transitions 1, 2, 3, and 4 make dominant contributions to CD and UV spectra.

The analysis of the numerical calculation further clarifies that transitions 1 and
2 were, to a first approximation, exciton-coupled partners, as seen from the oppo-
site signs of their rotational strengths. These transitions are composed of the
transition a of the monomeric flavone chromophore 62 (Fig. 19). The transition
moment of the monomeric transition a is almost parallel to the long axis of the
p-methoxycinnamoyl moiety. For determination of the nature of the transition a,
the molecular orbital distribution was investigated.
The dominant excitation configuration contained in transition a is HOMO
(No. 12) → LUMO (No. 13), illustrated in Fig. 20. The HOMO is mainly distrib-
TM

Figure 18 Rotational and dipole strengths of the transitions of biflavone, (aR)-
4′,4⵮,7,7″-tetra-O-methylcupressuflavone [(aR)-(⫺)-61], calculated by the π-electron
SCF-CI-DV MO method. Transitions 1, 4, 7, and 10 are polarized along the C2-symmetri-
cal axis, whereas the transition moments of 2, 3, 8, and 9 are perpendicular to the axis.
SCF-CI-DV MO, self-consistent field/configuration interaction/dipole velocity molecular
orbital. (Source: Ref. 13.)
uted in the p-methoxycinnamoyl subunit. Therefore, it may be concluded that

the transitions 1 and 2 are composed of exciton coupling between the long axis–
polarized transitions of two p-methoxycinnamoyl chromophores. However, the
transitions 1 and 2 deviate from pure exciton coupling because the rotational
strength of transition 1 (R ⫽ ⫹681 ⫻ 10⫺40 cgs unit) is larger than that of transi-
tion 2 (R ⫽ ⫺525 ⫻ 10⫺40 cgs unit). In the ideal case of exciton coupling, the
rotational strengths of the two exciton transitions should be opposite in sign but
equal in absolute value to each other. The imbalance of rotational strengths pro-
duces the positive Cotton effect at 359.7 nm.
TM

Figure 19 Polarization and relative intensity of dipole strength of monomeric flavone
62 calculated by the π-electron SCF-CI-DV MO method. SCF-CI-DV MO, self-consistent
field/configuration interaction/dipole velocity molecular orbital. (Source: Ref. 13.)
It was found that transitions of 3 and 4 were also exciton-coupled partners.

These transitions are composed of the monomeric transition b, the transition mo-
ment of which inclines as shown in Fig. 19. The major excitation configuration
of transition b is MO (No. 10) → LUMO (No. 13), where MO (No. 10) is delocal-
ized over the entire monomeric flavone subunit (Fig. 20). Thus, it is difficult to
assign transition b to a specific part of the chromophore. Since the rotational
strength of the transition No. 3 (R ⫽ ⫺1041 ⫻ 10⫺40 cgs unit) is larger than that
of its exciton-coupled partner, transition No. 4 (R ⫽ ⫹899 ⫻ 10⫺40 cgs unit),
summation of the two rotational strengths gives rise to the negative Cotton effect
at 317.5 nm. It was thus clarified that the bisignate Cotton effects at 359.7 and
317.5 nm originate from the composition of two pairs of exciton couplings: one
pair originates from transitions 1 and 2, and the second pair derives from transi-
Figure 20 Molecular orbital distribution of monomeric flavone 62 calculated by the

π-electron SCF-CI-DV MO method. SCF-CI-DV MO, self-consistent field/configuration
interaction/dipole velocity molecular orbital (Source: Ref. 13.)
TM

tions 3 and 4. Therefore, it became obvious that the CD exciton chirality method
could not be applied in a straightforward way to these Cotton effects. For determi-
nation of the absolute stereochemistry of biflavone (⫺)-61, the theoretical calcu-
lation of the CD curve was thus needed.
The analysis of the negative CD shoulder at 290 nm and the positive Cotton
effect at 263.2 nm is a little simpler than the case of the Cotton effects at 359.7
and 317.5 nm. Transitions 9 and 10, which are exciton-coupled partners, are
responsible for these Cotton effects (Fig. 18). Transitions 9 and 10 are composed
of the monomeric flavone transition e, which is polarized along the x axis (Fig.
19). The main excitation configuration of transition e is MO (No. 10) → LUMO
(No. 13) (Fig. 20). Since transition e is perpendicular to the 8-8″ linkage bond,
and the two component transitions e of exciton coupling constitute a counter-
clockwise twist for the (aR) absolute configuration, exciton Cotton effects of
negative chirality would be expected. In accordance with this expectation, the
CD calculations gave a negative CD shoulder at 290 nm and a positive Cotton
effect at 263.2 nm. The analysis of CD Cotton effects around 290–260 nm thus
also supported the (aR) absolute configuration of biflavone (⫺)-61.
D. Designation of the Enantiomers of the Biflavone 61 by

Their Circular Dichroism Data
Traditionally, enantiomers are designated by the sign of optical rotation, [α]D.
For example, the designation (aR)-(⫺)-61 means that the enantiomer of 61 with
negative rotation has the (aR) absolute configuration. However, it is also well
known that [α]D values are dependent on the sample concentration and solvent
used. In some extreme cases, the sign of [α]D is known to be inverted by changing
solvents [54], as shown in Table 1; biflavone 61 is such a case. The [α]D of the
natural product 61 first measured in methanol showed a negative rotation [13].
Later it was reported by Lin et al. that the [α]D of natural 61 is positive when
measured in MeOH and EtOH [15]. Therefore, the use of the [α]D value to define
a specific enantiomer has caused confusion.
We have proposed the use of CD data to designate enantiomers in addition
to [α]D values in such cases of ambiguity [55]. For example, the designation
‘‘[CD(⫹)362.0]-(aR)-61’’ means the enantiomer of 61 showing a positive CD
Cotton effect at 362.0 nm has the (aR)-absolute configuration. The importance
of including the sign of an invariant optical property becomes apparent when the
absolute stereochemistry of a chiral molecule is unknown. This method is very
useful in the case of (⫺)-61, because the CD spectrum of 61 always exhibits a
positive CD Cotton effect around 362.0 nm irrespective of the solvent, as shown
in Table 1. Therefore, the natural biflavone is unambiguously defined as
[CD(⫹)362.0]-(aR)-61. To maintain consistency with the previous [α]D designa-
TM

Table 1 Solvent Dependence of [α]D and CD Data of Biflavone 4′,4⵮,7,7″-tetra-O-
Methylcupressuflavone {[CD(⫹)362.0]-(aR)-(⫺)-61}a
Solvent [α] D CD data

Natural sample
MeOH ⫺25.3°(c 0.3) b
EtOH λext 362.0 nm (∆ε ⫹25.6), 326.2 (⫺54.4),
267.5 (⫹21.3)b
Synthetic sample
MeOH ⫺11.3°(c 0.261) λext 362.6 nm (∆ε ⫹22.0), 326.4 (⫺51.3),
(lit. ⫹1°(c 0.2)) c 268.6 (⫹19.4), 226.8 (⫹19.4), 216.6
(⫺4.5), 207.4 (⫹11.6)
EtOH ⫹43.9°(c 0.259) λext 363.0 nm (∆ε ⫹25.5), 327.0 (⫺55.3),
269.2 (⫹21.3), 226.4 (⫹9.1), 216.8
(⫺2.6), 208.0 (⫹11.8)
(lit. ⫹77°(c 0.2))c (lit. λext 360 nm (∆ε ⫹28.1), 324 (⫺66.4),
267 (⫹24.8))c
CHCl3 ⫹77.0°(c 0.257) λext 361.0 nm (∆ε ⫹31.5), 326.2 (⫺64.0),
269.8 (⫹24.7)
c: concentration.
a
From Ref. 18. The sign of optical rotation was taken from the [α]D value of the natural sample in
MeOH.
b
From Ref. 13.
c
From Ref. 15.
tion, we adopt both methods here; the natural product (⫺)-61 is thus defined
as [CD(⫹)362.0]-(aR)-(⫺)-4′,4⵮,7,7″-tetra-O-methylcupressuflavone. Since it is
also often useful to include P and M designations for these enantiomers, the (aR)
enantiomer has M helicity and the (aR) isomer has P helicity.
E. Total Synthesis of [CD(ⴙ)362.0]-(aR )-(ⴚ)-4′,4ⵯ,7,7″-

Tetra-O-Methylcupressuflavone [(aR )-(ⴚ)-61]
1. Synthesis of (⫾)-3,3′-Diacetyl-4,4′,6,6′-
Tetramethoxybiphenyl-2,2′-diol [65] by the Solid-State
Phenol Coupling Reaction
As a chiral synthetic building block for the total synthesis of (⫺)-61, we selected
3,3′-diacetyl-4,4′,6,6′-tetramethoxybiphenyl-2,2′-diol (65) (Scheme 14). To be-
gin the synthesis of biphenyldiol 65, phloroacetophenone (63) [56] was selec-
tively methylated with dimethyl sulfate and K2CO3 in acetone, giving 2,4-dimeth-
ylphloroacetophenone (64) in 92% yield [57]. Unfortunately, the phenol coupling
TM

Scheme 14 Synthesis and resolution of biphenyl (⫾)-65.
reaction of 64 using typical oxidants (FeCl3, (t-Bu)2O2, and others) in various

solvents was completely unsuccessful.
We subsequently turned to the solid-state reaction conditions reported by
Toda and coworkers [58]: heating a mixture of phenol 64 and FeCl3 (1.0 eq.) to
50°C for 12 h produced the desired coupled product 65 in 27% yield (Scheme
14). When phenol 64 was heated with excess FeCl3 (5.2 eq.) at 50°C for only
5.5 h, aryl chloride 66 was obtained in 33% yield, but no desired coupled product
65 was detected. We explored the reaction conditions further and found that the
solid-state phenol coupling reaction in the presence of FeCl3 /silica gel [59] gave
the best yield of biphenyldiol 65. When a mixture of phenol 64 and 2.8 equivalent
TM

of FeCl3 /silica gel (1:2) was heated at 43–45°C for 6 days, the desired coupled
product (⫾)-65 was produced in 81% yield.
2. Enantioresolution of (⫾)-3,3′-Diacetyl-4,4′,6,6′-
Tetramethoxybiphenyl-2,2′-diol as the bis(Camphanate)
Diesters 68 and X-Ray Crystallographic Analysis
Resolution of (⫾)-65 using the recently developed chiral dichlorophthalic acid
method [60–62] was unsuccessful, presumably because of the steric hindrance
around the phenol groups of 65, which prevented acylation. However, application
of the camphanate method to 65 (Scheme 14) using (1S)-(⫺)-camphanic acid
chloride [63] and 4-dimethylaminopyridine in pyridine (refluxed for 3 days)
yielded a mixture of diastereomeric diesters 68a and 68b (ca. 700 mg), which
were separated by HPLC on silica gel (CH3CN/CHCl3 1 :19, separation factor
α ⫽ 1.18, resolution factor Rs ⫽ 1.62) to give the faster-eluting (⫹)-68a (42%)
followed by (⫺)-68b (46%).
Recrystallization of diester 68b from hexane/EtOAc (1:1) gave large, col-
orless single crystals suitable for x-ray diffraction: orthorhombic, space group
P212121; R ⫽ 0.0974 and Rw ⫽ 0.1061 [19]. It was found that the crystal contained
EtOAc molecules as crystalline solvent, the crystal structure of which deviated
from the ideal structure of EtOAc. Therefore, the position of hydrogen atoms
could not be determined and the final R value remained at 9.74%. Nevertheless,
the absolute stereochemistry of the biphenyl portion of diester 68b could be un-
ambiguously assigned as (aR) by the internal reference method using the known
absolute configuration of the camphanate ester moieties.
It was difficult to determine the structure of the EtOAc molecules contained
as the crystalline solvent, and hence the R value remained at a high level. The
main reason for this difficulty was presumably the high volatility of EtOAc (bp
76.8°C), which may vaporize during the x-ray diffraction experiment. To elimi-
nate this problem and obtain a more precise molecular structure, we attempted
to replace EtOAc with another, higher-boiling ester solvent yet with a molecular
size still similar to that of EtOAc such as propyl acetate (PrOAc, bp 101.6°C).
Although PrOAc was not incorporated as a crystalline solvent, it was exciting
to find that the large prismatic crystals obtained by recrystallization from PrOAc
were composed of 68b only and were single crystals suitable for x-ray diffraction.
A single crystal of 68b obtained by recrystallization from propyl acetate
was subjected to x-ray analysis: colorless prism, monoclinic, space group P21;
R ⫽ 0.0346 and Rw ⫽ 0.0409 [18]. The absolute stereochemistry of the biphenyl
moiety of diester 68b could be unambiguously determined to be (aR) by the
internal reference method using the known absolute configuration of the campha-
nate ester moieties (Fig. 21). The absolute configuration of diester 68a was there-
fore determined to be (aS).
TM

Figure 21 ORTEP drawing of diester (aR)-(⫺)-68b. The atoms are drawn as 50% prob-
ability ellipsoids. (Source: Refs. 18 and 19.)
3. Completion of the Synthesis of Biflavone [CD(⫹)362.0]-

(aR)-(⫺)-61
Completion of the enantioselective total synthesis of the natural enantiomer of
biflavone 61 then commenced with chiral biphenyldiol (aR)-65 obtained from
diester (⫺)-68b after acidic hydrolysis (Scheme 15). When diester 68b was
treated with aqueous LiOH in EtOH, deacetyl products were obtained instead of
65. Therefore, diester (⫺)-68b was heated in aqueous 6 M HCl/EtOH to 78°C for
18 h, affording optically active biphenyldiol 65 (64% yield), whose CD spectrum
showed a negative CD Cotton effect at 301.2 nm. Although we had expected to
obtain enantiopure biphenyldiol (aR)-65, it was found that 65 had partially race-
mized during the acid-catalyzed hydrolysis (79% ee). The enantiomeric excess
of biphenyldiol 65 was determined by 1H NMR spectroscopy, using the chiral
shift reagent europium tris[3-(trifluoromethylhydroxymethylene)-(⫹)-camphor-
ate] [Eu(tfc)3].
Biphenyldiol (aR)-(⫺)-65 was next converted to bichalcone (aR)-(⫺)-69
(Scheme 15) by heating (70–75°C) with 4-anisaldehyde in 10% aqueous KOH/
TM

Scheme 15 Synthesis of biflavone [CD(⫹)362.0]-(aR)-(⫺)-61.
EtOH overnight, giving (aR)-(⫺)-69 in 65% yield. Bichalcone (aR)-(⫺)-69 was

then cyclized with iodine in the presence of a catalytic amount of H2SO4 in DMSO
[31] to afford 4′,4⵮,5,5″,7,7″-hexa-O-methylcupressuflavone [(aR)-(⫹)-70] in
52% yield. Finally, hexamethyl derivative (aR)-(⫹)-70 was selectively demethy-
lated with BCl3 in CH2Cl2, giving the desired natural product (aR)-(⫺)-61 in 80%
yield (78% ee). In the course of this work it was noticed that racemic biflavone
61 was barely soluble in MeOH. Therefore, the enantiomerically enriched bifla-
vone (aR)-61 (78% ee) was recrystallized from methanol, providing enantiomer-
ically pure biflavone (aR)-61. The enantiopurity was confirmed by 1H NMR spec-
TM

troscopy after conversion to the (S)-MTPA ester (aR)-71. The physical data of
100% ee biflavone 61 were mp 147–149°C (MeOH), [α]D22 ⫺11.3° (c 0.261,
MeOH), and CD (EtOH) λext 363.0 nm (∆ε ⫹25.5).
4. Experimental Verification of the Absolute Stereochemistry of

Biflavone [CD(⫹)362.0]-(aR)-(⫺)-61 Theoretically
Determined by the Total Synthesis of Its
Natural Enantiomer
As discussed in the examples of (⫹)-1,8a-dihydro-3,8-dimethylazulene [(⫹)-1]
and the chiral halenaquinol compounds, the theoretically predicted absolute ste-
Figure 22 Circular dichroism and ultraviolet spectra of the synthetic sample of biflav-
one,4′,4⵮,7,7″-tetra-O-methylcupressuflavone {[CD(⫹)362.0]-(aR)-(⫺)-61]} in EtOH.
(Source: Ref. 18.)
TM

reochemistry of [CD(⫹)362.0]-(aR)-(⫺)-4′,4⵮,7,7″-tetra-O-methylcupressufla-
vone [(aR)-(⫺)-61] was confirmed by enantioselective total synthesis. The case
of biflavone [CD(⫹)362.0]-(aR)-(⫺)-61 was rendered more intriguing, though,
since it had once been claimed that the absolute configuration of (⫺)-61 deter-
mined by x-ray crystallography disagreed with our theoretical prediction dis-
cussed earlier [15]. However, this claim was later retracted [16,17].
The CD and UV spectra of synthetic biflavone (aR)-(⫺)-61 are illustrated
in Fig. 22, and the CD curve is identical to that of the natural sample of 61 shown
in Fig. 15. Therefore, the absolute stereochemistry of the natural atropisomer,
[CD(⫹)362.0]-(⫺)-4′,4⵮,7,7″-tetra-O-methylcupressuflavone [(⫺)-61], was de-
termined to be (aR). This conclusion is consistent with our previous theoretical
determination of the absolute stereochemistry of biflavone [CD(⫹)362.0]-(⫺)-
61 by the molecular orbital calculation of CD spectra.
VI. CONCLUSIONS
We have established the absolute stereostructures of natural, chiral dihydroazu-

lenes, novel marine natural products of the halenaquinol family with a new penta-
cyclic skeleton, and the natural atropisomeric biflavone (⫺)-61 by theoretical
calculation of CD spectra using the π-electron SCF-CI-DV MO method. The
calculated CD data were in excellent agreement with the observed CD data. We
have also clarified the mechanism of their CD Cotton effects by analyzing their
circular dichroic power, which made the theoretical determination of the absolute
stereostructures more reliable. The absolute stereostructures of natural products
theoretically predicted were experimentally confirmed by the total syntheses of
these natural products and pertinent model compounds. This theoretical method-
ology has thus become a promising tool for the determination of the absolute
stereochemistry of various optically active compounds, including many natural
products with twisted π-electron systems.
REFERENCES
1. A. Moscowitz, Tetrahedron, 13: 48 (1961).

2. (a) C. M. Kemp and S. F. Mason, Tetrahedron, 22: 629 (1966). (b) A. Brown,
C. M. Kemp, and S. F. Mason, J. Chem. Soc. A, 751 (1971).
3. N. Harada and K. Nakanishi, Circular Dichroic Spectroscopy: Exciton Coupling in
Organic Stereochemistry, University Science Books, Mill Valley, Calif., and Oxford
University Press, Oxford (1983).
4. N. Harada, in Circular Dichroism, Principles and Applications (K. Nakanishi, N.
Berova, and R. W. Woody, eds), VCH Publishers, New York, p. 335 (1994).
TM

5. N. Harada, J. Kohori, H. Uda, K. Nakanishi, and R. Takeda, J. Am. Chem. Soc.,
107: 423 (1985).
6. N. Harada, J. Kohori, H. Uda, and K. Toriumi, J. Org. Chem., 54: 1820 (1989).
7. M. Kobayashi, N. Shimizu, I. Kitagawa, Y. Kyogoku, N. Harada, and H. Uda, Tetra-
hedron Lett., 26: 3833 (1985).
8. N. Harada, H. Uda, M. Kobayashi, N. Shimizu, and I. Kitagawa, J. Am. Chem. Soc.,
111: 5668 (1989).
9. N. Harada, T. Sugioka, Y. Ando, H. Uda, and T. Kuriki, J. Am. Chem. Soc., 110:
8483 (1988).
10. N. Harada, T. Sugioka, H. Uda, and T. Kuriki, J. Org. Chem., 55: 3158 (1990).
11. N. Harada, T. Sugioka, H. Uda, T. Kuriki, M. Kobayashi, and I. Kitagawa, J. Org.
Chem., 59: 6606 (1994).
12. N. Harada, T. Sugioka, T. Soutome, N. Hiyoshi, H. Uda, and T. Kuriki, Tetrahedron:
Asymmetry, 6: 375 (1995).
13. N. Harada, H. Ono, H. Uda, M. Parveen, N. U.-D. Khan, B. Achari, and P. K. Dutta,
J. Am. Chem. Soc., 114: 7687 (1992).
14. M. Ilyas, J. N. Usmani, S. P. Bhatnagar, M. Ilyas, W. Rahman, and A. Pelter, Tetra-
hedron Lett., 5515 (1968).
15. F.-J. Zhang, G.-Q. Lin, and Q.-C. Huang, J. Org. Chem., 60: 6427 (1995).
16. F.-J. Zhang, G.-Q. Lin, and Q.-C. Huang, J. Org. Chem., 61: 5700 (1996).
17. G.-Q. Lin and M. Zhong, Tetrahedron Lett., 38: 1087 (1997).
18. H.-Y. Li, T. Nehira, M. Hagiwara, and N. Harada, J. Org. Chem., 62: 7222 (1997).
19. N. Harada, H.-Y. Li, T. Nehira, and M. Hagiwara, Enantiomer, 2: 353 (1997).
20. N. Harada, H. Uda, T. Nozoe, Y. Okamoto, H. Wakabayashi, and S. Ishikawa, J.
Am. Chem. Soc., 109: 1661 (1987).
21. N. Harada, J. Iwabuchi, Y. Yokota, and H. Uda, Croatica Chem. Acta, 62: 267
(1989).
22. D. Gargiulo, F. Derguini, N. Berova, K. Nakanishi, and N. Harada, J. Am. Chem.
Soc., 113: 7046 (1991).
23. N. Berova, D. Gargiulo, F. Derguini, K. Nakanishi, and N. Harada, J. Am. Chem.
Soc., 115: 4769 (1993).
24. N. Harada, N. Hiyoshi, and K. Naemura, Recl. Trav. Chim. Pays-Bas, 114: 157
(1995).
25. N. Harada, A. Saito, N. Koumura, H. Uda, B. de Lange, W. F. Jager, H. Wynberg,
and B. L. Feringa, J. Am. Chem. Soc., 119: 7241 (1997).
26. N. Harada, A. Saito, N. Koumura, D. C. Roe, W. F. Jager, R. W. J. Zijlstra, B. de
Lange, and B. L. Feringa, J. Am. Chem. Soc., 119: 7249 (1997).
27. N. Harada, N. Koumura, and B. L. Feringa, J. Am. Chem. Soc., 119: 7256 (1997).
28. N. L. Allinger, QCPE, 11: 318 (1976); QCPE, Program No. 318.
29. MOPAC 93: J. J. P. Stewart, Fujitsu Limited, Tokyo, Japan (1993).
30. N. Harada, the computer program for CD and UV calculation, CD/NH-Sendai
(1978).
31. N. Mataga and K. Nishimoto, Z. Physik. Chem., 13: 140 (1957).
32. R. Takeda and K. Katoh, J. Am. Chem. Soc., 105: 4056 (1983).
33. M. Kobayashi, B. W. Son, M. Kido, Y. Kyogoku, and I. Kitagawa, Chem. Pharm.
Bull., 31: 2160 (1983).
TM

34. B. F. Bowden, J. C. Coll, and D. M. Tapiolas, Aust. J. Chem., 36: 211 (1983).
35. R. R. Izac, W. Fenical, and J. M. Wright, Tetrahedron Lett., 25: 1325 (1984).
36. M. Kobayashi, B. W. Son, Y. Kyogoku, and I. Kitagawa, Chem. Pharm. Bull., 32:
1667 (1984).
37. N. Harada, T. Sugioka, H. Uda, and T. Kuriki, Synthesis, 53 (1990).
38. N. Harada, T. Sugioka, H. Uda, and T. Kuriki, Collect. Czech. Chem. Commun.,
57: 1459 (1992).
39. (a) Z. G. Hajos and D. R. Parrish, J. Org. Chem., 39: 1615 (1974). (b) U. Eder, G.
Sauer, and R. Wiechert, Angew. Chem., 83: 492 (1971); Angew. Chem. Int. Ed.
Engl., 10: 496 (1971). (c) J. Gutzwiller, P. Buchschacher, and A. Furst, Synthesis,
167 (1977). (d) P. Buchschacher and A. Furst, Org. Synth., 63: 37 (1986). (e) P.
Buchschacher, A. Furst, and J. Gutzwiller, Org. Synth. Coll. 7: 368 (1990).
40. D. M. Roll, P. J. Scheuer, G. K. Matsumoto, and J. Clardy, J. Am. Chem. Soc., 105:
6177 (1983).
41. M. Kobayashi, N. Shimizu, Y. Kyogoku, and I. Kitagawa, Chem. Pharm. Bull., 33:
1305 (1985).
42. H. Nakamura, J. Kobayashi, M. Kobayashi, Y. Ohizumi, and Y. Hirata, Chem. Lett.,
713 (1985).
43. F. J. Schmitz and S. J. Bloor, J. Org. Chem., 53: 3922 (1988).
44. N. Harada, K. Nakanishi, and S. Tatsuoka, J. Am. Chem. Soc., 91: 5896 (1969).
45. N. Harada and K. Nakanishi, Accounts Chem. Res., 5: 257 (1972).
46. A. L. Gemal and J. L. Luche, J. Am. Chem. Soc., 103: 5454 (1981).
47. N. Harada and T. Sugioka, in Studies in Natural Products Chemistry (A. Rahman,
ed.), Elsevier, Amsterdam, p. 33 (1995).
48. L. F. Fieser and M. Fieser, Reagents for Organic Synthesis, Vol. 1, John Wiley &
Sons, New York, p. 921 (1967).
49. A. C. Baille and R. H. Thomson, J. Chem. Soc. (C), 2184 (1966).
50. G. Stork and P. D’Angelo, J. Am. Chem. Soc., 96: 7114 (1974).
51. W. G. Salmond, M. A. Barta, and J. L. Havens, J. Org. Chem., 43: 2057
(1978).
52. M. Oda and Y. Kanao, Chem. Lett., 37 (1981).
53. M. Parveen, N. U. Khan, B. Achari, P. K. Dutta, Abstracts of Papers, 3rd Interna-
tional Symposium on Flavonoids in Biology and Medicine, Singapore, Abstract p. 22
(Nov. 13–17, 1989).
54. (a) D. H. S. Horn and Y. Y. Pretorius, J. Chem. Soc., 1460 (1954). (b) A. Horeau,
Tetrahedron Lett., 3121 (1969). (c) Y. Kumata, J. Furukawa, and T. Fueno, Bull.
Chem. Soc. Jpn., 43: 3920 (1970). (d) D. Parker Chem. Rev., 91: 1441 (1991).
55. (a) N. Harada, J. Iwabuchi, Y. Yokota, H. Uda, Y. Okamoto, H. Yuki, and Y. Ka-
wada, J. Chem. Soc. Perkin. Trans. 1, 1845 (1985). (b) N. Harada, Enantiomer, 1:
81 (1996).
56. K. C. Gulati, S. R. Seth, and K. Venkataraman, in Organic Syntheses (A. H. Blatt,
ed), John Wiley & Sons, New York, Vol. II, p. 522 (1966).
57. K. Nakazawa, Chem. Pharm. Bull., 10: 1032 (1962).
58. F. Toda, K. Tanaka, and S. Iwata, J. Org. Chem., 54: 3007 (1989).
59. (a) K. Keinan and Y. Mazur, J. Org. Chem., 43: 1020 (1978). (b) T. C. Jempty,
L. L. Miller, and Y. Mazur, J. Org. Chem., 45: 751 (1980).
TM

60. (a) N. Harada, N. Hiyoshi, V. P. Vassilev, and T. Hayashi, Chirality, 9: 623 (1997).
(b) N. Harada, V. P. Vassilev, and N. Hiyoshi, Enantiomer, 2: 123 (1997).
61. (a) N. Harada, T. Soutome, S. Murai, and H. Uda, Tetrahedron Asymmetry, 4: 1755
(1993). (b) N. Harada, T. Hattori, T. Suzuki, A. Okamura, H. Ono, S. Miyano, and
H. Uda, Tetrahedron Asymmetry, 4: 1789 (1993). (c) N. Harada, T. Soutome, T.
Nehira, H. Uda, S. Oi, A. Okamura, and S. Miyano, J. Am. Chem. Soc., 115: 7547
(1993). (d) T. Hattori, N. Harada, S. Oi, H. Abe, and S. Miyano, Tetrahedron Asym-
metry, 6: 1043 (1995). (e) H. Hagiwara, T. Okamoto, N. Harada, and H. Uda, Tetra-
hedron, 51: 9891 (1995).
62. (a) N. Harada, T. Nehira, T. Soutome, N. Hiyoshi, and F. Kido, Enantiomer, 1: 35
(1996). (b) N. Harada, N. Koumura, and M. Robillard, Enantiomer, 2: 303 (1997).
(c) H. Nemoto, J. Miyata, K. Fukumoto, H. Ehara, N. Harada, H. Uekusa, and Y.
Ohashi, Enantiomer, 2: 127 (1997).
63. W. L. Meyer, A. P. Lobo, and R. N. McCarty, J. Org. Chem., 32: 1754 (1967).
TM

7
Recent Applications of Circular
Dichroism to Carbohydrate
Conformational Analysis and Direct
Determination of Drug Levels
Jesús Trujillo Vázquez

Universidad de La Laguna, Tenerife, Spain
This chapter is intended to show that circular dichroism (CD), although a well-
recognized spectroscopic technique, is still being developed, and that new appli-
cations are yet to be discovered. In spite of the existence of many useful empirical
rules and of the nonempirical CD exciton chirality method for determination of
absolute configuration and conformational analysis, this technique remains to be
developed fully.
Two recent CD applications are described: (1) First is the conformational
analysis of carbohydrates, exemplified by the CD and 1 H-nuclear magnetic reso-
nance (1 H-NMR) study of the rotational population dependence of the hydroxy-
methyl group in alkyl glucopyranosides on the aglycon and its absolute con-
figuration, a clear correlation between the rotamer distributions and the
stereoelectronic exo-anomeric effect being observed. As a consequence of these
results the absolute configuration of secondary alcohols can be determined by
either CD or 1 H NMR, as a single enantiomer is necessary for this purpose;
(2) The direct determination of drug levels in pharmaceutical formulations (oral
suspensions, injections, and capsules), as well as in human biological fluids
(urine and serum), with specific reference to the β-lactam antibiotics, is de-
scribed.
TM

I. INTRODUCTION
Circular dichroism (CD) is a powerful technique that has been mainly used for
the determination of absolute configuration of a great number of compounds of
both natural and synthetic origin. This technique has its origins in the 1960s
through important contributions by scientists such as C. Djerassi, P. Crabbé, and
G. Snatzke, who introduced many effective empirical rules for absolute configu-
ration determinations of organic compounds [1]. Further developments came with
the exciton chirality method [2], mainly through the seminal scientific contribu-
tions of N. Harada and K. Nakanishi. This method, which allows the determina-
tion of absolute configurations in a nonempirical way, offers great analytical po-
tential and, in my opinion, establishes a second era for CD. Applications of
circular dichroism have also been widely used in the conformational studies in
solution of small to medium size molecules, as well as macromolecules, espe-
cially proteins [3].
The development of the CD exciton chirality method [2] has extended the
utility of this technique for determining the absolute configuration of organic
compounds. The high sensitivity and straightforward spectral interpretation of
this method, as a consequence of the general validity of the pairwise additivity
in exciton-coupled systems [4], have not been fully exploited. CD is still the
technique that is least known and used by most organic chemists.
The CD exciton chirality method is based on the interaction through space
of the electric transition moments of two chromophores, which gives rise to an
excited state split into two energy levels. Excitations to these levels lead to a CD
spectrum with two Cotton effects of opposite signs, namely, to a ‘‘split’’ CD
curve [2]. The chiral environment of the two chromophores determines the sign
of the Cotton effects; the sign of the exciton chirality is that of the first Cotton
effect, the one at longer wavelength. Furthermore, the existence of an additivity
relation in multiple-chromophoric systems [4] allows an easy spectral interpreta-
tion of this type of compound, since the observed CD spectrum is the sum of
CDs arising from all pairwise interactions present in the system: namely, the CD
spectrum of a system composed of three interacting chromophores XYZ is the
sum of the CDs arising from X/Y, X/Z, and Y/Z interactions. The general valid-
ity of the pairwise additivity in exciton-coupled systems is valid independently
of whether the interacting chromophores are the same or not [4].
In the belief that CD is a spectroscopic technique that offers a series of
analytical advantages over other techniques, our group led the research to apply
the CD exciton chirality method to the conformational analysis of carbohydrates,
namely, to study the rotational population dependence of the hydroxymethyl
group in alkyl glycopyranosides on the structural nature of the aglycon and its
absolute configuration.
TM

CD has proved to be an extremely sensitive technique to carry out confor-
mational analysis, and many studies have been performed using this approach [1].
II. CARBOHYDRATE CONFORMATIONAL ANALYSIS
Recognition of the biological importance of carbohydrates has increased dramati-

cally over the last two decades, as a result of the fact that they possess many
biological functions. In cells, carbohydrates are normally attached to proteins or
lipids, forming glycoproteins or glycolipids, and it has been clearly demonstrated
that the carbohydrate moiety has an enormous effect on the properties of these
glycoconjugates. Glycoproteins are involved in many processes, such as cell sur-
face recognition, immune defense, cell growth, and cell differentiation [5].
As a consequence of the extremely important biological role of glycoconju-
gates, and taking into account that their oligosaccharides normally define the
biological function of these compounds, any contribution to a better knowledge
of the conformational properties of these sugar moieties would help to understand
how these macromolecules interact with other biomolecules.
Most of the oligosaccharides involved in glycoconjugates are made up of
hexopyranoses. The overall conformation is determined by the torsion angles (φ
and ψ) around the glycosidic linkages [6]. When (1–6) linkages are involved,
the corresponding torsion angle about C5–C6 bonds (ω), the preferred rotamers
around the C5–C6 bond being the gauche-gauche (gg), the gauche-trans (gt),
and the trans-gauche (tg) (Fig. 1) must be considered.
Figure 1 Torsion angles about the glycosidic linkage and around the C5–C6 bond (top);
Newman projections of the gg, gt, and tg rotamers around the C5–C6 bond (bottom).
TM

The conformational analysis of carbohydrates is not a new application of
CD, as may be inferred from the title. The CD exciton chirality method has
already been successfully used in the conformational study of the 2-(N-acetyl-
p-bromobenzamido) group in 2-amino-2-deoxy-galactopyranosides [7]; NMR
and CD techniques are a good tandem for this type of study.
An attempt will be made later to demonstrate this last point, by showing
the CD and 1 H NMR study of the rotational population dependence of the hydro-
xymethyl group in alkyl glucopyranosides on the structural nature of the aglycon
and its absolute configuration. Furthermore, the study revealed a clear correlation
between the rotamer distributions and the stereoelectronic exo-anomeric effect
and also revealed that as a consequence of these results the absolute configuration
of secondary alcohols can be determined, as a single enantiomer is necessary for
this purpose.
The studies on the additivity principle in circular dichroism performed with
hexopyranosides have shown that the rotational population of the hydroxymethyl
group at C6 depends mainly on the configuration of the substituent at C4 (axial
or equatorial) [8], and also on the bulkiness of this substituent [4a]. Recently,
we have reported, on the basis of 1 H NMR and CD data [9], that the rotational
population of the hydroxymethyl group in chiral and nonchiral alkyl β-d-gluco-
pyranosides also depends on the aglycon and its absolute configuration. It was
observed that the population of the gg and gt rotamers decreased and increased,
respectively, as the pK a of the aglycon increased, whereas the tg population re-
mained almost constant.
Fig. 2 shows the easily interpretable CD spectra of different nonchiral alkyl
2,3-bis-O-( p-bromobenzoyl)-4,6-bis-O-( p-methoxycinnamoyl)-β-d-glucopyran-
osides. This bichromophoric system is composed of two homo interactions, one
benzoate–benzoate (2B/3B), centered about the p-bromobenzoate λmax 245 nm,
and one cinnamate–cinnamate (4C/6C), centered about the p-methoxycinnamate
λ max 311 nm, as well as four hetero interactions (2B/4C, 2B/6C, 3B/4C, 3B/6C),
which cover both the benzoate and cinnamate regions and are weaker in intensity
than the homo interactions. The spectral differences were located in the cinna-
mate-cinnamate coupling region centered about the cinnamate λ max 311 nm, as
can be observed in the dotted box in Fig. 2 (left) or in its expanded box in the
same figure (right). The intensities of the first and second Cotton effects of the
CD spectra of the glucosylated nonchiral alcohols 1a–e gradually decreased from
methyl (A C value 36.6) [10], to ethyl (35.4), to iso-propyl (31.3), to cyclohexyl
(31.3), and to tert-butyl (23.6) glucopyranoside derivatives. These differences in
the CD curves clearly indicated that only the interactions involving the chromo-
phore at the 6 position are significantly affected, mainly the homo cinnamate–
cinnamate (4C/6C) pairwise interaction, and, therefore, that the rotamer distribu-
tions have changed.
TM

Figure 2 Circular dichroism spectra (CH3CN) of model alkyl 2,3-bis-O-( p-bromoben-
zoyl)-4,6-bis-O-( p-methoxycinnamoyl)-β-d-glucopyranosides: 400- to 200-nm region
(left), and 350- to 264-nm region (right).
Moreover, these spectral differences are only consistent with a gradual de-
crease and an increase in the population of the gg and gt rotamers, respectively,
as can be deduced by analyzing the pairwise interactions involved in these com-
pounds. Fig. 3 shows for the glucopyranosyl system those pairwise interactions
involving the chromophore at the 6 position, the 2/6, the 3/6, and the 4/6 interac-
tion, for the three rotamers. As can be observed, those interactions involving the
gg rotamer have a net positive contribution; those involving the gt rotamer have
a net negative contribution; and those interactions involving the tg rotamer have
a nil contribution. Therefore, the general decreases in the CD Cotton effects can
be explained by a decrease in the population of the gg rotamer (net positive
contribution) and an increase in the population of the gt rotamer (net negative
contribution). The other three existing pairwise interactions, the positive 2/3, the
negative 3/4, and the nil 2/4, having constant intensities and signs, do not affect
this CD interpretation. Furthermore, the striking increase in the first positive and
second negative Cotton effects observed on these spectra by lowering the temper-
ature (MeOH, ⫺80°C) can only be explained by an increase in the rotational
population of the energetically favored gg rotamer.
Rotamer distribution is normally determined by 1 H NMR analysis of the
coupling constants of the prochiral protons at C6 [11], which are assigned on
TM

Figure 3 Pairwise interactions involving the chromophore at the 6 position in each of
the three stable rotamers. (Source : Ref. 9.)
the basis of their chemical shifts. That is, the coupling constant values of the
usually well-differentiated doublet of doublets for H6R and H6S are substituted
in empirical equations to calculate the rotamer distribution [12].
With the exception of the cyclohexyl derivative 1d, a doublet, rather than
the usual clearly differentiated doublet of doublets for each proton at C6, was
obtained for the bichromophoric compounds. Therefore, their rotamer distribu-
tions could not be calculated by means of 1 H NMR coupling constants. However,
the coupling constant J H5,H6, which gradually increased from methyl (3.9 Hz), to
ethyl (4.1 Hz), to iso-propyl (4.4 Hz), and to tert-butyl (4.6 Hz) glucopyranoside
derivatives (1a–c,e), corroborated the increase in the gt population.
Since the results show that the gt population increases as the pK a of the
aglycon increases, and it is accepted that the stereoelectronic exo-anomeric effect
[13] increases with increasing ease for charge delocalization from the aglycon
to the anomeric carbon [14], this stereoelectronic effect must be responsible for
the rotational population dependence of the hydroxymethyl group on the aglycon,
because for these low-molecular-size alcohols the existence of nonbonded inter-
actions with the chromophore at C6 cannot be expected.
This correlation was confirmed by the CD and NMR data comparison of
TM

the methyl and the acetyl tetra-O-p-bromobenzoyl-β-d-glucopyranosides, com-
pounds 2a and 2b, respectively (Scheme 1). The spectrosopic data (Table 1)
showed a higher gg population for the acetyl glucopyranoside derivative than
that of the methyl glucopyranoside 2a, in agreement with the nil exo-anomeric
effect reported for the acetyl glucopyranoside 2b [15].
Rotational studies performed with nonchiral alkyl α-d-glucopyranosides
[16] and β-d- and α-d-galactopyranosides [17] have revealed that the gt popula-
tion also increases as the pK a of the aglycon increases; this behavior seems to
be general in hexopyranosides.
Spectroscopic analysis of chiral alkyl 2,3,4,6-tetrakis-O-( p-bromoben-
zoyl)-β-d-glucopyranosides 2c–h (Scheme 1) showed another correlation, that
is, between the rotational population of the hydroxymethyl group and the absolute
configuration of the aglycon. Higher CD A values and smaller J H5,H6R coupling
constants were obtained for the R-alkyl glucopyranosides than for their S-alkyl
glucopyranoside counterparts.
CD spectra of these glucosylated secondary alcohols showed a general de-
crease in the positive first and negative second Cotton effects with respect to
those of the methyl derivative 2a (Table 1), as occurred with the glucosylated
Scheme 1 Structures of model alkyl β-d-glucopyranosides.
TM

Table 1 Circular Dichroism Data (CH 3CN), J H5,H6 Coupling Constants (CDCl 3, 400
MHz), and Calculated Rotameric Populations (Percentage) around the C5-C6 Bond for
Model 2,3,4,6-Tetrakis-O-(p-bromobenzoyl)-β-d-glucopyranosides 2a–h
∆ε at
Compd Aglycon Cl′ 250/232 nm A Value J H5,H6R J H5,H6S P gg P gt P tg
2a Methyl — 23.7/–6.2 29.9 4.7 3.4 57 26 17
2b Acetyl — 26.2/–5.1 31.3 4.7 3.0 59 28 13
2c (⫺)-2-Octyl R 18.6/–5.0 23.6 5.1 3.6 53 29 18
2d (⫹)-2-Octyl S 17.8/–4.6 22.4 5.3 3.6 50 32 18
2e (⫺)-Menthyl R 17.7/–4.1 21.8 5.2 3.6 51 31 18
2f (⫹)-Menthyl S 9.5/–0.5 10.0 6.1 3.4 45 40 15
2g (⫺)-Bornyl R 17.2/–4.8 22.0 5.2 3.7 51 30 19
2h (⫹)-Bornyl S 16.8/–3.3 20.1 5.4 3.6 50 32 18
nonchiral secondary alcohols 1c–d with respect to 1a. Moreover, CD spectra of

those stereoisomers having an R absolute configuration at the aglyconic carbon
(C1′) (2c, 2e, and 2g) showed higher amplitudes (A values) than those obtained
from stereoisomers with the opposite absolute configuration (2d, 2f, and 2h).
Fig. 4 shows the CD spectra of methyl, (⫹)- and (⫺)-2-octyl, and (⫹)- and (⫺)-
menthyl 2,3,4,6-tetrakis-O-( p-bromobenzoyl)-β-d-glucopyranosides (2a,c–f, re-
spectively); their spectral differences are analyzed on the basis of their pairwise
interactions (Fig. 3).
In agreement with CD data, analysis of the 1 H NMR coupling constants
Figure 4 Circular dichroism spectrum (CH 3CN) of methyl, (⫹)- and (⫺)-2-octyl (left),
and (⫹)- and (⫺)-menthyl (right) 2,3,4,6-tetrakis-O-( p-bromobenzoyl)-β-d-glucopyrano-
sides. (Source : Ref. 9.)
TM

J H5,H6 of these chiral alkyl glucopyranosides showed smaller J H5,H6R coupling con-
stants for those stereoisomers having an R absolute configuration at the aglyconic
carbon than for their glucopyranoside counterparts (Table 1).
Circular dichroism and 1 H NMR data comparison indicated the existence
of an excellent correlation between the magnitudes of the rotamer populations
obtained by 1 H NMR and the CD A values. Furthermore, analysis of these spec-
troscopic data established that those stereoisomers having an R absolute configu-
ration at the aglyconic carbon possess higher gg and smaller gt populations than
their stereoisomers with the opposite absolute configuration.
The differences between each pair of stereoisomers may be due to steric
interactions, since the pK a of the enantiomers bonded is the same. There are two
possible explanations: (1) The first is the existence of nonbonded interactions
between the aglycon and the chromophore at C6. In agreement with the rotational
dependence of the torsion angle Ψ on the structure of the aglycon [6] higher
nonbonded interactions can be expected between aglycons having the bulkiest
substituent syn to O5 (2c,e, and 2g) and the chromophore at C6 in the gt confor-
mation than for those stereoisomers having the opposite configuration at the agly-
conic carbon (2d, 2f, and 2h) (Fig. 5). (2) There exists a more stabilizing exo-
anomeric effect for the diastereoisomers with an S absolute configuration at the
aglyconic carbon, those with a higher gt population. Since the study performed
with the glucosylated nonchiral alcohols revealed that a higher gt population
correlates with a higher value of the exo-anomeric effect, these diastereoisomers
may have a more favored disposition of the orbitals involved in this stereoelec-
tronic effect [13] than their corresponding diastereoisomers.
Chiral alkyl galactopyranosides [17] have shown behavior similar to that
of alkyl glucopyranosides.
Figure 5 Schematic representation of the rotational dependence of the torsion angle ψ

on the structure of the aglycone, and corresponding Newman projections.
TM

We can conclude from these experiments that the different intensities of
the CD spectra and values of the 1 H NMR coupling constants J H5,H6R and J H5,H6S
of alkyl glycopyranosides are mainly due to different values of the stereoelec-
tronic exo-anomeric effect.
As a consequence of these findings, two methods for the absolute configu-
ration determination of secondary alcohols were proposed, one by applying CD
[9,18] and the other by means of 1 H NMR [18].
As can be observed in Tables 1 and 2, a simple comparison of the intensities
of the CD curves of the corresponding diastereoisomeric alkyl 2,3,4,6-tetrakis-
O-benzoyl- or 2,3,4,6-tetrakis-O-( p-bromobenzoyl)-β-d-glucopyranosides shows
that compounds of R absolute configuration at the aglyconic C-atom exhibit a
higher A value, or rather, a higher intensity of the Cotton effect at longer wave-
length, than their corresponding diastereomers with S configuration at the same
stereogenic C1′. The ∆ε values shown for both glucopyranosides of testosterone
were obtained after subtraction of the overlapping π → π* transition of the enone
system at 234 nm (⫹7.7).
Table 2 Circular Dichroism Data (CH 3CN) and 1 H Nuclear Magnetic Resonance
Coupling Constants JH5,H6 (CDCl 3, 400 MHz) for 2,3,4,6-Tetrakis-O-Benzoyl-β-
Glucopyranosides of Secondary Alcohols
Glucose ∆ε
Entry Alcohol series Cl′ 233/220 nm A Value J H5,H6R J H5,H6S
1 (⫺)-2-Octanol D R 7.3/–0.9 8.2 5.6 3.3

2 (⫹)-2-Octanol D S 6.3/–0.6 6.9 5.8 3.2
3 (⫺)-Menthol D R 5.8 5.8 5.7 3.3
4 (⫹)-Menthol D S 4.2 4.2 6.8 3.1
5 (⫺)-Neomenthol D R 6.6 6.6 5.7 3.3
6 (⫹)-Neomenthol D S 6.2 6.2 5.9 3.3
7 (⫺)-Borneol D R 8.7 8.7 5.7 3.6
8 (⫹)-Borneol D S 7.5/–0.3 7.8 5.8 3.2
9 Cholesterol D S 7.4 7.4 5.9 3.3
10 Cholesterol L S ⫺7.8 ⫺7.8 6.1 3.1
11 Cholestanol D S 6.2/–0.9 7.1 6.0 3.4
12 Cholestanol L S ⫺7.2 ⫺7.2 6.1 3.2
13 Testosterone D S 8.0 8.0 5.6 3.1
14 Testosterone L S ⫺8.1 ⫺8.1 4.7 —
15 Dimethyl d-malate D R 8.0 8.0 5.6 3.5
16 Dimethyl l-malate D S 5.8 5.8 5.0 3.2
17 (⫺)-Methyl 3-hydroxy D R 10.8 10.8 5.2 3.3
butyrate
18 (⫹)-Methyl 3-hydroxy D S 6.2/–0.8 7.0 5.6 3.2
butyrate
Source: Ref. 9.
TM

In order to apply this method it is not necessary to have both enantiomers,
since for CD spectra comparison the CD spectrum of the unavailable diasteromer
can be obtained by glycoside formation of the available enantiomer with the
easily prepared 2,3,4,6-tetrakis-O-benzoyl- or 2,3,4,6-tetrakis-O-( p-bromoben-
zoyl)-α-l-glucopyranosyl bromide and multiplying its CD spectrum by ⫺1
(Fig. 6).
The 1 H NMR data obtained from the glucosylated secondary alcohols also
allow the absolute configuration of these secondary alcohols to be determined in
two different ways: (1) in agreement with a higher gt population for stereoisomers
with an S absolute configuration at the aglyconic carbon, a higher J H5,H6R coupling
constant (CDCl 3) has been observed for these stereoisomers than for their corre-
sponding diastereomers (Tables 1 and 2). However, the 2,3,4,6-tetrakis-O-benzo-
ylglucopyranosides of cholesterol, cholestanol, and dimethyl malate (Table 2)
did not exhibit that behavior, probably because of the absence of a substituent
at the β-positions in the first two cases, and of the appearance of new anisotropic
effects from the carbonyl groups in the last case. (2) The dramatic shifts in the
aglycon 1 H NMR peaks, upon tetra-O-benzoyl-β-glycosylation of secondary al-
cohols, as a consequence of anisotropic effects and glycosylation-induced 1 H-
Figure 6 Calculation of the circular dichroism curve of an ‘‘available’’ S-alkyl 2,3,4,6-

tetrakis-O-( p-bromobenzoyl)-β-d-glucopyranoside (dotted line) by multiplying by ⫺1 the
CD curve (dashed line) of its R-alkyl β-l-glucopyranoside derivative resulting from bond-
ing the available enantiomer (R) to the 2,3,4,6-tetrakis-O-( p-bromobenzoyl)-α-l-gluco-
pyranosyl bromide. (Source : Ref. 9.)
TM

Figure 7 Configurational correlation models for secondary alcoholic (a) β-d- and (b)
β-l-glucopyranosides. (Source : Ref. 18.)
NMR shifts (Fig. 7), constituted the basis of a new method to determine the
absolute configuration of secondary alcohols [18]. The differences between the
proton chemical shifts of the d-glucosylated derivative and the free alcohol (∆δ ⫽
δ D ⫺ δ ROH ) or, more significantly, between their chemical shifts in the d- and l-
glucosylated derivatives (∆δ ⫽ δD ⫺ δL) are characteristic of the absolute config-
uration of the secondary chiral alcohol. Moreover, this method involves the use
of one enantiomer and generally a single derivatization is sufficient.
Fig. 8 shows for (⫺)-menthol (in hertz, at 500 MHz, 25°C, CDCl 3) the
chemical shift differences (δD ⫺ δL) obtained by applying our method, on the
basis of the tetra-O-benzoylglucosylation, as well as those (δ S ⫺ δ R) acquired
by the advanced Mosher’s method [19]. The larger values obtained by the former
can make it a suitable alternative method to determine the absolute configuration
of secondary alcohols.
III. DIRECT DETERMINATION OF DRUGS
In order to be CD-active, a compound must fulfill two requisites: it must be

optically active and absorb electromagnetic light in the region under study. This
Figure 8 Comparison of the δ S ⫺ δ R and the δ D ⫺ δ L values of the MTPA [19d] (left)
and the tetrakis-O-benzoyl-glucopyranoside (right) derivatives of (⫺)-menthol, respec-
tively. MTPA, 1-methoxy-1-phenyl-1-trifluoromethylacetic acid. (Source: Ref. 18.)
TM

is due to the fact that two physical requirements (ellipticity and absorbance) are
measured simultaneously. Although this limits the applicability of CD, it also
confers on this technique a high degree of analytical selectivity, allowing an
optically active absorbing drug in a mixture to be analyzed directly.
A. Direct Determination of Drugs in Pharmaceutical

Formulations
A clear example of the advantage of the high selectivity of CD can be found in
the pharmaceutical field [20,21], in which an optically active absorbing drug can
be directly analyzed from its pharmaceutical preparations, with additives not in-
terfering with the CD measurement [22], as well as in biological fluids (urine
and serum). It is interesting to note that although the chiral drugs greatly outnum-
ber achiral ones, CD has only been used infrequently in pharmaceutical develop-
ment [20,21].
Thus, we have reported a method for the discrimination and accurate and
precise determination of cephalosporins in pure form, as well as for direct deter-
mination of cephalosporins in commercial oral suspensions, injections, and cap-
sules [23]. The study, performed with 15 commercial cephalosporins in common
clinical use (Scheme 2), first revealed sufficient CD spectral dissimilarities to
discriminate among the cephalosporin homologues and classify these antibiotics,
on the basis of the wavelengths of their Cotton effects, in five spectroscopic
groups (Fig. 9 and Table 3). Furthermore, distinguishing between the β-lactam
antibiotics, penicillins [22], cephalosporins, and cephamycins (R 3 ⫽ OCH 3) on
the basis of their CD spectral data was found to be straightforward [23]. The
observed CD spectral dissimilarities come from red shifts of the two transitions
(at 260 and at 230 nm) of the 3-cephem chromophore, the basic skeleton of
cephalosporin antibiotics [24], produced by the substituents directly attached to
it (R 2 and R 3), and/or the overlapping of new transitions originating from the
chromophores present in R 1 and R 2.
The validity of the method for the determination of cephalosporins by CD
was confirmed by analysis of the variance (ANOVA). This is a statistical analysis
technique that allows us to overcome the ambiguity that the estimation of signifi-
cant differences represents when more than one comparison is performed. Table
4 shows the slopes and the intercepts of the equations of the regression lines [25]
calculated for one member of each spectroscopic group, as well as the excellent
correlation coefficients and values of the root mean square of the ANOVA error
term, S θc, for these antibiotics. The acceptable limit of quantitation was found to
be 5 µg/mL, the overall accuracies for cefadroxil, cefapirin, cefamandole, cefoxi-
tin, and ceftriaxone were 99.9, 100.1, 101.1, 100.1, and 100.1, with precisions
of 1.10, 1.46, 1.69, 1.19, and 1.66, respectively.
The method was then applied to the direct determination of cephalosporins
in pharmaceutical formulations. The samples were dissolved and CD measured
TM

Scheme 2 Structures of the cephalosporins analyzed.
directly, without performing any derivatization, filtration, or chromatographic

separation steps. The CD spectral patterns of the analyzed drugs were identical
to those of the standard solutions, showing the noninterference of additives, since
these are non-CD-active. As can be observed in Table 5, the mean amount esti-
mated in each case, by using either the CD or the HPLC [26] technique, is in
good agreement with the amount claimed on the label. Therefore, CD can be of
TM

Figure 9 Circular dichroism spectra for one member of each group: cephradine (first
group), cephalotin (second group), cephazolin (third group), cefoxitin (fourth group), and
ceftriaxone (fifth group). (Source: Ref. 23.)
Table 3 Circular Dichroism Data of the Commercial Cephalosporins Analyzed

Group Name λ ext /∆ε λ ext /∆ε λ ext /∆ε λ ext /∆ε ∆ε ⫽ 0
First Cefadroxil 255/9.2 211/–16.3 235, 202

Cefoperazone 255/3.1 223/–17.4 207/–17.7 238
Cephalexin 255/12.1 223/–18.3 237
Cephradine 255/10.6 221/–14.4 202/–15.2 237
Second Cefotaxime 306/–0.6 258/10.5 228/–18.3 289, 242, 206
Cephalotin 258/12.4 228/–17.4 242
Cefuroxime 296/–1.1 258/11.4 227/–16.4 296, 242, 207
Cefaclor 259/11.0 228/–15.8 242
Cefapirin 259/14.6 229/–17.5 242, 208
Third Cefsulodin 263/6.3 231/–16.9 247, 210
Cefamandole 263/6.6 230/–16.9 248
Cephazolin 296/–1.6 263/7.4 228/–16.6 285, 247
Fourth Cefoxitin 262/17.2 238/–18.7 215/3.5 250, 222, 208
Moxalactam 265/8.0 239/–28.8 255, 215
Fifth Ceftriaxone 284/–5.2 236/–17.4 206
Source: Ref. 23.
TM

Table 4 Parameters of Calibration Graphs for One Member of Each Circular Dichroism Spectroscopic Group in Distilled Water
Range CD λ ext Correlation

Compound (µg/mL) n (nm) Slope (m) Intercept (z) coefficient (r) Sθc
Cefadroxil 1.0–80.0 30 255 0.8575 ⫾ 0.0016 ⫺0.0656 ⫾ 0.060 0.9999 0.265

Cefapirin 1.0–60.0 25 259 0.9960 ⫾ 0.0047 0.2014 ⫾ 0.1583 0.9998 0.537
Cefamandole 5.0–80.0 24 263 0.4228 ⫾ 0.020 ⫺0.0521 ⫾ 0.0912 0.9997 0.277
Cefoxitin 1.0–80.0 28 262 1.3209 ⫾ 0.0051 ⫺0.1915 ⫾ 0.2281 0.9998 0.806
Ceftriaxone 5.0–40.0 24 236 ⫺0.9968 ⫾ 0.0065 0.3867 ⫾ 0.1642 0.9995 0.410
Source: Ref. 23.
TM

Table 5 Determination of Cephalosporins in Pharmaceutical Formulations
Mean amount estimated ⫾ SD
(%) a
Amount of labeled
Dosage form claim (g/unit) CD HPLC
Oral suspension (n ⫽ 3) 3.0 (Cefadroxil) 104.5 ⫾ 2.13 99.2 ⫾ 2.45

35.0 (Saccharose)
Capsules (n ⫽ 4) 0.5 (Cefadroxil) 104.8 ⫾ 1.48 102.8 ⫾ 3.26
Injection (n ⫽ 5) 1.0 (Cefoxitin) 103.1 ⫾ 1.24 97.93 ⫾ 0.74
a
CD, circular dichroism; HPLC, high-performance liquid chromatography.
Source: Ref. 23.
great utility for the direct determination of drugs in pharmaceutical preparations;

its principal advantages, besides those mentioned, are quickness and simplicity.
In addition, the analysis is inexpensive and can be performed in a few minutes;
all these properties are very practical for quality control laboratories. Moreover,
the spectroscopic classification facilitates the discrimination of cephalosporin ho-
mologues and allows their correct identification in most cases.
We wish to emphasize that although we have used the β-lactam antibiotics
as model compounds to prove that CD is a suitable technique for the direct deter-
mination of these drugs in pharmaceutical formulations, this result can be ex-
tended to any optically active absorbing drug.
B. Direct Determination of Drugs in Human Biological

Fluids (Urine and Serum)
Since any research that, directly or indirectly, contributes to health improvement
is of great interest, we decided to perform an exhaustive study of the applicability
of CD to the quantitative determination of optically active absorbing drugs in
biological fluids [27]. The high selectivity of this technique should allow CD-
active drugs to be determined among other non-CD active ones, since prac-
titioners usually prescribe several drugs simultaneously.
1. Human Urine
At the time the study that follows [27] was performed, only two reports had been
published for the determination of drugs in human biological fluids by CD, one
describing the determination of tetracycline in urine [28] and the other the deter-
mination of tri- and tetra-(hydroxyethyl)-rutosides in urine and serum [29].
In collaboration with the urological department of the Hospital Universita-
TM

rio de Canarias, over 200 urine samples from 10 healthy adult subjects and 51
hospitalized patients were analyzed by CD. The standard dilution used in this
work for CD analysis of drugs in urine was prepared by diluting 20 µL of urine
in a 5-mL calibrated flask with distilled water: a 250-fold dilution. The CD spectra
of the urine samples belonging to healthy volunteers and to some patients under
the administration of multiple nonoptically active absorbing drugs showed no
Cotton effects in the 400- to 200-nm range. This dilution offers the best chance
of eliminating any possibility of interference from the urine metabolites. There-
fore, any Cotton effect that appears in this region under the specified conditions
must be due to the presence of urinary proteins and/or CD-active drugs in the
urine.
The optically active absorbing drugs vitamin C, cephalexin, cefoxitin, am-
picillin, and amoxicillin were detected in the CD spectra of the urine samples of
the patients under treatment with one of these drugs. Urinary proteins were also
detected in the CD spectra of the urine samples of 31 of 51 patients, whose
routine urinalysis had already revealed significant amounts of proteins (15–300
mg/dL). In many cases the patients under treatment with one β-lactam antibiotic
were under multiple-drug administration. None of the drugs shown in Table 6
interfered with the CD analyses of the β-lactam antibiotics. These drugs are non-
absorbing and/or nonoptically active or possess an ellipticity too low to be de-
tected in our standard dilution. Interference was only observed in those cases in
which the patients were also under treatment with vitamin C, an optically active
absorbing compound. Therefore, vitamin C must be avoided when another CD-
active drug is analyzed.
The pattern of the CD spectra of the urine samples depends on the presence
Table 6 Noncircular Dichroism–Active Drugs in Urine

Antimicrobial agents Tobramycin, gentamycin, netilmycin,
norfloxacin, pipemidic acid, ciprofloxacin,
cotrimoxazole
Hypnotics and sedatives Alprazolam, triazolam
Antidepressants Maprotiline hydrochloride
Analgesics–antipyretics Acetaminophen, amidopyrine
Nonsteroidal antiinflammatory agents Acetylsalicylic acid, diclofenac
Methylxanthine bronchodilators Theophylline
Coronary vasodilators Nitroglycerin, isosorbide dinitrate, nifedipine
Antihypertensive agents Prazosin
Antiemetics Metoclopramide
Antihistamines Astemizole, ranitidine
Source : Ref. 27.
TM

of proteins and/or a CD-active drug; thus, (1) the CD spectra of those protein-
free samples containing a CD-active drug were superimposable with those of the
standard dilutions [30] after matching the concentrations (Fig. 10); (2) the CD
spectra of those samples containing only urinary proteins exhibited clear protein-
type CD spectra [31], namely, a broad negative Cotton effect with two extrema
(around 220 and 209 nm) superimposable in most cases with that of human albu-
min [30], the main human urinary protein (Fig. 11); (3) the CD spectra of those
samples containing, in addition to proteins, a CD-active drug exhibited ‘‘com-
plex’’ CD spectra identical to those of the standard dilution in the range 400–
250 nm, although with modified patterns below 250 nm, as a consequence of the
overlapping of Cotton effects of the urinary proteins on those of the CD-active
drugs (Figs. 12 and 13).
The direct determination of β-lactam antibiotics in human urine by CD
was validated by using as model drugs the antibiotics cephalexin, cefoxitin, and
ampicillin. The linearity of the present method was confirmed by ANOVA of
the linear regression line equations of these drugs (Table 7) [25]. In addition, the
validation study performed confirmed the accuracy and precision of the method;
the overall accuracy for ampicillin, cefoxitin, and cephalexin was 100.5%, 99.6%,
and 100.1%, with precision of 1.69, 1.18, and 1.68, respectively.
In the case of urinary proteins it was assumed that the ellipticity of each
Figure 10 Circular dichroism (CD) spectrum of standard CD-active drugs: cefoxitin

(6.98 µg/mL), cephalexin (9.32 µg/mL), amoxicillin (9.91 µg/mL), and ampicillin (8.50
µg/mL). (Source : Ref. 30.)
TM

Figure 11 Circular dichroism spectrum of standard human albumin (10.48 µg/mL).
(Source : Ref. 27.)
Figure 12 Circular dichroism spectra of the urine of three patients under cefoxitin ther-
apy (250 times dilutions): without proteins, solid line (23.25 µg/mL of cefoxitin); with
proteins, dashed line (15.29 µg/mL of cefoxitin, and 11.19 µg/mL of proteins); dotted
line, (7.48 µg/mL of cefoxitin, and 17.16 µg/mL of proteins). (Source: Ref. 27.)
TM

Table 7 Parameters of Calibration Graphs for Model Drugs and Human Albumin in Urine
Range CD λ ext Correlation

Compound (µg/mL) n (nm) Slope (m) Intercept (z) coefficient (r) Sθc
Ampicillin 1.0–40.0 25 234 0.9582 ⫾ 0.0043 ⫺0.1769 ⫾ 0.0888 0.9998 0.288

Cefoxitin 1.0–80.0 24 262 1.1500 ⫾ 0.0036 ⫺0.2219 ⫾ 0.1671 0.9999 0.534
Cephalexin 1.0–50.0 25 255 1.1766 ⫾ 0.0067 ⫺0.0717 ⫾ 0.2000 0.9996 0.634
Albumin 1.0–50.0 25 220 ⫺1.5407 ⫾ 0.0046 ⫺0.3232 ⫾ 0.1229 0.9999 0.391
Source: Ref. 27.
TM

TM
C
Figure 13 Complex circular dichroism (CD) spectrum of the urine of a patient under
ampicillin therapy (250 times dilutions) (solid line), its estimated protein CD curve (dotted
line) (2.54 µg/mL), and its estimated ampicillin CD spectrum (dashed line) (14.55 µg/
mL). (Source: Ref. 27.)
urinary protein was identical with that of the main human urinary protein, human
albumin. Similarly to the β-lactam antibiotics, the analysis of the variance, (AN-
OVA) of the linear regression of the calibration line carried out with human
albumin at 220 and 209 nm confirmed the linearity of the method (Table 7); the
statistical analyses showed precise and accurate values for both wavelengths,
although slightly better for the one measured at 220 nm.
The direct quantitative determination of a CD-active drug in protein-free
urine samples from type-1 CD spectra or the determination of the total urinary
proteins in urine samples lacking CD-active drugs from type-2 CD spectra is
achieved in a straightforward manner by measuring in each case the ellipticity
angle (mdeg) at the selected wavelength and using the corresponding regression
line equation (Table 7).
The procedure to determine a CD-active drug from a ‘‘complex’’ type-3
CD spectrum depends on the wavelength of the first Cotton effect of the drug
to be analyzed: (1) If the drug exhibits a Cotton effect above 250 nm, its determi-
nation can be performed by direct measurement of the ellipticity at the wave-
length of this Cotton effect, regardless of the presence of proteins (Fig. 12). This
can also be determined by CD spectral subtraction of the CD-active drug contri-
bution to the ‘‘complex’’ CD spectrum; (2) the simultaneous determination of a
TM

CD-active drug, having only Cotton effects below 250 nm, and the total urinary
proteins required to perform a CD spectral subtraction, due to the overlapping
of their Cotton effects. In this case it is necessary to carry out a regression fit
using the shapes of the drug and the albumin spectra (Fig. 13). The individual
regression line equations are then used to determine their concentration.
CD analyses carried out in the presence of proteins did not exhibit, under
our standard dilution, overlapping of the very weak signals, from aromatic resi-
dues of proteins [3], and/or extrinsic Cotton effects from drug–protein binding
[32]. Therefore, the quantitation of CD-active drugs is not interfered with by
the possible existence of the Cotton effects mentioned. Fig. 14 shows how the
wavelength of the first Cotton effect of ‘‘complex’’ CD spectra of in-house mix-
tures of cefoxitin (80, 60, 40, 20, 5, and 1 µg/mL) and a constant concentration
of albumin (48 µg/mL) remained unaltered at 262 nm, the wavelength of the
first Cotton effect of cefoxitin. Furthermore, CD spectra shown in Fig. 15 exhibit
a constant ellipticity at this wavelength independent of the concentration of uri-
nary proteins.
These ‘‘complex’’ CD spectra become clear protein-type spectra (Fig. 11)
by removal of the corresponding CD drug contribution, so the total urinary pro-
teins (or human albumin for in-house mixtures) can be determined by measuring
at 220 nm the resulting ellipticity angle (Table 8).
Figure 14 Complex circular dichroism spectra, from top to bottom, of in-house mix-
tures of albumin (48 µg/mL) and cefoxitin (80, 60, 40, 20, 5, and 1 µg/mL), respectively
[30]. (Source: Ref. 27.)
TM

Figure 15 Circular dichroism spectra of three in-house samples with a constant concen-
tration of cefoxitin (20 µg/mL): without albumin (solid line); with albumin: 24 µg/mL
(dashed line) and 48 µg/mL (dotted line) [30]. (Source: Ref. 27.)
The present method for the direct determination of a CD-active drug in

human urine was also confirmed by performing a comparative analysis by means
of CD and HPLC. As can be observed in Table 9, the concordance between the
values obtained by these techniques was excellent.
The present method can be very useful to perform pharmacokinetic studies
Table 8 Determinations of in-House Lactam–Albumin Mixtures
Prepared concentration Determined

(µg/mL) concentration (µg/mL)
Entry Compound Drug Albumin Drug Albumin

1 Cefoxitin 20.00 24.00 19.98 24.10
2 20.00 48.00 20.05 48.19
3 Cephalexin 20.00 24.00 19.78 24.14
4 20.00 48.00 19.87 48.51
5 Ampicillin 20.00 24.00 20.04 24.63
6 20.00 48.00 20.51 48.93
Source : Ref. 27.
TM

Table 9 Comparative Analysis by HPLC and
CD Spectroscopy a
Determined
concentration
(mg/mL)
Entry Compound CD HPLC
1 Ampicillin 14.58 13.75

2 12.23 12.61
3 17.12 16.62
4 Cefoxitin 2.22 2.30
5 19.18 20.48
6 23.15 23.32
7 Cephalexin 11.86 11.85
8 17.20 16.11
9 17.78 17.88
a
HPLC, high-performance liquid chromatography; CD,
circular dichroism.
Source: Ref. 27.
of optically active absorbing drugs. As can be observed in Fig. 16, a clear gradual
decrease in the intensity of the Cotton effects of the urine samples along a 6-h
period was obtained from a patient receiving intravenously (i.v.) the first dose of
1 g of cefoxitin. The corresponding CD hourly determinations were in excellent
agreement with those obtained from HPLC, showing the applicability of this
method to monitor drugs. The total amounts of cefoxitin recovered from the first
up to the sixth hour were 349.4, 227.1, 113.4, 74.1, 59.7, and 39.3 mg, respec-
tively.
2. Human Serum
The most complex and important of the biological fluids is blood, and thus once
the study in urine was completed, the possibility of applying CD to establish a
method to determine CD-active drugs in this biological fluid was analyzed [33].
Since the concentration of proteins in human serum or plasma is very high,
it was necessary to dilute the serum sample to 1 : 1000, in order to have a suitable
voltage in the CD photomultiplier (below 460 V). Consequently, the direct deter-
mination of a CD-active drug is not possible from these very dilute solutions;
deproteinization of the serum sample is necessary before CD analysis.
The precipitation of the serum/plasma proteins by means of acetonitrile
proved to be an excellent way to perform CD analysis of drugs. The method
TM

Figure 16 Circular dichroism spectra recorded along a 6-h period from the urine sam-
ples of a patient under cefoxitin therapy (250 times dilution). From the first up to the sixth
hour the following concentrations of cefoxitin were determined: 53.54, 37.68, 25.11,
21.10, 14.88, and 7.44 µg/mL, respectively. (Source: Ref. 27.)
consists of: (1) adding 3 mL of CH 3CN to 1 mL of serum; (2) centrifuging the

mixed solution at 5000 rpm for 15 min; (3) measuring the CD spectrum of the
protein-free supernatant (about 3.5 mL).
Following this method, the CD spectrum of several drug-free serum sam-
ples showed no Cotton effect in the 400- to 200-nm range, clearly showing the
total removal of the proteins, although important absorptions below 250 nm re-
mained. Nonchiral absorbing compounds, such as creatinine and uric acid, must
be responsible for these absoptions. Therefore, only drugs having Cotton effects
at wavelengths above 250 nm can be measured by the procedure.
On the other hand, the CD spectra of protein-free supernatants containing
different cephalosporins (Fig. 17), all having Cotton effects above 250 nm,
showed Cotton effects in agreement (⫾2 nm) with those of the standard solutions
in the 400- to 250-nm range. Therefore, the determination of CD-active drugs
in human serum/plasma samples could be easily performed by measuring the
ellipticity angle of their first Cotton effects and replacing their values in the corre-
sponding regression line equations. Currently we are performing successfully the
validation and recovery studies of the present method [33].
It can be concluded that CD is a valid alternative spectroscopic technique
to perform, in a simple, precise, and accurate manner, the quantitative determina-
TM

Figure 17 Circular dichroism spectra of the resulting protein-free serum samples con-
taining 19.1 µg/mL of cephalotin, 16.7 µg/mL of cephalexin, and 19.0 µg/mL of ceftriax-
one (10-mm cylindrical quartz cell).
tion of drugs in biological fluids. Its high degree of selectivity, together with the
fact that it is quick and economical, could make CD a common technique in
clinical laboratories in the near future.
ACKNOWLEDGMENTS
I wish to thank all the persons who worked out the results described: Drs. J. I.
Padrón, E. Q. Morales, P. Gortázar, M. Ravina, and Mrs. M. Trujillo. Support
of this work by the Dirección General de Investigación Científica y Técnica
(DGICYT), Ministerio de Educación y Ciencia (Spain), through grant PB93-
0559, and by the Gobierno de Canarias is gratefully acknowledged.
REFERENCES AND NOTES
1. (a) F. Ciardelli and P. Salvadori, eds., Fundamental Aspects and Recent Develop-
ments in Optical Rotatory Dispersion and Circular Dichroism, Heyden & Son, Lon-
don (1973). (b) K. Nakanishi, N. Berova, and R. W. Woody, eds., Circular Dicho-
ism, Principles and Applications, VCH Publishers, New York (1994).
2. (a) N. Harada and K. Nakanishi, Circular Dichroic Spectroscopy-Exciton Coupling
in Organic Stereochemistry, University Science Books, Mill Valley, Calif. (1983).
(b) Ref. 1b, p. 361–398.
3. C. D. Fasman, ed., Circular Dichroism and the Conformational Analysis of Biomo-
lecules, Plenum Press, New York, p. 25–157 (1996).
4. (a) W. T. Wiesler, J. T. Vázquez, and K. Nakanishi, J. Am. Chem. Soc., 109: 5586
TM

(1987). (b) J. T. Vázquez, W. T. Wiesler, and K. Nakanishi, Carbohydr. Res., 176:
175 (1988).
5. (a) M. L. Phillips, E. Nudelman, F. C. A. Gaeta, M. Perez, A. K. Shingai, S. Hako-
mori, and J. C. Paulson, Science, 250: 1130 (1990). (b) L. A. Lasky, Science 258:
964 (1992). (c) R. A. Dwek, Chem. Rev., 96: 683 (1996).
6. R. U. Lemieux and S. Koto, Tetrahedron 30: 1933 (1974).
7. L-C. Lo, N. Berova, K. Nakanishi, E. Q. Morales, and J. T. Vázquez, Tetrahedron
Asymmetry, 4: 321 (1993).
8. H.-W. Liu and K. Nakanishi, J. Am. Chem. Soc., 104: 1178 (1982).
9. E. Q. Morales, J. I. Padrón, M. Trujillo, and J. T. Vázquez, J. Org. Chem., 60: 2537
(1995).
10. We define A C and A B values as the amplitudes of split CD Cotton effects in the
cinnamate–cinnamate and in the benzoate–benzoate coupling regions centered about
the cinnamate λ max 311 nm and about the benzoate λ max 245 nm, respectively.
11. (a) G. D. Wu, A. S. Serianni, and R. Barker, J. Org. Chem., 48: 1750 (1983). (b)
Y. Nishida, H. Ohrui, and H. Meguro, Tetrahedron Lett., 25: 1575 (1984). (c) H.
Ohrui, Y. Nishida, M. Watanabe, H. Hori, and H. Meguro, Tetrahedron Lett., 26:
3251 (1985). (d) H. Hori, Y. Nishida, H. Ohrui, H. Meguro, and J. Uzawa, Tetrahe-
dron Lett., 29: 4457 (1988). (e) Y. Nishida, H. Hori, H. Ohrui, H. Meguro, J. Uzawa,
D. Reimer, V. Sinwell, and H. Paulsen, Tetrahedron Lett., 29: 4461 (1988). (f) K.
Bock, and J. Duus, J. Carbohydr. Chem., 13: 513 (1994).
12. The following set of three equations is used to calculate the ratio of P gg , P gt , and
P tg : 1.3 P gg ⫹ 2.7 P gt ⫹ 11.7 P tg ⫽ J H5,H6S ; 1.3 P gg ⫹ 11.5 P gt ⫹ 5.8 P tg ⫽ J H5,H6R ;
P gg ⫹ P gt ⫹ P tg ⫽ 1. See Ref. 11.
13. The stereoelectronic exo-anomeric effect is the preference for the gauche [(sc):
synclinal] conformation about the glycosidic C-OR bond of sugar derivatives.
(a) R. U. Lemieux, A. A. Pavia, J. C. Martin, and K. A. Watanabe, Can. J. Chem.,
47: 4427 (1969). (b) Deslongchamps, Stereoelectronic Effects in Organic Chemistry
(J. E. Baldwin, ed.), Organic Chemistry Series, Vol. I, Pergamon Press, Oxford
(1983). (c) A. J. Kirby, The Anomeric Effect and Related Stereoelectronic Effects
at Oxygen (K. Hafner, C. W. Rees, B. M. Trost, J. M. Lehn, P. R. Schleyer, and
R. Zahradnik, eds.), Reactivity and Structure Concepts in Organic Chemistry, Vol.
15, Springer-Verlag, Berlin (1983). (d) The Anomeric Effect and Associated Stereo-
electronic Effects (G.R.J. Thatcher, ed.) ACS Symposium Series 539, Washington,
D.C. (1993).
14. J.-P. Praly and R. U. Lemieux, Can J. Chem., 65: 213 (1987).
15. A. Cossé-Barbi, D. G. Watson, and J. E. Dubois, Tetrahedron Lett., 30: 163 (1989).
16. J. I. Padrón and J. T. Vázquez, Chirality, 9: 626 (1997).
17. J. I. Padrón, E. Q. Morales, and J. T. Vázquez, J. Org. Chem., 63: 8247 (1998).
18. M. Trujillo, E. Q. Morales, and J. T. Vázquez, J. Org. Chem., 59: 6637 (1994).
19. (a) J. A. Dale and H. S. Mosher, J. Am. Chem. Soc., 95: 512 (1973). (b) G. R.
Sullivan, J. A. Dale, and H. S. Mosher, J. Org. Chem., 38: 2143 (1973). (c) T.
Kusumi, I. Ohtani, Y. Inouye, and H. Kakisawa, Tetrahedron Lett., 29: 4731 (1988).
(d) I. Ohtani, T. Kusumi, O. M. Ishitsuka, and H. Kakisawa, Tetrahedron Lett., 30:
3147 (1989). (e) I. Ohtani, T. Kusumi, Y. Kashman, and H. Kakisawa, J. Org.
Chem., 56: 1296 (1991). (f) I. Ohtani, T. Kusumi, Y. Kashman, and H. Kakisawa,
TM

J. Am. Chem. Soc., 113, 4092 (1991). (g) T. Pehk, E. Lippmaa, M. Lopp, A. Paju,
B. C. Borer, and R. J. K. Taylor, Tetrahedron: Asymmetry, 4: 1527 (1993).
20. (a) N. Purdie, and K. A. Swallows, Anal. Chem., 61: 77A (1989). (b) N. Purdie,
Analytical Applications of Circular Dichroism in Techniques and Instrumentation
in Analytical Chemistry, Vol. 14 (N. Purdie and H. G. Brittain, eds.), Elsevier Sci-
ence B. V., Amsterdam, p. 241–278. (1994).
21. A. Gergely, J. Pharm. Biomed. Anal., 7: 523 (1989).
22. N. Purdie and K. A. Swallows, Anal. Chem., 59: 1349 (1987).
23. P. Gortázar, and J. T. Vázquez, J. Pharm. Sci., 83: 1204 (1994).
24. R. Nagarajan and D. O. Spry, J. Am. Chem. Soc., 93: 2310 (1971).
25. The regression line equations of the linear relationship between the ellipticity angle
and the concentration of the antibiotic were defined as θ ⫽ mc ⫹ z, where θ is the
ellipticity angle (millidegree [mdeg]) at the selected wavelength, c is the concentra-
tion (µg/mL), m is the slope of the fitted line, and z is the θ intercept of the regression
line.
26. A. M. Brisson and J. B. Fourtillan, J. Chromatogr., 223: 393 (1981).
27. P. Gortázar, M. Ravina, and J. T. Vázquez, J. Pharm. Sci., 84: 1316 (1995).
28. J. M. Bowen, and N. Purdie, J. Pharm. Sci., 71: 836 (1982).
29. G. Jung, M. Ottnad, and W. Voelter, Eur. J. Drug Metab. Pharmacokinet., 3: 131
(1977).
30. Standard solutions were prepared by dissolving the accurately weighed compound
in calibrated flasks with control human urine. Measured volumes of the standard
solutions were diluted into calibrated flasks with human urine and 20 µL of the
resulting solutions diluted into 5-mL calibrated flasks with distilled water. For CD
measurement a 10-mm cylindrical quartz cell was used. For further details see
Ref. 27.
31. See, for example, Ref. 1a, pp. 352–372.
32. I. Sjöholm and T. Sjödin, Biochem. Pharmacol., 21: 3041 (1972).
33. P. Gortázar, A. Roën, and J. T. Vázquez, Chirality, 10: 507 (1998).
TM

8
Furan-Terminated Cationic
π-Cyclizations in the Synthesis
of Natural Products
Steven P. Tanis
Pharmacia & Upjohn, Inc., Kalamazoo, Michigan
I. INTRODUCTION
Approximately 23 years ago the sesquiterpene warburganal (1) was isolated from
the East African medicinal plant Warburgia ugandensis [1], and we had the op-
portunity to embark on a synthesis prior to the publication of the structure of 1.
Our initial approach constructed the sesquiterpene framework through a Diels-
Alder cycloaddition of 1-vinyl-2,6,6-trimethyl-1-cyclohexene with dimethyl acet-
ylenedicarboxylate, as shown in Scheme 1 [2]. The Diels-Alder adduct 2 (83%)
resisted the development of the desired trans-ring fusion via catalytic hydrogena-
tion, affording instead only cis-fused products. While we were wrestling with
this problem, which was eventually solved [2], we considered alternatives that
might directly construct the desired trans-fused sesquiterpene skeleton.
The cationic π-cyclization, which has been widely utilized in the construc-
tion of polycyclic ring systems, has been the object of intense study since the
early 1950s [3]. Initial forays into this arena demonstrated the syntheses of fused
ring terpenoid-type systems, and later efforts demonstrated the construction of
spiro and bridged ring carbocyclic systems [4]. These studies have demonstrated
that a wide variety of initiating functions (e.g., epoxides, allylic alcohols, enones,
olefins, carbinolamides) and terminating moieties (e.g., aromatic rings, acety-
Dedicated to Professor Koji Nakanishi on the occasion of his 70th birthday.
TM

Scheme 1 Approach to warburganal 1.
lenes, allylsilanes, allenes, olefins) can be incorporated into the cyclization sub-
strate to lead to terpenoids and alkaloids. In the synthesis of simple fused ring
systems, such as the perhydronaphthalene moiety of warburganal (1), this method
has produced the target 4,4,8a-trimethyl-trans-fused skeleton as shown in Eq. (1)
[5]. The issue to be considered prior to examining a cationic π-cyclization for
the preparation of 1 was the identity of the 1,4-dialdehyde equivalent as the termi-
nator function for the cyclization.
(1)
In principle, the 1,4-dialdehyde unit found in warburganal (1) could be

prepared by the hydrolysis of a furanoid precursor. This suggested the possibility
of terminating a cyclization such as that depicted in Eq. (1) with the π-excessive
aromatic ring of furan. The natural propensity of furan to suffer electrophilic
aromatic substitution at an available α-position (Fig. 1, Path a, X ⫽ H) dictates
that a blocking group X (Fig. 1, Path b, X ⫽ Me, TMS, SPh, etc.) be employed
in order to have access to the desired target substitution pattern.
When we were considering the chemistry of Fig. 1, we were faced with a
paucity of precedent in the literature. We surmised that it was likely due to the
relative inaccessibility of suitable substrate [6], the questionable nucleophilic
character of the furyl residue relative to more standard terminator functions [7],
and the increased acid liability of the more highly substituted product compared
with the starting material [7]. As a result, careful choice of both the initiating
TM

␲
Figure 1 A furan-based approach to the synthesis of warburganal. (Source: Ref. 1.)
moiety and the reaction conditions would be required if the projected chemistry
were to be successful.
Our interest in developing substituted furans as nucleophilic synthons in
annulative processes stemmed from the variety of useful functional groupings
that might be realized from the relatively unreactive furyl nucleus. As mentioned
(Fig. 1), a furan can serve as the operational equivalent of a 1,4-dialdehyde; Fig.
2 indicates additional acyclic, heterocyclic, and carbocyclic subunits that can be
derived from a furan after standard chemical manipulations.
Figure 2 Furan functional equivalencies.
TM

Scheme 2 Possible disconnections to provide linearly fused-, spirocyclic-, and bridged-
ring-containing compounds.
With a variety of potentially sensitive functionalities that might be derived

from the furan unit, we designed the connections shown in Scheme 2 to provide
access to linearly fused-, spirocyclic-, and bridged-ring-containing systems. For
the electronically favored furan 3-to-2 closure, we can consider the reaction of
the hypothetical furan 7 with a variety of doubly reactive synthons (8–10). The
interaction of a reactive side chain nucleophile/electrophile of 7 with 8–10 will
provide a coupling product; subsequent activation of the nascent-electron-defi-
cient center of 8–10, followed by aromatic substitution, could provide fused-
(11), spirocyclic- (12), and bridged- (13) ring systems. Manipulation of the furan
nucleus (Fig. 2) and other residual functionality would provide complex interme-
diates for natural products synthesis. The regiochemistry of furan termination of
the reactions of Scheme 2 could be altered by connecting the furyl tether to the
furan 2-position. Success in this electronically less favored cyclization paradigm
would lead to the isolation of the compounds of Scheme 2, regioreversed with
respect to the furan residue.
II. LINEARLY FUSED TERPENOID COMPOUNDS

(SCHEME 2, PATH A)
A. Epoxide Initiated Furan 3-to-2 Cyclizations
The epoxide moiety, among other groups, has been widely employed as an initia-
tor function for cationic π-cyclizations [8]. Workers in the field have employed
TM

a wide variety of Lewis acids to initiate the cyclization sequence, and when this
is coupled with the ease of precursor olefin epoxidation or introduction intact, it
suggested the epoxide as our first cyclization initiator. The cyclization substrates
that were examined were designed to permit entry into five-, six-, or seven-mem-
bered ring systems. In order to prevent ambiguity in the choice of a given ring
system available from a given oxirane, the epoxide function was biased to favor
one mode of C-O bond polarization over the alternative bond [8]. We also studied
the placement of the initiating function within the forming cycle (endocyclic) or
outside the forming cycle (exocyclic) [9]. On the basis of the work of Baldwin,
we anticipated a successful cyclization for 5-, 6-, and 7-exo systems, although
only 6-endo should proceed readily.
The epoxy furan substrates selected for this study were constructed as
shown in Eq. (2). A Grignard reagent 14, prepared from 3-chloromethylfuran,
was coupled with a haloalkene, in the presence of a catalyst, to give furyl-olefins
15, which were epoxidized with m-chloroperoxybenzoic acid (MCPBA) to afford
furyl-epoxides 16 [10]. Although the furan moiety is known to be susceptible to
oxidation, the relative rates of furan vs. olefin attack as a function of the degree
of substitution had not been reported. The compounds of Table 1 were prepared
by the chemistry of Eq. (2).
(2)
As shown in Table 1, the coupling reactions proceed smoothly when 14

was reacted with alkyl and allylic halides (runs 2–5, Li2CuCl4 as catalyst) [7a,11].
The synthesis of olefin 17, utilizing a vinyl halide, required anhydrous FeCl3 as
a catalyst [11]. Furyl-olefins 17–21 were submitted to standard epoxida-
tion conditions (1.05 eq. MCPBA, CH2Cl2, 0°C) to furnish good yields of
epoxides 22 (86%), 24 (85%), and 26 (81%), all derived from trisubstituted ole-
fins. Furyl-olefin 18 led to epoxide 23 in only 25% yield, and the epoxidation
of olefin 20 provided none of the expected epoxide 25. Closer examination of
run 2 indicated that epoxide 23 was accompanied by unreacted 18 (23%) and
the anhydride derived from 18 (37%). Olefin 20 led only to anhydride and recov-
ered 20.
The three remaining epoxides required in quantity for the study were pre-
pared as shown in Eq. (3) and (4). 3-Furylmethyllithium [7a,12] [Eq. (3)] was
reacted with epoxy iodides 27 and 28 [hexamethylphosphorus triamide (HMPA),
⫺25°C] [13] to give epoxy furans 25 (73%) and 29 (68%), respectively. Mono-
substituted epoxide 30 was prepared in 37% yield from a coupling of 3-furylmeth-
yllithium and a protected iododiol, as shown in Eq. (4).
TM

Table 1 Synthesis and Oxidation of 3-Furyl Olefins [Eq. (2)]
Run Olefin Catalyst Fulyl-alkene (yield) Furyl-epoxide (yield)
(3)
(4)
1. Cyclization Studies
Relatively potent Lewis acids such as boron trifluoride etherate are often selected
to catalyze epoxy olefin cyclizations [8]. Given the acid lability of the starting
furans and the increased acid sensitivity of the products, the choice of Lewis acid
should have a profound effect on the partitioning of the reaction between a fruitful
cyclization and acid-mediated decomposition. Six Lewis acids were utilized
TM

␲
the initial study [7a], BF3⋅OEt2 [8], EtAlCl2 [14], Et2AlCl [14], Al2O3 [13], Ti(Oi-
Pr)3Cl [15], and ZnI2 [16]. The choice of Lewis acid was dictated by (1) the
ability to modify the potency of a group of Lewis acids with a common metal
center readily and (2) the possibility of moderating the Brønsted acidity of the
medium through the choice of the Lewis acid. Adventitious protic acid might be
scavenged by a Lewis acid possessing a metal–carbon bond releasing an alkane;
alternatively with the proper choice of metal, the product metal–alcohol complex
should be a much weaker protic acid compared to a BF3 –alcohol complex.
The substrate epoxy furans were exposed to boron trifluoride etherate (0.3
eq.) in methylene chloride at ⫺25°C as the standard cyclization conditions (Table
2). Five-membered ring precursors 22, 23, and 30, when treated with BF3⋅OEt2,
afforded only allylic alcohols 32 (62%), 34 (53%), and 36 (49%), respectively.
Only epoxy furans 24 and 25 gave appreciable quantities of cyclized products,
37 (47%) and 39 (30%), respectively. The majority of the material isolated from
the BF3⋅OEt2-mediated cyclizations consisted of the depicted allylic alcohols, and
the mass balances were poor (ca. 60% or less). The poor yields of cyclized materi-
als and low mass balances are in agreement with the surmises regarding the
Lewis/Brønsted acidity requirements for these reactions.
The aluminum-based Lewis acids, EtAlCl2, Et2AlCl, and Al2O3, provided
better mass balance (ca. 70% or more), and variable amounts of cyclized prod-
ucts. The treatment of 22, 23, and 30, five-membered ring precursors, with either
EtAlCl2 or Et2AlCl led to good yields of allylic alcohols. Only furans 24 (6-
endo), 25 (6-exo), and 26 (7-endo) gave any cyclized material (10%–22%) with
EtAlCl2 or Et2AlCl. Alumina (Al2O3) provided only allylic alcohol for all epoxy
furans except 24, which furnished a 32% yield of 37 and a 51% yield of allylic
alcohol 38. The modification of the Lewis acid to provide a protonolyzable M-C
bond, thus reducing Brønsted acidity of the medium, did lead to improved mass
balance; however, elimination was the dominant path with the alkylaluminum
halides. Further modification of the aluminum-centered Lewis acid to alumina
also provided improved mass balance but little cyclization.
The next most Lewis acidic compound in the series Ti(Oi-Pr)3Cl [15]
proved to be an efficient and useful promoter of epoxy furan cyclization. As
before, five-membered-ring precursor oxiranes 22, 23, and 30 did not lead to
desired cyclized products, with elimination products 32 (80%) and 34 (72%)
coming from 22 and 23, respectively. The monosubstituted epoxide 30 could not
be induced to react, even after exposure to 3 eq. of Ti(Oi-Pr)3Cl for 24 h at room
temperature. Similar treatment of epoxy furan 24 led to the formation of the
desired cyclized adduct 37 in 78% yield, with no elimination product observed.
6-exo-Epoxide 25 and 7-endo-precursor 26 afforded cyclized adducts 39 (89%)
and 41 (87%), respectively, with the latter accompanied by 8% of allylic alcohol
42. Even 7-exo-precursor epoxide 29 gave a respectable yield of cyclized product
43 (36%) when treated with Ti(Oi-Pr)3Cl.
Epoxides 22, 23, and 30, five-membered ring precursors, were next exposed
TM

Table 2 Initial Cyclization Studies, Effect of Ring Size, Epoxide Placement, and
Lewis Acid
Furyl-epoxide Product(s)
22 31 32
BF3-OEt2 (0.3 eq.) 0% 62%
Et2AlCl (2 eq.) 0% 85%
Al2O3 0% 83%
Ti(OiPr)3Cl (3 eq.) 0% 80%
ZnI2 (3 eq.) 0% 76%
23 33 34
BF3-OEt2 (0.3 eq.) 0% 53%
Et2AlCl (2 eq.) 0% 85%
Ti(OiPr)3Cl (3 eq.) 0% 72%
ZnI2 (3 eq.) 0% 70%
30 35 36
BF3-OEt2 (0.3 eq.) 0% 49%
Et2AlCl (2 eq.) 0% 78%
Ti(OiPr)3Cl (3 eq.) no reaction
ZnI2 (3 eq.) 25% 44%
24 37 38
BF3-OEt2 (0.3 eq.) 47% 0%
EtAlCl2 (2 eq.) 16% 57%
Et2AlCl (2 eq.) 22% 49%
Al2O3 32% 51%
Ti(OiPr)3Cl (3 eq.) 78% 0%
ZnI2 (3 eq.) 71% 0%
TM

Table 2 Continued
Furyl-epoxide Product(s)
25 39 40
BF3-OEt2 (0.3 eq.) 30% 10%
EtAlCl2 (2 eq.) 0% 73%
Et2AlCl (2 eq.) 10% 70%
Al2O3 0% 81%
Ti(OiPr)3Cl (3 eq.) 89% 0%
ZnI2 (3 eq.) 70% 0%
26 41 42
BF3-OEt2 (0.3 eq.) 0% 41%
EtAlCl2 (2 eq.) 0% 76%
Et2AlCl (2 eq.) 10% 69%
Al2O3 0% 83%
Ti(OiPr)3Cl (3 eq.) 87% 8%
ZnI2 (3 eq.) 88% 9%
29 43 44
BF3-OEt2 (0.3 eq.) 10% 12%
EtAlCl2 (2 eq.) 0% 64%
Et2AlCl (2 eq.) 0% 73%
Al2O3 0% 79%
Ti(OiPr)3Cl (3 eq.) 36% 47%
ZnI2 (3 eq.) 23% 52%
TM

to the final Lewis acid in this initial study ZnI2 (3 eq. CH2Cl2, room temperature).
Allylic alcohols 32 (76%) and 34 (70%) were produced from 22 and 23, respec-
tively; however, furyl-epoxide 30 gave a mixture of the five-membered ring target
35 (25%) and allylic alcohol 36 (44%). Epoxides 24, 25, 26, and 30 afforded
modest to excellent yields of cyclic products.
The results presented in Table 2 suggested that the furan- (3-to-2 closure)
terminated/epoxide-initiated cationic π-cyclization is useful for the construction
of six- and seven-membered rings. Good to excellent yields can be realized with
a judicious choice of Lewis acid. However, closure to form five-membered rings
is difficult.
2. Syntheses of Pallescensin A and Aphidicolin

A more rigorous test of the epoxy furan cyclization would require the formation
of two or more rings, with the development of stereochemistry and/or transmis-
sion of chirality. Pallescensin-A (45) [7a,17,18] was selected as the initial test
of the former, and a formal total synthesis of aphidicolin (46) [19] would be used
to illustrate the latter.
Scheme 3 presents the total synthesis of (⫾)-pallescensin A (45). 6,7-Epoxygera-

nyl chloride [10,20] was coupled with 3-furylmethylmagnesium chloride to give
7,8-epoxydendrolasin (47) in 79% yield [10]. Cyclization with BF3⋅OEt2 gave
3 β-hydroxypallescensin A (48) in 47% yield [21]. As expected from the results
of Table 2, triisopropoxytitanium chloride and zinc iodide afford 48 in higher
yields, 62% and 65%, respectively, and the reaction mixtures were cleaner and
less complex. Compound 48 is smoothly converted to pallescensin A (45), as
described by Nasipuri and Das [21].
(⫹)-Aphidicolin (46), a diterpene tetraol produced by the mold Cepha-
losporium aphidacola Petch [19a,b], has provoked interest as a synthetic target
over the past 20 years [22]. (⫹)-Aphidicolin (46) has been associated with a
range of biological activities, including antiviral [19c,d], antitumor [19e], and
antileukemic [19f], activity as well as activity as a reversible inhibitor of deoxyri-
TM

Scheme 3 The synthesis of (⫾)-pallescensin A (45).
bonucleic acid (DNA) polymerase-α [19g]. McMurry [23] has reported the syn-
thesis of (⫾)-46 from diketone 49. We envisioned accomplishing a synthesis of
(⫾)-49 and (⫺)-49 [24] via furan-terminated cationic π-cyclization, as shown in
Scheme 4.
The chemistry outlined in Scheme 4 would exploit a furan-terminated cat-
ionic-π-cyclization to establish the carbon framework of 49 rapidly with the cor-
rect relative and absolute configuration created at carbons 4, 5, and 10; a furan
to dione conversion would provide 49. The potential availability of the modified
epoxygeranyl chloride precursor to furan 51 in optically pure form from a
Sharpless asymmetric epoxidation [25] is an additional advantage to the route
depicted.
In the event, geraniol was converted to the related benzoate (Scheme 5),
which readily underwent allylic hydroxylation (SeO2, TBHP) [26] to give alcohol
52 (70%). Sharpless asymmetric epoxidation of hydroxybenzoate 52 gave (⫺)-
53 in 71% yield [25a]. The optical purity of (⫺)-53 was judged to be ⱖ95% ee
after an examination of (⫺)-53 by nuclear magnetic resonance (NMR) in the
presence of Eu(hfpc)3 [25a] and high-performance liquid chromatography
(HPLC) analysis of the related Mosher ester [27]. Alternatively, (⫺)-53 could
be prepared in 83% yield and ⱖ95% ee via the Sharpless catalytic asymmetric
epoxidation protocol [25b]. We considered a number of blocking groups with
which to protect the free hydroxyl group of (⫺)-53 during the halogenation/
coupling/cyclization steps. Initially we selected a t-butyldimethylsilyl (TBDMS)
moiety for this function. This proved to be somewhat difficult at the cyclization
TM

Scheme 4 A furan-terminated cationic π-cyclization approach to aphidicolin (46).
Scheme 5 The formal total synthesis of (⫾)-aphidicolin (46).
TM

step, when widely variable yields of products were obtained. The required group
must be smaller than the TBDMS group, survive the chlorination/coupling/cycli-
zation steps, yet be readily removed to facilitate a chelation-directed reduction
of the C-3 ketone. We selected a benzyl ether to block the free hydroxyl of (⫺)-
53. Toward that end (⫺)-53 was treated with sodium hydride (NaH) and benzyl
bromide in tetrahydrofuran (THF) with added n-Bu4NI to provide (⫺)-54. Benzo-
ate hydrolysis (NaOMe, MeOH, n-Bu4NI) furnished the related alcohol, which
was immediately converted to the allylic chloride (⫹)-55 (n-BuLi, p-TsCl, LiCl)
[28] in 84% yield for the two steps. Chloride (⫹)-55 was smoothly coupled with
(3-furyl)methylmagnesium chloride, giving (⫺)-56 in 79% yield. In studies de-
scribed in Table 2 and Scheme 3, ZnI2 and Ti(Oi-Pr)3Cl stood out as the Lewis
acids of choice for epoxy furan cationic π-cyclizations. In the present case,
exposing the more highly oxygenated (⫺)-56 to ZnI2 and Ti(Oi-Pr)3Cl led to
highly variable (0%–49%) yields of (⫹)-57. After considerable experimentation
it was discovered that treatment of (⫺)-56 with BF3⋅OEt2 (6 eq.) and Et3N (3 eq.)
in CH2Cl2 /PhH/hexanes (1: 1 :1) at ⫺78°C gave an excellent 72% yield of (⫹)-
57. The successful completion of the synthesis then required the inversion of
stereochemistry at the C-3 position.
Alcohol (⫹)-57 was oxidized under Swern conditions [29] to afford ketone
(⫹)-58 in 97% yield, which then had to be selectively reduced. The reduction
of (⫹)-58, with L-Selectride, was examined with and without added metal salts.
It was assumed that the proper choice of metal salt would conspire to provide a
chelated intermediate, which would be selectively reduced [30]. Precomplexation
of (⫹)-58 with ZnI2, MgBr2, Ti(Oi-Pr)3Cl, or Ti(Oi-Pr)4 followed by the addition
of L-Selectride (⫺78°C, Fig. 3) gave 3α:3β alcohols 63 and 64 in ratios ranging
Figure 3 Reduction selectivity of (⫾)-58 with L-Selectride in the presence and absence
of added metal salts.
TM

from 1.2 : 1 to 8.5 :1. The optimal conditions (Fig. 3, entry d) employed 2 equiva-
lents of MgBr2 ⋅ OEt2 and provided a very respectable 8.5 : 1 ratio of 63 and 64
in 95% combined yield.
With optimal reducing conditions in hand, the reduction of (⫹)-58 on a
larger scale was examined as shown in Scheme 5. Ketone (⫹)-58 was treated
with MgBr2 ⋅ OEt2 and L-Selectride to give an 8.5 : 1 mixture of alcohols 63 and
64 (Fig. 3). Reductive cleavage of the benzyl ether, as described by Kutney [31],
led to (⫹)-59 in 89% yield, as an 8.5 :1 mixture at C-3. The mixture was readily
separated after the conversion of the cis-diol 63 to the related acetonide (⫹)-60
(acetone, H⫹; 93%). Furan (⫹)-60 was brominated at the available α-position
(NBS, DMF) [32], the bromide was immediately subjected to metal–halogen
exchange (n-BuLi), and the lithiofuran was alkylated (MeI) to provide 61 in 65%
overall yield. Furan oxidation with MCPBA [24,33] furnished the ene-dione (⫺)-
62 (97%), which was hydrogenated (H2 /Pd-C) to give the (⫺)-49 (96%), complet-
ing the formal total synthesis of (⫺)-aphidicolin (46) in 16 steps and 10.7%
overall yield from geraniol.
B. Allylic Alcohol–Initiated Furan 3-to-2 Cyclizations

To expand the usefulness of furan-terminated cationic π-cyclizations in the syn-
thesis of terpenoids, we wished to extend the scope of cyclization initiators to
allylic alcohols, enones, and acid derivatives. In the case of allylic alcohols, the
nature of the species shown in Scheme 2, Path A, could be altered to facilitate
selectivity in the initial bond formation. The utilization of a vinyl epoxide [Eq.
(5)] [34] or the enol ether of an α,β-epoxy ketone [Eq. (6)] [34a] as bis-electro-
philic synthons would provide selectivity, as the second electrophilic center
would result after the initial SN2′-like addition of an organometallic reagent to
65 [35], 66 [34a,36], and 67 [37] [Eq. (7)] [34a].
(5)
(6)
TM

(7)
Initial attempts to construct linearly fused terpenoid precursors via furan

3-to-2 terminated cationic π-cyclization utilized simple variants of 66 and 67
prepared from cyclohexenone. Scheme 6 illustrates the syntheses of exocyclic
allylic alcohol cyclization precursors from 66 (n ⫽ 1, Scheme 2, Path A). Cyclo-
hexenone was converted to the spirovinyl epoxide 66 by the addition of methyl-
thiomethyl lithium to cyclohexenone, to give the related alcohol (89%), methyla-
tion of sulfur (MeI, 99%), and epoxide closure (KOt-Bu, 91%) [34a,36]. The
Grignard reagents derived from 2-(3-furyl)-1-bromoethane and 3-(3-furyl)-1-bro-
mopropane were treated with CuCN followed by 66 to give the SN2′ addition
products 68 (m ⫽ 2, 56%) and 69 (m ⫽ 3, 58%), respectively. For anticipated
ease of ionization, alcohols 68 and 69 were converted to their respective second-
ary allylic alcohols 70 (75%) and 71 (66%), respectively, via pyridinium chloro-
chromate (PCC) oxidation and MeLi addition to the derived enals.
The synthesis of endocyclic allylic alcohol cyclization substrates from
vinyl-epoxide 67 (n ⫽ 1) is presented in Scheme 7. The Grignard reagents derived
from 2-(3-furyl)-1-bromoethane and 3-(3-furyl)-1-bromopropane were treated
with CuCN followed by enol-ether epoxide 67 [37] to give SN2′ addition products.
Acid treatment produced enones 72 (m ⫽ 2, 75%) and 73 (m ⫽ 3, 72%), the
products of enol ether hydrolysis and dehydration. Methyl lithium addition then
afforded the endocyclic tertiary allylic alcohol cyclization substrates 74 (90%)
Scheme 6 The synthesis of linearly fused–allylic alcohol substrates—I.
TM

Scheme 7 The synthesis of linearly fused–allylic alcohol substrates—II.
and 75 (90%), respectively. The inclusion of only six- and seven-membered pre-
cursor chain lengths (m ⫽ 2, 3) in the compounds of Schemes 6 and 7 was based
upon the results presented in Table 2.
1. Cyclization Studies
Allylic alcohols have been extensively employed as initiators in cationic π-cycli-
zations [3], and the reaction conditions that have been employed generally in-
volve a protic acid of reasonable strength in a solvent in which it is soluble. Of
the many conditions reported in the literature, the two-phase mixture of anhy-
drous formic acid and cyclohexane [38] was selected for this study as the mildest
method for initiating the cyclization of allylic alcohols 70, 71, 74, and 75 (Scheme
8). Alcohols 70 and 71, designed to form six- and seven-membered rings, respec-
tively, from an allylic alcohol with the hydroxyl center exocyclic to the existing
ring, were separately exposed to formic acid in cyclohexane for 20 min at room
temperature. Alcohols 70 and 71 smoothly cyclized to give the six-membered
target 76 and the seven-membered target 77 in 68% and 61% yields, respectively,
as mixtures of exo-ethylidene double-bond isomers. The endocyclic allylic alco-
hols 74 and 75 were next treated with formic acid in cyclohexane to give tricyclic
furans 77 (73%) and 78 (56%), respectively. The studies of Scheme 8 firmly
established allylic alcohols as suitable initiators in furan-terminated cationic π-
cyclizations. Next we examined the cyclization of enals 79 and 80, which were
prepared as intermediates from the oxidation of 68 and 69, as previously illus-
trated in Scheme 6.
Enals 79 and 80 [Eq. (8)] were treated with BF3⋅OEt2, SnCl4, TiCl4, EtAlCl2,
Et2AlCl, MgBr2⋅OEt2, ZnI2, and Ti(Oi-Pr)3Cl, in a variety of solvents at various
temperatures, to no avail. Extensive decomposition was observed when BF3⋅OEt2,
SnCl4, TiCl4, EtAlCl2, and Et2AlCl were employed; the use of MgBr2⋅OEt2, ZnI2,
and Ti(Oi-Pr)3Cl led to recovered starting materials. Acylative-type enal cycliza-
tions similar to those reported by Andersen [39], Marshall [40], and Harding [41]
TM

Scheme 8 Linearly fused–allylic alcohol initiated–furan terminated cyclizations.
were also examined. The treatment of 79 and 80 with either Ac2O/HClO4 /EtOAc
or (CF3CO)2 O/CF3CO2H resulted in a facile and high-yield acylation of the furyl
nucleus at the 2-position. Protic acid treatment (HCOOH, c-C6H12) of 79 and 80
led to recovered starting material or, after extended treatment, extensive decom-
position.
(8)
C. Acylium Ion–Initiated Furan 3-to-2 Closure: The

Syntheses of (ⴞ)- and (ⴚ)-Fastigilin-C
1. First-Generation Approach
The cytotoxic, antineoplastic helenanolide (⫺)-fastigilin C (81) [42] was an at-
tractive target for total synthesis via furan-terminated cationic π-cyclization, as
TM

illustrated in Eq. (9). The butyrolactone moiety of 81 would be derived from the
furan of acyl-furan 82, which would be derived from cyclopentanone 83, which
possessed a cyclization initiator (acylium ion precursor) and the terminator furan.
An attractive benefit, in addition to the regiospecific protolactone introduction,
of incorporating a furan into dione 82 was the possibility of routine control of
stereochemistry about the periphery of the bicyclo[5.3.0]decane by inducing the
normally flexible seven-membered B-ring to adopt a well-defined chairlike con-
formation [43].
(9)
Syntheses of cyclopentanones such as 83 require control of two exocyclic

stereocenters and concomitant trans-addition of the elements of propionate and
3-furaldehyde to the 3- and 2-positions, respectively, of 2-methyl-2-cyclopenten-
one [44]. Our first-generation approach to (⫾)-81 is shown in Scheme 9 [44].
Mukaiyama has described a trityl salt–catalyzed, silicon transfer, tandem
conjugate addition–aldol condensation sequence [45] to form trans-2,3-disubsti-
tuted cyclopentanones with predictable exocyclic stereochemistry. Such a proto-
Scheme 9 First-generation approach to (⫾)-fastigilin C (81).
TM

col was ideally suited for the synthesis of the target cyclopentanone 83. Toward
that end, t-butyl thiopropionate was converted to the related TBDMS-enol ether
(TBDMSOTf, i-PrNEt2) [46] 84 (80%, Z/E ⬎95: 5), which was combined with
2-methyl-2-cyclopentenone (CH2Cl2, ⫺95°C bath), and the resulting mixture was
treated with 5 mol% of Ph3CSbCl6. After 20 min, 3-furaldehyde was added and
the mixture was warmed to room temperature overnight to give a mixture of 83
and pro-C-6-iso-83 (73%, 6 :1), which was difficult to separate. The ratio of 83
to pro-C-6-iso-83 is in agreement with the reports of Mukaiyama [45] and is
temperature-dependent, falling to 2.5 :1 at ⫺80°C. The stereochemical outcome
has been rationalized [45,47] as being the result of consecutive conjugate addition
and aldol reactions, which proceed in a trans-fashion across the cyclopentenone 2-
and 3-positions, through an open transition state. The B-ring closure was readily
accomplished by exposing 83 to Hg(OTFA)2 (2 eq., anhydrous CH3CN, room
temperature) [48] to afford dione 82 (R ⫽ TBDMS) in 65% yield. The fifth of
the seven B-ring stereocenters of fastigilin C (81) was selectively introduced by
reduction of the 9-ketone with NaBH4 (EtOH) to provide the 9β-alcohol 85 (99%)
as a single stereoisomer. This alcohol was protected as the related 2-(trimethylsi-
lyl)ethoxymethyl ether (SEM), giving 86 (99%). At this juncture it was necessary
to consider the completion of the endeavor. That is, it would be necessary to
develop the A-ring double bond prior to the development of the fragile butyrolac-
tone moiety, to avoid the difficulties encountered by Lansbury et al. [49]. This
strategy would require an A-ring enone protection before furan-to-butyrolactone
elaboration, followed at some point by a deblocking. The advantages offered
by the Scheme 9 route, (1) robust intermediates and (2) a well-defined B-ring
conformation that aids in B-ring development, could not outweigh the problems
to be overcome, which were (1) when and how to develop the A-ring and (2)
difficulties in transforming the chemistry of Scheme 9, a route to (⫾)-81, to
access the natural optical antipode (⫺)-81.
2. The Synthesis of (⫾)-Fastigilin C[(⫾)-81]

The introduction of an A-ring 2,3-double bond surrogate, which would obviate
the anticipated end-game problems of incorporating this unit of unsaturation and
also provide for the development of a route to (⫺)-81, was examined next. This
might be achieved by substituting 4-methoxy-2-methyl-2-cyclopentenone into the
Michael-aldol chemistry of Scheme 9. The synthesis of (⫾)-fastigilin C [(⫾)-
81] is shown in Scheme 10 [47]. Racemic 4-methoxy-2-methyl-2-cyclopentenone
[(⫾)-87] [50] was reacted with the TBDMS-enol ether of t-butyl thiopropionate
and 3-furaldehyde, mediated by trityl hexachloroantimonate in CH2Cl2, at ⫺22°C
for the Michael reaction, followed by cooling to ⫺78°C for the silicon transfer
aldol condensation (addition of 3-furaldehyde). This protocol gave (⫾)-88 in 90%
yield, uncontaminated by stereoisomers at any of the derived stereogenic centers.
TM

Scheme 10 The synthesis of (⫾)-fastigilin C [(⫾)-81].
The relative orientation of the methoxyl group of the cyclopentanone, to the exo-
cyclic stereocenters, could not be ascertained spectroscopically; therefore we
elected to continue the synthesis and determine the relative orientation when a
suitable crystalline derivative was obtained. Initial attempts to cyclize (⫾)-88
with Hg(OTFA)2 gave only a 12% yield of (⫾)-89, with the bulk of the recov-
ered material corresponding to the carboxylic acid equivalent of (⫾)-88. Ex-
posing (⫾)-88 to mercuric triflate/N,N-dimethylaniline complex [51], with a less
nucleophilic triflate counteranion, afforded an excellent 96% yield of the target
bicyclo[5.3.0]decane (⫾)-89. The reduction of (⫾)-89 was modified to employ
Luche conditions (NaBH4, MeOH, CeCl3 ⋅ 7H2O, ⫺78°C [52], after sodium boro-
hydride in ethanol afforded mixtures. This modification led to the isolation of
(⫾)-90 in 93% yield as a single stereoisomer, establishing the fifth stereocenter
about the seven-membered B-ring. The C-9-OH was protected as the correspond-
ing TBDMS-ether, giving (⫾)-91 (99%).
Numerous attempts to block the 4-ketone of (⫾)-91 as the related
TM

dioxolane/ketal were unsuccessful. However, (⫾)-91 was readily reduced with
diisobutylaluminum hydride (DIBAL) to the corresponding 4α-alcohol (⫾)-92
(97%). Alcohol (⫾)-92 was nicely crystalline and was submitted to single-crystal
x-ray analysis, which established the relative configuration in this series to be
that shown for (⫾)-92 in Scheme 10. The x-ray stereostructure of (⫾)-92 (Fig.
4) indicated that the C-2-OMe relative orientation is that expected from a steric
controlled approach of the Michael nucleophile in the Mukaiyama conjugate ad-
dition–aldol protocol [53] and established (S)-4-methoxy-2-methyl-2-cyclopen-
tenone as the requisite starting material for the synthesis of (⫺)-81.
Alcohol (⫾)-92 was protected as the corresponding methoxyethoxymethyl
(MEM) ether (MEM-Cl, i-Pr2NEt) to furnish (⫾)-93 (97%), which was desily-
lated (n-Bu4NF, THF), giving (⫾)-94 (98%). Furan-diol 94 was selectively mo-
nosilylated (TBDMSCl, imidazole) to give (⫾)-95 (98%), and the 6-OH was
protected as the readily removable ethoxyethyl ether analogue (⫾)-96 (96%).
The furan unit of (⫾)-96 was smoothly silylated (i. n-BuLi; ii. Me3SiCl), yielding
(⫾)⫹97 (95% overall) after careful ethoxyethyl ether cleavage ( p-TsOH). As
desired, silylfuran (⫾)-97 was readily oxidized (CH3CO3H) [54] to give buteno-
lide (⫾)-98 (87%), setting the stage for the crucial hydroxyl-directed hydrogena-
tion required to set the required stereochemistry at C-7 and C-8.
After examining the Wilkinson catalyst [(Ph3P)3RhCl] [55], Crabtree’s cata-
lyst [Ir(COD)py(P(Cy)3)PF6] [55,56], and [Rh(NBD)(DIPHOS-4)]BF 4 [55,57],
it was discovered that the cationic rhodium catalyst readily reduced (⫾)-98, at
1000 psi, affording (⫾)-99 in 82% yield. Lactone (⫾)-99 was converted to α-
methylene lactone (⫾)-100 (83%) via the procedure of Lansbury et al. [49], car-
boxylation followed by treatment with Eschenmoser’s salt. Lactone (⫾)-100 was
treated with the symmetrical anhydride derived from 3-methylcrotonic acid [4-
dimethylaminopyridine (DMAP), Et3N], and the mixture was heated in refluxing
xylenes to provide senecioate ester (⫾)-101 (92%). Zinc bromide treatment of
(⫾)-101 smoothly and selectively deprotected the C-4-α-oxygen function, lead-
ing to alcohol (⫾)-102 (89%). Oxidation, to the 4-one, and α-elimination to fur-
nish the 2,3-double bond, remained before a total synthesis of (⫾)-81 was
achieved. Pyridinium chlorochromate (PCC) oxidation of (⫾)-102 gave β-me-
thoxyketone (⫾)-103 (81%), setting the stage for double-bond introduction via
Figure 4 The relative stereochemistry of (⫾)-92.
TM

Scheme 11 The preparation of (S)-(⫺)-4-hydroxy-2-methyl-2-cyclopentenone.
β-elimination. Ketone (⫾)-103 was heated to reflux in ether with Amberlyst-

15 to produce (⫾)-fastigilin C [(⫾)-81, 96%], a product of β-elimination and
deprotection of the 9-OH. The sequence from 4-methoxy-2-methyl-2-cyclopen-
tenone (⫾)-87 to (⫾)-81 was accomplished in 17 steps and an overall yield of
24.6%.
3. The Synthesis of (⫺)-Fastigilin C [(⫺)-81]

The synthesis of (⫺)-81 required a source (S)-(⫹)-4-methoxy-2-methyl-2-cyclo-
pentenone (⫹)-87. An enzymatic resolution of (⫾)-4-hydroxy-2-methyl-2-cyclo-
pentenone [(⫾)-104], shown in Scheme 11, was developed to provide gram quan-
tities of (S)-(⫹)-87.
Racemic 4-hydroxy-2-methyl-2-cyclopentenone [(⫾)-104] [50] and β,β,β-
trifluoroethyl butyrate [58] were dissolved in ether, and the resulting mixture was
treated with porcine pancreatic lipase (PPL) [59]. After 6 days the reaction had
proceeded to 51% conversion, providing (S)-(⫺)-104 in 43% yield and 68% ee,
and butyrate (R)-105 (44%) after chromatography. The butyrate (R)-105 was
cleaved to the alcohol (R)-104, as described by Wong [59] (guanidine, MeOH)
(78%, 46% ee), and the stereocenter was inverted (Ph3P, diethylazodicarboxylate,
HCO2H, MeOH, Al2O3) [59], furnishing an additional quantity of (S)-alcohol (S)-
(⫺)-104 (combined 66% yield, 60% ee). The combined (S)-(⫺)-104 was again
TM

exposed to β,β,β-trifluoroethyl butyrate and PPL in ether to afford (S)-(⫺)-4-
hydroxy-2-methyl-2-cyclopentenone [(S)-(⫺)-104] in 52% isolated yield and
ⱖ98% ee. The alcohol was easily converted to the target (S)-(⫹)-4-methoxy-2-
methyl-2-cyclopentenone (⫹)-87 as outlined in Eq. (10).
(10)
The chemistry of Scheme 10 was then undertaken, on a large scale, starting

with ca. 50 g of (S)-(⫹)-87. The 17-step sequence from (S)-(⫹)-87 to (⫺)-fastigi-
lin C [(⫺)-81] was accomplished in 14% overall yield.
III. SPIROCYCLIC TERPENOID COMPOUNDS (SCHEME 2,

PATH B)
A. Allylic Alcohol– and Enone-Initiated Furan 3-to-2 and
Furan 2-to-3 Cyclizations
As has been previously discussed, allylic alcohols and enones can serve as initia-
tors for cationic π-cyclizations. These moieties can be prepared by nucleophilic
addition of an organometallic reagent to vinyl epoxides as was illustrated in Eqs.
(5), (6), and (7). Those equations defined by the disconnection illustrated in
Scheme 2, Path A, for the preparation of linearly fused ring-containing com-
pounds by equating vinyl epoxides with a cycloalkyl dication [Eq. (7)], wherein
the electron-deficient centers reside on adjacent carbons, are realized sequentially,
the first as a result of the SN2′-type reactivity of a vinyl epoxide, and the second
as a consequence of the ionization of the allylic alcohol produced from the initial
addition. This paradigm ensures regiochemical integrity in the annulation relative
to resident markers. In order to prepare spirocyclic compounds by Path B of
Scheme 2, one must alter the placement of the electron-deficient centers, placing
the first exocyclic, which will allow closure via furan termination at the tertiary
pro-spiro center.
The addition of an organometallic reagent to an exo-vinyl cycloalkyl epox-
ide [Eq. (11)] would result in the preparation of an allylic alcohol by SN2′ addition
[34,60]. Subsequent acid treatment would reveal the second electron-deficient
center, thus equating an exo-vinyl cycloalkyl epoxide with the dication 9 [Eq.
(12)].
TM

(11)
(12)
To examine the utility of allylic alcohols and enones in the synthesis of

spirocyclic systems via furan 3-to-2 cyclizations, a variety of exo-methylene 2,3-
epoxycycloalkanes were prepared [34a,60] and reacted with (3-furyl)alkyl Grig-
nard reagents 104a–c in the presence of CuCN (Scheme 12). This reaction af-
forded the target allylic alcohol cyclization substrates 105a–e in 59%–82%
yields. The desired related enones were smoothly prepared via PCC oxidation
(CH2Cl2) of 105a–e to give enones 106a–e (72%–87%).
The cyclizations of furyl-allylic alcohols 105a–e were attempted with
HCOOH/cyclohexane by analogy to the chemistry of Scheme 8. In the event
Scheme 12 Spirocyclic furan 3-to-2 cyclization precursor synthesis.
TM

Scheme 13 Allylic alcohol–initiated spirocyclic furan 3-to-2 cyclization.
(Scheme 13), alcohols 105a and 105b were exposed to the two-phase mixture
of HCOOH/cyclohexane to give spiro[4,5]decane 107a (58%) and spiro[4,6]un-
decane 107b (53%), respectively. Similarly, alcohols 105c–e were treated with
HCOOH/cyclohexane to furnish the formate of 105c (84%), spiro[5,5]undecane
107d (72%), and spiro[5,6]dodecane 107e (58%), respectively.
The enones 106a–e, produced as illustrated in Scheme 12, were dissolved
in cyclohexane and treated with formic acid. Of the five substrates shown in
Scheme 14, only enones 106a and 106d, leading to spiro[4,5]decane 108a (72%)
and spiro[5,5]dodecane 108d (66%), respectively, provided any cyclized prod-
ucts. Enones 106b,c and 106e were recovered unchanged. Compounds 106b,c
Scheme 14 Enone-initiated spirocyclic furan 3-to-2 cyclization.
TM

Scheme 15 Spiro-ring formation, furan-2-to-3-closure.
and 106e also proved resistant to cyclization under a wide variety of other reac-
tion conditions including Lewis acids and acylating agents.
In order to examine whether the formation of a six-membered ring is the
necessary factor in a successful enone-initiated/furan-terminated cyclization, the
furan 2-to-3 closure related to the chemistry of Scheme 13 was examined. The
Grignard reagent derived from 3-(2-furyl)-1-bromopropane was added to the iso-
butyl enol ether of 2-methyl-1,3-cyclohexanedione (Scheme 15), [60] to give
enone 109 (85%) after acidic workup. Enone 109 was exposed to a variety of
Brønsted and Lewis acids, as well as acylating agents, to no avail. The target
spiro[5,5]undecane furan 110, regioisomer of 108d, was not observed. This rela-
tively less favored (electronically) furan 2-to-3 cyclization [61] was observed
when allylic alcohol 111, prepared from 109 in 93% yield (NaBH4, CeCl3), was
treated with formic acid/cyclohexane, affording olefin 111 (68%).
IV. BRIDGED-RING-CONTAINING TERPENOID

COMPOUNDS (SCHEME 2, PATH C)
A simple drawing defining a possible disconnection for the synthesis of bridged,

ring-containing compounds is presented in Scheme 2, Path C. An existing ring
would present a reactive electrophilic/nucleophilic center to interact with a furan-
containing moiety possessing a reactive nucleophilic/electrophilic center, thereby
creating a ring system with a tethered furan. Cyclization to form the bridged
system would result when an electron-deficient center was developed at a noncon-
tiguous carbon center. Three strategies were considered to develop this reactivity
TM

pattern: (1) create the second reactive center as a result of the initial bond con-
struction; (2) utilize differentially reactive electrophilic centers and maximize
selectivity [such as was described in the syntheses of pallescensin A (45) and
aphidicolin (46)]; and (3) perform the first bond construction, then introduce the
desired electrophilic center.
Nakafuran 9 (112), a fish antifeedant isolated from the marine sponge
Dysedea fragilis and the sponge predacious nudibranchs Hypselodoris godeffroy-
ana and Chromodoris maridadilus [62], was selected as the test case for bridged-
ring synthesis via furan-terminated cationic π-cyclization. A retrosynthesis of
112, presented in Scheme 16, suggests that the bicyclo[4.3.1]decane skeleton
would be prepared from the readily available 3-furylmethyl dianion (discussed
previously) and a highly substituted dication.
The synthesis of (⫾)-nakafuran-9 [112] is presented in Scheme 17 [34a].
The Grignard reagent derived from 3-chloromethylfuran was coupled with the
exo-methylene vinyl epoxide, prepared from 2-methyl-2-cyclohexenone, in the
presence of CuCN to give allylic alcohol 113 in 62% yield, thus establishing
the C-4, C-5 bond of 112. Oxidation (PCC, 89%) and treatment of the derived
enone with MeCu ⋅ BF3 [63] introduced the C-6 methyl group, providing ketone
114 (70%) as a 60 :40 mixture at pro-C-7, in 62% overall yield from 113. The
second electrophilic center needed for closure at C-10 was easily introduced as
the enone via selenylation [64] of the kinetic enolate, followed by oxidation and
elimination (H2O2, Et3N) of the selenoxide-yielding enone 115 (72%). It was
discovered that triethylamine was a useful addend to the oxidation mixture as it
provided a sink for the phenylseleninic acid produced in the elimination. In its
absence, enone 115 was produced admixed with the desired cyclization product
116, albeit in greatly reduced yields. Cyclization of 115 was effected with
HCOOH-c-C6H12, affording the desired bicyclo[4.3.1]decanone 116 in excellent
yield (79%) as a 60:40 mixture at C-7. A methyl equivalent and double bond
were simultaneously introduced via a Wittig olefination of 116 using the condi-
tions of Conia and Limasset [65] (Ph3P⫹CH3 I⫺, K-t-amylate) to give 117 (80%)
as a 60:40 mixture at C-7. After considerable experimentation [34a] it was found
that exposure of 117 to a solution of p-toluenesulfonic acid in refluxing benzene
Scheme 16 Retrosynthetic analysis of nakafuran 9 (112).
TM

Scheme 17 The synthesis of (⫾)-Nakafuran-9 (112).
for 16 min provided a 95:5 mixture of nakafuran-9 (112) and ∆8,9-isonakafuran-

9 (118) in 80% yield.
V. ALKALOID SYNTHESIS VIA FURAN-TERMINATED

CATIONIC ␲-CYCLIZATION
The successful investigations involving the use of furan-terminated cyclizations

in the synthesis of carbocycles (discussed previously) led to the suggestion of
creating azacycles under the same, or similar, conditions. Alkaloid natural prod-
ucts should be easily and efficiently accessed through a furan-terminated cationic
π-cyclization by employing a nitrogen-containing cationic initiator such as an N-
acyliminium ion. Much of the chemistry surrounding the synthesis, reactivity,
and utility of the acyliminium ion has been very well documented in the pioneer-
ing work of Speckamp and coworkers [66]. On the bases of this work and the
work of others, such as Chamberlin [67a] and Evans [67b], it was assumed that
the ease of synthesis, and high reactivity of acyliminium ions were a good match
for a furan terminator. A variety of linearly fused, spirocyclic, and bridged aza-
cycles could be synthesized by simply altering the placement of the furan tether
TM

Figure 5 An approach to alkaloid synthesis via furan-terminated cationic-π-cyclization.
on the N-acyliminium ion precursor as illustrated in Fig. 5. The first foray into
alkaloid synthesis via N-acyliminium ion furan-terminated cyclization [68] was
directed to the preparation of the spirolactam intermediate 119 in Kishi’s synthe-
sis of perhydrohistrionicotoxin (120) [69].
The Kishi spiropiperidine 119 [69] was approached by the spirocyclic discon-
nection presented in Fig. 5. For the present application it was necessary to alter
the substrate from the illustrated carbinolamide-3-substituted furan to a 3-(5-
ethyl-2-furyl)propyl carbinolamide in order to provide a proper ketone–side chain
location. In the event (Scheme 18), the Grignard reagent prepared from 1-bromo-
3-(5-ethyl-2-furyl)propane was added to the iodomagnesium salt of glutarimide
[67b] to furnish the related carbinolamide 121 in excellent crude yield. Without
TM

Scheme 18 A formal total synthesis of (⫾)-perhydrohistrionicotoxin 120.
purification 121 was cyclized (HCOOH/c-C6H12) to afford spiropiperidine 122 in

72% overall yield. The furan ring was oxidatively cleaved with MCPBA [24,33],
yielding 123 (70%) after reduction (H2-Pd/C; EtOAc, aq. HOAc) of the relatively
unstable ene-dione. The completion of the synthesis requires dione differentiation
and removal of the unwanted side chain oxygen.
Ketone 123 was treated under Noyori kinetic ketalization conditions [70]
(TMSS(CH2)2STMS, TMSOTf ) to lead, almost exclusively, to the side chain thio-
ketal 124 (67%, ⱖ98 :2). Reductive cleavage was accomplished with Raney
nickel in refluxing ethanol, leading to the target spiropiperidine 119 in 78% yield.
The chemistry of Scheme 18 describes a concise (6 steps, 26%) formal total
synthesis of perhydrohistrionicotoxin [120] and illustrates the utility of the furan-
terminated cationic π-cyclization protocol in alkaloid synthesis. The application
of furan-terminated cationic π-cyclizations to the synthesis of linearly fused- and
bridged-ring-containing alkaloids is currently under study. These results will re-
ported in due course.
VI. CONCLUSIONS
The furan-terminated cationic π-cyclization has been developed as a useful reac-

tion sequence for the construction of appropriate linearly fused-, spirocyclic-,
and bridged-ring-containing terpenoids. Through the careful selection of non-
Brønsted acidic reaction conditions, high yields of cyclized products have been
realized with epoxides, allylic alcohols, enones, and thioesters as initiating func-
TM

tions. Ring size restrictions, for the forming cycle, have been discovered, and
the cyclization has been demonstrated to be somewhat sensitive to the position
of attachment of the furan to the tether (3 vs. 2). Asymmetry transfer from the
cyclizing chain has been realized, and the use of exogenous chirality to induce
asymmetry has been demonstrated. Finally, alkaloid synthesis, through the
agency of an acyliminium ion initiator, has been demonstrated in a limited sense
(spirocycles), with much work remaining to bring this area to the level of under-
standing enjoyed in the terpene field.
ACKNOWLEDGMENT
I wish to thank those people who made all this possible; the grad students, post-
docs, and coworkers at Michigan State University, the Upjohn Co., and Phar-
macia and Upjohn Inc., who did the overwhelming majority of the work men-
tioned. Sincere thanks go to Yu-Hwey Chuang, Mark Collins, Melissa Deaton,
Lisa Dixon, Dave Head, Paul Herrinton, Mark McMills, Tim Parker, and Ed
Robinson. Thanks also to the Camille and Henry Dreyfus Foundation, the Na-
tional Institutes of Health, and the Upjohn Postdoctoral Research Scholar Pro-
gram for financial support.
REFERENCES
1. I. Kubo, Y.-W. Lee, M. J. Pettei, F. Pilkiewicz, and K. Nakanishi, J. Chem. Soc.

Chem. Comm., 1013 (1976).
2. S. P. Tanis and K. Nakanishi, J. Am. Chem. Soc., 101: 4398 (1979).
3. (a) J. K. Sutherland, Comprehensive Organic Synthesis, Vol. 3 (B. M. Trost, ed.),
Pergamon, Oxford, p. 341 (1991). (b) S. R. Haring and T. Livinghouse, J. Org.
Chem., 62: 6388 (1997).
4. (a) S. P. Tanis and P. M. Herrinton, J. Org. Chem., 50: 3988 (1985). (b) S. P. Tanis,
P. M. Herrinton, and L. A. Dixon, Tetrahedron Lett., 26: 5347 (1985).
5. (a) G. Stork and A. W. Burgstahler, J. Am. Chem. Soc., 77: 5068 (1955). (b) P. A.
Stadler, A. Eschenmoser, H. Schinz, and G. Stork, Helv. Chim. Acta, 40: 2191
(1957).
6. S. P. Tanis, Tetrahedron Lett., 23: 3115 (1982).
7. (a) S. P. Tanis and P. M. Herrinton, J. Org. Chem., 48: 4572 (1983). (b) F. M.
Dean, Adv. Heterocycl. Chem., 30: 167 (1982). (c) M. V. Sargent and T. M. Crisp,
Comprehensive Organic Chemistry, Vol. 4 (P. G. Sammes, ed.), Pergamon, Oxford,
p. 693 (1979).
8. (a) E. J. Corey, and H. B. Wood, Jr., J. Am. Chem. Soc., 118: 11982 (1996). (b)
E. J. Corey; and S. Lin, J. Am. Chem. Soc., 118: 8765 (1996).
TM

9. J. E. Baldwin, R. C. Thomas, L. I. Kruse, and L. X. Silberman, J. Org. Chem., 42:
3846 (1977).
10. S. P. Tanis, Tetrahedron Lett., 23: 3115 (1982).
11. M. Tamura and J. K. Kochi, Synthesis, 303 (1971).
12. L. T. Burka, L. J. Felice, and J. S. Jackson Phytochemistry, 20: 647 (1981).
13. R. K. Boeckman, Jr., K. J. Bruza, and G. R. Heinrich, J. Am. Chem. Soc., 100: 7101
(1978).
14. (a) B. B. Snider, D. J. Rodini, M. Karras, and J. van Straten, Tetrahedron, 37: 3927
(1981). (b) B. B. Snider, D. J. Rodini and J. van Straten J. Am. Chem. Soc., 102:
5872 (1980). (c) B. B. Snider, Encyclopedia of Reagents for Organic Synthesis,
Vol. 3 (L. A. Paquette, ed.), Wiley, New York, p. 1777 (1995). (d) B. B. Snider,
Encyclopedia of Reagents for Organic Synthesis, Vol. 4 (L. A. Paquette, ed.), Wiley,
New York, p. 2388 (1995).
15. (a) Y. E. Raifel’d, A. A. Nikitenko, and B. M. Arshava Tetrahedron Asymmetry, 2:
1083 (1991). (b) N. A. Petasis, Encyclopedia of Reagents for Organic Synthesis,
Vol. 2 (L. A. Paquette, ed.), Wiley, New York, p. 1221 (1995).
16. (a) J. A. Marshall and P. G. M. Wuts, J. Org. Chem., 42: 794 (1977). (b) M. T.
Reetz, S. Hüttenhain, and F. Hübner, Synth. Commun., 11: 217 (1981). (c) G. J.
McGarvey, Encyclopedia of Reagents for Organic Synthesis, Vol. 8 (L. A. Paquette,
ed.), Wiley, New York, p. 5569 (1995).
17. G. Cimino, S. De Stefano, A. Guerriero, and L. Minale, Tetrahedron Lett., 1417
(1975). (b) G. Cimino, S. De Stefano, A. Guerriero, and L. Minale, Tetrahedron
Lett., 1425 (1975).
18. For additional syntheses of 45 see: (a) U. Lange, and S. Blechert, Synthesis, 1142
(1995). (b) L. A. Paquette and R. E. Maleczka, Jr., J. Org. Chem., 57: 7118 (1992).
(c) K. Shishido, K. Umimoto, and M. Shibuya, Heterocycles, 31: 597 (1990).
19. (a) K. M. Bundret., W. Dalziel, B. Hesp, J. A. J. Jarvis, and S. Neidle, J. Chem.
Soc. Chem. Commun., 1027 (1972). (b) W. Dalziel, B. Hesp, K. M. Stevenson, and
J. A. J. Jarvis, J. Chem. Soc. Perkin Trans. 1, 2841 (1973). (c) R. A. Bucknall, J.
Moores, R. Simms, and B. Hesp, Antimicrob. Agents Chemother., 4: 294 (1973).
(d) S. Ikegami, T. Taguchi, M. Ohashi, M. Oguro, H. Nagano, and Y. Mano, Nature
(London), 275: 458 (1978). (e) J. Douros, and M. Suffness, New Anticancer Drugs
(S. K. Carter and Y. Sakurai, eds.), Springer-Verlag, Berlin, p. 29 (1980). (f) G.
Pedrali-Noy, M. Belvedere, T. Crepaldi, F. Focher, and S. Spedari, Cancer Res.,
42: 3810 (1982); (g) S. Spadari, F. Sals, and G. Pedrali-Noy, Trends Biochem. Soc.,
7: 29 (1982).
20. J. H. H. Chan, Stauffer Chemical Co.; Ger. Offen. 2,300,261 (c1. C 07d); U.S. Appl.
219,866; Chem. Abstr., 79: 105441d (1973).
21. D. Nasipuri and G. Das, J. Chem. Soc. Perkin Trans. I, 2776 (1979).
22. For recent syntheses of 46 see (a) R. Marini-Bettolo, Korean J. Med. Chem., 6: 329
(1996). (b) M. Toyota, Y. Nishizawa, and K. Fukumoto, Tetrahedron, 52: 10347
(1996). (c) D. G. Hall, and P. Deslongchamps, J. Org. Chem., 60: 7796 (1995).
23. (a) J. E. McMurry, A. Andrus, G. M. Ksander, G. H. Musser, and M. A. Johnson,
J. Am. Chem. Soc., 101: 1330 (1979). (b) J. E. McMurry, A. Andrus, G. M. Ksander,
G. H. Musser, and M. A. Johnson, Tetrahedron 1981, 37, Suppl. i, 319 (1981).
TM

24. (a) S. P. Tanis, Y.-H. Chuang, and D. B. Head, Tetrahedron Lett., 26: 6147 (1985).
(b) S. P. Tanis, Y.-H. Chuang, and D. B. Head, J. Org. Chem., 53: 4929 (1988).
25. (a) Stoichiometric asymmetric epoxidation: T. Katsuki and K. B. Sharpless, J. Am.
Chem. Soc., 102: 5974 (1980). (b) Catalytic asymmetric epoxidation: R. M. Hanson
and K. B. Sharpless, J. Org. Chem., 51: 1922 (1986).
26. K. B. Sharpless and M. A. Umbreit, J. Am. Chem. Soc., 99: 5526 (1977).
27. J. A. Dale, D. L. Dull, and H. S. Mosher, J. Org. Chem., 34: 2543 (1969).
28. G. Stork, P. A. Grieco, and M. Gregson, Org. Synth., 54: 68 (1975).
29. A. Mancuso, and D. Swern, J. Org. Chem., 43: 2480 (1978).
30. For some examples of diastereoselective reductions of 3-alkoxy- and 3-hydroxyke-
tones see: (a) D. A. Evans and K. T. Chapman, Tetrahedron Lett., 27: 5939 (1986).
(b) T. Oishi, New Synthetic Methodology and Functionally Interesting Compounds.
Proceedings of the 3rd International Kyoto Conference on New Aspects of Organic
Chemistry (Z-i. Yoshida, ed.), Elsevier, Amsterdam, p. 81 (1986). (c) K. Narasaka
and F.-C. Pai, Tetrahedron, 40: 2233 (1984).
31. J. P. Kutney, N. Abduraham, C. Gletsos, P. LeQuesne, E. Piers, and I. Vlattas, J.
Am. Chem. Soc., 92: 1727 (1970).
32. R. H. Mitchell, Y. H. Lai, and R. V. Williams, J. Org. Chem., 44: 4733 (1979).
33. P. D. Williams and E. LeGoff, J. Org. Chem., 46: 4143 (1981). (b) P. D. Williams,
and E. LeGoff, Tetrahedron Lett., 26: 1367 (1985).
34. (a) S. P. Tanis, and P. M. Herrinton, J. Org. Chem., 50: 3988 (1985). (b) J. P. Marino
and H. Abe, J. Am. Chem. Soc., 103: 2907 (1981). (c) E. Ziegler and M. A. Cady,
J. Org. Chem., 46: 122 (1981).
35. For a synthesis and reactions of 65 when n ⫽ 1 see: S. Chang, R. M. Heid, and
E. N. Jacobsen, Tetrahedron Lett., 35: 669 (1994).
36. For a synthesis and reactions of 66 when n ⫽ 1 see ref. 34a, and S. P. Tanis, M. C.
McMills, and P. M. Herrinton, J. Org. Chem., 50: 5887 (1985).
37. For syntheses and reactions of 67 when n ⫽ 1 see: (a) J. P. Marino, and J. C. Jaen,
J. Am. Chem. Soc., 104: 3165 (1982). (b) P. A. Wender, J. M. Erhardt, and L. Leten-
dre, J. Am. Chem. Soc., 103: 2114 (1981).
38. See for example: ref. 34a, and E.-J. Brunke, J.-J. Hammerschmidt, and H. Struwe,
Tetrahedron, 37: 1033 (1981).
39. (a) N. H. Andersen and H. Uh, Tetrahedron Lett., 2079 (1973). (b) N. H. Andersen,
D. W. Ladner, and A. L. Moore, Synth. Commun., 8: 437 (1978).
40. J. A. Marshall, and P. G. M. Wuts, J. Org. Chem., 42: 1794 (1977).
41. (a) J. L. Cooper, and K. E. Harding, Tetrahedron Lett., 3321 (1977). (b) K. E. Har-
ding, J. L. Cooper, P. M. Puckett, and J. D. Ryan, J. Org. Chem., 43: 4363 (1978).
42. (a) W. Herz, S. Rajappa, S. K. Roy, J. J. Schmid, and R. N. Mirrington, Tetrahedron,
22: 1907 (1966). (b) G. R. Pettit, C. H. Herald, D. Gust, D. L. Herald, and L. D.
Vanell, J. Org. Chem., 43: 1092 (1978).
43. E. S. Glazer, R. Knorr, and J. D. Roberts J. Am. Chem. Soc., 94: 6026 (1972).
44. S. P. Tanis, M. C. McMills, T. A. Scahill, and D. A. Kloosterman, Tetrahedron
Lett., 31: 1977 (1990).
45. (a) T. Mukaiyama, M. Tamura, and S. Kobayashi, Chemistry Lett., 743 (1987). (b)
T. Mukaiyama, M. Tamura, and S. Kobayashi, Chemistry Lett., 1817 (1986). (c) T.
TM

Mukaiyama, M. Tamura, and S. Kobayashi, Chemistry Lett., 1805 (1986). (d) T.
Mukaiyama, M. Tamura, and S. Kobayashi, Chemistry Lett., 1017 (1986).
46. C. Gennari, M. G. Beretta, A. Berbardi, G. Moro, C. Scolastico, and R. Todeschini,
Tetrahedron, 42: 893 (1986).
47. S. P. Tanis, E. D. Robinson, M. C. McMills, and W. Watt, J. Am. Chem. Soc., 114:
8349 (1992).
48. (a) S. Masamune, Y. Hayase, W. Schilling, W.-K. Chan, and G. S. Bates, J. Am.
Chem. Soc., 99: 6756 (1977). (b) S. Masamune, S. Kanata, and W. Schilling, J. Am.
Chem. Soc., 97: 3515 (1975).
49. P. T. Lansbury, T. E. Nickson, J. P. Vacca, R. D. Sindelar, and J. M. Messinger,
II, Tetrahedron, 43: 5583 (1987).
50. P. G. Baraldi, G. P. Pollini, D. Simoni, B. Zanirato, A. Barco, and S. Benetti, Synthe-
sis, 781 (1986).
51. (a) M. Nishizawa, H. Takenaka, H. Nishide, and Y. Hayashi, Tetrahedron Lett., 24:
2581 (1983). (b) M. Nishizawa, H. Yamada, and Y. Hayashi, J. Org. Chem., 52:
4878 (1987).
52. J.-L. Luche, J. Am. Chem. Soc., 100: 2226 (1978).
53. (a) K. Chow, and S. Danishefsky, J. Org. Chem., 54: 6016 (1989). (b) L. Jeroncic,
M. Cabal, S. Danishefsky, and G. M. Schulte, J. Org. Chem., 56: 387 (1991) for an
interesting discussion of the factors which might be involved in such stereoselection.
54. (a) S. P. Tanis and D. B. Head, Tetrahedron Lett., 25: 4451 (1984). (b) I. Kuwajima
and H. Urabe, Tetrahedron Lett., 22: 5191 (1981).
55. J. M. Brown, Angew. Chem. Int. Ed. Eng., 26: 190 (1987).
56. G. Stork and D. E. Kahne, J. Am. Chem. Soc., 105: 1072 (1983).
57. (a) D. A. Evans and M. M. Morrissey, J. Am. Chem. Soc., 106: 3866 (1984). (b)
D. A. Evans, M. M. Morrissey, and R. L. Dow, Tetrahedron Lett., 26: 6005 (1985).
58. T. M. Stokes, and A. C. Oehschlager, Tetrahedron Lett., 28: 2091 (1987).
59. N. A. Babiak, J. S. Ng, J. H. Dygos, C. L. Weyker, Y.-F. Wang, and C.-H. Wong,
J. Org. Chem., 55: 3377 (1990).
60. S. P. Tanis, P. M. Herrinton, and L. A. Dixon, Tetrahedron Lett., 26: 5347 (1985).
61. The related thiophene and pyrrole 2-to-3 cyclizations have been examined: (a) B. E.
Maryanoff, D. F. McComsey, and B. A. Duhl-Emswiler, J. Org. Chem., 48: 5062
(1983). (b) C. G. M. Janssen, A. A. Macco, H. H. M. Buck, and E. F. Godefroi,
Recl. Trav. Chim. Pays-Bas., 98: 448 (1979). (c) A. A. Macco, R. J. De Brouwer,
P. M. M. Nossin, E. F. Godefroi, and H. M. Buck, J. Org. Chem., 43: 1591 (1978).
(d) A. Corvers, P. C. H. Scheers, J. W. deHaan, and H. M. Buck, Recl. Trav. Chim.
Pays-Bas., 96: 448 (1979). (e) J. Gourler, and P. Cannone, Can. J. Chem., 48: 2587
(1970).
62. G. Schulte, P. J. Scheuer, and O. J. McConnel, Helv. Chim. Acta, 63: 2159 (1980).
63. Y. Yamamoto, and K. Maruyama, J. Am. Chem. Soc., 100: 2183 (1978).
64. (a) K. B. Sharpless, R. F. Lauer, and A. Y. Teranishi, J. Am. Chem. Soc., 95: 6137
(1973). (b) H. J. Reich, J. M. Renga, and I. L. Reich, J. Am. Chem. Soc., 97: 5434
(1975).
65. J. M. Conia and J. C. Limasset, Bull. Soc. Chim. Fr., 6: 1936 (1967).
66. For recent work from this laboratory see: (a) T. Luker, H. Hiemstra and W. N.
TM

Speckamp, J. Org. Chem., 62: 3592 (1997). (b) N. M. Teerhuis, H. Hiemstra, and
W. N. Speckamp, Tetrahedron Lett., 38: 155 (1996).
67. (a) A. R. Chamberlin, H. D. Nguyen, and J. Y. Chung, J. Org. Chem., 49: 1682
(1984). (b) D. A. Evans, E. W. Thomas, and R. E. Cherpeck, J. Am. Chem. Soc.,
104: 3695 (1982).
68. S. P. Tanis and L. A. Dixon, Tetrahedron Lett., 28: 2495 (1985).
69. T. Fukuyama, L. V. Dunkerton, M. Aratani, and Y. Kishi, J. Org. Chem., 40: 2011
(1975).
70. T. Tsunoda, M. Suzuki, and R. Noyori, Tetrahedron Lett., 21: 1357 (1980).
TM

9
Chemistry and Biology of
Semisynthetic Avermectins
Timothy A. Blizzard
Merck Research Laboratories, Rahway, New Jersey
I. INTRODUCTION
The avermectins are a family of macrocyclic natural products discovered at

Merck in the late 1970s with useful anthelmintic and pesticidal activities [1–3].
After their discovery, a medicinal chemistry program was initiated to discover
analogues with improved activity against a broader spectrum of parasites, an
improved safety profile, and/or increased stability. The primary fermentation
product, avermectin B1 (1) (Fig. 1), was isolated from the fermentation of Strepto-
myces avermitilis as a mixture of two components. The major a component
(ⱖ80%) contains a sec-butyl side chain at C-25, whereas the minor b component
(ⱕ20%) has an isopropyl group at this carbon. Although the components can be
separated by high-performance liquid chromatography (HPLC) [4], in production
this is not normally done since the a and b isomers have essentially identical
biological activities. Thus, all compounds referred to in this article are actually
mixtures of a and b isomers, but for the sake of clarity only the a component is
shown [i.e., the structure (1) shown for avermectin B1 is actually the structure
of avermectin B1a]. Avermectin B1, also known as abamectin, is currently used
as an agricultural pesticide. Ivermectin (2) [5], the semisynthetic 22,23-dihydro
analogue of avermectin B1, is a widely used anthelmintic agent in animal health
[1]. Ivermectin also plays an important role in efforts to control onchocerciasis,
a serious human health problem in some parts of the world. Since the avermectins
not only are complex natural products with interesting structures but also have
major economic importance, there has been considerable interest in their chemi-
cal modification and total synthesis [6]. The field of avermectin chemistry has
TM

Figure 1 Structures of avermectin B1 (1) and ivermectin (2).
been widely reviewed [1–4,6] and will not be comprehensively reviewed again
here. This article will describe only work in the late stages of the medicinal
chemistry program with which I was involved, with reference to other work only
as necessary to place our work in context.
II. CHEMISTRY
A. Avermectin Epoxides
1. Avermectin B1 8,9-Epoxide
Avermectin B1 8,9-epoxide (3) (Fig. 2) is a highly active anthelmintic that was
synthesized at Merck as a potential agricultural pesticide [7]. Although 3 is more
stable than 1 because of the elimination of the photosensitive diene, the presence
of a potentially reactive epoxide functionality still raised concerns about the
chemical stability of 3. In order to address these concerns we explored the reactiv-
ity of 3 with various nucleophiles. Not surprisingly, we found that the epoxide
reacted readily with strong nucleophiles such as thiophenoxide [8]. This result
was not of great importance by itself; however, the fact that the thiophenol adduct
4 crystallized from methanol turned out to be highly significant. When we began
our studies on 3, the epoxide stereochemistry had not been unambiguously estab-
lished. X-ray diffraction analysis of the crystals of 4 allowed the definitive assign-
ment of the stereochemistry of 4 as 8S,9R. Since 4 is derived from 3 with inver-
sion at C-9, then 3 must be the α-epoxide. Although 4 proved to be useful for
structural determination, epoxide-opened analogues of 3, including 4, were uni-
formly inactive (see Section III.B for a discussion of bioactivity).
TM

Figure 2 Epoxide-opening reactions of avermectin B1 8,9-epoxide (3). (Source:
Adapted from Ref. 8.)
2. Avermectin B1 3,4-Epoxide
Although disappointed by the inactivity of epoxide-opened derivatives of 3, we
were encouraged by the activity of intact 3 and decided to explore the possibility
of preparing additional avermectin epoxides. Since 3 had been prepared by util-
izing a hydroxyl-directed epoxidation, we felt that it might be possible to ap-
ply similar chemistry to epoxidize the 3,4-double bond of 1. In fact, once the
C-7 hydroxyl group was blocked as the trimethylsilyl (TMS) ether 6 (Fig. 3) by
persilylation of 1 with bis(trimethylsily)trifluoroacetamide (BSTFA) followed by
selective hydrolysis of the secondary TMS ethers, the C-5 hydroxyl group effi-
ciently directed epoxidation of the 3,4 double bond. The β-3,4-epoxide 7 was
thus readily prepared [9].
As observed with the 8,9-epoxide, the 3,4-epoxide 7 also reacted readily with
strong nucleophiles to afford inactive ring-opened adducts (e.g., 8). In another par-
Figure 3 Synthesis of avermectin B1 3,4-epoxide (7) and reaction of 7 with thiophenol.

(Source: Adapted from Ref. 9.)
TM

allel with the 8,9-epoxide series, we were able to use the epoxide-opened analogue
8 to determine the stereochemistry of the epoxide, this time by careful analysis of
nuclear magnetic resonance (NMR) coupling constants [9]. In the 1H NMR spec-
trum of 8, the coupling constant J2,3 was found to be ⬇4 Hz, consistent with an α
orientation of the phenylthiolate group at C-3 (the corresponding isomer with a β
phenylthio substituent at C-3 should have J2,3 ⬇ 9 Hz). Since 8 is derived from 7
with inversion at C-3, 7 must be the β-epoxide, a conclusion that is also consistent
with the mechanism of the hydroxyl-directed epoxidation used to prepare 7.
Although both epoxides turned out to be somewhat of a disappointment as
starting materials for the synthesis of active avermectin analogues, the epoxide
work did result in some interesting chemistry. One especially interesting re-
arrangement of the 8,9-oxide is discussed in the next section.
B. Reactions of Natural Avermectins

1. Fragmentations and Rearrangements
Some of the most interesting avermectin chemistry involved unexpected fragmen-
tation or rearrangement reactions. An example from earlier work with avermec-
tins is shown in Fig. 4 [10]. Treatment of unprotected avermectin B1 8,9-oxide
3 with p-toluenesulfonic acid in wet THF afforded the expected diol 5 (Fig. 2).
However, under the same reaction conditions, the corresponding TMS ether 9
afforded the unexpected rearrangement product 10 (60%–70%). Since the two
secondary TMS ethers of 9 are cleaved almost immediately under these condi-
Figure 4 Acid-catalyzed rearrangement of avermectin B1 8,9-epoxide 9. (Source:

TM

tions whereas the more hindered tertiary silyl ether at C-7 survives for several
hours, it is likely that the C-7 OTMS group somehow facilitates the re-
arrangement, possibly via silyl transfer to the epoxide oxygen.
In addition to the epoxide work described, we were interested in exploring
the chemistry of avermectin analogues wherein the macrocyclic lactone had been
cleaved. Treatment of avermectin B1 (1) with methanolic diazabicycloundecene
(DBU) resulted in an unexpected fragmentation reaction (Fig. 5) [10]. Along with
the expected ring-opened product 11, we isolated a small amount of fragmenta-
tion product 12 in which a significant portion of the molecule had been removed.
Unfortunately, the destruction occurred in that part of the molecule that is most
important for biological activity and 12 was totally inactive.
Another novel fragmentation reaction was observed during our effort to
synthesize 13-epi-avermectins [11]. Glycosylation of the protected 13-epi-aver-
mectin B2a aglycone 13, prepared from avermectin B2a aglycone by inversion at
C-13 [12], using conditions similar to those employed by Nicolaou et al. in their
partial synthesis of avermectin B1 [13] resulted in only low yields of the desired
glycoside. We hypothesized that the free hydroxyl group at C-7 was interfering
in the glycosylation reaction so we prepared the 5,7,23-tris-protected aglycone
14 by persilylation of 13 with BSTFA followed by selective hydrolysis of the
secondary TMS ether at C-13 [11]. We were pleased to find that glycosylation
of 14 did, in fact, cleanly and rapidly afford one major product under the same
reaction conditions. However, the product was not the desired glycoside, but was
instead aldehyde 15 (Fig. 6). This unexpected fragmentation product is probably
formed by a vinylogous fragmentation/elimination reaction similar to the Grob
fragmentation [14–16]. Additional experiments established that the glycosyl do-
Figure 5 Base-catalyzed fragmentation of avermectin B1 (1). (Source: Adapted from

Ref. 10.)
TM

Figure 6 Fragmentation of 13-epi-avermectin B2 aglycones 13 and 14. (Source:
nor, in this case a glycosyl fluoride, was not involved in the reaction. However,
both SnCl2 and AgClO4 were required, for reasons that we do not currently under-
stand. Furthermore, the reaction was independent of C-13 stereochemistry, but
was greatly accelerated by the presence of a TMS ether at C-7, possibly as a
result of relief of steric strain. It was subsequently shown that 13 also produced
15 under the glycosidation conditions. Recall that we had previously observed
a similar accelerating effect of a 7-OTMS ether in the rearrangement reaction of
avermectin B1-8,9-oxide derivative 9 described.
2. Disaccharide Excision
Although the fragmentation and rearrangement reactions discussed previously
are potentially useful as sources of intermediates for avermectin synthesis, we
did not utilize them for that purpose. We did, however, apply the concept of using
a natural avermectin as a source of synthetic intermediates when we required a
source of the avermectin disaccharide. By modifying the multistep procedure
developed by Hanessian et al. [17], we were able to prepare the desired disaccha-
ride 17 successfully from bis-silylated avermectin B1 (16) on a multigram scale
in 57% yield via a one-pot procedure (Fig. 7) [18]. Reaction of disaccharide 17
with diethylaminosulfur trifluoride (DAST) provided the glycosyl fluoride 18.
This convenient preparation of large quantities of the disaccharide proved to be
essential in our later synthetic studies, as described later.
C. Avermectins with Modified Glycosidic Linkages

The availability of the avermectin disaccharide enabled us to pursue several inter-
esting avenues of research that had previously been inaccessible. For example,
TM

Figure 7 Preparation of the avermectin disaccharide 18. (Source: Adapted from Ref.
18.)
we were able to study the effect of modifying the linkage between the avermectin
aglycone and the disaccharide. Starting with ivermectin aglycone 19, we intro-
duced a variety of spaces (e.g., the ethylene glycol spacer shown in Fig. 8), then
attached the disaccharide to the new spacer to afford, after deprotection, ‘‘spacer-
mectin’’ analogues such as 23 and 24 [19]. Interestingly, some of the ‘‘spacer-
mectins’’ had activities approaching those of the natural compounds.
D. Stereoisomers of Natural Avermectins

1. 13-epi-Avermectins
In addition to allowing synthesis of the ‘‘spacermectins,’’ the efficient disaccha-
ride preparation enabled us to undertake the considerably more challenging syn-
thesis of 13-epi-avermectins. It was known that epimerization at C-2 of an aver-
mectin results in substantial loss of activity [1], so we were interested in exploring
the effect of epimerization at other stereocenters. We therefore prepared several
13-epi-avermectins, including 13-epi-avermectin B1 (30) and 1′,13-bis-epi-aver-
mectin B1 (31), by inversion at C-13 of the corresponding aglycone followed by
glycosylation with the avermectin disaccharide [12]. Although the synthesis of
these analogues, outlined in Fig. 9, was relatively straightforward, it was during
this work that we discovered the interesting fragmentation reaction discussed
earlier. We were pleased to find that the 13-epi-avermectins not only were as
TM

Figure 8 Synthesis of ivermectin analogues, the ‘‘spacermectins’’ 23 and 24, with a
spacer between the aglycone and the disaccharide. (Source: Adapted from Ref. 19.)
active as the natural isomers, but also had a substantially better therapeutic index,
which was due to their reduced mammalian toxicity [12].
2. 19-epi-Avermectin B1 (36)
Encouraged by the outstanding activity of the 13-epi-avermectins,we decided to
undertake the synthesis of 19-epi-avermectin B1 (36) (Fig. 10). The synthesis of
36 was complicated somewhat by the propensity of the avermectin C-3,4 double
bond to move into conjugation with the lactone carbonyl under basic conditions.
Thus, although formation of the seco-acid 32 and inversion of the C-19 stereo-
chemistry via Mitsunobu lactonization to give 33 are relatively straightforward,
reestablishment of the C-2 stereocenter is more difficult. After considerable ex-
perimentation, we were able to accomplish the deconjugation successfully by
TM

Figure 9 Synthesis of 13-epi-avermectin B1 (30). (Source: Adapted from Ref. 12.)
using a procedure based on methods previously developed by Hanessian et al.

[20,21] and refined by Danishefsky et al. [22] in their avermectin total-synthesis
studies. Thus, quenching of the vinylogous enolate formed upon deprotonation
of 7-OTMS-derivative 34 with the highly hindered pivalic acid gave the deconju-
gated penultimate intermediate 35. Subsequent deprotection of 35 completed our
synthesis of 19-epi-avermectin B1 (36) [23]. A synthesis of the related 19-epi-
avermectin A1 via a very similar route was independently accomplished by
Hanessian et al. [24]. Unfortunately, 36 turned out to be substantially less active
than avermectin B1.
III. BIOLOGY
Although much interesting chemistry was discovered during this avermectin proj-
ect, our primary interest was in the biological activity of the novel avermectin
analogues that we prepared. The next section describes some of our efforts to
measure biological activity and discusses the interesting structure–activity rela-
tionships that were uncovered.
TM

Figure 10 Synthesis of 19-epi-avermectin B1 (36). (Source: Adapted from Ref. 23.)
A. Development of a Brine Shrimp Assay for Avermectin

Analogues
The success of any medicinal chemistry project depends on the timely availability
of biological data for each new analogue prepared by the medicinal chemists.
The avermectin project was not exceptional in this regard. What was exceptional
about this project was that we were able to develop and implement an assay that
could be performed by chemists with no biological training or facilities. Since
the use of brine shrimp larvae as an assay for general toxicity is well known
[25–28], it seemed likely that we might be able to use a similar assay to evaluate
new avermectin analogues. A moderate amount of experimentation soon led to
a simple, highly reproducible assay that employed brine shrimp larvae to evaluate
new avermectin analogues rapidly [29]. As we had hoped, the assay was generally
predictive of activity in our more elaborate and time-consuming biological assays
and thus provided valuable feedback on a very timely basis. This proved most
TM

helpful in developing structure–activity relationships that greatly aided our selec-
tion of subsequent synthetic targets.
B. Structure–Activity Relationships of Semisynthetic

Avermectins
The brine shrimp assay results for the avermectin derivatives discussed in this
article are summarized in Table 1. The data in Table 1 allow several conclusions
to be drawn regarding which structural features of the avermectins are necessary
for optimal biological activity. For example, epimerization of the stereocenter at
C-13 results in compound 30, which is nearly as active as avermectin B1 (1)
(compare items 1 and 10). However, epimerization at C-2 (Item 13), C-19 (Item
12), or C-1′ (compare Items 10 and 11) leads to analogues that are much less
active. Also, epoxidation of the double bonds at C-3,4 (Item 6) or C-8,9 (Item
3) results in little, if any, loss of activity, but opening the resulting epoxide with
a nucleophile causes substantial loss of activity. The complete inactivity of 7-
OTMS-avermectin B1 (6) (Item 5) demonstrates the importance of the C-7 hy-
droxyl group, the moderate activity of the ‘‘spacermectin’’ 23 (Item 9) suggests
some flexibility in the exact spatial orientation of the disaccharide. The brine
shrimp assay thus allowed us quickly to delineate useful structure–activity rela-
Table 1 Activity of Avermectins vs. Brine Shrimp Larvae
Item Compound IC100 (µg/mL)a

1 Avermectin B1 (1) 335
2 Ivermectin (2) 430
3 Avermectin B1 8,9-epoxide (3) 650
4 8-SC6H5,-9-OH-8,9-H2-Avermectin B1 (4) ⬎55,500
5 7-OTMS-Avermectin B1 (6) ⬎55,500
6 Avermectin B1 3,4-epoxide (7) 430
7 3-SC6H5,4-OH-3,4-H2-Avermectin B1 (8) ⬎55,500
8 Avermectin B2 fragmentation product (15) ⬎55,500
9 Spacermectin (23) 1,730
10 13-epi-Avermectin B1 (30) 870
11 1′,13-bis-epi-Avermectin B1 (31) 10,415
12 19-epi-Avermectin B1 (36) 13,900
13 2-epi-Avermectin B1 ⬎55,500
a
IC100 ⫽ minimum concentration at which 100% of brine shrimp are immobile.
OTMS ⫽ o-trimethylsilyl.
TM

tionships that were, in general, confirmed by activity in our more traditional
assays.
IV. CONCLUSIONS
The most enjoyable aspects of the avermectin project were the rich diversity of
the chemistry and the multidisciplinary nature of the project, which incorporated
elements of both chemistry and biology. In this respect, the avermectin project
at Merck resembled many of the projects in Professor Nakanishi’s lab at Colum-
bia. In both cases, effective collaboration with scientists in other fields was not
only critical to success, but also a key element in making the project a lot more
fun for all concerned.
ACKNOWLEDGMENTS
I would like to thank Professor Koji Nakanishi for a highly educational and very
enjoyable postdoctoral experience. My years in his laboratory were the best possi-
ble preparation for the interdisciplinary nature of research in the pharmaceutical
industry. I am also indebted to the many excellent scientists at Merck who were
involved in the avermectin project, especially Dr. Helmut Mrozik and Ms. Gaye
Margiatto.
REFERENCES
1. W. C. Campbell, ed., Ivermectin and Abamectin, Springer-Verlag, New York (1989).

2. H. G. Davies and R. H. Green, Nat. Prod. Rept., 3: 87 (1986).
3. T. Blizzard, M. H. Fisher, H. Mrozik, and T. L. Shih, in Recent Progress in the
Chemical Synthesis of Antibiotics (G. Lukacs and M. Ohno, eds.), Springer-Verlag,
Berlin, pp. 65–102 (1990).
4. T. W. Miller and V. P. Gullo, in Natural Products Isolation—Journal of Chromatog-
raphy Library Series, Vol. 43 (G. H. Wagman and R. Cooper, eds.) Elsevier, Amster-
dam, chapter 9, pp. 347–376 (1989).
5. J. C. Chabala, H. Mrozik, R. L. Tolman, P. Eskola, A. Lusi, L. Peterson, M. Woods,
M. H. Fisher, W. C. Campbell, J. R. Egerton, and D. A Ostlind, J. Med. Chem., 23:
1134 (1980).
6. T. Blizzard, Org. Prep. Proc. Int., 26: 617 (1994).
7. H. Mrozik, U.S. Patent 4,530,921 (1985).
8. T. A. Blizzard, H. Mrozik, and M. H. Fisher, Bioorg. Med. Chem. Lett., 3: 2093
(1993).
TM

9. T. A. Blizzard, H. Mrozik, F. A. Preiser, and M. H. Fisher, Tetrahedron Lett., 31:
4965 (1990).
10. T. A. Blizzard, H. Mrozik, and M. H. Fisher, Tetrahedron Lett., 29: 3163 (1988).
11. T. A. Blizzard, G. Margiatto, H. Mrozik, and M. H. Fisher, J. Org. Chem., 58: 3201
(1993).
12. T. A. Blizzard, G. Margiatto, H. Mrozik, W. L. Shoop, R. A. Frankshun, and M. H.
Fisher, J. Med. Chem., 35: 3873 (1992).
13. K. C. Nicolaou, R. E. Dolle, D. P. Papahatjis, and J. L. Randall, J. Am. Chem. Soc.,
106: 4189 (1984).
14. C. A. Grob, Angew. Chem. Int. Ed. Eng., 8: 535 (1969).
15. M. Ochiai, T. Ukita, S. Iwaki, Y. Nagao, and E. Fujita, J. Org. Chem., 54: 4832
(1989).
16. H. B. Henbest and B. B. Millward, J. Chem. Soc., 3575 (1960).
17. S. Hanessian, A. Ugolini, P. J. Hodges, and D. Dube, Tetrahedron Lett., 27: 2699
(1986).
18. T. Blizzard, G. Marino, H. Mrozik, and M. H. Fisher, J. Org. Chem., 54: 1756
(1989).
19. T. Blizzard, G. Margiatto, B. Linn, H. Mrozik, and M. H. Fisher, Bioorg. Med.
Chem. Lett., 1: 369 (1991).
20. S. Hanessian, A. Ugolini, D. Dube, P. J. Hodges, and C. Andre, J. Am. Chem. Soc.,
108: 2776 (1986).
21. S. Hanessian, D. Dube, and P. J. Hodges, J. Am. Chem. Soc., 109: 7063 (1987).
22. S. J. Danishefsky, D. M. Armistead, F. E. Wincott, H. G. Selnick, and R. Hungate,
J. Am. Chem. Soc., 111: 2967 (1989).
23. T. Blizzard, L. Bostrom, G. Margiatto, H. Mrozik, and M. Fisher, Tetrahedron Lett.,
32: 2723 (1991).
24. S. Hanessian and P. Chemla, Tetrahedron Lett., 32: 2719 (1991).
25. A. S. Michael, C. G. Thompson, and M. Abramovitz, Science, 123: 464 (1956).
26. J. Harwig and P. M. Scott, Appl. Microbiol., 21: 1011 (1971).
27. M. G. Prior, Can. J. Comp. Med., 43: 352 (1979).
28. B. N. Meyer, N. R. Ferrigni, J. E. Putnam, L. B. Jacobsen, D. E. Nichols, and J. L.
McLaughlin, Planta Med., 45: 31 (1982).
29. T. A. Blizzard, C. L. Ruby, H. Mrozik, F. A. Preiser, and M. H. Fisher, J. Antibiotics,
42: 1304 (1989).
TM

10
Chemical and Biological
Approaches to Molecular Diversity:
Applications to Drug Discovery
Harold V. Meyers
New Chemical Entities, Inc., Framingham, Massachusetts
I. INTRODUCTION
Despite recent scientific advances in the pharmaceutical industry, drug discovery

costs have steadily increased while the rate of introduction of useful pharmaceuti-
cals has decreased. The average cost for introducing a new drug entity into the
marketplace has recently been estimated at more than $400 million, nearly a
third of which goes to the preclinical tasks of discovery and optimization of lead
chemical compounds. The synthetic preparation of each novel molecule in the
traditional serial fashion alone has been estimated to cost between $5,000 and
$10,000. From the economic viewpoint, there is clearly room for significant im-
provement in the discovery process.
An exciting advance in the industry, the ability to screen hundreds of thou-
sands to millions of compounds in relatively short time frames, has been made
possible with the recent advent of automated, high-throughput screening (HTS)
technologies. Such capacity often exceeds the supply of available compounds.
Compound libraries typically used in mass screening consist of either historical
collections synthesized in drug programs, or natural products collections. Al-
though historical collections can number in the hundreds of thousands of com-
pounds at large pharmaceutical firms, they often have limited structural diversity
(e.g., many β-lactams, steroids), whereas natural product collections can be disad-
vantageous because of their structural complexity, which renders making syn-
thetic analogues to generate drug candidates potentially very difficult.
TM

During the last decade a new source of compounds has arisen with the
advent of combinatorial chemistry, which has the effect of providing large num-
bers of compounds for biological testing at a greatly reduced cost. In conjunction
with HTS, combinatorial chemistry can enhance the speed to market of a drug
by reducing development timelines. For the development of novel and proprietary
drug candidates, combinatorial chemistry enables many unique chemical struc-
tures to be synthesized rapidly, thereby allowing for broader patent coverage
around lead compounds. This can diminish the reliance on ‘‘me too’’ drug strate-
gies of many companies, whereby clinical compounds are derived from typically
conservative structural variations of competitors’ compounds having narrow pat-
ent coverage. As the marketplace rewards innovation and places a premium on
novel, efficacious drugs, being the first to market with a novel compound can
provide enormous financial rewards. Furthermore, by substantially reducing the
period for drug candidate selection, the useful patent life of a product can be
significantly extended. For a big-selling drug (e.g., $1 billion/year) this can trans-
late into hundreds of millions of dollars of additional sales [1].
The bottom line is, combinatorial chemistry accelerates the drug discovery
process in a more economical manner. This chapter will provide an overview of
combinatorial chemistry methodologies as applied to drug discovery, concluding
with an assessment of future directions.
II. SYNTHESIS AND SCREENING OF COMBINATORIAL

LIBRARIES
This section discusses some of the key biological and chemical methodologies
available for the creation of molecular diversity. In particular, the molecular bio-
logical generation of peptide and protein libraries by phage display techniques
is discussed; then an overview of chemical strategies and methods developed for
the combinatorial synthesis of peptide and small molecule libraries used in drug
discovery and lead optimization processes is presented. Several good review arti-
cles describing molecular biological and chemical methodologies are available
for a more comprehensive discussion of these topics [2–6].
A. Biological Approaches
Several techniques for creating large libraries of filamentous phage clones dis-
playing unique peptides or proteins on their protein coat surface have been devel-
oped [7–11]. By inserting fully random cassettes of synthetic oligonucleotides
into targeted loci, tens of millions of mutant strains can be isolated [12–15].
Each unique peptide is structurally encoded by its respective single-stranded viral
deoxyribonucleic acid (DNA) sequence and can be expressed as copies as a fusion
TM

peptide with either the minor-coat protein (pIII) or the major-coat protein (pVIII),
respectively. Phage display methods have been widely used in the area of epitope
mapping, peptide drug discovery, and protein engineering to study structure/func-
tion relationships.
As phage libraries are large enough to preclude analysis of individual mu-
tants, population selection schemes are necessary. In the ‘‘colony-lift’’ technique,
a polymeric membrane is placed on a high-density clone colony and lifted off
to adsorb the expressed macromolecules that mirror their location on the growth
colony. A labeled protein or antibody that binds to the membrane then reveals the
location of the clone, which can be further analyzed [16]. An affinity purification
technique termed biopanning [8] allows the affinity selection of peptides from
phage display libraries by biological targets (ligates). A biotinylated ligate is
incubated with the phage library in solution, and then a ‘‘panning’’ step captures
the phage-bound ligate with a streptavidin-coated polystyrene Petri dish. Alterna-
tively the library is panned using an immobilized ligate that has been coated
directly in the Petri dish [11]. The unbound phage is washed away, and the bound
phage can be eluted and propagated in Escherichia coli for DNA sequencing to
reveal the identity of active peptides. Often several iterations of biopanning are
performed in a selection/amplification process to find high-affinity peptides to
the target(s) of interest.
An application of phage display peptide libraries, ‘‘epitope libraries,’’ have
been used in mapping the specificity of antibodies. Peptides that mimic antibody-
binding determinants through continuous [9,10] and discontinuous epitopes [17],
the latter composed of adjacent residues in a folded or conformationally depen-
dent structure but distant in primary sequence, have been discovered. Epitope
mapping enables ligands to be found for antibodies whose specificity is not
known in advance.
The screening of phage peptide libraries against receptors of pharmaceuti-
cal interest has been an active area of research. Conformationally constrained
peptide libraries, e.g., loops derived from cysteine disulfide formation, have also
been used to find higher-affinity ligands to receptors such as gpIIb/IIIa [18], as
cyclic peptides provide better leads for drug development.
The application of phage display libraries in the field of protein engineering
has made possible the study of structure/function relationships, for example, with
human growth hormone [19] and alkaline phosphatase [20]. The proper folding of
protein domains on the phage surface has been demonstrated to afford functional
receptors with respect to binding [19] and catalytically active enzymes [20]. The
construction of numerous variants of a parental protein by phage display tech-
niques has allowed the alteration of catalytic properties of enzymes and binding
properties of proteins, as exemplified with potent protein inhibitors of human
neutrophil elastase [21], through iterative affinity selection/amplification meth-
ods against selected targets (protein biopanning), an area of research that has
TM

come to be known as ‘‘directed evolution.’’ Applied evolution methods will un-
doubtedly improve industrial processes that utilize enzymes, among other things.
In the antibody engineering area, the development of customized antibod-
ies, especially completely humanized antibodies, with high affinities and selectiv-
ities holds considerable promise for their use as therapeutic agents. In a process
that mimics an immune response for high-affinity antibody selection in vivo,
‘‘in vitro affinity maturation’’ utilizes the introduction of sequence variation and
recombination of both light and heavy chains as well as mutagenesis in the com-
plementarity determining regions (CDRs) to find enhanced binding with an anti-
gen. This selection process may be repeated with an enriched antibody pool until
suitable candidates are isolated [22]. Other applications include screening for
catalytic and metal coordinating antibodies [23].
Limitations of phage display techniques for drug discovery purposes in-
clude the use of only naturally occurring amino acids, and little is known about
how the phage environment affects molecular recognition events. These de-
scribed techniques are suitable to isolate clones from cDNA or genomic libraries
of up to ⬃107 clones, but are not practical for larger libraries, which are better
suited for drug discovery projects (107 –109 clones). Toward that end more power-
ful biological peptide display methods have recently been described. One method
involves the display of peptides on plasmids, an alternative scheme that links the
peptide and DNA by the DNA-binding protein LacI [24]. An in vitro system that
displays nascent peptides in polysome complexes has been very recently de-
scribed [25]; it has the potential to create polysome libraries of 1014 –1015 mem-
bers.
Perhaps the biggest drawback of utilizing biological display methods for
drug discovery is that if affords peptides. It is well known in the pharmaceutical
industry that peptides are not generally well suited as drugs because of in vivo
instability and lack of oral absorption. Furthermore, converting a peptide lead
into a peptide mimic is far more difficult than identifying the peptide lead, since
a general solution for designing pharmaceutically useful, orally available peptide
mimics has yet to be developed [26]. These drawbacks have provided the impetus
for the pharmaceutical industry to apply combinatorial chemistry methods to the
synthesis of small-molecule, nonpeptidyl libraries for drug discovery.
B. Chemical Approaches
Early chemical combinatorial efforts were focused on synthesizing libraries of
peptides [27], followed by the progression to peptidelike oligomeric libraries
[28], and most recently to small-molecule, nonpeptidyl libraries [29,30]. The use
of solid-phase chemistry, originally developed for peptide synthesis [31] and later
amply demonstrated for organic synthesis [32–34], has experienced a renaissance
of late spurred by combinatorial chemistry. Solid-phase synthesis precludes often
TM

tedious reaction workup and purifications typically performed after each synthetic
step. These advantages are amplified when constructing even relatively small
libraries of hundreds of compounds. In addition, reactions can be driven to com-
pletion on solid supports by employing a large excess of reagents. Hence, many
combinatorial approaches rely on solid-phase chemistry, but isolation and purifi-
cation techniques recently applied to parallel solution synthesis are enhancing
the utility of this approach (discussed later).
The key issues that differentiate the various combinatorial approaches,
then, involve the use of solid-phase or solution chemistry, the synthesis of indi-
vidual or mixtures of compounds, the size of the library and application of encod-
ing strategies, the extent of automation for various syntheses, purification, and
analytical tasks, and the screening of libraries, on or off solid supports. Regardless
of the approach, it is desirable to have each compound in a library present in
a roughly equimolar concentration for screening purposes. What follows is a
description of some key combinatorial chemistry technologies and their inherent
advantages and disadvantages.
1. Synthesis of Mixtures
Combinatorial libraries can be synthesized and screened as mixtures. The synthe-
sis of large mixtures of compounds (105 –108) was made possible by employing
the split synthesis approach on solid supports.
a. Split Synthesis (Split-Pool, Split-Mix, or Portion Mixing). Furka first dem-
onstrated the split-mix strategy on solid supports to synthesize large mixtures of
peptides in equimolar quantities rapidly [35]. This one bead/one unique peptide
approach is schematically represented (Fig. 1). A solid support material is split
into equal-sized portions and placed in separate reaction vessels. Each portion is
coupled with excess building blocks (e.g., unique amino acid derivatives), which
provide uniform coupling since competition between reactants is eliminated. The
coupled resins are then pooled together into one reaction vessel for the removal
of common protecting groups. The resin is partitioned again into separate pools,
each containing mixtures of unique peptides, for coupling new reactants. This
iterative split-couple-mix strategy theoretically enables libraries of oligomers of
Xn members to be assembled, where X is the number of monomers, amino acids
in this example, in the basis set and n is the length of the oligomer. A pure
statistical distribution of sequences results.
When screening split-pool libraries on solid supports, bioactive compounds
can be physically isolated through incubation with a tagged ligate. Polymeric
beads possessing peptides bound to a ligate have been identified by visual inspec-
tion, physical removal, and microsequencing of the active peptides [36]. Alterna-
tively, the supported peptides are cleaved into solution for screening. When
screening mixtures of compounds in solution from split-synthesis libraries, sev-
TM

Figure 1 The split-pool method: the shaded circles represent the beads (solid support)
and the lettered squares are individual chemical building blocks, for example, individual
amino acids.
TM

eral deconvolution methods for determining the structures of compounds possess-
ing the most bioactivity have been developed. These include ‘‘iterative’’ or ‘‘re-
cursive’’ deconvolution methods involving resynthesis of bioactive pools [37,38],
methods that preclude the resynthesis of bioactive pools such as ‘‘positional scan-
ning’’ [39] and ‘‘orthogonal’’ libraries [40], and encoded methods (discussed
late). It has been demonstrated, however, that increasingly larger pool sizes (ca.
100 or more) result in higher rates of false-positive and false-negative bioactives
in screens. The testing of mixtures creates the potential for compounds to interact
with one another, leading to possible increased measured potencies through syn-
ergistic effects (false-positive) or decreased or antagonistic effects (false-nega-
tive) on the measured potencies as compared with those of pure compounds.
Also, as the number of compounds in a library increases, the concentration of
each compound typically decreases to maintain solubility of all members, and
that can preclude detecting members with moderate activities. Nevertheless, split
synthesis is generally considered one of the most powerful tools for lead discov-
ery, as the large number of compounds it provides should statistically offer better
chances of uncovering lead compounds against biological targets of interest.
Encoding strategies. The assembly of very large libraries (e.g., 106 –108)
generally results in minute quantities of compound (e.g., 10⫺9 –10⫺15 mole), neces-
sitating an encoding scheme to identify structurally any component of interest
from a biological assay. Several methods that utilize molecular tagging systems
to encode structures in combinatorial libraries have been developed. In these
approaches a molecular ‘‘tag’’ is attached to the solid support and used as an
identifier for each monomer used in split synthesis. Hence, an orthogonal and
compatible solid-phase synthesis scheme is needed for both the library members
and their respective tags. One method utilizes single-stranded oligonucleotides
to encode each library member, wherein a unique oligonucleotide sequence en-
codes each library member. Oligonucleotides have the advantage of allowing
their amplification through polymerase chain reaction (PCR) for delineating their
sequence by DNA sequencing methods [41–43]. The limited stability of oligonu-
cleotides, however, can preclude the application of some chemistries and hence
the accessibility of certain compound classes. In another method, natural amino
acids are used to construct peptide tags for a library of peptide-based oligomers
(‘‘binding’’ strands), whereby bioactive constituents in a library are decoded
through Edman degradation [44]. The binding strand can be cleaved from the
solid support for solution assays prior to decoding. A potential limitation of this
approach is that peptide tags can potentially interact with biological targets. The
use of halocarbon derivatives as molecular tags in a binary encoding scheme was
more recently demonstrated; in this method each tag is photolytically cleaved
and identified by electron capture capillary gas chromatography [45]. Unlike the
oligonucleotide and peptide encoding approaches, which record the order of as-
TM

sembly of chemical building blocks in the tag sequence, the binary approach uses
uniquely defined mixtures of tags to represent each building block of the binding
strand at a particular synthetic step.
Recent reports on the use of radiofrequency-encoded tags preclude the need
for molecular tags altogether. In these methods a microchip capable of receiving,
storing, or emitting radiofrequency signals is encapsulated with the synthesis
resin in a polymeric vessel to record each unique synthesis site with a unique
signal. Radiofrequency encoding is an easy and rapid method for decoding chemi-
cal structures of interest and does not interfere with any chemical steps in the
library synthesis [46,47].
b. Mixture Synthesis. In this approach an equimolar mixture of reactants is
placed in a reaction vessel with a single substrate theoretically to generate an
equimolar mixture of products. This strategy was employed in an early peptide
library synthesis [27]. Although this is the most direct method to construct mix-
tures, a limitation of this approach is that reactivity is highly dependent on the
structure of reactants, so equimolarity will only be achieved if reactants have
comparable reactivity with the substrate. An equimolar mixture of products can
also result in principle from reacting 1 equivalent of an equimolar mixture of
reactants with a substrate; however, this requires very efficient coupling reactions
to take place [48].
2. Multiple Simultaneous Synthesis (Parallel, Discrete, Array,

or Spatially Separate Synthesis)
The multiple simultaneous synthesis approach ideally constructs one target com-
pound per reaction vessel, and therefore is not very practical for libraries con-
taining more than 104 –105 members. Although more labor-intensive than split
synthesis, automation can largely diminish the tedious, repetitive nature of con-
structing these libraries. Parallel synthesis libraries are useful tools for lead dis-
covery, especially if they are well designed and of good quality, and are the
method of choice in the drug industry for optimizing lead compound identifica-
tion. Structure/activity data are more reliable on individual compounds than on
pools of compounds, and the screening of single, structurally defined molecules
has a proven track record in the industry. Moreover, the need for encoding strate-
gies is precluded as micromolar quantities of each compound are available from
many parallel synthesis methods, enabling routine spectroscopic techniques to
be applied directly to obtain structural information. As little as 10 micromoles
of each compound is sufficient to perform multiple HTS assays over a period of
several years.
a. Pin Technology. Geysen first demonstrated the multiple simultaneous syn-
thesis of peptides on polyacrylic acid–grafted polyethylene pins arranged in a
microtiter plate format, which allows the construction of 96 unique peptides per
TM

plate [49]. This technique utilizes conventional peptide synthesis methods and
has been used to construct thousands of peptides within several weeks. The pep-
tides can be screened on the pins, enabling them to be reused for additional tests,
or they can be cleaved and screened in solution [50].
b. Tea Bag Method. In this version of multiple peptide synthesis, porous poly-
propylene containers, dubbed ‘‘tea bags,’’ are used to enclose the synthesis beads
[51]. Each bag contains a unique peptide resin, which is immersed in individual
solutions of activated amino acid derivatives to effect coupling reactions, while
washing and deprotection steps are all carried out by pooling the bags together,
then portioning the bags for subsequent coupling steps. Cleavage of the peptides
from their solid support provides a unique, soluble peptide from each individual
bag. This technique has been partially automated to permit the synthesis of up
to 150 different peptides [52], and sufficient quantities of material can be gener-
ated for purification and complete characterization (approx. 0.5 mmol).
c. Spatially Addressable, Light-Directed Synthesis. A method combining
solid-phase synthesis with photolithography has been developed for the parallel
synthesis of more than 100,000 discrete peptides [53]. This was accomplished
by anchoring amine-blocked alkyl groups on a silicon chip with photolabile pro-
tecting groups. By manipulating a photolithographic mask, specific regions of
the chip are deblocked by light, exposing the reactive amine functionality at select
locations for subsequent coupling reactions with activated amino acids. The ‘‘ad-
dress’’ or location of the compound on the chip, determined by the pattern of
masks and sequence of reactants, reveals its structure. A limitation of this ap-
proach is that the library must be screened on the silicon chip. This requires the
use of a tagged, soluble ligate, precluding certain receptors and enzyme targets.
The screening of any compound on a solid support also introduces inherent uncer-
tainty regarding the validity of molecular recognition events as compared to its
soluble counterpart since the density of compounds on the support, the support
itself, and other factors may perturb relevant solution conformations, binding
kinetics, and other critical factors.
d. Solution Phase Synthesis. Methods for parallel solution phase synthesis
have recently been reported [54,55]. These methods essentially rely upon classic
aqueous acid and aqueous base extractions for purification of intermediates and
final products with selective unmasking of acidic and basic functional groups
within the intermediates and final products to control solubility. The major limita-
tions to performing solution phase parallel synthesis have been the isolation and
purification of the products of large libraries. The use of parallel purification
techniques such as liquid/liquid extraction or solid-phase extraction, however,
makes solution phase parallel synthesis an attractive alternative to solid-phase
synthesis. The numerous solid supported reagents and catalysts for organic syn-
TM

thesis that are available [56] and scavenging resins for selective removal of prod-
ucts or excess reagents in solution will strengthen the utility of solution phase
synthesis, especially for multistep sequences. Multicomponent condensation re-
actions like the Ugi reaction [57] have also been demonstrated in parallel synthe-
sis to generate relatively complex molecules [58]. The advantages of performing
parallel synthesis in solution include the larger scales possible for performing
reactions, which provide more of each compound. In addition, the development
or utilization of functionalized resins and of cleavable linker groups, the attach-
ment and cleavage of substrates, and any capping steps are precluded. Solution
phase parallel manipulations are readily automated as well [59].
In a related process, a soluble polymer support for conducting synthesis in
solution has been reported [60]. Precipitation of the polymer after each synthetic
transformation for washing provides the advantages of both solution and solid-
phase synthesis.
e. Sphinx Approach. An approach taken at Sphinx was to create large numbers

of small, nonpeptidyl molecules in a parallel format as a general discovery tool.
The concept of a ‘‘universal library’’ has driven our efforts in this field (discussed
late). The libraries are synthesized on solid supports but are made available in
solution for screening and are generated in sufficient quantities (milligrams) for
use in multiple screens.
The initial parallel synthesis demonstration at Sphinx involved the prepara-
tion of several hundred substituted phenols in good yields and high purity levels
[61]. A simple, economical apparatus for multiple simultaneous synthesis was
constructed and successfully utilized for the rapid preparation of organic mole-
cules via multistep resin-based procedures (Fig. 2). A 96-well format was chosen,
Figure 2 Reaction plate and clamp assembly (side view).
TM

as used in numerous biological applications, including automated high-
throughput screens. Plate synthesis was carried out on cross-linked polystyrene
resin placed in sterile polypropylene deep well plates, which were modified for
filtration by drilling a small hole in the bottom of each well, then placing a porous
polyethylene frit into the bottom of each well. An aluminum plate clamp was
made as a two-piece assembly, consisting of a solid base clamp fitted with four
removable corner stainless steel studs and a frame clamp that fits atop the plate
and is secured with wing nuts. A Viton gasket was utilized on the base clamp
to prevent leakage of well contents. A vacuum plenum can be used to facilitate
the filtration of resin washings and isolation of solution libraries. This is accom-
plished by removing the modified plate from the clamp assembly after the syn-
thetic transformation and placing it atop the vacuum plenum designed to accom-
modate the plate. The well caps are then removed and the vacuum source
connected to an outlet in the plenum floor. The vacuum plenum is deep enough
to place a rack of 96 microdilution tubes beneath the plate to collect the individual
product solutions. The inexpensive equipment and limited human resources re-
quired for this effort illustrate the economic and productivity advantages of this
technology.
Universal library concept. Biological macromolecules (e.g., receptor, en-
zyme, antibody) recognize binding substrates or ligands through a number of
precise physicochemical interactions. These interactions can be divided into a
number of different parameters: size, hydrogen bonding ability, hydrophobic in-
teractions, dipole interactions, and others. To explore multiparameter space rap-
idly, the Sphinx libraries were designed to orient groups responsible for these
binding interactions at unique locations in space through a scaffolding approach.
A major challenge was to select a class of target molecules of sufficient generality
to allow for wide structural variations, preferably utilizing well-precedented syn-
thetic methodologies. The biphenyl scaffold met these criteria and was chosen
as the basic structure for a class of target molecules that allow for facile introduc-
tion of three or four functional groups in a large number of spatial arrangements
(Fig. 3). The use of aromatic templates such as the biphenyl moiety also enables
the size, shape, and other properties of these molecules to be changed readily by
incorporating different linker groups between the aromatic rings while varying
Figure 3 General structure for a universal library using a biphenyl scaffold.
TM

their substitution patterns, and derivatizing these scaffolds with a diverse set of
functional groups using related sequences of chemical reactions. A large number
of compounds prepared around each scaffold can thus explore unique sizes,
shapes, and volumes. A collection of such libraries will represent a ‘‘universal’’
library designed to explore multiparameter space (‘‘diversity’’ space) incremen-
tally and will result in the effective identification of chemical leads for any biolog-
ical target of interest. An initial report on the solid-phase synthesis of function-
alized biphenyls, which utilizes Mitsunobu chemistry to introduce functional
groups via ether bond formation, has appeared recently [62].
The universal library is being generated using a double combinatorial strat-
egy, a novel approach developed at Sphinx (Fig. 4). In this scheme functional
groups are introduced onto the first scaffold building block (the first phenyl ring)
in the first combinatorial cycle bound to a solid support. Then, the second scaffold
building block (another phenyl ring) is covalently attached and the second round
of functional group introduction takes place. The final target molecule is cleaved
from the solid support to afford the desired product in solution and available for
screening. The key step in the biphenyl library synthesis was a Stille coupling
to attach the second phenyl scaffold, thereby forming the biphenyl structure (Fig.
5). This double combinatorial approach enables large numbers of highly function-
alized, low-molecular-weight molecules to be assembled rapidly from a small
collection of building blocks.
Figure 4 Synthetic strategy for the ‘‘double combinatorial’’ approach to the biphenyl
library.
TM

Figure 5 Solid-phase Stille coupling to complete biphenyl scaffold.
III. CONCLUSIONS
The field of chemical diversity generation is experiencing enormous growth

within the pharmaceutical industry. Almost every major pharmaceutical and bio-
technology company has initiated, collaborated on, or acquired an effort in this
area. Successes have been disclosed in both the lead generation and optimization
of potential drugs, and an increase in the number of clinical candidates arising
from combinatorial chemistry can be expected. Improved methods are evolving
for the design, synthesis, and analysis of combinatorial libraries, particularly in
the realm of small organic molecules. Solution chemistry will continue to be
adapted to solid supports to expand the repertoire of solid-phase reactions. Even-
tually one can expect that most solution reactions will have a solid-phase counter-
part, and some solid-phase chemistry having no solution phase counterpart will
develop. New types of solid supports, including organic polymers and inorganic
materials, are being developed for organic synthesis to, among other things, in-
crease loading capacities and product yields. In the field of natural products,
recent demonstrations of combinatorial analoging may provide a resurgence for
natural product screening in the pharmaceutical industry. More active research
in the areas of combinatorial biosynthesis to generate diverse and structurally
complex libraries, particularly through the genetic manipulation of polyketide
biosynthetic pathways [63–65], can also be expected.
Advances in automation and instrumentation are also rapidly evolving for
the synthesis of chemical libraries. Commercial vendors now provide robotic
workstations for partially or fully integrated synthesis solutions, albeit at substan-
tial cost in some cases, but more practical and affordable robotic systems are
being developed for eventual use by each and every bench chemist. Parallel, high-
throughput purification and spectroscopic analysis and detection techniques (e.g.,
high-performance liquid chromatography–mass spectrometry [HPLC-MS], high-
performance liquid chromatography–nuclear magnetic resonance [HPLC-NMR];
TM

see Chapter 4) under development will eventually provide high-quality, well-
characterized libraries. A trend toward better integration of synthesis and screen-
ing, whereby both chemical and biological methods of diversity generation can
be coupled directly to screening protocols, is emerging as well. Furthermore,
progress will likely drive combinatorial synthesis and screening toward miniatur-
ization.
New computational methods will also provide a more complete understand-
ing of ‘‘diversity’’ space, what it is and how it is measured, for more effective
drug design. For example, the desirable physicochemical properties of known
drugs, such as bioavailability, can be computationally culled to design ‘‘drug-
like’’ libraries of compounds. Empirically derived structure/activity data can also
be used in computational analyses to optimize the properties of lead compounds
more rapidly. Genetic algorithms are also being applied in library design strate-
gies to uncover potent compounds more efficiently through iterative search and
selection protocols of ‘‘virtual libraries’’ to determine which subset populations
to synthesize [58,66]. As an explosion of chemical, physical, biological, and com-
putational data is resulting from combinatorial libraries, it will be critical to man-
age, integrate, and analyze these data effectively and efficiently to maintain a
competitive edge.
Advances in the field of genomics, in which genetic information derived
from high throughput sequencing and analysis (‘‘bioinformatics’’), will likely
result in an enormous number of new and relevant biological targets for screen-
ing. Combinatorial libraries will be useful in not only identifying new drug candi-
dates, but evaluating and validating the physiological relevance of new targets
in a much shorter time frame than traditional discovery protocols have demon-
strated.
With the rapid growth in the chemical generation of molecular diversity,
combinatorial libraries will significantly decrease the time and cost of discovering
novel therapeutic agents for multiple disease states. The development of combina-
torial chemistry methods and their application in the discovery process represents
one of the most significant paradigm shifts in the pharmaceutical industry in
decades and has become an important and indispensable tool.
REFERENCES
1. P. van Eikeren, in Chiral Separations, Applications and Technology (S. Ahuja, ed.)
American Chemical Society, Washington, D.C., chapter 2 (1997).
2. W. H. Moos, G. D. Green, and M. R. Pavia, Annu. Rep. Med. Chem., 28: 315 (1993).
3. M. R. Pavia, T. K. Sawyer, and W. H. Moos, Bioorg. Med. Chem. Lett., 3: 387
(1993).
TM

4. M. A. Gallop, R. W. Barrett, W. J. Dower, S. P. A. Fodor, and E. M. Gordon, J.
Med. Chem., 37: 1233 (1994).
5. E. M. Gordon, R. W. Barrett, W. J. Dower, S. P. A. Fodor, and M. A. Gallop, J.
Med. Chem., 37: 1385 (1994).
6. L. A. Thompson and J. A. Ellman, Chem. Rev., 96: 555 (1996).
7. G. P. Smith, Science, 228: 1315 (1985).
8. S. F. Parmley and G. P. Smith, Gene, 73: 305 (1988).
9. S. Cwirla, E. A. Peters, R. W. Barrett, and W. J. Dower, Proc. Natl. Acad. Sci. USA,
87: 6378 (1990).
10. J. K. Scott and G. P. Smith, Science, 249: 386 (1990).
11. J. J. Devlin, L. C. Panganiban, and P. E. Devlin, Science, 249: 404 (1990).
12. M. D. Matteucci and H. L. Heyneker, Nucleic Acids Res., 11: 3113 (1983).
13. J. A. Wells, M. Vasser, and D. B. Powers, Gene, 34: 315 (1985).
14. M. S. Z. Horwitz and L. A. Loeb, Proc. Natl. Acad. Sci. USA, 83: 7405 (1986).
15. J. F. Reidhar-Olsen and R. T. Sauer, Science, 241: 53 (1988).
16. R. A. Young and R. W. Davis, Science, 222: 778 (1983).
17. M. Balass, Y. Heldman, S. Cabilly, D. Givol, E. Katchalski-Katzir, and S. Fuchs,
Proc. Natl. Acad. Sci. USA, 90: 10638 (1993).
18. K. T. O’Neil, R. H. Hoess, S. A. Jackson, N. S. Ramachandran, S. A. Mousa, and
W. F. DeGrado, Proteins Struct. Funct. Genet., 14: 509 (1992).
19. S. Bass, R. Green, and J. A. Wells, Proteins Struct. Funct. Genet., 8: 309 (1990).
20. J. McCafferty, R. H. Jackson, and D. J. Chiswell, Prot. Eng., 4: 955 (1991).
21. B. L. Roberts, W. Markland, A. C. Ley, R. B. Kent, D. W. White, S. K. Guterman,
and R. C. Lander, Proc. Natl. Acad. Sci. USA, 89: 2429 (1992).
22. E. Soderlind, A. C. Simonsson, and C. A. K. Borrebaeck, Immunol. Rev., 130: 109
(1992).
23. C. Barbas, J. Bain, D. Hoekstra, and R. Lerner, Proc. Natl. Acad. Sci. USA, 89:
4452 (1992).
24. M. G. Cull, J. F. Miller, and P. J. Schatz, Proc. Natl. Acad. Sci. USA, 89: 1865
(1992).
25. L. C. Mattheakis, R. R. Bhatt, and W. J. Dower, Proc. Natl. Acad. Sci. USA, 92:
9022 (1994).
26. A. Giannis, and T. Kolter, Angew. Chem. Int. Ed. Engl., 32: 1244 (1993).
27. H. M. Geysen, S. J. Rodda, and T. J. Mason, Mol. Immunol., 23: 709 (1986).
28. R. J. Simon, R. S. Kania, R. N. Zuckerman, V. D. Huebner, D. A. Jewell, S. Banville,
S. Ng, L. Wang, S. Rosenberg, C. K. Marlowe, D. C. Spellmeyer, A. D. Frankel,
D. V. Santi, F. E. Cohen, and P. A. Bartlett, Proc. Natl. Acad. Sci. USA, 89: 9367
(1992).
29. B. A. Bunin and J. A. Ellman, J. Am. Chem. Soc., 114: 10997 (1992).
30. S. H. DeWitt, J. S. Kiely, C. J. Stankovic, M. C. Schroeder, D. M. R. Cody, and
M. R. Pavia, Proc. Natl. Acad. Sci. USA, 90: 6909 (1993).
31. R. B. Merrifield, J. Am. Chem. Soc., 85: 2149 (1963).
32. F. Camps, J. Cartells, and J. Pi, Anales Quim., 70: 848 (1974).
33. H. Rapoport and J. I. Crowley, Acc. Chem. Res., 9: 135 (1976).
34. C. C. Leznoff, Acc. Chem. Res., 11: 327 (1978).
TM

35. A. Furka, F. Sebestyen, M. Asgedom, and G. Dibo, Int. J. Pept. Protein Res., 37:
487 (1991).
36. K. S. Lam, S. E. Salmon, E. M. Hersh, V. J. Hruby, W. M. Kazmierski, and R. J.
Knapp, Nature, 354: 82 (1991).
37. C. T. Dooley, N. N. Chung, B. C. Wilkes, P. W. Schiller, J. M. Bidlack, G. W.
Pasternak, and R. A. Houghten, Science, 266: 2019 (1994).
38. E. Erb, K. D. Janda, and S. Brenner, Proc. Natl. Acad. Sci. USA, 91: 11422 (1994).
39. C. T. Dooley and R. A. Houghten, Life Sci., 52: 1509 (1993).
40. B. Deprez, X. Williard, L. Bourel, H. Coste, F. Hyafil, and A. Tartar, J. Am. Chem.
Soc., 117: 5405 (1995).
41. S. Brenner, and R. A. Lerner, Proc. Natl Acad. Sci. USA, 89: 5381 (1992).
42. J. Nielsen, S. Brenner, and K. D. Janda, J. Am. Chem. Soc., 115: 9812 (1993).
43. M. C. Needels, D. G. Jones, E. H. Tate, G. L. Heinkel, L. M. Kochersperger,
W. J. Dower, R. W. Barrett, and M. A. Gallop, Proc. Natl. Acad. Sci. USA, 90:
10700 (1993).
44. J. M. Kerr, S. C. Banville, and R. N. Zuckerman, J. Am. Chem. Soc., 115: 2529
(1993).
45. M. H. J. Ohlmeyer, R. N. Swanson, L. W. Dillard, J. C. Reader, G. Asouline, R.
Kobayashi, M. Wigler, and W. C. Still, Proc. Natl. Acad. Sci. USA, 90: 10922
(1993).
46. K. C. Nicolau, X. Y. Xiao, Z. Parandoosh, A. Senyei, and M. P. Nova, Angew.
Chem. Int. Ed. Eng., 34: 2289 (1995).
47. E. J. Moran, S. Sarshar, J. F. Cargill, M. M. Shahbaz, A. Lio, A. M. M. Mjalli, and
R. W. Armstrong, J. Am. Chem. Soc., 117: 10787 (1995).
48. T. Carell, E. A. Wintner, A. Bashirhashemi, and J. Rebek, Angew. Chem. Int. Ed.
Engl., 33: 2059 (1994).
49. H. M. Geysen, R. H. Meleon, and S. J. Barteling, Proc. Natl. Acad. Sci. USA, 81:
3998 (1984).
50. A. M. Bray, N. J. Maeji, and H. M. Geysen, Tetrahedron Lett., 31: 5811 (1990).
51. R. A. Houghten, Proc. Natl. Acad. Sci. USA, 82: 5131 (1985).
52. A. G. Beck-Sickinger, H. Durr, and G. Jung, Pept. Res., 4: 88 (1991).
53. S. P. A. Fodor, J. L. Read, M. C. Pirrung, L. Stryer, A. T. Lu, and D. Solas, Science,
251: 767 (1991).
54. D. L. Boger, C. M. Tarby, P. L. Myers, and L. H. Caporale, J. Am. Chem. Soc.,
118: 2109 (1996).
55. S. Cheng, D. D. Comer, J. P. Williams, P. L. Myers, and D. L. Boger, J. Am. Chem.
Soc., 118: 2567 (1996).
56. J. J. Parlow, Tetrahedron Lett., 36: 1395 (1995).
57. I. Ugi, A. Domling, W. Horl, Endeavour, 18: 115 (1994).
58. L. Weber, S. Wallbaum, C. Broger, and K. Gubernator, Angew. Chem. Int. Ed. Engl.,
34: 2280 (1995).
59. P. L. Myers, Proceedings of the 13th International Symposium of Laboratory Auto-
mation Robotics, 1995: 235 (1995).
60. H. Han, M. M. Wolfe, S. Brenner, and K. D. Janda, Proc. Natl. Acad. Sci. USA,
92: 6419 (1995).
TM

61. H. V. Meyers, G. J. Dilley, T. L. Durgin, T. S. Powers, N. A. Winssinger, H. Zhu,
and M. R. Pavia, Molec. Div., 1: 13 (1995).
62. M. R. Pavia, M. P. Cohen, G. J. Dilley, G. R. Dubuc, T. L. Durgin, F. W. Forman,
M. E. Hediger, G. Milot, T. S. Powers, I. Sucholeiki, S. Zhou, and D. G. Hangauer,
Bioorg. Med. Chem., 4: 659 (1996).
63. J. Rohr, Angew. Chem. Int. Ed. Engl., 24: 881 (1995).
64. R. McDaniel, C. M. Kao, H. Fu., P. Hevezi, C. Gustafsson, M. Betlach, G. Ashley,
D. E. Cane, C. Khosla, J. Am. Chem. Soc., 119: 4309 (1997).
65. C. M. Kao, M. McPherson, R. N. McDaniel, H. Fu, D. E. Cane, C. Khosla, J. Am.
Chem. Soc., 119: 11339 (1997).
66. R. P. Sheridan and S. K. Kearsley, J. Chem. Inf. Comput. Sci., 35: 310 (1995).
TM

11
Imidazoline Receptors and Their
Endogenous Ligands
Colin J. Barrow
The University of Melbourne, Parkville, Victoria, Australia
Ian F. Musgrave
Prince Henry’s Institute for Medical Research, Clayton, Victoria,
Australia
I. INTRODUCTION
The development of clonidine as an antihypertensive agent over 20 years ago

opened new vistas in receptor research with the elucidation of multiple α-adreno-
ceptor subtypes (α1A-D, α2A-D). Paradoxically it is now accepted that clonidine also
defines distinct, nonadrenergic receptors or sites that recognize its imidazoline
structure [1–3]. These sites have been previously termed imidazoline preferring
sites, imidazoline-guanidinium receptive sites (IGRSs), or nonadrenergic imida-
zoline-binding sites (NAIBSs) and are now designated as imidazoline receptors
[1–3]. Imidazoline receptors appear to be of physiological importance [1–3],
but whether these sites are receptors in the pharmacological sense remains to be
determined.
II. PHARMACOLOGY AND FUNCTION OF IMIDAZOLINE

RECEPTORS
Imidazoline receptors are classified principally by their ligand binding character-

istics. Like α2-adrenoceptors, imidazoline receptors have high affinity for com-
pounds with an imidazoline [e.g., clonidine (3) and idazoxan (10)], oxazoline
TM

[e.g., rilmenidine (8)], or guanido structure [e.g., guanabenz (13), see Table 2
for structures] [4–7]. Because of this, it was originally thought that imidazoline-
binding sites were a subtype of the α2-adrenoceptors. However, as imidazoline-
binding sites have negligible affinity for phenylethylamines such as epinephrine
and norepinephrine, or for α2-adrenoceptor ligands that are structurally unrelated
to imidazolines, such as yohimbine and BHT 933 [4–7] this is now considered
unlikely. Furthermore, imidazoline receptors have been shown to be physically
distinct from α2-adrenoceptors [8], and it is now established that the imidazoline
receptors represent distinct entities [3].
Imidazoline receptors can be subdivided into at least two types of imidazo-
line-binding sites. These are the I1-sites, which have a high affinity for clonidine
[4,9], and I2-sites, which have a high affinity for idazoxan [10] (see Table 1 for
rank orders of potency; see also Table 2). Both the I1- and the I2-sites may be
further subdivided on the basis of drug affinity. There is also at least one further
site, the so-called I3 binding site [13]. In addition, some imidazoline-preferring,
nonadrenergic sites that cannot be classified as I1-imidazoline-binding sites or I2-
imidazoline-binding sites exist and so are collectively called non-I1, non-I2-sites
[13,14].
A. I1-Imidazoline-Binding Sites
The I1-imidazoline-binding sites are characterized by having nanomolar affinity
for clonidine (3), naphazoline (2), cirazoline (4), and idazoxan (10), with cloni-
dine having a higher affinity than idazoxan [7,15–18] (see Table 2). Moxonidine
(7) and rilmenidine (8), recently developed antihypertensives, are relatively selec-
tive for the I1-imidazoline-binding sites over I2-imidazoline-binding sites and α2-
adrenoceptors, although the exact degree of selectivity varies in different reports
[2,15,17]. There may be more than one form of I1-imidazoline-binding site; for
example, human I1-imidazoline-binding sites differ from bovine sites in having
lower affinity for guanabenz (13) and moxonidine [2,15–18]. The binding of
ligands to the I1-imidazoline-binding sites is strongly inhibited by cations such
as K⫹ and tetraethyl ammonium (TEA) [7,15] (see Table 1).
I1-Imidazoline-binding sites are mostly distributed in the brain (corpus stri-
atum, hippocampus) and the brain stem (medulla oblongata, rostral ventrolateral
medulla) [2,11]. I1-Sites have also been found in the prostate [19] and the kidney
[16], in both cases localized to the epithelium.
Virtually nothing is known about the structure of I1-imidazoline-binding
sites. They are present in the plasma membrane, and although some investigators
have reported that [3H]-clonidine binding to these sites is modulated by guanosine
triphosphate (GTP) suggesting a G-protein-linked receptor [19,20], to date there
is no consensus [7,15,21]. Furthermore, binding of ligands to I1-sites does not
TM

Table 1 Classification of Imidazoline Receptorsa
Non-I1, non-I2-imidazoline-
I1-Imidazoline-binding sites I2-Imidazoline-binding sites binding sites References
Selectivity Cirazoline ⫽ clonidine ⫽ napha- Cirazoline ⫽ idazoxan ⬎⬎ cloni- Various, often idazoxan ⱖ 6,7,14,74,77,
zoline ⬎ idazoxan ⬎ phentol- dine ⬎ naphazoline ⫽ phentol- clonidine 80,81
amine ⬎⬎⬎ adrenaline amine ⬎⬎⬎ adrenaline
Cation 4-AP ⬎ CsCl ⫽ TEA ⬎ RbCl 4-AP ⬎ CsCl ⬎ TEA ⬎ RbCl Unknown 7,15,77,81
sensitivity ⬎ KCl ⬎ LiCl ⬎ NaCl ⬎ KCl ⬎ NaCl ⬎ LiCl
Location Plasma membrane Outer mitochondrial membrane Unknown 7,40,80,82
G-protein Unknown No Unknown 77,78,79,81,
linked 82
Subtypes Probably A,B Probably a number of sites 7,77,80,81,
82
Structure Unknown Monoamine oxidase, other? Unknown 41
Tissue location Brain stem, basal ganglia, hippo- Adipose tissue, liver, placenta Brain, stomach 2,11,12,16,19,
campus, prostate, kidney 14,37,78,81
Action Modulation of blood pressure Growth factor–like effects? Unknown, modulation of 29,30,31,33,
Modulation of ocular pressure? gastric secretion? 34,44,71
Modulation of gastric secretion?
Modulation of renal secretion?
a
TEA, tetraethyl ammonium.
TM

TM
C
Table 2 Structure/Affinity Relationships for Ligands at I1-Imidazole-Binding Sitesa
Compound Ki vs. I1 Ki vs. I2 Ki vs. α2A Structure Reference
Oxymetazoline (1) 6.2 19,500 5.9 53,80
Naphazoline (2) 8.4 1,451 14.3 7,53,77
Clonidine (3) 14.8 272 8.7 7,53,77
Cirazoline (4) 17.0 4.3 193 7,53,77
BU 224 (5) 41.5 20.8 4,020 53
Efaroxan (6) 52.4 44,800 9.8 53
Moxonidine (7) 55 36,600 150 53,83
Rilmenidine (8) 59.2 65.1 36.2 53
2-BFI (9) 66.7 9.0 8,490 53
Idazoxan (10) 103 4.1 12.2 7,53,77
Phentolamine (11) 148 1,424 51.3 7,53,77
Agmatine (12) 127 74,440 46,980 53
TM

Table 2 Continued
Compound Ki vs. I1 Ki vs. I2 Ki vs. α2A Structure Reference
Guanabenz (13) 35,100 12.9 2.0 53,80
Amiloride (14) ⬎10,000 30b n.d 7,77
a
Ki Values are nanomolar.
b
Amiloride distinguishes between two forms of the I2-imidazoline, binding site, one with a high
affinity for amiloride (I2A) and one with a low affinity (⬎1000 nM: I2B); n.d., Not determined.
activate G-protein-mediated signal transduction pathways such as adenylate cy-

clase or phospholipase C [22–25]. Other possible targets are the ion channels.
However, imidazolines do not appear to inhibit voltage operated calcium chan-
nels or nonselective cation channels in the neuronal model, rat pheochromocy-
toma (PC12) cells [24], or nonselective cation channels in a nonneuronal model
[26]. Ligand-gated ion channels, which are important in modulation of neuronal
activity, have been shown to be inhibited by imidazolines [24,25,27], in particular
the nicotinic acetylcholine receptor [24,25], the 5HT3 receptor [27], and the aden-
osine triphosphate– (ATP)-dependent K⫹ channel [28]. However, the profile of
potency of imidazoline ligands against the latter two channels suggests a non-I1,
non-I2 site [27,28].
The I1-imidazoline-binding site has been implicated in the hypertension
produced by clonidine and other imidazolines. When injected into the brain stem
baroreflex center; the rostral ventrolateral medulla (RVLM) of the cat [29], rat,
and rabbit [30,31]; imidazolines produced hypotension through a nonadrenergic
mechanism [29–31]. This observation suggests an I1-imidazoline-binding site-
mediated effect on the basis of the relative potencies of the imidazolines, and
the efficacy of the relatively I1-site-selective agents moxonidine and rilmenidine
[1–4]. The actual mechanism by which I1-imidazoline-binding sites act is unclear.
However, there is a reduction in the activity of catecholaminergic neurons in the
brain stem when imidazoline antihypertensives are injected [31,32].
The I1-site has also been implicated in modulation of gastric acid secre-
tion [33], modulation of intraocular pressure via central nervous system (CNS)
action [34], and modulation of electrolyte secretion in the rat kidney [35]. How-
ever, full pharmacological characterization of these actions has not been per-
formed.
TM

B. I2-Imidazoline-Binding Sites
I2-Imidazoline-binding sites have a distinct binding profile, with idazoxan and
cirazoline having nanomolar affinity for these sites, whereas clonidine and mox-
onidine are much weaker, typically having micromolar affinity (see Tables 1
and 2). Relatively selective ligands for I2-imidazoline-binding sites include 2-(2-
benzofuranyl)-2-imidazoline (2-BFI) (9) and 2-(4,5-dihydroimidaz-2-yl-quino-
line) (BU224) (5) [36]. The I2-imidazoline-binding sites can be further subdivided
on the basis of their affinity for amiloride (14) into I2A-sites and I2B-sites. Similar
to I1-imidazoline-binding sites, the I2-imidazoline-binding sites are sensitive to
cations (Table 1). The I2-imidazoline-binding sites are widespread, found in
astrocytes, kidney, liver, adrenal medulla, adipocytes, placenta, and urethra [37].
There is considerable detailed information about the structure of the I2-
sites. These sites are proteins of molecular weight 60–70 kDa, insensitive to
GTP, present in the outer mitochondrial membrane, and physically distinct from
the α2-adrenoceptor [10,38–40]. The molecular size and mitochondrial location
suggest that I2-sites could be located on monoamine oxidases. Recently it has
been demonstrated that cloned monoamine oxidases (MAOs) A and B do indeed
possess I2-sites [41], and that the two forms of imidazoline-binding protein corre-
spond to MAO-A and MAO-B. Whether all I2-imidazoline binding sites represent
MAO remains to be determined.
In contrast to that for the I1-imidazoline-binding sites, no definitive physio-
logical role for I2-sites has yet been determined. Binding of imidazolines to mono-
amine oxidase does inhibit these enzymes, and thus they may have a role in
modulating mitochondrial function [42]. Prolonged incubation of tissues with
imidazolines also down-regulates I2-imidazoline-binding sites, suggesting that
they are under physiological control and therefore have a functional role [43].
Furthermore, idazoxan also acts as an antiproliferative agent in vascular smooth
muscle [44]. This action occurs presumably through I2-imidazoline-binding sites,
since the potency of compounds as antiproliferative agents correlates with their
affinity at the I2-imidazoline-binding site in blood vessels [44].
III. THE SEARCH FOR ENDOGENOUS IMIDAZOLINE

RECEPTOR LIGANDS
A. Extraction and Isolation of Clonidine-Displacing
Substance
More than 10 years of research has been devoted to the search for endogenous
ligands for imidazoline receptors. As early as 1984 an endogenous compound
that competes for [3H]-clonidine binding was partially purified from both bovine
and rat brain [45]. This compound, labeled ‘‘clonidine-displacing substance’’
TM

(CDS), was determined to be a heat-stable, methanol-soluble, nonpeptidic small
molecule. The molecular weight of CDS was estimated at 500 daltons using size
exclusion chromatography [45,46], and at 588 Da using plasma desorption mass
spectrometry [47]. More recently, CDS has been partially purified from human
plasma [48], human cerebrospinal fluid [49], and bovine lung [50]. However, it
has not been unambiguously determined whether the CDS compound or com-
pounds responsible for the observed biological activities are the same in each
case.
There has been considerable disagreement in the literature on both the phys-
icochemical and the biological properties of CDS, possibly because of the range
of extraction procedures, assay systems, and purification strategies employed by
the various groups in obtaining CDS [4,45,46,50–52]. For example, partially pu-
rified CDS initially isolated from bovine brain by Atlas and Burstein, eluted from
reversed-phase chromatography with 39%–45% propanol in 10 mM ammonium
bicarbonate [45], indicated a hydrophobic molecule. In contrast, a recent investi-
gation found that neither lung nor brain CDS was retained on reversed-phase
chromatography, with all activity eluting with water, suggesting a hydrophilic
compound [50].
Atlas and Burstein initially obtained CDS extracts from bovine brain by
homogenizing at pH 7.7 in aqueous solution, boiling the supernatant for 15 min
to coagulate proteins, centrifuging and removing solvent, then extracting the resi-
due with methanol. These authors reported that CDS is pH- and heat-stable. A
recent study, however, indicated that CDS degraded in brain homogenates at high
temperature and that 50% of the CDS activity was lost over 3 h when methanol
extracts were diluted with HEPES-magnesium-EGTA assay medium at pH 7.7
[53]. These inconsistent properties reported for CDS indicate that multiple CDS
compounds are probably present in brain extracts. The presence of multiple CDS
compounds together with possible instability could be responsible for the varying
biological activities observed for extracts reputedly containing CDS.
The initial detection and partial purification of an endogenous CDS were
performed by monitoring binding to α2-adrenergic receptors rather than imidazo-
line receptors [44]. The major reasons for this were the availability of sensitive
assays for α2-adrenoceptors, the uncertainty as to the existence of specific imida-
zoline receptors, and the activity of CDS, like that of clonidine, at both α2-adren-
ergic and imidazoline receptors. However, the recently isolated CDS, agmatine
(12), is considerably more potent at the I1-receptor than at the α2-adrenergic re-
ceptor [53], indicating that bioassay directed isolation using a single binding
assay to detect CDS activity could be misleading. Thus, if CDS activity in brain
extract is due to multiple compounds with differing receptor selectivity, then
multiple binding assays should be used during fractionation to distinguish activi-
ties due to different CDS molecules. Using a single binding assay for monitoring
isolation makes it impossible to determine whether activity spread during frac-
TM

tionation is due to a single compound running poorly or the separation of multiple
active components.
Use of single concentrations of extract in binding assays during CDS isola-
tion was also recently reported to be misleading in that some fractions with potent
activity at a single concentration showed steep binding curves when multiple
concentrations were examined [50], suggesting that the activity was an artefact
or, alternatively, that the receptors exhibit strong positive cooperativity. Thus, it
is strongly advisable to monitor binding at multiple concentrations during frac-
tionation. In addition, compounds that have CDS binding activity do not necessar-
ily have corresponding functional activity, indicating that fractionation should
also be monitored at least periodically using an appropriate functional assay. For
example, agmatine, recently isolated as an endogenous CDS with both imidazo-
line and α2-adrenergic receptor-binding activity, does not appear to be function-
ally active at the α2-receptor (see Section V.C). It should also be noted that func-
tional activity reported for methanolic brain extracts containing CDS activity may
be misleading. For example, norepinephrine present in crude methanolic bovine
brain extracts may be responsible for the reported ability of CDS to inhibit electri-
cally evoked contraction of the rat isolated vas deferens by activation of α2-
adrenoceptors [50,54]. The levels of norepinephrine in the crude methanolic ex-
tract of brain appear to be too low to account for all the CDS binding activity
but could account for the observed functional activity in a system with a high
receptor reserve where even low concentrations of norepinephrine could produce
a response [50].
Any effort to isolate new endogenous CDS compounds needs to eliminate
the possibility that activity is due to known contaminants such as monovalent
cations, histamine, catecholamines, and the reported CDS agmatine. In addition,
once a semipurified CDS has been obtained, then a variety of assays need to be
performed to confirm the presence of both binding and functional activity. Further
purification should be monitored using full binding curves and functional activity,
at least periodically, to eliminate false-positive results. However, not until puri-
fied CDS is isolated and structurally characterized will the biological properties
of endogenous CDS compounds be unambiguously determined.
B. Biological Activity of Clonidine-Displacing Substance

In addition to displacing [3H]-clonidine from various membrane preparations,
CDS has been shown to have functional effects when applied to several prepara-
tions. Microinjection of partially purified CDS derived from brain into the RVLM
of the rat or cat modifies arterial pressure [55–57]. However, the results are not
consistent: either an increase [56,57] or a decrease [55] in arterial pressure may
be produced. It is not clear whether the variations are due to differences in the
CDS produced or the presence of contaminants such as biogenic amines and K⫹
[50]. Contaminants may not be necessarily responsible for the hypertensive
TM

fect, as a nonhypertensive dose of the CDS was able to inhibit the hypotensive
effect of clonidine microinjected into the RVLM of rats and cats [57] and the
hypotensive effects of rilmenidine microinjected into the RVLM in rabbits [58].
These results are consistent with an action at a nonadrenergic site in the brain
stem.
Partially purified CDS also has functional effects that may be due to an
action at α2-adrenoceptors. For example, brain-derived CDS blocks electrically
stimulated contraction of rat vas deferens [54]. This effect was reversed by yo-
himbine and thus is consistent with an agonist action at α2-adrenoceptors [54].
In this CDS preparation a possible contribution from norepinephrine cannot be
excluded, but in other preparations such contamination is unlikely. For example,
brain-derived CDS antagonizes the epinephrine-induced aggregation of human
platelets but is unable to stimulate aggregation by itself [59]. Thus, blockade of
aggregation is unlikely to be due to contaminating biogenic amines. If there was
a significant amount of norepinephrine/epinephrine present, then aggregation
should have been seen, or epinephrine-induced agregation potentiated.
Brain-derived CDS can also contract urethral smooth muscle but has no
effect on vascular smooth muscle [60]. This is a nonadrenergic effect that could
not be blocked by antagonists to muscarinic, serotonergic, angiotensin II, or bra-
dykinin receptors [60]. Furthermore, as the contractile effect was not seen in
vascular smooth muscle, this eliminates the possibility that cations such as K⫹
are responsible [60].
In contrast, plasma-derived CDS was able to produce concentration-depen-
dent contraction of rat aortic rings [61]. This appears to be due to an action at
α2-adrenoceptors as the effect was inhibited by the α2-adrenoceptor antagonist
rauwolscine, but not by the α1-adrenoceptor antagonist prazosin [61]. These re-
sults suggest that the effect of the CDS is not due to contaminating biogenic
amines, which should have activated α1-adrenoceptors, or cations such as K⫹,
which would cause depolarization.
Thus, despite a wide variety of preparation procedures and biological
sources, in some cases CDS has actions that can be attributed to effects at α2-
adrenoceptors, independent of contaminating biogenic amines or cations.
Whether or not there is more than one CDS or whether other non-CDS material in
the partially purified extracts contributes to activity is unclear from these results.
IV. AGMATINE: THE FIRST ENDOGENOUS

CLONIDINE-DISPLACING SUBSTANCE
A. Isolation and Identification of Agmatine
Recently agmatine (12), the decarboxylation product of arginine, was isolated as
the first endogenous CDS compound from bovine brain [62]. Although agmatine
does not appear to have all the biological properties attributed to CDS, it clearly
TM

binds to imidazoline receptors and has some functional activity [62]. Agmatine
was isolated from bovine brain with a bioassay-directed isolation strategy, using
displacement of [3H]-p-aminoclonidine from rat cerebral cortical membranes to
track activity. The extraction method began with homogenization of fresh brain
in chilled water, then centrifugation. Protein was then precipitated from the super-
natant with 70% ethanol, then the extract was concentrated, centrifuged, and
passed through a Dowex 50 (H⫹) column eluting with 3N hydrochloric acid [62].
This method prevents the temperature extremes employed by Atlas and Burstein
in their initial isolation and identification of a CDS [45–47], which appears not
to be agmatine [63,64].
After ion-exchange chromatography, agmatine was partially purified by
size-exclusion chromatography, followed by reversed-phase C18 high-perfor-
mance liquid chromatography (HPLC), but because of the high polarity of this
compound and lack of an ultraviolet (UV) chromophore, attempts to purify ag-
matine further were unsuccessful. Agmatine was eventually isolated as its 9-
fluorenylmethyl chloroformate (FMOC) derivative. Structural confirmation of
this CDS compound was performed using electrospray mass spectroscopy and
direct comparison with authentic agmatine [62].
During the decomposition of dead tissue, polyamines such as agmatine are
formed from amino acids by the action of bacterial decarboxylases. Indeed, detec-
tion of agmatine has previously been used as an indicator of the freshness of fish
[65]. Therefore, when agmatine was initially isolated in minor quantities from
bovine brain there was concern that it may be a bacterial contaminant rather than
an endogenous compound. However, arginine decarboxylase activity was shown
to be present in the brain by measuring the conversion of l-[14C]arginine to 14CO2
[62]. The presence of arginine decarboxylase activity in rat brain indicated that
agmatine is indeed a compound endogenous to mammalian brain, and therefore
a true endogenous CDS.
B. Distribution of Agmatine in Mammalian Tissue

In a more recent study, agmatine levels in a variety of rat tissues were measured
by a standardized method consisting of tissue homogenization, derivatization,
extraction on a C18 cartridge, and HPLC chromatographic analysis with fluores-
cence detection [66]. Agmatine was found in all tissues examined with the great-
est concentration in the stomach (Table 3). It is interesting to note that agmatine
was present in relatively low concentration in brain tissue (2.4 ng/g) and that
agmatine concentrations in the lung were approximately 5 times higher than in
the brain, whereas those in the stomach were approximately 20 times higher than
in the brain [66]. Singh et al. recently reported the presence of a noncatechola-
mine CDS in methanolic extracts of bovine lung in a threefold higher concentra-
tion than that found in bovine brain [50]. However, it was concluded that this
TM

Table 3 Distribution of Agmatine
in Various Tissues of Adult Male
Sprague-Dawley Rats
Tissue Wet weight (ng/g)a
Stomach 71.00 ⫾ 10.33

Aorta 57.41 ⫾ 12.74
Small intestine 55.35 ⫾ 9.39
Large intestine 27.86 ⫾ 6.73
Spleen 17.38 ⫾ 3.17
Lung 10.23 ⫾ 2.82
Vas deferens 9.45 ⫾ 2.08
Adrenal gland 6.97 ⫾ 3.29
Kidney 6.45 ⫾ 1.40
Heart 6.45 ⫾ 0.79
Liver 5.63 ⫾ 0.87
Skeletal muscle 5.30 ⫾ 0.72
Brain 2.40 ⫾ 0.60
Testes 2.04 ⫾ 0.22
Plasma 0.45 ⫾ 0.05
a
Concentrations are expressed in wet weight (ng/g).
Data are averaged from either four or five experi-
ments.
Source: Ref. 66.
CDS was not agmatine, again indicating the presence of multiple CDS com-
pounds in multiple tissue types [50].
Thus, agmatine and other CDS compounds appear to be widely distributed
in mammalian tissue. Agmatine present in the stomach appears to be contained
in endothelium and vascular smooth muscle and so may play a role in modulating
gastric secretion [66]. However, there is substantial variation in the quantity of
agmatine in individual animals [66], greater than 10-fold in some cases, raising
concern about a functional role for agmatine. Furthermore, the imidazoline-bind-
ing site in the stomach appears to be a non-I1, non-I2-site [14].
V. BIOLOGICAL EVALUATION OF AGMATINE

A. Cardiovascular Effects of Agmatine
The effect of agmatine on sympathetic cardiovascular regulation in conscious
rabbits was recently characterized [67]. Agmatine injected intracisternally caused
central sympathoexcitation by increasing renal sympathetic nerve activity, plasma
TM

catecholamine concentrations, and arterial blood pressure [67]. However, a de-
crease rather than an increase in heart rate was observed. It is interesting to note
that extracts containing CDS have been reported to cause either an increase
[56,57] or a decrease [55] in blood pressure (see Section III.B), illustrating the
difficulties of working with semipurified material. Agmatine could have been the
active compound in the studies reporting an increase in blood pressure, but in
the study reporting a decrease in blood pressure, a second compound would have
to be responsible.
The mechanism by which agmatine exerts these cardiovascular effects is
unclear, although mediation via imidazoline receptors is a possibility. Agmatine
is able to pass and block the intrinsic-ion channel of the nicotinic receptor [68].
However, this is unlikely to play a role in sympathoexcitation by agmatine as
high millimolar concentrations of agmatine are required to block the nicotinic
receptor [68]. It is also unlikely that agmatine acts via α2-adrenoceptors for a
number of reasons. First, agmatine does not appear to be functionally active at
α2 adrenoceptors [53,67]; that is, it is neither an antagonist nor an agonist (see
later discussion). Second, intracisternally injected agmatine causes bradycardia,
in contrast to intracisternally injected yohimbine which is known to act via block-
ade of α2-adrenoceptors [67].
It has been suggested that the I1-receptor agonist clonidine acts on I1-recep-
tors in the RVLM to produce its sympathoinhibitory effect [69]. Thus, it is possi-
ble that the sympathoexcitatory effects of agmatine are due to antagonist effects
at I1-receptors. However, a recent report confirming the sympathoexcitatory activ-
ity of agmatine injected intracisternally showed that agmatine, unlike clonidine,
does not alter arterial pressure when microinjected into the RVLM [69]. Ionto-
phoresis of agmatine onto defined vasomotor neurons in the RVLM was also
without effect [69]. Therefore, although agmatine has a high affinity for I1-recep-
tors [53], its cardiovascular effects elicited by intracisternal injection are probably
not modulated by these receptors. Also, because inhibition of the RVLM vasomo-
tor neurons represents the major mechanism by which clonidine produces its
hypotensive effects [69], agmatine and clonidine appear to have quite different
mechanisms of cardiovascular action.
In contrast to central administration of agmatine, intravenous administra-
tion of agmatine into rats produces sympathoinhibition [69] with a fall in arterial
blood pressure [69,70]. The mechanisms for the hypotensive actions of peripher-
ally administered agmatine are still unclear, but blockade of ganglionic transmis-
sion and vasodilatation from a direct action on vascular smooth muscle may be
involved [69,70].
B. Gastrointestinal Function of Agmatine

The relationship between hypertension and gastrointestinal disorders, such as gas-
tric ulcers, can be investigated by examining the actions of known antihyper-
TM

tensive drugs on gastric function or response to injury. For example, the anti-
hypertensive drug clonidine has been shown to decrease gastric acid secretions
while enhancing gastric adherent mucus levels [33]. The identity of the receptors
through which clonidine produces its gastrointestinal actions is unknown. How-
ever, the I1-imidazoline receptor agonist moxonidine is both a potent inhibitor
of basal gastric acid secretion and significantly raised intragastric pH, indica-
ting that I1-imidazoline receptors may be involved in gastrointestinal function
[33].
Agmatine, if an endogenous ligand for the imidazoline receptors, should
have similar gastrointestinal effects to those of I1-imidazoline receptor agonists
such as moxonidine. Agmatine was recently tested for effects on gastrointestinal
function and on experimental gastric mucosal injury and was found to augment
gastric acid and pepsin secretion, decrease gastric adherent mucus, and worsen
experimental gastric mucosal injury consequent to stress [71]. These effects are
opposite to the gastrointestinal effects observed for the I1-site agonist moxonidine
[33]. The observed ‘‘inverse agonist’’ activity of agmatine at imidazoline recep-
tors could be a result of the involvement of different subtypes of the imidazoline
receptor at the same site. Alternatively, agmatine may modulate gastrointestinal
function via novel receptor subtypes [14,71].
C. Interaction of Agmatine with ␣2-Adrenoceptors

and Imidazoline-Binding Sites
Agmatine has been shown to produce a concentration-dependent inhibition of
[3H]-clonidine binding to both rat and bovine cerebral cortex membranes with a
Ki of approximately 10 µM [18,62,72]. However, in functional studies agmatine
failed to act as either an α2-adrenoceptor antagonist or an agonist in a wide variety
of tissues [67,72,73]. For example, agmatine did not inhibit electrically evoked
contractions of the rat vas deferens [72], in contrast to activity originally
described for partially purified CDS [54] and the action of the α2-adreno-
ceptor agonist clonidine [72]. This indicates that in studies with CDS-contain-
ing extracts, agmatine may be responsible for the activity detected in the radi-
oligand binding assay, but other compounds such as catecholamines [74] or a
second CDS compound must be responsible for the observed functional
activity.
Furthermore, at concentrations of 100 µM, agmatine did not shift the con-
centration response curve to clonidine in the rat vas deferens even though a 10-
fold shift was expected [72], suggesting that agmatine is not an α2-adrenoceptor
antagonist. The same results were seen separately with both clonidine and the
selective α2-adrenoceptor agonist UK 14304 in rat thoracic aorta, rat cerebral
cortical slices, porcine palmar veins, and guinea pig ileum [72]. Similar results
have also been seen in rabbit occipital cortex slices [67] or rat locus ceruleus
slices [73]. Thus, agmatine would appear to be neither an agonist nor an antago-
TM

Table 4 Comparison of Clonidine-Displacing Substance and Agmatine
Inhibition Constants at α2-Adrenergic and Imidazoline-Binding Sites from
Human and Bovine Tissuesa
CDS IC50
Binding sites (Diluted ⫻ 10⫺3) Agmatine Ki (nM)
Human α2A 51 ⫾ 4 46,980 ⫾ 2,600
Human α2B 48 ⫾ 3 164,400 ⫾ 21,412
Human α2C 20 ⫾ 1 26,300 ⫾ 3,969
Human I2b 23 ⫾ 1 74,400 ⫾ 14,152
Human I1 49 ⫾ 2 Highb 33 ⫾ 19
Low 284,600 ⫾ 37,903
Bovine I1 9⫾2 High 127 ⫾ 33
Low 275,500 ⫾ 51,900
a
CDS, clonidine-displacing substance.
b
Human and bovine I1-binding sites have both a high- and a low-affinity site for agmatine.
Source: Ref. 51.
nist at α2-adrenoceptors, even though it binds to these receptors. However, it may

be that agmatine is a much weaker antagonist than its Ki value would suggest.
Jurkiewitz et al. [74] found that agmatine acted as an α2-adrenoceptor antagonist
in the rat vas deferens, but only at concentrations of 0.3 and 1 mM.
Because crude methanolic bovine brain extracts show CDS activity consis-
tent with agonist activity at α2-adrenoceptors, and this activity is attributable
to neither agmatine nor histamine, a second CDS compound is probably present
in these extracts [75]. Furthermore, direct comparison of relative potencies of
CDS-containing brain extract and agmatine at all three human α2-adrenergic
receptor subtypes and at human imidazoline sites suggests that CDS and agmatine
are not equivalent (Table 4). Crude bovine CDS was potent at all sites, although
a 2.5- to 6-fold higher potency for bovine imidazoline I1-sites was observed when
compared with all human sites, indicating some selectivity of CDS for I1-sites
[53]. Agmatine, in contrast to CDS, bound to two sites, one with high affinity
(33% of all sites) and one with micromolar affinity (Table 4). The reason for
the apparent presence of both high- and low-affinity binding sites for agmatine
at I1-receptors is still under investigation [53]. Again in contrast to CDS, agmatine
has a much lower affinity for the α2B-adrenoceptor vs. the α2A-adrenoceptor (Ta-
ble 4). The differences between the potencies of methanolic CDS extract from
brain and agmatine further suggest that at least one component other than agmat-
ine is present in the CDS extract. However, the high affinity of agmatine for a
subset of I1-sites still supports agmatine as a true endogenous ligand for I1-sites
[53].
TM

VI. RECENT PROGRESS IN THE SEARCH FOR
ENDOGENOUS CLONIDINE-DISPLACING
SUBSTANCES
The purification and characterization of endogenous CDSs other than agmatine

remain elusive. Major reasons for this include the absence of a definitive biologi-
cal action for CDS, the presence of multiple CDSs and biologically active con-
taminants, and the limited abundance of CDS in tissue. A recent study has shown
that tissue other than brain contains CDS activity [75, see also 44]. As previously
noted, bovine lung is reported to have approximately three times the CDS activity
found in the brain [75]. Although it has not been confirmed that the same CDSs
are responsible for activity observed in different tissue types, it appears that
known contaminants such as monovalent cations, histamine, catecholamines, and
the known CDS agmatine cannot explain all observed binding activity found in
either lung or brain extracts. After eliminating activity due to these contaminants
it appears that the remaining CDS activity in both lung and brain samples is due
to an unidentified compound with agonist activity at α2-adrenoceptors [72].
Chemical properties of the unidentified CDSs from bovine lung and brain
indicate that both tissues contain the same CDSs [50]. Both lung and brain CDS
are poorly soluble in nonpolar solvents at room temperature and are retained on
cation exchange columns, indicating that the CDS from both tissues has a positive
charge. Furthermore, CDS was recovered from similar fractions when both lung
and brain extracts were chromatographed using size-exclusion chromatography
[50]. The authors of this study found that neither lung nor brain CDS is retained
on C18 chromatography, a result contrasting with previous work, when CDS from
brain was strongly retained on C18 material and only eluted with 35% propanol/
water [46,47].
In our group we have recently attempted to simplify the search for new
endogenous CDS by (1) developing a large-scale extraction method to overcome
the difficulties in obtaining enough CDS material for structure determination;
(2) determining whether CDS activity in lung is due to the same compound that
is responsible for CDS activity in brain; (3) developing an HPLC detection
method for CDS; and (4) obtaining structural information for a CDS other than
agmatine. In the following section we detail some of these results [76].
A. Development of a Large-Scale Extraction Method

The standard method typically used for extraction of CDS activity from tissue
is difficult to scale up to the kilogram quantities needed to obtain enough
CDS for nuclear magnetic resonance (NMR) studies, for several reasons. First,
the method uses time-consuming homogenization and filtration of the homoge-
TM

nate. Second, large quantities of aqueous filtrate then need to be freeze-dried.
Finally, the water extraction procedure exposes CDS to enzymatic degradation,
and the use of boiling water could potentially degrade heat-sensitive CDS mole-
cules.
We have modified the standard extraction method by replacing the time-
consuming filtration with centrifugation and eliminating heating, water extrac-
tion, and use of a homogenizer. We have thus developed an extraction method
that can be routinely used for extraction of CDS activity from a variety of tissues
on a kilogram scale. This method involves lyophilization of wet tissue followed
by extraction of inert fats by blending with hexane. After centrifugation to remove
the hexane solubles the tissue is blended with methanol and the methanol-insolu-
ble material removed by centrifugation. The methanol is removed by rotary evap-
oration to yield a CDS extract. Quantitative comparison of this method with the
previously used standard extraction method has shown that considerably more
CDS activity was extracted from both lung and brain tissue by using our opti-
mized method.
B. Development of a Large-Scale Clonidine-Displacing

Substance Isolation Method
During the isolation procedure, CDS activity was monitored by using an α2-
adrenoceptor radioligand membrane binding assay, with displacement of [3H]-
clonidine. Naphazoline was used to calculate nonspecific binding. Clonidine, nor-
epinephrine, and yohimbine were used as positive controls [45,46]. Methanol
extracts were preabsorbed onto C18 reversed-phase material and passed through
a C18 column, eluting first with water and subsequently with increasing ratios of
methanol/water. In addition to ease of scale-up, using C18 chromatography in the
first purification step retained the majority of CDS activity on the stationary phase
after elution with water, so cations responsible for nonspecific activity were
readily removed.
As reported previously [50], CDS activity was retained on a cation ex-
change column, indicating that CDS is positively charged. We found that CDS
activity was enriched after eluting the partially purified C18 fraction from an anion
exchange column, probably as a result of retention of inactive anionic com-
pounds. Therefore, an anion exchange step was incorporated after the C18 fraction-
ation. Subsequent size-exclusion chromatography, using Biogel P2, gave good
enrichment of CDS activity in fractions corresponding to a molecular weight of
less than 600 Da. Final purification of a new CDS was achieved on HPLC using
a C18 reversed-phase semipreparative column, eluting with 0.1% triflouroacetic
acid in water (Fig. 1). Fifty micrograms of pure CDS material was obtained from
2.2 Kg of wet lung [76].
TM

Figure 1 Strategy for the isolation of CDS from both bovine lung and brain. CDS,
clonidine-displacing substance.
C. Development of a High-Performance Liquid

Chromatography Detection Method
for Clonidine-Displacing Substance
Using a variety of acidic methanol/water or acetonitrile/water gradients it was
found that the majority of CDS activity eluted from C18 reversed-phase chroma-
tography with the solvent front. However, using isocratic elution with 0.1%
triflouroacetic acid in water, the majority of CDS activity eluted in a sharp band.
The CDS activity was associated with a single peak with a UV maximum at 280
nm (Fig. 2).
Application of the HPLC method to the detection of the new CDS com-
pound in extracts of bovine lung and brain indicates that this compound is present
in both tissues and is approximately threefold more abundant by weight in lung
than in brain. Semipreparative isolation and bioassay of this peak from the initial
extract confirmed its CDS activity and indicated that the compound was responsi-
ble for approximately 50% of the CDS activity observed in both lung and brain.
Norepinephrine, epinephrine, histamine, and agmatine were eliminated as being
responsible for the observed CDS activity by comparing the retention times and
UV profile of each with those observed for the new CDS.
TM

Figure 2 HPLC profile with UV spectrum of purified CDS. HPLC, high-performance
liquid chromatography; UV, Ultraviolet; CDS, clonidine-displacing substance.
D. Partial Structure Determination of a New

Clonidine-Displacing Substance
Using tandem mass spectrometry–mass spectrometry (MSMS) and 1H nuclear
magnetic resonance (NMR) spectroscopy, a partial structure for the new CDS
was obtained. Electrospray MS indicated a molecular weight of 275 Da and
MSMS showed two consecutive losses of 84 Da, which were not readily assign-
able. The 1H NMR spectrum in dimethyl sulfoxide-d6 (DMSO-d6) indicated the
presence of a sugar moiety and a heteroaromatic group. The 1H NMR and UV
data were consistent with the presence of a substituted imidazole ring. Although
a final structure has not been assigned, a probable partial structure (15) shows a
close relationship to guanosine (16). However, the new CDS was shown not to be
guanosine or its mono-, di-, or triphosphate analogues, by comparison of HPLC
retention times, MS, NMR spectra, and clonidine-displacement binding data with
those of authentic samples.
TM

It has not been determined whether the new CDS was the same as the
original compound partially purified by Atlas et al. [45], or one of the other
partially purified CDSs [72,75]. However, it is clear that at least one CDS other
than agmatine exists in both brain and lung tissue. Full structure determination
is needed to unravel the relationship between the various CDS activities being
observed by a variety of groups and also to allow full biological characterization
of the proposed endogenous CDS.
REFERENCES
1. P. Bousquet, G. Bricca, M. Dontenwill, J. Feldman, H. Greney, A. Belcourt, A.

Stutzman and J. Stutzmann, Fundam. Clin. Pharmacol., 6, suppl, 1: 15S (1992).
2. H. De Vos, G. Bricca, J. De Keyser, J. P. De Backer, P. Bousquet, and G. Vauquelin,
Neuroscience, 59: 589 (1994).
3. P. Ernsberger, M. E. Graves, L. M. Graff, N. Zakieh, P. Nguyen, L. A. Collins, and
K. L. Westbrooks, Ann. NY Acad. Sci., 763: 22 (1995).
4. P. Ernsberger, R. Giuliano, R. N. Willette, A. R. Granata, and D. J. Reis, J. Hyper-
tens., suppl. 6: S554 (1988).
5. I. Limon, I. Coupry, S. M. Lanier, and A. Parini, J. Biol. Chem., 267: 21645 (1992).
6. P. R. King, A. L. Gundlach, B. Jarrott, and W. J. Louis, Am. J. Hypertens., 5: 64S
(1992).
7. G. J. Molderings, D. Moura, K. Fink, H. Bönisch, and M. Göthert, Naunyn Schmie-
debergs Arch. Pharmacol., 348: 70 (1993).
8. M. C. Michel, J. W. Regan, M. A. Gerhardt, R. R. Neubig, P. A. Insel, and H. J.
Motulsky, Mol. Pharmacol., 37: 65 (1990).
9. M. C. Michel and P. Ernsberger, Trends Pharmacol. Sci., 13: 369 (1992).
10. F. Tesson, C. Prip-Buus, A. Lemoine, J. P. Pegorier, and A. Parini, J. Biol. Chem.,
266: 155 (1991).
11. Y. Kamisaki, T. Ishikawa, Y. Takao, H. Omodani, N. Kuno, and T. Itoh, Brain Res.,
514: 15 (1990).
TM

12. D. Atlas, S. Diamant, and R. Zonnenschein, Am. J. Hypertens., 5: 83S (1992).
13. P. R. King, S. Suzuki, W. J. Louis, and A. L. Gundlach, Ann. NY Acad. Sci., 763:
194 (1995).
14. G. J. Molderings, K. Donecker, and M. Göthert, Naunyn Schmiedebergs Arch. Phar-
macol., 351:561 (1995a).
15. G. Bricca, J. Zhang, H. Greney, M. Dontenwill, J. Stutzmann, A. Belcourt, and P.
Bousquet, Br. J. Pharmacol., 110: 1537 (1993).
16. P. Ernsberger, G. Feinland, M. P. Meeley, and D. J. Reis, Am. J. Hypertens., 3: 90
(1990).
17. P. Ernsberger, T. H. Damon, L. M. Graff, S. G. Schäfer, and M. O. Christen, J.
Pharmacol. Exp. Ther., 264: 172 (1993).
18. J. E. Piletz, D. N. Chikkala, and P. Ernsberger, J. Pharmacol Exp. Ther., 272: 581
(1995).
19. D. Felsen, P. Ernsberger, P. M. Sutaria, R. J. Nejat, P. Nguyen, M. May, D. S.
Breslin, D. N. Marion, and E. D. Vaughan, Jr., J. Pharmacol. Exp. Ther., 268: 1063
(1994).
20. P. Ernsberger, K. Westbrooks, and L. Graff, FASEB J., 5: A1066 (1991).
21. G. Bricca, H. Greney, J. Zhang, M. Dontenwill, J. Stutzmann, A. Belcourt, and P.
Bousquet, Eur. J. Pharmacol., 266: 25 (1994).
22. S. Regunathan, M. J. Evinger, M. P. Meeley, and D. J. Reis, Biochem. Pharmacol.,
42: 2011 (1991).
23. S. Regunathan and D. J. Reis, Eur. J. Pharmacol., 269: 273 (1994).
24. I. F. Musgrave, D. Krautwurst, J. Hescheler, and G. Schultz, Ann. NY Acad. Sci.,
763: 272 (1995).
25. I. F. Musgrave, unpublished results.
26. I. F. Musgrave and R. Seifert, Biochem. Pharmacol., 49, 187 (1995).
27. G. J. Molderings, K. Ruppert, H. Bönisch, and M. Göthert, Ann. NY Acad. Sci., 763:
420 (1995).
28. K. Lee, W. J. Groh, T. A. Blair, J. G. Maylie, and J. P. Adelman, Eur. J. Pharmacol.,
285: 309 (1995).
29. P. Bousquet, J. Feldman, and J. Schwartz, J. Pharmacol Exp. Ther., 230: 232 (1984).
30. R. E. Gomez, R. Ernsberger, G. Feinland, and D. J. Reis, Eur. J. Pharmacol., 195:
181 (1991).
31. E. Tibirica, J. Feldman, C. Mermet, L. Monassier, F. Gonon, and P. Bousquet, Eur.
J. Pharmacol., 209: 213 (1991).
32. C. Mermet and L. Quintin, Eur. J. Pharmacol., 204: 105 (1991)
33. G. B. Glavin and D. D. Smyth, Br. J. Pharmacol., 114: 751 (1995).
34. W. R. Campbell, and D. E. Potter, J. Ocul. Pharmacol., 10: 393 (1994).
35. D. D. Smyth, F. K. Darkwa, and S. B. Penner, Ann. NY Acad. Sci., 763: 340
(1995).
36. L. A. Lione, D. J. Nutt, and A. L. Hudson, Eur. J. Pharmacol., 304: 221 (1996).
37. D. Atlas, S. Diamant, and R. Zonnenschein, Am. J. Hypertens., 5: 83S (1992).
38. I. Coupry, D. Atlas, R. A. Podevin, I. Uzielli, and A. Parini, J. Pharmacol. Exp.
Ther., 252: 293 (1990).
39. M. C. Michel, J. W. Regan, M. A. Gerhardt, R. R. Neubig, P. A. Insel, and H. J.
Motulsky, Mol. Pharmacol., 37: 65 (1990).
TM

40. H. Greney, G. Bricca, M. Dontenwill, J. Stutzmann, P. Bousquet, and A. Belcourt,
Neurochem. Int., 25: 183 (1994).
41. F. Tesson, I. Limon-Boulez, P. Urban, M. Puype, J. Vandekerckhove, I. Coupry, D.
Pompon, and A. Parini, J. Biol. Chem., 270: 9856 (1995).
42. A. Parini, C. G. Moudandos, N. Piinat, and S. M. Lanier, Trends Pharmacol. Sci.,
17: 13 (1996).
43. C. A. Hamilton, E. Jardine, and J. L. Reid, Eur. J. Pharmacol., 243: 95 (1993).
44. S. Regunathan, C. Youngson, W. Raasch, H. Wang, and D. J. Reis, J. Pharmacol.
Exp. Ther., 276: 1272 (1996).
45. D. Atlas and Y. Burstein, Eur. J. Biochem., 144: 287 (1984).
46. D. Atlas and Y. Burstein, FEBS Lett., 170: 387 (1984).
47. D. Atlas, S. Diamant, H. M. Fales, and L. Pannell, J. Cardiovasc. Pharmacol., 10,
suppl. 12: S122 (1987).
48. D. Synestos, V. G. Manolopoulos, D. Atlas, and E. Pipili-Synestos, J. Auton. Phar-
macol., 11: 343 (1991).
49. H. Goldberg Stern, D. Atlas, L. Schwartz, A. Achiron, I. Ziv, R. Djaldetti, Y. Zoldan,
and E. Melamed, Brain Res., 601: 325 (1993).
50. G. Singh, J. F. Hussain, A. MacKinnon, C. M. Brown, D. A. Kendall, and V. G.
Wilson, Naunyn Schmiedebergs Arch. Pharmacol., 351: 17 (1995).
51. J. Feldman, E. Tibirica, G. Bricca, M. Dontenwill, A. Belcourt, and P. Bousquet,
Br. J. Pharmacol., 100: 600 (1990).
52. D. J. Reis, S. Regunathan, and M. P. Meeley, Am. J. Hypertens., 5: 51S (1992).
53. J. E. Piletz, H. Zhu, and D. N. Chikkala, J. Pharmacol. Exp. Ther., 279: 694 (1996).
54. S. Diamant and D. Atlas, Biochem. Biophys. Res. Com., 134: 184 (1986).
55. M. P. Meeley, P. R. Ernsberger, A. R. Granata, and D. J. Reis, Life Sci., 38: 1119
(1986).
56. P. Bosquet, J. Feldman, and D. Atlas, Eur. J. Pharmacol., 124: 167 (1986).
57. P. Bousquet, J. Feldman, and D. Atlas, J. Cardiovasc. Pharmacol., 10, suppl. 12:
S167. (1987).
58. P. Bousquet, J. Feldman, E. Tibirica, G. Bricca, A. Molines, M. Dontenwill, and A.
Belcourt, Am. J. Med., 87: 10S (1989).
59. S. Diamant, A. Eldor, and D. Atlas, Eur. J. Pharmacol., 144: 247 (1987).
60. D. Felsen, P. Ernsberger, M. P. Meeley, and D. J. Reis, Eur. J. Pharmacol., 142:
453 (1987).
61. D. Synetos, V. G. Manolopoulos, D. Atlas, and E. Pipili-Synetos, J. Auton. Pharma-
col., 11: 343 (1991).
62. G. Li, S. Regunathan, C. J. Barrow, J. Eshraghi, R. Cooper, and D. J. Reis, Science,
263: 966 (1994).
63. D. Atlas, Science, 266: 462 (1994).
64. D. J. Reis, G. Li, S. Regunathan, C. J. Barrow, and R. Cooper, Science, 266: 463
(1994).
65. T. Kawabata, H. Ohshima, and M. Ino, J. Agric. Food. Chem., 26: 334 (1978).
66. W. Raasch, S. Regunathan, G. Li, and D. J. Reis, Life Sci., 56: 2319 (1995).
67. B. Szabo, R. Urban, N. Limberger, and K. Starke, Naunyn Schmiedebergs Arch.
Pharmacol., 351: 268 (1995).
68. R. H. Loring, Br. J. Pharmacol., 99: 207 (1990).
TM

69. M. K. Sun, S. Regunathan, and D. J. Reis, Clin. Exp. Hypertens., 17: 115 (1995).
70. Y. Gao, B. Gumusel, G. Koves, A. Prasad, Q. Hao, A. Hyman, and H. Lippton, Life
Sci, 57: 83 (1995).
71. G. B. Glavin, M. A. Carlisle, and D. D. Smyth, J. Pharmacol. Exp. Ther., 274: 741
(1995).
72. D. Pinthong, I. K. Wright, C. Hanmer, P. Millns, R. Mason, D. A. Kendall, and
V. G. Wilson, Naunyn Schmiedebergs Arch. Pharmacol., 351: 10 (1995).
73. J. Pineda, J. A. Ruiz-Ortega, R. Martin-Ruiz, and L. Ugedo, Neurosci. Lett., 219:
103 (1996).
74. N. H. Jurkiewicz, L. G. Docarmo, H. Hirata, W. D. Santos, and A. Jurkiewicz, Eur.
J. Pharmacol., 307: 299 (1996).
75. D. Pinthong, J. F. Hussain, D. A. Kendall, and V. G. Wilson, Br. J. Pharmacol.,
115: 689 (1995).
76. M. Grigg, I. F. Musgrave, and C. J. Barrow, unpublished results.
77. G. J. Molderings, L. Kundt, and M. Göthert, Naunyn Schmiedebergs Arch. Pharma-
col., 350: 252 (1994).
78. J. E. Piletz, A. C. Andorn, J. R. Unnerstall, and A. Halaris, Biochem. Pharmacol.,
42: 569 (1991).
79. J. E. Piletz and K. Sletten, J. Pharmacol. Exp. Ther., 267: 1493 (1993).
80. C. M. Brown, A. C. MacKinnon, J. C. McGrath, M. Spedding, and A. T. Kilpatrick,
Br. J. Pharmacol., 99: 803 (1990).
81. R. Zonnenschein, S. Diamant, and D. Atlas, Eur. J. Pharmacol., 190: 203 (1990).
82. J. E. Wikberg, S. Uhlen, and V. Chhajlani, Pharmacol. Toxicol., 70: 208 (1992).
83. S. A. Munk, R. K. Lai, J. E. Burke, P. E. Arasasingham, A. B. Karhlamb, C. A.
Manlapa, E. U. Padillo, M. K. Wijono, D. W. Hasson, L. A. Wheeler, and M. E.
Garst, J. Med. Chem., 39: 1193 (1996).
TM

12
Oxidoredox Suppression of
Fungal Infections by
Novel Pharmacophores
Valeria Balogh-Nair
The City College of the City University,
New York, New York
I. INTRODUCTION
Over the last decade, the incidence of fungal infections has increased dramati-
cally. This can be attributed to drug resistance, the emergence of new pathogens
and resurgence of old ones, and the lack of effective therapeutics. As patients
become severely immunocompromised because of underlying diseases such as
leukemia or acquired immunodeficiency syndrome (AIDS), and as immune sup-
pression is becoming routine in cancer and organ transplantation patients, oppor-
tunistic pathogens have begun preying on a growing population, causing life-
threatening systemic fungal infections. Candida species, the ordinarily harmless
denizens of the gastrointestinal and genitourinary tracts, are now becoming one
of the largest threats in hospitals. Infections by Candida albicans and other oppor-
tunistic fungi are often the most devastating in AIDS patients [1]. Pneumocystis
carinii pneumonia [2] affects ⬎70% of AIDS patients and is one of the most
common lethal infections among them. Cryptococcus neoformans causes poten-
tially fatal meningitis [3] in 7%–10% of AIDS patients and Candida albicans
causes oral and esophageal candidiasis [4] in ⬎70% of them. In this chapter,
focus will be on the development of novel oxidoredox pharmacophores and their
effectiveness against the three selected pathogens, C. albicans, C. neoformans,
and P. carinii.
This chapter is dedicated to the memory of Professor M. S. Ramachandran Nair.
TM

Because fungi are eukaryotes, the search for antifungals selectively detri-
mental to the fungi, but safe to the host, is difficult. Investigations into the modes
of action of existing antifungals have led to a better understanding of the bio-
chemical characteristics of the fungal cell membrane and have provided a target
for drug design. Advances in antifungal therapy have been dominated for many
years by a single large class of synthetic compounds, which have as their common
feature an azole nucleus. Fluconazole [5], like other N-substituted azoles, inhibits
fungal sterol biosynthesis, resulting in the accumulation of 14-methyl fungal
sterols and ultimately in damage to the fungal cell membrane. However, the
unhindered nitrogen lone pair in fluconazole interacts not only with the heme
of cytochrome P-450–linked monooxygenase, which catalyzes lanosterol C-14
demethylation, inhibiting fungal sterol biosynthesis, it also interferes with steroid
metabolism in humans. Although possessing structures entirely different from
the azoles, amphotericin B and nystatin are prominent representatives of polyene
macrolide antibiotics. For more than 30 years, amphotericin B has been the preemi-
nent drug for the treatment of serious systemic fungal infections [6]. The sterol-
dependent ion channel formation in membranes, favoring however slightly the er-
gosterol-rich fungal membranes, is responsible for its potent activity. However,
because of its only slight preference to disrupt fungal membranes rather than mam-
malian ones, its therapeutic value is greatly diminished. Although it can be lifesav-
ing, its side effects tend to make amphotericin B an agent of last resort.
Among the opportunistic infections associated with AIDS, cryptococcosis
can be suppressed with fluconazole, after primary treatment with amphotericin
B (or preferably with flucytosine). About 50% of the AIDS patients do not survive
treatment with flucytosine, and mortality rate during therapy is 33% and 40%
for amphotericin B and fluconazole, respectively. Mucocutaneous candidiasis in
human immunodeficiency virus– (HIV)-infected patients can be treated with flu-
conazole, but its continued use to prevent recurrence leads to both clinical treat-
ment failure and antifungal resistance, especially in highly immunodeficient pa-
tients. Pneumocystis carinii pneumonia (PCP) responds to antiprotozoal agents;
therefore P. carinii has been considered by experts to be a protozoan. More re-
cently, however, on the basis of phylogenetic classification of its ribosomal ribo-
nucleic acid (RNA), it has been reclassified as a fungus [7]. The most effective
drug regimen used against PCP infections is a combination of trimethoprim/
sulfamethoxazole [8,9]. For about 20% of patients, for whom this combination
of a sulfa drug with a dihydrofolate reductase fails, requiring a switch to the
alternate drug pentamidine, the prognosis is poor, with a high mortality rate.
These examples illustrate the shortcomings of the antifungals in use and demon-
strate the acute need for novel, safer, and more effective drugs.
The main approaches to the development of better antifungal therapies in-
clude high-throughput screening programs for novel antifungal antibiotics,
screening of synthetic combinatorial libraries for new leads, and rational drug
TM

design to identify lead compounds based on three-dimensional understanding of
drug receptor interactions. Another approach, the subject of this review, is hy-
pothesis-driven drug development. In this approach, unlike those that led to the
development of azoles and polyene antifungals, the fungal cell membrane is not
the target for drug design. The types of drugs sought are intended to modulate
or mimic the body’s own defense mechanisms against fungal pathogens. We
report herein on the discovery of novel oxidoredox pharmacophores and their
antifungal activities.
II. THE MOLECULAR BASIS OF NATURAL DEFENSES

AGAINST FUNGAL INFECTIONS
A. Macrophage-Derived Defense Molecules
Until the late 1980s, nitric oxide (•NO), a highly toxic gaseous free radical, was
considered just one of the environmental pollutants, a destroyer of ozone, and a
suspected carcinogen. Therefore, the first reports [10,11] identifying it with the
elusive endothelial relaxing factor [12], a vasodilatory messenger, were greeted
by many with skepticism. However, by 1992, •NO had become the ‘‘Molecule
of the Year,’’ emerging as an essential and unifying thread linking neuroscience,
physiology, and immunology [13]. At present, •NO is recognized as a major
messenger molecule in the cardiovascular and nervous systems, and as a major
effector molecule released by murine macrophages and other cells after immuno-
logical activation [14]. As an effector molecule, •NO has a role in the immune
system that is quite different from its function in either neurons or blood vessels.
The •NO released by macrophages is cytotoxic to intracellular microorganisms,
pathogens such as fungi and helminths, and tumor cells. It is in this role that
•
NO, highly reactive and freely diffusing through cell membranes, participates
in the host defense mechanism of eukaryotes. However, since •NO is toxic, its
release by macrophages to control microorganisms, pathogens, and tumors is also
linked to tissue destruction that occurs in a number of diseases, including arthritis,
diabetes, septic shock, transplant rejection, and multiple sclerosis [15]. Control
of the synthesis of •NO, catalyzed by nitric oxide synthases in vivo, thus becomes
a key issue in regulating its activity and all of its physiological functions.
Immune-stimulated macrophages express inducible nitric oxide synthases
(iNOSs). These homodimeric enzymes contain a reductase domain similar to P-
450 reductase, and a P-450-like oxygenase domain that contains heme, H4biop-
terin, and a binding site for l-arginine. These two domains interact to catalyze
the second step in the conversion of l-arginine to citrulline and •NO, as shown
in Scheme 1 [16].
There is a consensus that when acting as a signal transducer, through oxida-
tion of thiols, hemes, Fe clusters, and nonheme prosthetic groups, •NO regulates
TM

Scheme 1 Oxidative conversion of l-arginine to •NO and citrulline catalyzed by iNOS.
the target proteins that evoke the functional response. It is also understood that
coordination to the enzyme-bound ferrous heme by •NO triggers guanylate cy-
clase-catalyzed generation of cyclic guanosine monophosphate (GMP) from gua-
nosine triphosphate (GTP). The increases in intracellular second messenger cyclic
GMP concentrations are then responsible for the cellular responses, such as vas-
cular smooth muscle contraction. However, there is no consensus concerning the
role of •NO in the immunological response leading to the microbicidal effects
and accompanying tissue injury. Along with the mechanisms of action associated
with •NO itself, it has been postulated that redox-related forms of •NO play a
major role. Beckman et al. [17] were the first to point out that under pathophysio-
logical conditions, both •NO and superoxide anion (O •⫺ 2 ) are produced by macro-
phages at high rates. These can react to form the more potent oxidant peroxynitrite
(ONOO⫺), the conjugate base of peroxynitrous acid.
Peroxynitrite is known to react with a variety of biologically important
molecules. Nitric oxide or O •⫺ 2 reacts only slowly with the iron–sulfur cluster in
mitochondrial aconitase, a major target of oxidant-mediated toxicity in cells, but
peroxynitrite rapidly inactivates it [18]. Peroxynitrite oxidizes sulfhydryls, ascor-
bate, α-tocopherol, and lipids. Nitration of phenolic compounds, such as tyrosine,
in a metal-catalyzed process also can occur. Therefore, peroxynitrite is considered
sufficiently reactive to mediate nitric oxide–dependent microbial killing. During
macrophage activation, its precursors •NO and O •⫺ 2 are formed simultaneously,
in high concentrations, and peroxynitrite is formed at almost diffusion-controlled
rates (6.7 ⫻ 10 9 M⫺1s⫺1). This indicates that significant concentrations of peroxy-
nitrite should be attainable in vivo [19,20]. The observation that superoxide dis-
mutase (SOD), a known quencher of superoxide, enhances •NO levels in vivo,
and that nitrotyrosine is detected in biological fluids that produce high levels of
•
NO, casts peroxynitrite in a key role. However, these lines of evidence attributing
to peroxynitrite a key role in vivo are considered by some as circumstantial.
TM

Other interpretations [14] have been advanced, relegating peroxynitrite to a less
important role, if any. Thus, it has been suggested that enhancement of •NO levels
by SOD may not be due to the ability of SOD to dismutate superoxide; SOD could
oxidatively convert a reduced metabolite of •NO back to •NO instead. Alternate
mechanisms for nitrotyrosine formation that do not necessarily require the pres-
ence of peroxynitrite in biological fluids have been proposed. That peroxynitrite
is not the microbicidal species of phagocytes, but only an obligatory intermediate
for forming other cellular oxidants, was proposed on the ground that its rapid
reaction with CO 2 makes its existence in physiological fluids unlikely [21]. An
intrinsic difficulty with these, as well as with many other lines of evidence for
and against peroxynitrite, is that they are derived mainly from in vitro studies.
Progress in the development of methods to detect and identify highly reactive
intermediates in complex biological matrices has been relatively slow, limiting
advances in the in vivo characterization of radical reaction pathways in these
media. Sensors that can readily detect 10 ⫺20 mol of •NO, suitable for measuring
intracellular concentrations, are available [22], but these •NO-specific sensors do
not allow concomitant monitoring of other reactive species, sources, or targets
of •NO under physiological conditions. Further studies are required to establish
whether nitrosyl complexes are as important in cell defense as the reaction of
•
NO with superoxide radical.
B. Neutrophil-Derived Defense Molecules

Neutrophils constitute 50% to 70% of the total white blood cells in humans.
Together with the macrophages, they play a critical role in the immune response
against microorganisms. When triggered, reduced nicotinamide-adenine dinucle-
otide phosphate (NADPH) oxidase in the plasma membrane of neutrophils gener-
ates a superoxide-derived pool of ‘‘reactive oxygen derivatives’’ to serve as a
first line of defense. A second line of defense consists of microbicidal–cytotoxic
peptides, defensins [23], discharged from the neutrophils’ cytoplasmic granules
into the phagocytic vacuole and into the extracellular space. The focus of the
following discussion is on the ‘‘reactive oxygen derivatives’’ because they are
important in developing novel first-line therapies.
Unlike in the case of macrophages, in which the existence of iNOS is firmly
established, in neutrophils the occurrence of a nitric oxide–generating system
has been controversial. Depending on the assay used, it has been shown that
neutrophils can or cannot release •NO [24–30]. Further, there are differences in
the amounts of •NO released by neutrophils from different sources. It was argued
that human neutrophils produce substantially more O •⫺ 2 than do rat neutrophils
and, therefore, the resultant amount of •NO release in human neutrophils is low.
To rationalize the low levels of •NO release in human neutrophils, others invoked
its rapid reactions with active oxygen species or with myeloperoxidase, preclud-
TM

ing its detection. Careful assessment of the available data led us to conclude that
although •NO release might occur in neutrophils, at present there is no evidence
that neutrophil-derived •NO plays a role in the cytotoxic activity of these cells.
It is an alternative system, with myeloperoxidase cast in a central role, that sus-
tains the cytotoxic activity against both normal and tumor cells and may contrib-
ute to inflammatory tissue destruction by neutrophils.
At the start of World War I, solely on the basis of screening of compounds
for bactericidal action, and without awareness of their modes of action, HOCl
and chloramines were employed extensively as microbicides to prevent infection
of wounds. These were the very compounds that are at present held responsible
for the microbicidal action of neutrophils. With many of the crucial mechanistic
details still not well understood, HOCl and chloramines are produced in a com-
plex sequence of events, as follows [31]: The NADPH oxidase in stimulated
neutrophils produces superoxide anion. The bulk of the superoxide anion gener-
ated rapidly dismutates to yield hydrogen peroxide. Almost all of the hydrogen
peroxide produced is used up by the large amounts of myeloperoxidase dis-
charged from the neutrophil (up to 5% of the dry weight of the cell) to oxidize
halides or thiocyanate ions. Because at most sites in vivo the plasma concentra-
tion of chloride is more than one thousand times that of other halides and thiocya-
nate, hypochlorous acid (HOCl) is the major oxidation product. Although the
production of HOCl by human neutrophils can be sustained for up to 3h, HOCl
alone is not the only important oxidant produced by neutrophils. Rapid oxidation
of endogenous primary and secondary amines by HOCl under physiological con-
ditions yields a plethora of N-Cl derivatives, the chloramines [32]; amides are
also oxidized, yielding chloramides. The chloramines are less powerful, but long-
lived oxidizing agents, retaining the oxidizing equivalentce of HOCl. Their re-
lease is associated with the slow oxidation of microbial sulfhydryls, thereby en-
hancing and prolonging the microbicidal effect of neutrophils. The cytotoxicity
of chloramines appears to be modulated by their overall structures, rather than
by the N-Cl moiety alone. The balance between formation of lipophilic and hy-
drophilic N-Cl derivatives may regulate the biocidal activities of neutrophils [33].
About 30% to 40% of the endogenous amines in neutrophils consists of the β-
amino acid, taurine. It is converted by the myeloperoxidase/H 2 O 2 /Cl⫺ system
into the hydrophilic taurine chloramine, a milder oxidant than HOCl. Thus, tau-
rine chloramine formation could attenuate the biocidal activity of HOCl, pro-
tecting the cells from autolysis.
The tissue damage accompanying the release of oxidants has been corre-
lated with activation of latent metalloproteinases contained in the neutrophil
granules [31]. These proteolytic enzymes cause tissue damage by catalyzing
the hydrolytic degradation of the extracellular matrix. However, it appears that
neutrophils have both oxidative and nonoxidative paths for activation of their
metalloproteinases, and it has been found that the HOCl/enzyme ratio must fall
TM

in a narrow range for oxidative activation to occur [34]. Therefore, it is doubtful
whether effective regulation of metalloproteinases by HOCl could occur in vivo,
where various HOCl scavengers abound.
III. THE OXIDOREDOX HYPOTHESIS OF THE

ANTIFUNGAL PHARMACOPHORES
From the 1980s onward, the number of antimicrobial compounds in preclinical

and clinical development has declined and there is now a shortage of truly novel
antimicrobials. In the face of emerging multidrug-resistant microbes, new ave-
nues need to be explored to develop more effective weapons against the threat of
microbe-borne infectious diseases. The rapidly developing field known as ‘‘host-
defense peptide’’ biology, by exploring how various life forms employ antimicro-
bial substances, is offering new hope to solve the problem of microbial resistance
[35]. The natural host-defense peptides, e.g., cecropins, defensins, tachyplasmins,
magainins, and analogues derived from them, which kill microbes by permeabil-
izing their cell membranes, but not those of the host’s cell, are a new class of
drugs exploiting innate immunity. However, these host-defense peptides are not
immune to the common problems associated with peptide drug development,
e.g., stability, delivery, and manufacturing costs, rendering their pharmaceutical
development difficult.
To address the acute need for better antimicrobials, we chose to develop
drugs based on the understanding of host-defense processes occurring in macro-
phages and neutrophils. However, instead of seeking to enhance the drug poten-
tial of natural, peptidic defense molecules, entirely novel, nonpeptidic, and fully
synthetic pharmacophores were sought. Considering the nature of the reactive
species involved in the cytotoxic flux, we hypothesized that simple organic func-
tionalities with oxidoredox properties could modulate microbicidal activity. An
example of this modulation is seen with taurine, which is present in large amounts
in neutrophils and which is chlorinated to the N-chloro derivative by the neutro-
phil’s HOCl. Since the chloramine product is a more gentle oxidant than HOCl,
it attenuates the microbicidal activity. To attempt modulation of biocidal activity,
five different functionalities, but each with oxidoredox properties, were selected
as potential pharmacophores against the opportunistic fungal pathogens, Candida
albicans, Cryptococcus neoformans, and Pneumocystis carinii: Oxaziridines, by
virtue of the properties of their strained ring, are extensively employed in organic
synthesis as oxygen transfer agents. Sulfonyloxaziridines are among the most
effective oxidizing agents currently available for chiral epoxidations. Nitrones
are versatile building blocks employed in constructing heterocycles via 1,3-addi-
tions of the nitrone dipole; they form oxaziridines by electrocyclic ring closure,
and the oxaziridines can then be deoxygenated by various reducing agents. Mac-
TM

rocyclic polyamines could be converted to N-chloro compounds in neutrophils
or could undergo N-oxidation, and metallomacrocycles could form metal oxo
complexes or form adducts with •NO. Currently we are exploring the effect of
these pharmacophores on selected fungal pathogens.
IV. THE OXIDOREDOX CHEMISTRY OF THE

NOVEL PHARMACOPHORES
The controlled release of active oxygen species in an exact position in space in

an organism and at the exact time is critical for health. Since both high and low
levels of these small but vital molecules can be pathogenic, in an attempt to strike
a balance between utilization and elimination of them, drug development efforts
included the use of extracellular forms of SODs, synthetic SOD mimetics, •NO
releasers and scavengers, and a broad range of antioxidant compounds [36]. The
radical scavenging function of SODs appears to protect biological macromole-
cules from oxidative damage. These ubiquitous metalloenzymes have a redox
active cofactor to catalyze the one-electron redox cycle required for oxygen dis-
mutation:
(1)
The use of extracellular forms of SOD enzyme to inhibit oxygen radical–

mediated cell injury met with limited success, and its efficacy is controversial
[37]. Thus, SOD alone or in combination with catalase has been reported to afford
protection from ischemia/reperfusion injury, but recent studies in humans showed
inconsistent results, perhaps because protection occurs only in a narrow range of
doses. This controversy about the efficiency of SOD enzyme emphasized the
need for low-molecular-weight antioxidants, the SOD mimetics. Both in vitro
and in vivo studies with nonpeptidic SOD mimetics indicated that they perform
better than SOD in inhibiting tissue injury. However, because most SOD mimet-
ics catalyze the removal of superoxide anion alone without affecting the removal
of hydrogen peroxide, they offer limited protection from neutrophil-mediated
injury [38].
Since both elevated and reduced levels of •NO (necessary for physiological
function) result in diseased states, depending on the site/process affected, both
•
NO donors and •NO quenchers have therapeutic potential. The ever growing
family of compounds designed to control biological concentrations of •NO in-
cludes the well-known sodium nitroprusside and the sydnonimines that release
•
NO. Recently, a very interesting •NO releaser, possibly a useful source of •NO
in vivo was reported by Lerner’s group [39]. In an antibody-catalyzed retro Diels-
Alder reaction, nitroxyl is released from an anthracene-HNO adduct; the released
TM

nitroxyl is then oxidized to •NO in the presence of SOD. Most efforts to reduce
•
NO levels were in situations in which abnormally high levels of •NO produce
severe clinical problems, such as endotoxic shock, responsible for a high death
rate in surgical emergency units. Inhibitors of iNOS should have high therapeutic
potential to control •NO levels in vivo. The difficulty with approaches to reduce
•
NO levels that use arginine analogues is that the majority of these studies were
carried out using partially purified iNOS, or cells in culture; hence these must
be interpreted with great caution. The amino acid moiety is not a prerequisite for
either potency or selectivity. Thus, isothioureas, bisisothioureas, and a recently
reported amidine derivative showed various degrees of selectivity for human
iNOS [40]. However, the major problem is that for a therapeutic application,
isoform-selective inhibitors are a prerequisite, and the active sites’ structures in
these proteins are not yet known with sufficient precision to permit rational drug
design.
The strategy proposed here to employ oxidizing agents as potential thera-
peutics to treat systemic fungal infections can be viewed as problematic because
of the concern that the damage they may cause to the host will outweigh the
benefits of their microbicidal effects. However, careful selection of the oxidore-
dox functionalities may allow striking a balance between tissue destruction and
microbicidal activity. The rationale for selection of each type of oxidoredox phar-
macophore is outlined below.
A. Sulfonyloxaziridines
Sulfonyloxaziridines, known for two decades, are among the most versatile in
the repertory of oxidizing agents [41]. They are best known as the reagents of
choice for chiral oxidations, but it is other aspects of their chemical characteristics
that work in their favor in the context of pharmacophore development. Because
the oxygen in sulfonyloxaziridines is more electrophilic than is the oxygen in
oxaziridines, sulfonyloxaziridines, with few exceptions, are more reactive to
nucleophiles than are oxaziridines. Oxygen transfer by sulfonyloxaziridines
(Scheme 2) involves S N 2-type displacement of the sulfimine, facilitated by a rela-
tively weak oxygen–nitrogen bond and by the favorable enthalpy of the C N
π bond formed, with a slight bias in favor of a planar rather than a spiro transition
state. Their reactivity is modulated by the substituents attached to the carbon and
Scheme 2
TM

nitrogen, those attached to the nitrogen having the greater effect. However, in
the oxygen transfer reactions of interest here, the dominant factor appears to be
steric, not electronic. Thus, although most sulfonyloxaziridines are highly reac-
tive with nucleophiles, camphorsulfonyloxaziridine does not react with amines.
Sulfonyloxaziridines rapidly oxidize sulfides to sulfoxides, convert disulfides to
thiosulfinates at room temperature, and oxidize thiols via sulfenic acid interme-
diates [42]. Tertiary amines are oxidized to N-oxides, secondary amines to hy-
droxylamines and nitrones, and primary amines to nitroso compounds (Scheme
3) [43].
Of particular interest is the sulfonyloxaziridines’ enhanced reactivity to bio-
logically important nitrogen and sulfur nucleophiles. We hypothesized that these
reactions may be important for modulation of the oxidoredox species in vivo and
that they might be equally relevant to the pharmacological action of sulfonyloxa-
ziridines. In humans, the hepatic flavooxygenase plays a crucial role in the detoxi-
fication of natural and xenobiotic amines and sulfides by N and S oxidations;
thus N-oxides are apparently the straightforward metabolic reaction products of
tertiary amines. It is interesting that the N-oxide metabolites that sulfonyloxaziri-
dines might produce, were they to react with biological amines, are the same as
the metabolites formed in vivo. However, the enzymatic reactions are reversible
by reductases that are able to deoxygenate the N-oxide metabolites back to the
amines. If sulfonyloxaziridines react with biological secondary amines to form
hydroxylamines and/or nitrones, these products too might act as oxidoredox mod-
ulators (as discussed in the next section). In model reactions, when two equiva-
lents of sulfonyloxaziridine are employed to oxidize secondary amines, nitrones
are produced almost exclusively. Nitrones are known metabolites. Human hepatic
flavomonooxygenases produce them in detoxifying hydroxylamines (Scheme 4).
Scheme 3 Oxidation of sulfur and nitrogen nucleophiles by arylsulfonyloxaziridines,

or by flavoenzyme mixed function oxidase. (Source: From Ref. 43.)
TM

Scheme 4 Conversion of hydroxylamines to nitrones in enzymatic and model reactions.
(Source: From Ref. 43.)
Since sulfonyloxaziridines transfer an oxygen atom to the same types of sub-

strates as enzyme systems do, they should be excellent oxygenase mimics.
Marked interest in the mechanisms of biochemical reactions of peroxidases,
catalases, and cytochrome P-450s that modulate oxidant levels in vivo spawned
numerous studies on model systems in which various oxidants interact with the
metal centers of heme and nonheme prosthetic groups. This led to a consensus
that a chemical process similar to the ‘‘peroxide shunt’’ of cytochrome P-450
enzymes takes place when peroxidases react with H 2 O 2, alkylhydroperoxides, or
percarboxylic acids. In each case, the key intermediate is an enzyme-bound iron-
oxo(IV)-porphyrin π-cation radical species [44,45] containing an Fe(IV) O
unit at the iron center. In horseradish peroxidase, this iron-oxo species can react
with a number of electron donors, including amines and sulfides, before returning
to the Fe(III) state. However, in the case of chloroperoxidase and the myeloperox-
idase of the neutrophils, this iron-oxo species reacts with chloride to yield HOCl.
Thus, if sulfonyloxaziridines were competent to generate iron-oxo intermediates,
these in turn would increase the levels of the oxidizing species of neutrophils.
Unfortunately, most of the model studies were focused on the interaction of H 2 O 2
and other peroxides with the Fe(III) or Mn(III) centers in porphyrins [46] or
employed various other auxiliary oxidants in combination with metal centers, but
these did not include oxaziridines. Yuan and Bruice [47], however, reported stud-
ies in which a sulfonyloxaziridine was employed as an oxene transfer agent to
manganese (III) tetraphenylporphyrin chloride in the catalytic epoxidation of al-
kenes. This lends support to our hypothesis that the sulfonyloxaziridine pharma-
cophore may act by forming an oxene, which serves to regulate the levels of
biocidal oxidoredox species. Considering the remarkable reactivities of sulfonyl-
TM

Scheme 5
oxaziridines, together with their antifungal properties reported here, it is antici-

pated that studies in progress will help to determine whether oxo-transfer reac-
tions contribute to the antifungal effects or whether other mechanisms operate.
Since both photochemical and thermal rearrangements of oxaziridines yield
isomeric amides, it is tempting to speculate whether sulfonyloxaziridines undergo
similar types of rearrangements to form the corresponding sulfonylamides. A
precedent for this could be the isolation of N-(p-chlorophenylsulfonyl)acetamide
by Davis et al. [48] from an unsuccessful attempt to oxidize a sulfonylimine to
the sulfonyloxaziridine (Scheme 5). Although the major first-pass metabolites of
the sulfonyloxaziridine functionality are likely to originate from the sulfonyli-
mine, formed by loss of the active oxygen, the occurrence of rearrangements to
sulfamide-containing compounds in vivo should lead to metabolites with interest-
ing pharmacological properties [49].
B. Oxaziridines
Despite an idea advanced in the early seventies that the critical oxygenating spe-
cies of flavin oxygenases [50] is an oxaziridine, before our studies [51] there were
no efforts to explore the pharmacophore potential of the oxaziridine functionality.
These oxygenases are unique in that they carry out the only known non-metal-
ion-requiring oxygen activation reactions in biological systems. They bind and
activate molecular oxygen, ultimately transferring one oxygen atom to substrate
and releasing the second as water. On the basis of mechanistic studies of these
enzymes, in 1974 Orf and Dolphin [50] proposed an oxaziridine intermediate as
the monooxygenating species, derived by rearrangement of an initial intermedi-
ate, 4α-hydroperoxyflavin (Scheme 6). Working with nonenzymatic models, Ras-
tetter et al. [52] proposed that a nitroxyl radical derived from Dolphin’s oxaziri-
dine (Scheme 6) is a viable candidate for the oxygenating species. However,
many other mechanisms were proposed later, not involving the oxaziridine, and
these received more attention [43].
The lack of interest in possible biological roles for oxaziridines can be
attributed to the fact that at the time of the mechanistic proposals discussed not
much was known about the chemical properties of of oxaziridines. Despite the
substantial number of papers published since then, many details in the mechanism
TM

Scheme 6 Oxaziridine [50] and nitroxyl radical [52], the putative oxygenating species
in flavin monooxygenases.
of oxygen transfer from structures where the active site oxygen is part of a three-
membered ring, e.g., in metal peroxides, dioxiranes, oxaziridines, and sulfonylox-
aziridines, remain obscure. The similarity in the active site structures, however,
suggests that they may have a common mechanism of oxygen transfer, an S N2-
type displacement by the nucleophilic substrate (Z) on the electrophilic oxygen
atom (Scheme 7). Kinetic studies of deoxygenation of oxaziridines and sulfony-
loxaziridines [53], epoxidation of alkenes by sulfonyloxaziridines [54], and theo-
retical calculations on the transfer of oxygen from oxaziridines to ethylene [55]
support this mechanism.
Oxaziridines are active oxygen compounds. They transfer their ring oxygen
atom to tertiary phosphines and oxidize HI, but they are considered to be poor
reagents to oxidize sulfides to sulfoxides, or to epoxidize alkenes. However, their
reactivity is strongly dependent on the substitution pattern of the oxaziridine.
Thus, a bis-oxaziridine oxidized thiacycloalkanes to sulfoxides in only 5%–7%
yields, but an N-tert-butyloxaziridine converted dimethylsulfide to the sulfoxide
quantitatively [41]. An oxaziridine substituted with electron withdrawing groups,
2-(trifluoromethyl)-3,3-difluorooxaziridine, epoxidized alkenes under extremely
mild conditions (⫺50°C, ⬃1 h), and perfluorodialkyloxaziridines performed hy-
droxylation of unactivated tertiary aliphatic C-H bonds at room temperature, in
high yields [56,57]. Moreover, aza-aromatic N-oxides, when irradiated with ultra-
violet (UV) light, transfer their oxygen atom. It was proposed that oxaziridines,
formed by photoisomerization of the N-oxides, are the oxygen transfer agents,
Scheme 7 S N2-type mechanism of oxygen transfer.
TM

and the loss of oxygen is facilitated by rearomatization of the heterocycle [58].
Oxaziridines react with ferrous salts, leading to products resulting from homolytic
cleavage of the oxaziridine ring. Their broad range of reactivities and their poten-
tial to react with nitrogen and sulfur nucleophiles, ubiquitous in biological sys-
tems, make oxaziridines attractive candidates for modulating the oxidoredox pro-
cesses occurring in phagocytic cells.
C. Nitrones
Nitrones could directly modulate the levels of oxidoredox species important in
the biocidal action of macrophages and neutrophils by preferential trapping of
one or more of the biocidal species. However, it is more likely that a complex
series of events takes place, in which nitrone-derived radicals and/or nitrone-
derived hydroxylamines are major contributors to the biocidal effects:
1. Nitrones, by undergoing ring closure to the corresponding oxaziridines,
could act as precursors of these efficient oxygen transfer agents dis-
cussed in Section IV.B.
2. Nitrones are well known spin traps [59]. Depending on their structure,
they can react with C-, O-, N-, and S-centered radicals, including those
produced by macrophages and neutrophils, to modulate their in vivo
concentrations (Scheme 8). Recently, efficient spin trapping of O-, C-,
and S-centered radicals and peroxynitrite, but not superoxide, was re-
ported when using 2H-imidazole-1-oxides as spin traps [60]. Because
the radical trapping reactions of nitrones yield stable nitroxyl radicals,
these radicals might participate in controlling the levels of oxidoredox
species present in vivo by serving as traps for short-lived radicals. On
the other hand, stable nitroxyl radicals are SOD and catalase mimics,
Scheme 8
TM

Scheme 9
and hence they are potential agents of cytoprotection against postisch-

emic reperfusion injury [61].
Heteroaromatic N-oxides, some reactions of which resemble those
of nitrones, are currently being evaluated as hypoxia-selective cytotox-
ins in the clinical treatment of solid tumors [62]. It is believed that the
tumor selectivity is due to both tumor hypoxia and the expression of
high levels of reductive enzymes. It has been suggested that nontoxic,
free radical reduction products of the heteroaromatic N-oxides are in-
volved in the mechanism of cytotoxicity. Further, it was proposed that
these radicals are toxic to the hypoxic cells, but the drug, D, is restored
in the well-oxygenated tissues (Scheme 9). However, no mechanism
was proposed to explain how the N-oxides would inflict damage on
the target, except that drug is likely to abstract a hydrogen from the
sugar residue of DNA, generating a sugar radical.
3. We hypothesized that various reductants that are abundant in biological
systems could convert the nitrones to nitroxyl radicals and to hydroxyl-
amines (Scheme 10). The hydroxylamines produced by reduction of
Scheme 10
TM

nitrones could react with superoxide radicals, and with the peroxyni-
trite from macrophages. In support of this, very recently new spin traps,
TEMPONE-H and CP-H, containing hydroxylamine moieties were re-
ported to trap efficiently both superoxide radical and peroxynitrite,
yielding nitroxyl radicals [63].
4. VBN-3, a bisnitrone, is highly active against P. carinii, both in vitro
and in vivo. Among the several mechanisms by which VBN-3 could
work in vivo, two pathways are shown in Scheme 11. According to
pathway a, reductive activation would yield a diradical. This could
release two •OH/mol of VBN-3, and cause the antifungal effect. The
antifungal properties of heteroaromatic N-oxides were attributed to
their ability to produce hydroxyl radicals that are known DNA cleavers,
causing oxidative damage to the sugar–phosphate backbone of DNA
[64]. The DNA damage may also occur as a result of H• abstraction
by the diradical initially formed via pathway a. According to pathway
b, oxidative activation would yield the nitronyl nitroxide, a diradical.
Model experiments showed this radical can be obtained by slow air
oxidation of VBN-3, as a purple, crystalline, and stable material. In
vivo, oxidants of neutrophils, for example, H 2 O 2, could accomplish
this step. The nitronyl nitroxide should readily react with •NO, since
other nitronyl nitroxides have been employed to detect •NO in air [65].
The reaction with •NO would yield the imino nitroxide. The imino
nitrogen in imino nitroxides has a pronounced basic character and can
be reversibly protonated; the nitroxide is reducible, for example, by
ascorbate or SH groups, to the corresponding hydroxylamine. Reduc-
tion by SH groups, however, is still controversial. The nitronyl ni-
troxide could also be derived from pathway a, by dehydration/oxida-
tion steps (Scheme 11). The bond energy for the OH bond in
hydroxylamines is relatively low (⬃70 kcal/mol), and therefore reoxi-
dation to the nitroxides can be done with a variety of oxidants. In vivo,
the hydroxylamines could be reoxidized to nitroxides, for example,
by flavomonooxygenase (FlOOH), by a metal-catalyzed reaction that
proceeds through superoxide, or by myeloperoxidase (MPO) oxidants
of the neutrophils. Simple nitroxide spin traps, in which the nitroxide
is contained in five-membered and six-membered rings, protect mam-
malian cells from oxidative damage; hence they appear to function as
SOD mimics. Although the SOD activities of the nitroxides tested to
date are orders of magnitude lower than that of SOD, their low toxicity
and high cell permeability might permit employing them at higher con-
centrations to enhance their SOD activity. It is interesting to speculate
whether the nitroxides derived from VBN-3 types of compounds
(Scheme 11) might also confer SOD activity on them.
TM

Scheme 11 Pathways postulated for (a) reductive and (b) oxidative activation of
VBN-3.
D. Macrocyclic Amines
The rationale for selecting macrocyclic amines as potential microbicidal pharma-
cophores is based on consideration of the reactions of amines with the potent
oxidants produced by macrophages and neutrophils. The macrocyclic structure,
as carrier of the amine functionalities, was chosen because such structures can
TM

incorporate several amine moieties within the same molecule, thereby providing
high local concentrations of the pharmacophore, as in the ‘‘respiratory burst’’
characterized by high local concentrations of the reactive species. Further, we
envisaged that the macrocyclic structures will facilitate partition of the com-
pounds into membranes, and we have conducted preliminary toxicity studies to
select macrocyclic frameworks devoid of toxicity to human cell lines.
The mechanisms of amine oxidation by peroxynitrite or peroxynitrous acid
have been controversial. Peroxynitrous acid is both a one- and a two-electron
oxidant, but two-electron oxidations (that is, oxygen transfer) of amines by
HOONO have been observed experimentally [66].
From the limited set of macrocyclic amines tested to date, it is not likely that
N-oxidation of tertary amines confers antifungal activity. Another possibility is
that biologically abundant metal nitrosyl complexes acting as electrophilic ni-
trosating agents transfer NO⫹, thus converting the amines to N-nitrosamines.
However, the biological relevance of NO⫹ under physiological conditions has
been disputed [67]. An alternate, and most likely explanation for the mode of
action of macrocyclic amines is that they function as prodrugs; that is, they are
converted by the HOCl of neutrophils into N-Cl derivatives, potent and long-
lasting oxidizing agents. It is interesting to note that two marine alkaloids, pa-
puamine and haliclondiamine, containing unusual 13-membered macrocyclic
rings encompassing 1,3-diaminopropane moieties, display significant antifungal
activity against C. albicans. However, the recent total synthesis of papuamine
by Weinreb’s group [68] required 16 steps, whereas the macrocyclic imines re-
ported here, highly active against C. albicans, can be obtained from commercially
available starting materials, in 2 steps, in ⬎70% overall yields.
E. Metallomacrocycles
The rationale for employing metallomacrocycles as antifungals was based on
consideration of the reactivities of their biological counterparts, as follows:
1. Peroxidases contribute to the modulation of the physiological levels of

microbicidal oxidants. They have iron-porphyrin cofactors, and when
activated, their active site structures include the FeIV ⫽ O unit responsi-
ble for oxygen atom transfer to substrates. The electron needed to com-
plete the oxygen’s octet is drawn from the porphyrin, to yield porphy-
rin⫹•, or from an active site amino acid residue, to yield residue⫹•
TM

(Scheme 12). For example, horseradish peroxidase and catalase contain
porphyrin⫹•, but cytochrome c peroxidase contains residue⫹•. Besides
these well-characterized oxo-metal species, numerous catalytic sys-
tems mimicking the oxygen transfer step have been described.
The macrophages’ iNOSs, akin to the P-450 enzymes, are also well
characterized. They have their heme coordinated to the apoprotein
through an axial thiolate ligand. The molecular events involved in •NO
production and ligand release from the metalloporphyrin-NO com-
plexes in iNOS are known to a great extent. In contrast, although a
large amount of information on the parameters of activation of the
neutrophil NADPH oxidase complex has accumulated during the past
few years, the heme active site structure of its O •⫺2 generating redox
core, flavocytochrome b, remains elusive. However, an interesting ob-
servation was made in connection with the activation of the NADPH
oxidase. It was reported that the endogenous oxidative activity of mac-
rophages can be enhanced by peroxidases [69]. Thus, when horseradish
peroxidase was added to macrophages, it enhanced their phagocytic
and oxidative activity, even though some of the peroxidase added was
lost by proteolysis. This implied that the heme of the peroxidase might
be transferable to the cytochrome b of NADPH oxidase, to activate
the latter. This raised the question of whether synthetic metallomacro-
cycles might be employed, instead of the endogenous hemes, in drug
design to enhance the microbicidal effects of macrophages and neutro-
phils.
2. Further impetus to test whether synthetic metallomacrocycles might
have microbicidal activity was provided by consideration of the high
reactivity of certain hemoproteins to •NO. Reaction with protein cyste-
ine residues yields nitrosothiol adducts, and reaction with vicinal thiols
generates disulfide bonds. Reaction with a ferrous heme center, as in
deoxymyoglobin, yields Fe 2⫹-N O, that is, nitrosomyoglobin. When
•
NO reacts with the superoxide form of bound oxygen, Fe 2⫹ Oδ⫹-Oδ⫺,
as in oxymyoglobin, •NO diffuses into the heme pocket first, then reacts
Scheme 12
TM

rapidly to give Fe 3⫹-OH2, that is, metmyoglobin and nitrate [70]
(Scheme 13).
Ferrihemes also react with •NO. Thus, metmyoglobin, ferricyto-
chrome c, and catalase bind •NO reversibly, and methemoglobin reacts
with it irreversibly to yield nitrosohemoglobin [71]. The reactivity of
ferriheme nitrosyl adducts can be rationalized in terms of the charge
transfer from the •NO to the metal to give a structure best described
as Fe II ⋅ ⋅ ⋅ ⋅ NO⫹. This makes the nitrosyl more electrophilic, and
susceptible to attack by a variety of biological nucleophiles.
Other important reactions of •NO with hemes relevant to the antifun-
gal effect of metallomacrocycles are as follows: Binding of excess •NO
to the heme center of iNOS deactivates the enzyme, thereby providing
an unusual example of a self-regulating enzyme whose products deacti-
vate it. Binding of •NO to the soluble guanylate cyclase’s b type heme
activates this receptor to produce cGMP [72]. Removal of the heme
results in loss of ability of •NO to activate the enzyme, whereas recon-
stitution of the enzyme with its heme partially restores the ability of
•
NO to activate it. Ignarro, in the early 1980s, demonstrated that puri-
fied heme-deficient soluble guanylate cyclase can be activated by pre-
formed nitrosyl–heme complexes, and also by protoporphyrin IX [73].
He proposed that •NO activates soluble guanylate cyclase by binding
directly to the heme to form a 5-coordinate nitrosyl-heme complex
[74].
On the basis of the preceding considerations, it is likely that addition
of surrogate prosthetic groups might activate the otherwise latent mi-
crobicide-producing enzymes, such as NADPH oxidase and iNOS, to
spur production of endogenous oxidants. However, it is also possible
that by forming surrogate heme–oxidant complexes, better oxidants,
Scheme 13
TM

that is, microbices, become available in vivo. The synthetic metalloma-
crocycle, VBN-10, is highly active against P. carinii, both in vitro and
in vivo. Mode of action studies will establish whether its activity is
related to the formation of metal-oxo or metal-nitrosyl complexes in
vivo.
3. Remarkable findings about •NO binding to hemoglobin were recently
reported by Stamler’s group [75]. They demonstrated that S-nitrosoth-
iol formation endows hemoglobin with hitherto unsuspected allosteric
and electronic properties, which enable it to control tissue oxygenation
levels. Specifically, when oxygen is bound to heme iron, this binding
promotes S-nitrosylation of cysteineβ93 by •NO. Deoxygenation trig-
gers an allosteric transition from the oxygenated (R) state to the deoxy-
genated (T) state and concomitant release of •NO. The R state of the
protein contracts blood vessels; the T state relaxes them. Thus, hemo-
globin senses the oxygen gradient in tissues via conformational
changes and ensures that blood flow is in line with oxygen require-
ments. In hypoxic tissues •NO is released, but if oxygen supply exceeds
demand, the •NO is retained by the hemoglobin in the R structure with
the net effect of reducing blood flow. These newly discovered proper-
ties of hemoglobin will have far reaching therapeutic applications, not
limited to the control of blood pressure. The mode of action of any
drug candidate having structure enabling it to affect •NO levels must
take into account the •NO shuttle of hemoglobin. This is especially
relevant to those novel pharmacophores we are developing whose
mode of action might be tied to •NO levels, or to levels of oxidants
connected to it. It will be important to evaluate how the high, local
concentrations of •NO, or other oxidants produced by macrophages and
neutrophils in response to pathogen attack, fit the new image of the
•
NO shuttle.
V. THE SYNTHESIS OF ANTIFUNGAL DRUGS

CONTAINING OXIDOREDOX PHARMACOPHORES
Of the five pharmacophores selected for synthesis, at least one compound in each
class showed high activity against P. carinii, except the macrocyclic polyamines,
which showed activity only against Cryptococcus neoformans and Candida albi-
cans. Further, we recently established, in the case of the metallomacrocycle phar-
macophore VBN-10, active against P. carinii, that the in vitro activity closely
correlates with in vivo efficacy. Here we summarize the syntheses of the lead
compounds and their respective activities:
TM

Scheme 14 a, Amberlyst 15/toluene/reflux, 16 h; 47%; b, m-CPBA/aq. NaHCO3-
CH 2 Cl2 /BTEAC, 0°–2°C; 92%.
A. Synthesis of Sulfonyloxaziridines
Finding novel structural classes, hitherto not employed as therapeutic agents, is
usually difficult and seldom attempted. We overcame this hurdle by synthesizing
the lead compounds, a sulfonyloxaziridine podand, VBN-1, and the macrobicy-
clic oxaziridine, VBN-2 (see Section V.B). However, since these are truly new
structural leads and thus have not been employed in any drug development, it is
necessary to determine which types of molecular structures are best as vehicles
for delivery of the active pharmacophore.
Podand VBN-1 contains two sulfonyloxaziridine moieties; it was obtained
in 92% yield by m-CPBA (meta-chloroperbenzoic acid) oxidation of the corre-
sponding sulfonimine [76] in the presence of a phase transfer catalyst [77]
(Scheme 14). It is a stable white solid that can easily be obtained on ⬎10 g scale.
The two oxaziridine moieties in VBN-1 have identical geometry, as ascertained
by spectral data. In vitro, VBN-1 was active against P. carinii, at a concentration
of 2.1 µM, comparable to that of pentamidine. In vivo testing in mice indicates
it has no gross toxicity at 50 mg/kg/day intraperitoneally the highest concentra-
tion tested to date.
B. Synthesis of Oxaziridines
Since oxaziridine as a pharmacophore was unknown prior to our studies, in the
first attempt to establish its usefulness against P. carinii we prepared a compound
containing several oxaziridine units, so as to have as many active oxygens as
possible delivered per mole. We were aware of the formidable difficulties, since
one can theoretically obtain a large number of isomers. To gain some control
over the stereochemistry we chose a rigid macrocyclic framework as carrier for
the oxaziridine groups. The hexaimino macrobicyclic precursor of VBN-2, which
the spectral and x-ray crystallographic analysis showed to contain six trans-im-
ines [78], was deemed to be a good candidate. Inspection of its structure generated
with MacroModel and those of two related hexaimino macrobicycles allowed us
to visualize the likely mode of attack of the peracid on these imines. Even though
the initial trans-geometry of the imines is essential, this alone was not expected
to provide stereocontrol over the production of the oxaziridines, since in nonmac-
TM

rocyclic analogues oxaziridine ratios obtained were independent of the composi-
tion of the initial imine mixtures [79]. However, if one considers a Baeyer-
Villiger-type mechanism in which the peracid adds to the imine to form the tetra-
hedral intermediate, which then affords the oxaziridine by nucleophilic attack of
the nitrogen on the electron-deficient oxygen atom, then blocking attack by the
peracid on the C N bond from the ‘‘endo’’ direction by the macrocycle’s
framework should favor the production of only one oxaziridine diastereomer.
Examination of the structure of the hexaimino macrobicycle indicates this might
be the case. Further, it was reasonable to assume that N-O bond formation to
yield the oxaziridine versus C-N bond rotation will be strongly influenced by the
macrocyclic framework.
The molecular modeling studies employed to select the structure of a mac-
robicyclic imine precursor were also helpful in the interpretation of the nuclear
magnetic resonance (NMR) spectra of the macrocycles. For example, in the hexa-
imino macrobicycle prepared from 1,3-benzenedicarboxaldehyde, one group of
aromatic protons appeared at an unusually high field (5.3 ppm). We used Mac-
roModel to explain this observation and to confirm the structure. It showed that
in the energy-minimized conformation, the molecule is highly symmetric. The
high-field proton on each aromatic ring points directly to the center of an another
aromatic ring, thus explaining this shielding (Scheme 15).
We synthesized macrobicyclic oxaziridine, VBN-2, by m-CPBA oxidation
of a hexaimino macrocycle in the presence of BTEAC (benzyltriethylammonium
chloride), a phase transfer catalyst [51]. Starting material, the macrocyclic Schiff
base, was obtained by template-free [3 ⫹ 2] condensation of terephthaldicarbox-
Scheme 15 Energy-minimized conformation of a hexaimino macrobicycle containing

1,3-disubstituted aromatic rings.
TM

aldehyde with tris(2-aminoethyl) amine in excellent yield (88%, Scheme 16).
Two other groups have also synthesized this hexaimino macrocycle, using the
template-free route, in yields of 50% [80] and 80% [81], respectively. Selecting
the suitable solvent was critical; in CH 3CN as solvent, oligomers/polymers were
obtained instead of the desired [3 ⫹ 2] adduct [51,80]. The structure of the metal-
free macrocyclic Schiff base was determined by x-ray crystallography [78], which
showed a trans-arrangement for all the imine functions, a divergent (uncoordinat-
ing) arrangement of the N donors, and a tiny cavity of 4Å diameter. Oxidation
of the hexaimino macrocycle, using Davis’s method [82], gave VBN-2, the first
member of a new family of macrocycles carrying oxaziridine groups. VBN-2
transfers six oxygen atoms to triphenylphosphine quantitatively. Molecular mod-
eling using MacroModel along with 1 H and 13 C NMR spectroscopic analysis [51]
indicated that the molecule is highly symmetric and that all oxaziridine moieties
have E geometry.
The 1 H nmr of the VBN-2 consists of two sharp singlets in the low-field
region, indicating that all aromatic protons are equivalent, as are all six oxaziri-
dine protons; the methylene protons are well resolved. Oxaziridine isomers have
diagnostic values for the oxaziridine proton in the E isomer at δ 4.6–4.7 ppm,
whereas in the Z isomers this proton is deshielded by the nitrogen lone pair to
δ 5.3–5.4 ppm. Conformational motions are limited in VBN-2, as shown by the
well-resolved methylene signals. Thus, in VBN-2 all oxaziridines are E, and
hence control of stereochemistry by using the macrocycle is evident. A related
macrobicyclic Schiff base that has a larger cavity size and that, because of one
additional methylene group in each bridge, is more flexible than the precursor
of VBN-2 gave a complex mixture of products, from which no oxaziridine could
be isolated. There are now many examples that apply a similar control of three-
dimensional (3D) space, via macrocycles or molecular recognition constructs, to
achieve desired stereochemistry. An interesting example is an extended aromatic
Scheme 16 a, Anhydrous EtOH, reflux/3 h/N2; 88%, b, m-CPBA/CHCl3 /aq.

NaHCO3 /BTEAC/0°–2°C, 6 h; 25%.
TM

shelf (naphthyl group) that acts as an impenetrable barrier to the approach of the
reagent from the bottom of the structure [83].
Toxicity studies in human cell lines, CEM-SS and Jurkat, were reported
recently [77]. VBN-2, with an in vitro activity of 1.5 µM against P. carinii, is
close in activity to pentamidine, thus demonstrating the potential of the novel
oxaziridine pharmacophore (Table 1, Sec. VI). The Indiana University culture
method [84] demonstrated that VBN-2 at 2 µg/ml decreased growth by 50%
compared to that of untreated cultures, suggesting that VBN-2 may be effective
against P. carinii in vivo.
C. Synthesis of Nitrones
During our studies on retinoids, we synthesized highly conjugated open chain
nitrones [85] and found that this conversion reduced the teratogenicity of reti-
noids [86,87]. However, we decided to try a less conjugated molecule containing
two nitrone groups and an aromatic ring to employ as a pharmacophore against
P. carinii. Bis-(amidine oxides) were very attractive as target structures for test-
ing the nitrone pharmacophore. Nitrones are known to react with the biologically
important radicals, which we believe to be particularly relevant to the inhibition
of P. carinii (e.g., •OH, •OOH, O ⫺• •
2 , and NO), and are thereby themselves con-
verted to more stable aminoxyl (nitroxyl) radicals. These aminoxyls elicit height-
ened interest as potential therapeutic agents, spin traps, and synthetic intermedi-
ates [88]. The aminoxyl radicals generated from the reaction of nitrones are in
themselves of considerable interest because their drug potential is enhanced, as
evidenced by the observation that they are typically quite nontoxic in vivo [59].
Nitronyl nitroxides were also of interest because they react specifically with nitric
oxide [89], an important product of iNOS in macrophages, to yield imino nitroxyl
radicals.
We prepared the target nitrones, VBN-3 and VBN-4, and nitronyl ni-
troxides, VBN-3-ox and VBN-4-ox, by condensation of 2,6-pyridinedicarboxal-
dehydes and 1,4-benzenedicarboxaldehydes with 2,3-bis(hydroxylamino)-2,3-
dimethylbutane sulfate followed by oxidation of the condensation products
according to the procedure of Ullman [90,91]. In the case of 2,6-pyridinedicar-
boxaldehyde, yellow crystals of two products, VBN-3 and VBN-3′, were ob-
tained in 85% and 5% yields, respectively. The structures of both were ascer-
tained by UV, infrared (IR), 1 H, and 13 C NMR spectra and by MS. Interestingly,
these conformers do not interconvert in solution at room temperature and can be
readily distinguished by the low-field region 1 H NMR spectra. In VBN-3′ only
one of the pyridine protons (C-5H) shows the typical deshielding (δ 9.5 ppm)
[92] by the nitrone oxygen, whereas both 3 and 5 protons are deshielded in VBN-
3 (δ 9.44 ppm). In VBN-3′ the pyridine-3-proton appears at δ 8.21 ppm.
TM

Bisnitrones VBN-3 and VBN-4, containing the pyridine and 1,4-disubsti-
tuted benzene rings, respectively, showed both stability and nontoxicity to Jurkat
cell lines. Podand VBN-4 was inactive against P. carinii; it self-associates in solu-
tion [93], and this stacking may be the cause of its lack of activity. But VBN-3
showed excellent activity. At the lowest concentration tested, 2.8 µM (1 µg/mL),
79.4% inhibition was observed after 24 h, and after 48 h no live P. carinii could be
detected (Table 1, Sec. VI). In vivo tests of this nitrone in mice are in progress at
the National Institute of Allergy and Infectious Diseases (NIAID) (Table 4).
D. Synthesis of Macrocyclic Amines

and Metallomacrocycles
In order to assess the potential of the target metallomacrocycles as oxidoredox
pharmacophores against opportunistic infections (OIs), a prerequisite was the
development of synthetic methods by which these macrocycles can be obtained
metal-free (template-free) from easily accessible starting materials in reasonable
yields. However, traditionally, to prevent formation of oligomeric/polymeric
products, the majority of the macrocycles that interest us were obtained via tem-
plate-directed syntheses. For the planned purpose, the resulting template–ligand
complexes would have only limited usefulness in developing efficacious agents
against the OIs because the presence of the template would render modification of
the structures difficult. Although the ‘‘macrocyclic effect’’ [94,95] would confer
enhanced stability to the template–ligand complex, it would at the same time
preclude many of the further reactions intended to enhance in vivo efficacy of
the macrocyclic structure. Exchange of the template and attempts to dissociate
it have met with only limited success. In the few early cases when the macrocyclic
rings were obtained template-free, without recourse to a metal template or high
dilution techniques, some special circumstances were present. Examples include
hydrogen bonding to reduce lone-pair repulsions that would otherwise be domi-
nant within the macrocyclic ring, and successful high yield (⬎90%) syntheses of
macrocycles from sterically rigid precursors, which by promoting intramolecular
TM

Scheme 17 Synthesis of metal-free macrocyclic imines by [2 ⫹ 2] condensations.
(Source: From Ref. 77.)
hydrogen bond formation favored ring closure. Lehn was among the first in 1987
[96] to report efficient syntheses of macrocyclic Schiff bases via template-free
[3 ⫹ 2] condensations. Since that time, diverse needs for metal-free macrocyclic
ligands led to the development of methodologies to obtain them by simple [2 ⫹
2] and [3 ⫹ 2] macrocyclizations. We synthesized the Schiff base macrocycles,
such as the macrobicyclic imine precursor of VBN-2, and a series of iminomacro-
cycles, such as VBN-5 to VBN-9, and the macrocyclic amide precursor of VBN-
10 via these nontemplate syntheses [51,77]. These, together with the syntheses
of aminomacrocycles screened for antifungal activity against Candida albicans
and Cryptococcus neoformans, result to date in over 50 macrocycles made in
template-free manner. The success of these syntheses (yields ⬎ 50%), in which
we obtained the metal-free macrocycles by direct [3 ⫹ 2] and [2 ⫹ 2] condensa-
tions (step a, Scheme 16; Scheme 17), depends on the careful selection of inher-
TM

ently rigid reactants for condensations. This reduces the conformational degrees
of freedom in the reactants/intermediates, thereby optimizing entropy factors that
promote cyclization. Further, the target macrocycles are designed with rigid aro-
matic ‘‘head groups,’’ and some of them, e.g., VBN-5 to VBN-8, also contain
diphenylmethane ‘‘hinges’’ that contribute to the observed stability of these
Schiff base macrocycles.
The precursor of Schiff base macrocycle VBN-2 and imino macrocycles
VBN-5-9 were synthesized in high yields (⬎80%), template-free, via [3 ⫹ 2]
and [2 ⫹ 2] condensations [51,77]. However, it is important to notice that experi-
mental conditions will vary, depending on the type of macrocyclic structure that
contains the Schiff base groups. Thus, Lehn’s group obtained the macrobicyclic
hexaimino Schiff bases by simply adding dropwise an acetonitrile solution of the
dialdehydes to a solution of tren, [N(CH 2 CH 2NH 2 )3 ], in CH 3 CN and stirring the
solution at room temperature for 15 min, when ‘‘a precipitate of almost pure
macrobicycle hexaimine was formed’’ [96]. On the other hand, Menif and Martell
report [97] a complex NMR spectrum (possibly due to Schiff base isomers) when
a tetraimino macrocycle (unlike Lehn’s macrobicycle) was prepared (in 72%
yield) using Lehn’s reaction conditions, but longer reaction times. Nelson reports
synthesis of Schiff base macrobicyles via direct [3 ⫹ 2] condensations in reflux-
ing alcohol (⬃60% yields), but in this same series of compounds that contain
aromatic ‘‘head groups’’ a thiophene-containing one could not be obtained tem-
plate-free, and in the solvents employed condensation of furan-dicaroxaldehyde
with 2,6-bis(aminomethyl)pyridine did not give the macrocyclic [2 ⫹ 2] product
[98]. These authors indicate the importance of the solvent employed; thus a
trispyrrole BISTREN macrocycle could not be obtained when CH 3 CN was used
as solvent. A polymer formed instead. However, with MeOH as solvent a good
yield of this cryptand was obtained [98]. In 1993 Smith et al. [99] reported the
direct [3 ⫹ 2] synthesis in 50% yield of a hexaimino macrobicyclic Schiff base
from condensation of tris(2-aminoethyl)amine (tren) with an aldehyde in the ab-
sence of a template ion. This macrobicycle is similar to the precursor macrocycle
of VBN-2 (Scheme 16) but is much more flexible. They caution that the ‘‘essen-
tial element of this reaction appears to be the slow addition of glyoxal to tren at
low temperatures (0°C). Higher temperatures and faster addition rates lead to
variable yields, ranging from 10–30%.’’ All these reports demonstrate the interest
in template-free syntheses, and the importance of the appropriate reaction condi-
tions. We have synthesized over 19 macrocyclic Schiff bases by [2 ⫹ 2] and 3
macrobicyclic ones by [3 ⫹ 2] condensations, template-free, and in high yields.
Trans geometry of the Schiff bases was demonstrated by NMR spectros-
copy. The same observation has been reported for similar macrocyclic imines
based on x-ray crystallographic and NMR analyses [100,101]. The large ring size
(26 and 28 atoms) can easily accommodate trans-double bonds. This versatile
methodology was used to afford macrocycles containing different functionalities
TM

in their framework; thus macrocyclic imines were obtained by condensation of
various dialdehydes with diamines and macrocyclic amides by condensation of
the diamines with diacid chlorides. In addition, a wide array of ‘‘spacer-arms’’
(the R groups) can be introduced prior to the condensation step, or they can be
subsequently attached to the macrocycle. A further built-in feature of this syn-
thetic approach is that it not only allows introduction of functionalized spacer
arms but also can be used to obtain other macrocyclic skeletons; e.g., the imino
macrocycles can be converted to macrocycles containing amine, oxaziridine, or
nitrone groups. Moreover, the headgroups (shaded boxes in Scheme 17) can be
varied by changing the carbonyl component.
The imino macrocycle VBN-5 with the pyridine head group was toxic to
both CEM-SS and Jurkat cell lines. In contrast, macrocycles VBN-6-8 containing
furan and 1,3- and 1,4-disubstituted phenyl head groups and VBN-9 were not.
The furan headgroup in VBN-6 demonstrated a cell viability of 100%.
Cyanoborohydride reduction of the metal-free macrocyclic imines afforded
a series of aminomacrocycles in high yields. The ease of the two-step synthesis,
yields of ⬎50% in the [2 ⫹ 2] cyclization and almost quantitative yields in the
reduction step, as well as the facile derivatization of the spacer-arms, make these
macrocycles attractive target compounds. Nineteen aminomacrocycles, obtained
by cyanoborohydride reduction of the macrocyclic imines, were screened for
antifungal activity [102]. Among these, compound 4 showed outstanding
activity against Cryptococcus neoformans at a minimum inhibitory concentration
(MIC ⱕ 0.5 µM); compound 11 was fungistatic against Candida albicans at
a MIC ⱕ 0.5 µM and also showed modest fungistatic and fungicidal activity
against Cryptococcus neoformans strains 1 and 2 (Tables 2 and 3).
The metallomacrocycle, VBN-10 (Scheme 18), was obtained by applying
the [2 ⫹ 2] condensation procedure developed by us [51]. Thus, in a typical
procedure for the preparation of macrocyclic amides (e.g., VBN-10, R ⫽ C 10 H 21),
2,2′-methylenebis(4-nitrophenyl decyl ether) dissolved in aqueous EtOH-THF
(ethanol tetrahydrofuran), was reduced with Zn/NH 4 Cl. To the dried, crude re-
duction product dissolved in absolute EtOH was added the diacid chloride, and
Scheme 18 Spacer-armed metallomacrocyclic amides.
TM

the mixture was stirred at rt for 16 h under inert atmosphere. The precipitate of
the macrocycle was washed with ethanol and was dried to obtain the macrocyclic
amides in yields of 50%–55% (unoptimized). The nickel complex, VBN-10, was
obtained by complexation of the deprotonated macrocycle in THF with NiCl2 ⋅
6H 2 O, 12 h, rt. Complexation with nickel resulted in a new red-shifted absorption
band and a new peak in the FAB-MS (fast atom bombardment-mass spectrum).
The metallomacrocycle VBN-10 was not toxic to CEM-SS and Jurkat cells and
showed potent in vitro activity against P. carinii, causing inhibition of 89.9%
and 97.4% (24/48 h) at a concentration of 0.7 µM (Table 1, Sec. VI). Its in vivo
efficacy is comparable to that of trimethoprim/sulfamethoxazole (TMP/SMX)
(Table 4).
VI. IN VITRO AND IN VIVO STUDIES WITH DRUGS

CONTAINING NOVEL PHARMACOPHORES DIRECTED
AGAINST OPPORTUNISTIC INFECTIONS
Assessment of the efficacy profile of therapeutic agents presently in clinical use

against OIs indicates an acute need to develop better agents. This can be accom-
plished either by improving the efficiency of the known therapeutic agents or by
obtaining novel types of lead compounds and their analogues by syntheses. We
developed such lead compounds on the basis of the oxidoredox hypothesis out-
lined in the foregoing. The biological test results against the fungal pathogens
P. carinii, C. albicans, and C. neoformans are summarized in the following:
A. Activities Against Pneumocystis carinii

In vitro assays against P. carinii were carried out by the NIAID contractor’s
laboratory, following the protocol described by Chen and Cushion [103]. A total
of 14 compounds were tested against P. carinii isolates from Brown Norway rat
lungs with severe infection. The activity of four of these compounds, VBN-1,
VBN-2, VBN-3, and VBN-10, 2.1 µM, ⱕ1.5 µM, ⱕ2.8 µM, and 0.7 µM, respec-
tively, was comparable to that of pentamidine, 1.7 µM, the positive control (Table
1). For VBN-2 and VBN-3, no end points were determined (concentrations less
than 1.7 and 2.8 µM were not tested); thus the activity of VBN-2 and VBN-3
may well exceed that of pentamidine in vitro. Compounds VBN-4–9 and VBN-
11–14 showed no appreciable activity against P. carinii.
None of the active compounds was toxic to human (Jurkat and CEM-SS)
cell lines tested [77]. There are many compounds active against P. carinii, in
vitro. These activities, however, do not necessarily predict in vivo efficacy. The
rationale for our high expectations of in vivo activity in the case of compounds
VBN-1, -2, -3, and -10 is that despite their drastically different structures con-
TM

Table 1 Evaluation of VBN-1–VBN-14 Compounds Against Pneumocystis carinii a
1 µg/mL 10 µg/mL 100 µg/mL
Compound Untreated (% Decrease) (% Decrease) (% Decrease)
Pentamidine 61.6 0.6 (99.5) 1.7 µM b
VBN-1 14.9 1.3 (91.3) 2.1 µM b 0.2 (98.7) 0.2 (98.7)
VBN-2 12.2 ⋅⋅⋅⋅⋅ ⋅⋅⋅⋅⋅ ⋅⋅⋅⋅⋅
17.5 3.1 (82.3)/24 h 1.5 µM b 2.9 (83.3)/24 h 0.5 (97.3)/24 h
VBN-3 17.5 ⋅⋅⋅⋅⋅ ⋅⋅⋅⋅⋅ ⋅⋅⋅⋅⋅
17.5 3.6 (79.4)/24 h 2.8 µM b 0.7 (96.0)/24 h 1.5 (95.1)/24 h
VBN-10 38.5 1.0 (97.4) 0.7 µM b 1.4 (96.4) 0.5 (96.4)
VBN-4 78.7 67.2 (14.6) 55.6 (29.4) 0.9 (99.9)
VBN-5 78.7 56.7 (28.0) 61.8 (21.5) 0.7 (99.1)
VBN-6 78.7 77.0 (2.2) 58.9 (25.2) 11.1 (85.9)
VBN-7 63.7 46.3 (27.2) 42.4 (33.3) 0.5 (99.2)
VBN-8 63.7 48.1 (24.5) 31.4 (50.7) 0.4 (99.4)
VBN-9 33.3 (0) 33.3 (0) 33.3 (0) 33.3 (0)
VBN-11 30.4 25.1 (17.4) 28.8 (15.3) 16.2 (46.7)
VBN-12 30.4 30.0 (0) 28.3 (6.9) 0.9 (97.0)
VBN-13 30.4 31.3 (0) 32.0 (0) 18.3 (39.9)
VBN-14 30.4 30.4 (0) 28.6 (5.9) 5.6 (81.6)
a
ATP ⫻ 10 ⫺9 /10 8 M/10 8 nuclei (ATP decrease %). All results are after 48 h, 24 h results are also
given for VBN-2 and VBN-3, where ATP % was not detectable after 48 h. ⋅ ⋅ ⋅ ⋅ ⋅, ATP % below
the level of detection at 48 h. ATP, adenosine triphosphate.
b
For the first four (active) compounds, micromolar (µM) values given are equivalent to the concentra-
tions at 1 µg/mL.
taining unlike functionalities that render them chemically very different, each of
these compounds is able to cause oxidative damage to P. carinii; i.e., the only
common feature of these compounds, which are structurally so widely dissimilar,
e.g., VBN-10, molecular weight of 1412, a metallomacrocycle, and VBN-3, a
relatively small, molecular weight (MW) ⬍ 400 aromatic entity, is that all are
oxidizing agents. This gives credence to our hypothesis, the basis of our design
of these compounds, namely, that P. carinii is susceptible to oxidative damage.
The in vivo studies discussed in Section VI.C confirmed this.
B. Activities Against Cryptococcus neoformans and

Candida albicans
During the course of the synthesis of oxidoredox pharmacophores, we encoun-
tered several compounds that, although inactive against P. carinii, were active
against C. neoformans and C. albicans. Among the 14 compounds tested initially
TM

Table 2 In Vitro Assays Against Cryptococcus neoformans Strains 1 and 2
MIC 48/72 h (µg/mL) MLC 48/72 h (µg/mL)
Analogue Formula 1 2 1 2
1 C 28 H 38N6O4 5/⬎5 5/5 ⬎5/⫺ ⬎5/⫺
2 C 32 H 46N6O4 6.25/⬎6.25 6.25/6.25 6.25/⬎6.25 6.25/⫺
3 C 28 H 46N6O4 5/⬎5 5/5 ⬎5/⫺ ⬎5/⫺
4 C 32 H 54N6O4 ⱕ0.3/ⱕ0.3 ⱕ0.3/0.6 ⱕ0.3/0.6 0.6/2.5
5 C 34 H 50N6O4 ⬎5/⫺ 5/5 ⫺/⫺ ⬎5/⫺
6 C 28 H 38N6 ⬎5/⫺ ⬎5/⫺ ⫺/⫺ ⫺/⫺
7 C 40 H 46N6 5/⬎5 2.5/2.5 5/⬎5 ⬎5/⫺
8 C 30 H 42N6 ⬎5/⫺ 5/⬎5 ⫺/⫺ ⫺/⫺
9 C 44 H 54N6 5/5 2.5/2.5 ⫺/⫺ 5/5
10 C 32 H44N4O6 ⬎5/⫺ 5/⬎5 ⬎5/⫺ ⬎5/⫺
11 C 28 H 46N6 0.6/1.25 1.25/1.25 1.25/2.5 1.25/1.25
12 C 46 H 50N6 5/⬎5 1.25/2.5 ⬎5/⫺ ⬎5/⫺
13 C 40 H 38N6 5/⬎5 ⱕ0.3/2.5 ⬎5/⫺ ⬎5/⫺
14 C 44 H 46N6 ⬎5/⫺ 5/5 ⫺/⫺ 5/⬎5
15 C 44 H 44N4O 2 ⬎5/⫺ 2.5/2.5 ⫺/⫺ 2.5/⬎5
16 C 44 H 48N8 ⬎5/⫺ 1.25/2.5 ⫺/⫺ 5/⬎5
17 C 26 H 32N4O6 ⬎5/⫺ ⬎5/⫺ ⫺/⫺ ⫺/⫺
18 C 16 H 26N4O 2 ⬎5/⫺ ⬎5/⫺ ⫺/⫺ ⫺/⫺
19 C 18 H 30N4O4 ⬎5/⫺ ⬎5/⫺ ⫺/⫺ ⫺/⫺
(data not shown), one of the macrocyclic compounds showed encouraging activ-
ity in in vitro tests. Since this active compound could be synthesized in two easy
steps, and in high yields, using the methodology we developed, we prepared 19
analogues and tested them against these fungal species. The results of the assays
of these analogues, compounds 1–19, are shown in Tables 2 and 3.
Compound 4 (Table 2) showed activity against C. neoformans strains 1
and 2 at a minimum inhibitory concentration (MIC) ⱕ0.5 µM (ⱕ0.3 µg/ml).
This level of activity compares favorably with that of fluconazole, the positive
control, especially, since an end point was not established; i.e., the lowest concen-
tration tested was 0.5 µM. Compound 11 (Table 3) was found to be fungistatic
against C. albicans strains A–E at a MIC ⱕ 0.5 µM (ⱕ0.3 µg/mL) and also
showed both fungistatic and fungicidal activity, though more modest, against C.
neoformans strains 1 and 2 at a MIC of 1.3–2.6 µM (0.6–1.25 µg/mL) (Table 2).
Because of the high levels of in vitro activity, the syntheses of both compounds 4
and 11 were scaled-up to gram amounts for in vivo assays on mice. Initial toxicity
studies are encouraging, since all mice survived for at least 1 month when daily
doses of 15 mg/kg of compound 11 and 10 mg/kg of compound 4 were given
TM

Table 3 In Vitro Assays Against Candida albicans Strains A–E
MIC 24/48 (µg/mL) MLC 24/48 (µg/mL)
No. Formula A B C D E A B C D E
1 C 28 H 38N6O4 ⬎5/⫺ ⬎5/⫺ ⬎5/⫺ ⬎5/⫺ ⬎5/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺
2 C 32 H 46N6O4 6.25/⬎6.25 ⬎6.5/⫺ 6.25/⬎6.25 6.25/⬎6.25 6.25/⬎6.25 ⬎6.5/⫺ ⫺/⫺ ⬎6.5/⫺ ⬎6.5/⫺ ⬎6.5/⫺
3 C 28 H 46N6O4 ⬎5/⫺ ⬎5/⫺ ⬎5/⫺ ⬎5/⫺ ⬎5/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺
4 C 32 H 54N6O4 ⬎5/⫺ ⬎5/⫺ 5/⬎5 ⬎5/⫺ ⬎5/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺
5 C 34 H 50N6O4 ⬎5/⫺ ⬎5/⫺ ⬎5/⫺ ⬎5/⫺ ⬎5/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺
6 C 28 H 38N6 ⬎5/⫺ ⬎5/⫺ 5/⬎5 ⬎5/⫺ ⬎5/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺
7 C 40 H 46N6 ⬎5/⫺ ⬎5/⫺ 5/⬎5 ⬎5/⫺ ⬎5/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺
8 C 30 H 42N6 ⬎5/⫺ ⬎5/⫺ 5/⬎5 ⬎5/⫺ ⬎5/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺
9 C 44 H 54N6 ⬎6.5/⫺ ⬎6.5/⫺ ⬎6.5/⬎6.5 ⬎6.5/⫺ ⬎6.5/⫺ ⫺/⫺ ⫺/⫺ ⬎6.5/⫺ ⫺/⫺ ⫺/⫺
10 C 32 H 44N6O6 ⬎5/⫺ ⬎5/⫺ ⬎5/⫺ ⬎5/⫺ ⬎5/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺
11 C 28 H 46N6 ⱕ0.3/5 ⱕ0.3/5 ⱕ0.3/5 ⱕ0.3/5 ⱕ0.3/5 ⬎5/⫺ 5/5 5/⬎5 5/⬎5 ⬎5/⫺
12 C 40 H 50N6 ⬎5/⫺ ⬎5/⫺ 5/⬎5 ⬎5/⫺ ⬎5/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺
13 C 40 H 38N6 5/⬎5 ⬎5/⫺ 5/⬎5 ⬎5/⫺ ⬎5/⫺ ⬎5/⫺ ⫺/⫺ ⬎5/⫺ ⫺/⫺ ⫺/⫺
14 C 44 H 46N6 5/⬎5 ⬎5/⫺ 5/⬎5 ⬎5/⫺ 5/⬎5 ⫺/⫺ ⫺/⫺ ⬎5/⫺ ⫺/⫺ ⬎5/⫺
15 C 44 H 44N4O 2 ⬎5/⫺ ⬎5/⫺ 5/⬎5 ⬎5/⫺ ⬎5/⫺ ⫺/⫺ ⫺/⫺ ⬎5/⫺ ⫺/⫺ ⫺/⫺
16 C 44 H 48N8 5/⬎5 5/⬎5 5/⬎5 5/⬎5 ⱕ0.3/⬎5 ⬎5/⫺ ⬎5/⫺ ⬎5/⫺ ⬎5/⫺ ⬎5/⫺
17 C 26 H 32N4O6 ⬎5/⫺ ⬎5/⫺ 5/⬎5 ⬎5/⫺ ⬎5/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺
18 C 16 H 26N4O 2 ⬎5/⫺ ⬎5/⫺ 5/⬎5 ⬎5/⫺ ⬎5/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺
19 C 18 H 30N4O4 ⬎5/⫺ ⬎5/⫺ 5/⬎5 ⬎5/⫺ ⬎5/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺ ⫺/⫺
TM

TM
C
intraperitoneally. Should these structural leads we have identified prove success-
ful in the ongoing in vivo studies, subsequent drug development steps starting
with the establishment of pharmacokinetical parameters; e.g., clearance, distribu-
tion, bioavailability, will be initiated.
C. In Vivo Studies of VBN-1, VBN-3, and VBN-10 Against

Pneumocystis carinii
In vivo studies were carried out at the NIAID contractor’s laboratory according
to well-established protocols. Thus, male C3HeB/FeJ mice, 29 g average weight,
were given dexamethasone and a cephradine antibiotic to initiate their latent P.
carinii infections, then were submitted to drug therapy at 7 weeks into the immu-
nosuppression therapy. Stock solutions of VBN-1, VBN-2, and VBN-10 were
made in dimethyl sulfoxide (DMSO) and were diluted in sterile water. The mice
were given daily doses intraperitoneally. The results obtained to date are summa-
rized in Table 4. The efficacy of VBN-10 is comparable to that of TMP-SMX,
and it had no gross toxicity up to 50 mg/kg. The table shows efficacy for VBN-
10 at 50 → 20 mg/kg/day but no activity for VBN-1 and VBN-3 at this dosage.
Considering that VBN-1 and VBN-3 even at high concentrations are not toxic
to human cell lines, higher dosage levels of VBN-1 and VBN-3 were tested.
These in vivo studies now in progress indicate activity at a dosage of 150 mg/
kg/day.
VII. CONCLUSIONS
The existing drug regimens to treat opportunistic infections, devastating to the

immunocompromised, are of only limited use because of their high toxic effects
and the pathogens’ ability to develop resistance to them readily. The oxidoredox
hypothesis advanced here serves as foundation for a novel therapeutic approach
to develop clinically useful alternatives to prevent and cure OIs. Four novel,
oxidoredox pharmacophores were selected and incorporated into molecules hav-
ing disparate structures, to yield four types of lead compounds. The synthesis of
the lead compounds entailed development of methodologies that use very few
steps; the key steps are template-free [2 ⫹ 2] and [3 ⫹ 2] macrocyclizations.
Template-free [3 ⫹ 2] cyclization and subsequent oxidation afforded a novel
class of macrocycles having oxaziridine moieties. The in vitro and in vivo anti-
fungal activities of each of the four types of leads indicate the high therapeutic
potential of the novel pharmacophores. Thus, the feasibility of generating thera-
peutic agents based on the oxidoredox hypothesis is demonstrated.
TM

Table 4 In Vivo Studies of VBN-1, VBN-3, and VBN-10 Against Pneumocystis carinii in Mice
Cyst count
Treatment Dose regimen a Cyst count (median) P value (geometric mean) P value
Control steroid — 2.4 ⫻ 10 6 — 6.13 ⫾ 0.8c —
(2.23 ⫻ 104–1.83 ⫻ 10 7)b (1.35 ⫻ 10 6)d
VBN-1 0.1 mg/kg/d IP 2.62 ⫻ 10 6 N/S 6.42 ⫾ 0.50 N/S
(3.57 ⫻ 10 5 –1.23 ⫻ 10 7) (2.63 ⫻ 10 5)
VBN-1 10 mg/kg/d IP 1.701 ⫻ 10 6 N/S 6.21 ⫾ 0.46 N/S
(1.34 ⫻ 10 5 –5.90 ⫻ 10 6) (1.62 ⫻ 10 6)
VBN-1 50 mg/kg/d IP 4.15 ⫻ 10 6 N/S 6.43 ⫾ 0.49 N/S
(1.56 ⫻ 10 5 –1.09 ⫻ 10 7) (2.69 ⫻ 10 6)
VBN-3 0.1 mg/kg/d IP 7.37 ⫻ 10 5 N/S 5.89 ⫾ 0.60 N/S
(6.70 ⫻ 10 4 –1.54 ⫻ 10 7) (7.76 ⫻ 10 5)
VBN-3 10 mg/kg/d IP 5.18 ⫻ 10 5 N/S 6.63 ⫾ 0.34 N/S
(6.92 ⫻ 10 5 –1.30 ⫻ 10 7) (4.27 ⫻ 10 6)
VBN-3 50 mg/kg/d IP 3.27 ⫻ 10 6 N/S 6.50 ⫾ 0.32 N/S
(7.82 ⫻ 10 5 –8.15 ⫻ 10 6) (3.16 ⫻ 10 6)
VBN-10 0.1 mg/kg/d IP 3.90 ⫻ 10 6 N/S 6.29 ⫾ 0.77 N/S
(4.47 ⫻ 10 4 –1.23 ⫻ 10 7) (1.95 ⫻ 10 6)
VBN-10 10 mg/kg/d IP 1.65 ⫻ 10 6 N/S 5.94 ⫾ 0.84 N/S
(2.23 ⫻ 10 4 –5.85 ⫻ 10 6) (8.71 ⫻ 10 5)
VBN-10 50 → 20 mg/kg/d IP 2.23 ⫻ 10 4 ⬍0.001 4.52 ⫾ 0.37 ⬍0.001
(2.23 ⫻ 10 4 –3.57 ⫻ 10 5) (3.31 ⫻ 10 4)
Pentamidine 10 mg/kg/3 ⫻ wk IM 2.23 ⫻ 10 4 ⬍0.05 4.86 ⫾ 0.67 ⬍0.001
(2.23 ⫻ 10 4 –7.15 ⫻ 10 5) (7.24 ⫻ 10 4)
Atovaquone 100 mg/kg/d PO 1.34 ⫻ 10 5 N/S 5.23 ⫾ 0.97 ⬍0.05
(2.23 ⫻ 10 4 –2.55 ⫻ 10 6) (1.70 ⫻ 10 5)
Primaquine/clindamycin 2/225 mg/kg/d PO 3.13 ⫻ 10 6 N/S 6.11 ⫾ 0.82 N/S
(2.23 ⫻ 10 4 –5.00 ⫻ 10 6) (1.29 ⫻ 10 6)
TMP/SMXd 50/250 mg/kg/d PO 2.23 ⫻ 10 4 ⬍0.01 4.35 ⫾ 0.00 ⬍0.001
(2.23 ⫻ 10 4 –2.23 ⫻ 10 4) (2.23 ⫻ 10 4)
Control Normal — 2.23 ⫻ 104 — 4.35 ⫾ 0.00 —
(2.23 ⫻ 10 4 –2.23 ⫻ 10 4) (2.23 ⫻ 10 4)
a
IP, intraperioteneal; IM, intramuscular PO, oral; N/S, not significant; TMP/SMX, trimethoprim/sulfamethoxazole, the positive control. b Range. c Mean
⫾ 1 SD. d Antilog.
TM

TM
C
ACKNOWLEDGMENTS
The author thanks Drs. Nancy A. Roth and Ronald J. Roth for more than the
critical reading of the manuscript. She would also like to express her appreciation
for the contributions of her coworkers. Financial support from NIH, National
Institute of Allergy and Infectious Diseases grant ROI AI39418, is gratefully
acknowledged.
REFERENCES
1. N. M. Ampel, Emerging Infect. Dis., 2: 109 (1996).

2. J. A. Kovacs, J. W. Heimentz, A. M. Macher, D. Stover, H. W. Murray, J. Shel-
hamer, H. C. Lane, C. Urmarcher, C. Honig, D. Longo, M. M. Parker, J. E. Natan-
son, J. E. Parillo, A. S. Fauci, P. A. Pizzo, and H. Masur, Ann. Intern. Med., 100:
663 (1984).
3. R. A. Larsen, M.A.E. Leal, and L. S. Chan, Ann. Intern. Med., 113: 183 (1990).
4. W. G. Powderly, K. Robinson, and E. J. Keath, J. Infect. Dis., 168: 463 (1993).
5. K. Richardson, K. Cooper, M. S. Marriott, M. H. Tarbit, P. F. Troke, and P. J.
Whittle, Ann. NY Acad. Sci., 544: 4 (1988).
6. J. Brajtburg, W. G. Powderly, G. S. Kobayashi, and G. Medoff, Antimicrob. Agents
Chemother., 34: 183 (1990).
7. J. C. Edman, J. A. Kovacs, H. Masur, D. V. Santi, H. J. Elwood, and M. L. Sogin,
Nature, 334: 519 (1988).
8. Physicians’ Desk Reference, Medical Economics Data Production Company, Mont
Vale, N.J., p. 1918; Supplement B, p. 47 (1994).
9. R. Berkow, A. J. Fletcher, eds., The Merck Manual of Diagnosis and Therapy, 16th
ed., Merck Research Laboratories, Merck & Co., Inc., Rahway, N.J. (1992).
10. L. J. Ignarro, G. M. Buga, K. S. Wood, R. E. Byrns, and G. Chaudhuri, Proc. Natl.
Acad. Sci. USA, 84: 9265 (1987).
11. R. M. J. Palmer, A. G. Ferrige, and S. Moncada, Nature, 327: 524 (1987).
12. R. F. Furchgott and J. V. Zawadzki, Nature, 288: 373 (1980).
13. E. Culotta and D. E. Koshland Jr., Science, 258: 1862 (1992).
14. J. M. Fukuto and L. J. Ignarro, Acc. Chem. Res., 30: 149 (1997).
15. S. Moncada, R. M. J. Palmer, and E. A. Higgs, Pharmacol. Rev., 43: 109 (1991).
16. D. K. Ghosh, H. M. Abu-Saud, and D. J. Stuehr, Biochemistry, 34: 11316 (1995).
17. J. S. Beckman, T. Beckman, J. Chen, P. A. Marshall, and B. A. Freeman, Proc.
Natl. Acad. Sci. USA, 87: 1620 (1990).
18. L. Castro, M. Rodriguez, and R. Radi, J. Biol. Chem., 269: 29409 (1994).
19. H. Ischiropoulos, L. Zhu, and J. S. Beckman, Arch. Biochem. Biophys., 298: 446
(1992).
20. R. E. Huie and S. Padmaja, Free Radic. Res. Commun., 18: 195 (1993).
21. S. V. Limar and J. K. Hurst, J. Am. Chem. Soc., 117: 8867 (1995).
22. T. Malinski and Z. Taha, Nature, 358: 676 (1992).
23. S. H. White, W. C. Wimley, and M. E. Selsted, Curr. Opin. Struct. Biol., 5: 521
(1995).
TM

24. M. A. Marletta, P. S. Yoon, R. Iyengar, C. D. Leaf, and J. S. Wisnook, Biochemis-
try, 27: 8706 (1988).
25. H. H. H. W. Schmidt, R. Seifert, and E. Bohme, FEBS Lett., 244: 357 (1989).
26. C. D. Wright, A. Mulsch, R. Busse, and H. Osswald, Biochem. Biophys. Res. Com-
mun., 160: 813 (1989).
27. T. B. McCall, N. K. Boughton-Smith, R. M. J. Palmer, B. J. R. Whittle, and S.
Moncada, Biochem. J., 261: 293 (1989).
28. S. E. Malawista, R. R. Montgomery, and G. Van Blaricom, J. Clin. Invest., 90:
631 (1992).
29. M. Markert, B. Carnal, and J. Mauel, Biochem. Biophys. Res. Commun., 199: 1245
(1994).
30. S. J. Klebanoff and C. F. Nathan, Biochem. Biophys. Res. Commun., 192: 192
(1993).
31. S. J. Weiss, N. Engl. J. Med., 320: 365 (1989).
32. M. B. Grisham, M. M. Jefferson, D. F. Melton, and E. L. Thomas, J. Biol. Chem.,
259: 10404 (1984).
33. K. Fukuda, Y. Hirai, H. Yoshida, T. Nakajima, and T. Usui, Clin. Chem., 29: 1758
(1982).
34. J. Michaelis, M. C. Vissers, and C. C. Winterbourn, Arch. Biochem. Biophys., 292:
555 (1992).
35. K. J. Kelley, Nature Biotechnol., 14: 587 (1996).
36. J. V. Bannister and W. H. Bannister, eds. The Biology and Chemistry of Active
Oxygen, Elsevier Science, New York (1984).
37. R. Bolli, Cardiovasc. Drugs Ther., 5: 249 (1991).
38. S. C. Black, C. S. Schasteen, R. H. Weiss, D. P. Riley, E. M. Driscoll, and B. R.
Lucchesi, J. Pharmacol. Exp. Ther., 270: 1208 (1994).
39. N. Bahr, R. Guller, J.-M. Reymond, and R. A. Lerner, J. Am. Chem. Soc., 118:
3550 (1996).
40. E. P. Garvey, J. A. Oplinger, E. S. Furfine, R. J. Kiff, F. Laszlo, B. J. R. Whittle,
and R. G. Knowles, J. Biol. Chem., 272: 4959 (1997).
41. F. A. Davis and A. C. Sheppard, Tetrahedron, 45: 5703 (1989).
42. F. A. Davis, S. G. Lal, and H. D. Durst, J. Org. Chem., 53: 5004 (1988).
43. T. C. Bruice, in Mechanistic Principles of Enzyme Activity (J. F. Liebman and A.
Greenberg, eds.) VCH Publishers, New York, pp. 315–352 (1988).
44. M. Morrison and G. R. Schonbaum, Annu. Rev. Biochem., 45: 861 (1976).
45. H. B. Dunford and J. S. Stillman, Adv. Inorg. Chem., 119 (1979).
46. T. C. Bruice, Acc. Chem. Res., 24: 243 (1991).
47. L.-C. Yuan and T. C. Bruice, J. Am. Chem. Soc., 107: 512 (1985).
48. F. A. Davis, J. Lamendola, Jr., U. Nadir, E. W. Kluger, T. C. Sedergran, T. W.
Panunto, R. Billmers, R. Jenkins, Jr., I. J. Turchi, W. H. Watson, J. S. Chen, and
M. Kimura, J. Am. Chem. Soc., 102: 2000 (1980).
49. C.-H. Lee, J. D. Korp, and H. Kohn, J. Org. Chem., 54: 3077 (1989).
50. H. W. Orf and D. Dolphin, Proc. Natl. Acad. Sci. USA, 71: 2646 (1974).
51. C. E. Brathwaite, C.-X. Chen, and V. Balogh-Nair, Indian J. Chem., 31B: 810
(1992).
52. W. H. Rastetter, T. R. Gadek, J. P. Tane, and J. W. Frost, J. Am. Chem. Soc., 101:
2228 (1979).
TM

53. F. A. Davis, J. M. Billmers, D. J. Grosciniak, J. C. Towson, and R. D. Bach, J.
Org. Chem., 51: 4240 (1986).
54. F. A. Davis, N. F. Abdul-Malik, S. B. Awad, M. E. Harakal, Tetrahedron Lett.,
22: 917 (1981).
55. R. D. Bach and G. Wolber, J. Am. Chem. Soc., 106: 1410 (1984).
56. B. A. O’Brian, W. Y. Lam, and D. D. DesMarteau, J. Org. Chem., 51: 4466
(1986).
57. D. D. DesMarteau, A. Donadelli, V. Montanari, V. A. Petrov, and G. Resnati, J.
Am. Chem. Soc., 115: 4897 (1993).
58. M. N. Akhtar, D. R. Boyd, J. D. Neill, and D. Jerina, J. Chem. Soc. Perkin I, 1693
(1980).
59. N. Kocherginsky and H. M. Schwartz, Nitroxide Spin Labels: Reactions in Biology
and Chemistry, CRC Press, Boca Raton, Fla. (1995).
60. S. Dikalov, I. Kirilyuk, and I. Grigor’ev, Biochem. Biophys. Res. Commun., 218:
616 (1996).
61. P. Kuppusamy, P. Wang, and J. L. Zweier, Biochemistry, 35: 7051 (1996).
62. I. Priyadarsini, M. F. Dennis, M. A. Naylor, M. R. L. Stratford, and P. Wardman,
J. Am. Chem. Soc., 118, 5648 (1996).
63. S. Dikalov, M. Skatchkov, and E. Bassenge, Biochem. Biophys. Res. Commun.,
231: 701 (1997).
64. J. S. Daniels and K. S. Gates, J. Am. Chem. Soc., 118: 3380 (1996).
65. J. S. Nadeau and D. G. B. Boocock, Anal. Chem., 49: 1672 (1977).
66. K. N. Houk, K. R. Condroski, and W. A. Pryor, J. Am. Chem. Soc., 118: 13002
(1996).
67. D. C. Williams, in Nitrosation (D. C. Williams, ed.) Cambridge University Press,
New York, pp. 1–214 (1988).
68. R. Borzirelli, S. M. Weinreb, and M. Parvez, J. Am. Chem. Soc., 117: 10905 (1995).
69. R.-Q. Wei, S. S. Lefkowitz, J. Mone, and J. Everse, Proc. Soc. Exp. Biol. Med.,
182: 515 (1986).
70. R. F. Eich, T. Li, D. L. Lemon, D. H. Doherty, S. R. Curry, J. F. Aitken, A. J.
Mathews, K. A. Johnson, R. D. Smith, G. N. Phillips Jr., and J. S. Olson, Biochemis-
try, 35: 6976 (1996).
71. J. Am. Chem. Soc., 118, 5702 (1996).
72. J. R. Stone and M. A. Marletta, Biochemistry, 35: 1093 (1996).
73. L. J. Ignarro, K. S. Wood, and M. S. Wolin, Proc. Natl. Acad. Sci. USA, 79: 2870
(1982).
74. L. J. Ignarro, K. S. Wood, and M. S. Wolin, Adv. Cyclic Nucleotide Phosphoryla-
tion Res., 17: 267 (1984).
75. L. Jia, C. Bonaventura, J. Bonaventura, J. S. Stamler, Nature, 380: 221 (1996).
76. F. A. Davis, J. P. McCauley, Jr., S. Chattopadhyay, M. A. Harakal, W. H. Watson,
and I. Tavanaiepour, in Perspectives in the Organic Chemistry of Sulfur (B. Zwa-
nenburg and A. J. H. Klunder, eds), Elsevier Science, Amsterdam, p. 153 (1988).
77. V. Balogh-Nair, C. E. Brathwaite, C.-X. Chen, and J. Vargas Jr., Cell. Mol. Biol.,
41: S9 (1995).
78. V. McKee, M. R. J. Dorrity, J. F. Malone, D. Marss, and J. Nelson, Chem. Com-
mun., 383 (1992).
TM

79. F. A. Davis and R. H. Jenkins, Jr., in Asymmetric Synthesis (J. D. Morrison and
J. W. Scott, eds.), Academic Press, Orlando, Fla. p. 313 (1984).
80. D. MacDowell and J. Nelson, Tetrahedron Lett., 29: 385 (1988).
81. D. Chen and A. E. Martell, Tetrahedron, 47: 6895 (1991).
82. F. A. Davis and O. D. Stringer, J. Org. Chem., 47: 1774 (1982).
83. J. Rebek, Jr., in Molecular Recognition: Chemical and Biochemical Problems
(S. M. Roberts, ed.) The Royal Society of Chemistry, pp. 211 (1989).
84. M. M. Durkin, M. M. Shaw, M. S. Bartlett, and J. W. Smith, J. Protozol., 38: 210S
(1991).
85. V. Balogh-Nair and K. Nakanishi, Pharm. Res., 93 (1984).
86. C. C. Willhite and V. Balogh-Nair, Toxicology, 33: 331 (1984).
87. C. C. Willhite and V. Balogh-Nair, Teratog. Carcinog. Mutagen., 5: 355 (1985).
88. R. D. Hinton and E. G. Janzen J. Org. Chem., 57: 2646 (1992).
89. M. Shimomura, K. Abe and M. Hirota J. Chem. Soc. Perkin Trans. II, 795 (1988).
90. E. F. Ullman, J. H. Osiecki, D. G. B. Boocock, and R. Darcy, J. Am. Chem. Soc.,
94: 7049 (1972).
91. E. F. Ullman, L. Call, and J. H. Osiecki, J. Org. Chem., 35: 3623 (1970).
92. L. B. Volodarsky, Imidazoline Nitroxides, Vol. 1. Synthesis and properties, CRC
Press, Boca Raton, Fla. (1988).
93. C.-X. Chen, Synthesis of Biologically Active Molecules, Ph.D. Thesis, City Uni-
versity of New York, 1995.
94. H.-J. Buschmann, in Stereochemical and Stereophysical Behavior of Macrocycles
(I. Bernal, ed.), Elsevier Science, Amsterdam, p. 103 (1987).
95. L. F. Lindloy, The Chemistry of Macrocyclic Ligand Complexes, Cambridge Uni-
versity Press, Cambridge, p. 174 (1989).
96. J. Jazwinsky, J-M. Lehn, D. Lilienbaum, R. Ziessel, J. Guilhem, and C. Pascard,
J. Chem. Soc. Chem. Commun., 1691 (1987).
97. R. Menif and A. E. Martell, J. Chem. Soc. Chem. Commun., 1521 (1989).
98. D. MacDowell and J. Nelson, Tetrahedron Lett., 29: 385 (1988).
99. P. H. Smith, M. E. Barr, J. R. Brainard, D. K. Ford, H. Freiser, S. Muralidharan,
S. D. Reilly, R. R. Ryan, L. A. Silks III, and W. Yu, J. Org. Chem., 58: 7939
(1993).
100. V. McKee, M. R. J. Dorrity, J. F. Malone, D. Marss, and J. Nelson, J. Chem. Soc.
Chem. Commun., 383 (1992).
101. G. Bombieri, F. Benetollo, A. Pollo, K. K. Fonda, and L. M. Vallarino Polyhedron,
10: 1385 (1991).
102. C.-X. Chen, C. E. Brathwaite, and V. Balogh-Nair, unpublished results.
103. F. Chen and M. T. Cushion, J. Clin. Microbiol., 32: 2791 (1994).
TM

13
A Mechanistic Analysis of CEO
Bond Cleavage Events with a
Comparison to 3,6-Dideoxysugar
Formation
David A. Johnson and Hung-wen Liu

University of Minnesota, Minneapolis, Minnesota
I. INTRODUCTION
The making and breaking of carbon–oxygen bonds are ubiquitous in biological

systems by virtue of the organic nature of life. Most processes from energy metab-
olism to deoxyribonucleic acid (DNA) synthesis involve the formation and/or
cleavage of CEO bonds. Understandably, there is a wide variation in the struc-
tures of the molecules that undergo this transformation. As expected, a diverse
array of mechanisms has evolved to facilitate the CEO bond scission for these
different molecules. However, most of the reaction mechanisms appear to display
a characteristic feature in that the reaction is initiated by the cleavage of a CEH
bond before the target CEO bond is broken. For these enzyme systems, the
manner in which this critical CEH bond is labilized during catalysis is a defining
property of the mechanism. Some of the enzymes can catalyze the reaction using
binding activation without assistance from any cofactor, such as enoyl–coenzyme
A (enoyl-CoA) hydratase. Other enzymes, such as diol dehydrase and ribonucleo-
tide reductase, require a highly reactive radical-generating cofactor to break the
CEH bond and provide the necessary driving force for the reaction. Various
other cofactors are used in different mechanisms, including those that initiate a
CEO bond cleavage before a CEH bond is broken.
One particular enzyme, cytidine diphosphate-6-deoxy-l-threo-d-glycero-
TM

4-hexulose-3-dehydrase (E 1), depends on pyridoxamine 5′-phosphate (PMP), an
inorganic [2Fe-2S] center, and a separate reduced nicotinamide adenine dinucleo-
tide– (NADH) dependent reductase (E 3, formerly known as CDP-6-deoxy-∆ 3,4-
glucoseen reductase) to remove the C-3 hydroxyl of CDP-6-deoxy-l-threo-d-
glycero-4-hexulose irreversibly, leading to the biosynthesis of ascarylose, a
3,6-dideoxy sugar [1]. Abstraction of the pro-S hydrogen at C-4′ of the PMP–
substrate complex initiates this deoxygenation. Interestingly, the labile CEH
bond that promotes the catalysis in this reaction is part of the cofactor skeleton
instead of the substrate. Additionally, the fact that radical intermediates are in-
volved in this transformation makes E 1 the only member in a novel class of
hydrolases.
With emphasis on the cofactor requirements, this review summarizes the
catalytic mechanisms of several characterized enzymes involved with CEO bond
cleavage events, whether the reaction is irreversible or not. These systems are
later used in a comparative analysis to highlight the singular nature of E 1. To
narrow the focus, irreversible CEO bond formations, such as hydroxylations
and oxygenations, and hydrolysis reactions (replacement of one CEO bond with
another) are not discussed here. For the sake of discussion, each enzyme is placed
in a ‘‘mechanistic class,’’ which is defined by the cofactor requirements of the
deoxygenation reaction. Although enzyme systems representative of each general
mechanistic class are presented as examples, it is not the intent of this review to
list exhaustively all enzymes that would fit into each class.
II. GENERAL MECHANISMS OF CEO BOND

TRANSFORMATIONS
A. Reactions Catalyzed by Enzymes Without Cofactors
1. Enoyl–CoA Hydratase
Enoyl–CoA hydratase (crotonase, EC 4.2.1.17) is a member of the hydratase/
isomerase superfamily [2]. As shown in Fig. 1, it catalyzes the reversible hydra-
tion of a ∆ 2,3-unsaturated enoyl-CoA substrate to the corresponding 3-hydroxya-
cyl-CoA product [3]. This reaction is the second step in the β-oxidation pathway
of fatty acid metabolism and is also an important reaction in the catabolism of
branched-chain amino acids. Early studies (Fig. 1a) revealed that bond formation/
cleavage at the α- and β-positions proceeds in a concerted manner with a syn
stereochemistry [4,5]. In an attempt to explain the capability of this enzyme to
abstract the α-H, which has a pK a of 16–20 in free solution, a stepwise process
via an enolic intermediate (Fig. 1b) has also been suggested [6,7]. However, more
recent studies provided further evidence supporting the concerted mechanism by
demonstrating that the 3-hydroxyl leaving group is necessary for cleavage of the
TM

CEO Bond
Figure 1 (a) Concerted model for the β-elimination catalyzed by enoyl-CoA hydratase.
Evidence supports the assignment of a glutamate residue as the active-site amino acid
responsible for both protonation and deprotonation. (b) Stepwise model for the reaction.
In this model, an acidic residue protonates the carbonyl and lowers the pK a of the α-
proton so a single basic residue can both abstract the proton and protonate the leaving
hydroxyl.
CEH bond to occur [8]. In the hydration direction, the driving force of this
catalysis has been attributed to the capability of crotonase to polarize the π-elec-
trons of the α,β-unsaturated double bond, thus enhancing the electrophilicity of
the β-carbon [9]. Consequently, the corresponding active-site acid/base pair plays
a ‘‘push–pull’’ role in the reverse (dehydration) reaction and promotes the α-H
abstraction and the β-OH cleavage.
Sequence alignments with other members of this family implicated a highly
conserved glutamate (Glu) residue (Glu164 in rat liver enzyme) as the catalytic
residue. In fact, substitution of this residue with glutamine (Gln) led to a dramatic
decrease of k cat by more than 100,000-fold [10]. Since the K m remained essentially
unaffected by the mutation, it was concluded that Glu164 assumes the role as
the active-site base to abstract the α-proton. A similar conclusion was reached
in studying the enoyl-CoA hydratase activity associated with the large α-subunit
of the multienzyme complex (79 kDa) of fatty acid oxidation of Escherichia coli
[11–14]. Again, a glutamate residue (Glu139) was identified as the catalytic resi-
due by site-directed mutagenesis [15]. Although additional work is necessary to
elucidate the mechanistic details, sufficient data have been collected to establish
that the hydratase activity in this multifunctional protein proceeds through a gen-
eral acid–general base mechanism with no required cofactors.
TM

2. β-Hydroxydecanoylthioester Dehydrase
A similar reaction can be found in the fatty acid biosynthetic pathway (Fig. 2).
This reversible dehydration in the anaerobic pathway of unsaturated fatty acid
biosynthesis in bacteria is catalyzed by β-hydroxydecanoylthioester dehydrase
(EC 4.2.1.60, encoded by fabA), which is a tetrameric protein and contains no
cofactors [16]. Its catalysis consists of two parts: the dehydration of (R)-3-hy-
droxydecanoyl-ACP (acyl carrier protein) (1) to (E )-2-decenoyl-ACP (2), and
the subsequent isomerization to (Z )-3-decenoyl-ACP (3). Extensive studies have
shown that the dehydration is a syn process [17], whereas the isomerization is
a suprafacial allylic rearrangement [18]. Together, these results strongly imply
an active-site topology that includes a single, acidic/basic residue [19], which is
responsible for deprotonation and protonation at specific sites. Mechanism-based
inactivation and site-directed mutagenesis experiments have identified His70
(numbering of the E. coli enzyme) as an essential active-site base [20–23]. The
participation of a histidine residue (His70) in catalysis was further supported by
the newly released x-ray structure of the E. coli enzyme [24], however, the x-
ray data implicate the additional participation of an aspartate (Asp84′), a cysteine
(Cys80), and a glycine (Gly79) in the reaction mechanism. It is proposed that
Asp84′ acts as a second base and both Cys80 and Gly79 assist the stabilization
of the anionic intermediate via hydrogen bonds. Future experiments will be aimed
at exploring these proposals.
By comparative analysis, the role of His70 in the E. coli enzyme is assumed
by His878 in the multifunctional rat fatty acid synthase, which is a 272-kDa
homodimeric enzyme [25,26]. A mutation in which alanine (Ala) replaced His878
completely abolished the dehydrase activity, although none of the other activities
in the multifunctional mutant enzyme was significantly affected [25]. In con-
trast to the catalytic residue His70 in the E. coli dehydrase, His878 is not involved
in any isomerization, and all 3-hydroxyacyl-S-ACP intermediates are converted
to their corresponding 2-enoyl thioesters, which are subsequently hydrogenated
by the enoyl reductase activity of the enzyme. Therefore, this rat synthase gener-
ates only saturated fatty acids.
Figure 2 Fatty acid biosynthetic pathway involving β-hydroxydecanoylthioester dehy-

drase. Both the dehydration and isomerization steps are depicted. ACP, acyl carrier protein.
TM

CEO Bond
3. 3-Dehydroquinase
Enzymes with no cofactor requirements often recruit amino acid residues other
than general acids or bases to participate in the catalytic mechanism, and 3-dehy-
droquinase (DHQase, EC 4.2.1.10) is a good example. This enzyme catalyzes
the dehydration of 3-dehydroquinic acid (4) to 3-dehydroshikimic acid (5) in
the third step of the shikimate pathway, leading to the biosynthesis of aromatic
metabolites in microorganisms, fungi, and plants [27,28]. There are two distinct
classes of DHQase, designated type I and type II, and they exhibit different struc-
tural properties and mechanisms [29–31]. Chemical modification and differential
peptide mapping studies revealed that the type I enzyme, exemplified by the E.
coli enzyme, catalyzes a syn elimination of a water molecule [32] via an imine
intermediate linked to a conserved lysine residue (Lys170) [33,34], and it uses
a conserved histidine residue (His143) as a general base [35]. Subsequent experi-
ments with site-directed metagenesis of the monofunctional, homodimeric E. coli
enzyme [36] confirmed these conclusions, and the deduced mechanism is illus-
trated in Fig. 3. The covalent imine serves as an electron sink to stabilize the
carbanion intermediate after the pro-R hydrogen is abstracted from the C-2 posi-
tion. As a result of a proposed distortion of the carbocyclic ring of dehydroquinate
upon formation of the Schiff base [32], the critical CEH bond is apparently
reactive enough to be abstracted by His143. Interestingly, this histidine residue
Figure 3 Mechanism for 3-dehydroquinase (DHQase) from the third step of the shiki-
mate pathway in Escherichia coli.
TM

may also play a key role in the formation and breakdown of the imine interme-
diate [36].
Unlike the type I enzymes, for the type II enzymes not much is known
about the mechanism. However, it has been established that the catalysis does
not involve an imine intermediate [29], and the elimination is antistereospecific
[32,37]. Additionally, a recent study has identified a hyperactive arginine that
may be involved in substrate binding [38]. The crystal structure of a type II
DHQase has been determined [31], and when more information is available, a
detailed comparison of the type I and type II enzyme mechanisms should yield
useful insight.
B. Reactions Catalyzed by Divalent Metal Ion–Dependent

Enzymes
1. Carbonic Anhydrase
Although some enzymes such as those discussed previously are able to activate
the necessary CEH and/or CEO bonds, leading to dehydration without cofactor
assistance, many enzymes must rely on various cofactors to accomplish their cataly-
ses. The polarizing effect of a Lewis acid metal on the leaving group of a substrate
is commonly utilized by biological systems to facilitate eliminations, and many
CEO bond cleavage reactions catalyzed by divalent metal-containing enzymes are
known. The reversible hydration of CO 2 catalyzed by carbonic anhydrase (CA,
EC 4.2.1.1) is a good example. This zinc protein is expressed in almost all living
organisms, including bacteria, plants, and animals, and it is a critical participant in
a range of physiological processes [39,40]. Mammalian CAs are the best known,
and there are at least seven different isozymes, designated CA I through VII [40].
The currently available structural and kinetic data support a similar active-site to-
pology and a common mechanism (Fig. 4) for these CA isozymes.
X-ray data indicate that the active site consists of a cleft with a zinc ion
bound at the bottom [41]. In mammalian CA, the zinc ion is tetrahedrally coordi-
nated to three histidine imidazoles and to a water molecule, which exists in the
ionized, hydroxyl form with a pK a around 7 [42]. A threonine (Thr) and a Glu
in the active site participate in a hydrogen bonding network with the Zn-OH ⫺
(6), and this structural feature is conserved in all nonplant CA isozymes. This
network forcibly orients one of the hydroxyl lone electron pairs toward the CO 2
substrate, which is in a hydrophobic pocket [43,44]. The lone pair adds to CO 2
and forms a Zn-coordinated bicarbonate (7), which is rapidly replaced with H 2O
(Fig. 4). Loss of one water proton to the solvent regenerates the Zn-OH ⫺, and
this rate-limiting proton transfer is mediated by a His residue [42,45] and two
water molecules [46], which together form an intramolecular proton shuttle.
Whether the equilibrium favors hydration or dehydration depends significantly
TM

CEO Bond
Figure 4 Schematic representation of the reversible hydration of CO 2 catalyzed by car-

bonic anhydrase. A histidine shuttle (not shown) is involved in the transfer of a proton
between the solvent buffer and the catalytic ZnOH ⫺.
on the specific isozyme and tissue type [40], though both the forward and reverse
reactions proceed through the highly reactive Zn-OH ⫺.
In spinach CA, and perhaps in other plant CAs, extended x-ray absorption
fine structure (EXAFS) studies have shown that the coordination sphere for the
zinc ion includes at least one sulfur ligand [47]. Though the exact ligand structure
is not yet known, plant CAs apparently have a quite different active-site organiza-
tion than mammalian CAs. Regardless of the structural differences, the kinetics
of spinach CA are consistent with the mammalian zinc-hydroxide mechanism,
including a rate-determining proton transfer step that requires the participation
of buffer for maximum efficiency [47]. Despite the lack of conclusive evidence
for a proton-shuttle system in the spinach CA [47], it appears that a similar mech-
anism is shared by all known CAs. Further structural studies of the convergently
evolved mammalian and plant enzymes will be instrumental in developing a
deeper understanding of the catalysis of CA. Sly and Hu [40] summarize the
information on this enzyme more comprehensively.
2. Enolase and Related Enzymes

Enolase (EC 4.2.1.11), an enzyme in the glycolytic and gluconeogenesis path-
ways, catalyzes the reversible dehydration of 2-phospho-d-glycerate (8) (2-PGA,
Fig. 5) [48]. Its catalysis requires two divalent cations, one of which binds at the
high-affinity ‘‘conformational’’ site (site I); the other binds at the lower-affinity
‘‘catalytic’’ site (site II) [49]. Although the natural cofactor Mg 2⫹ provides the
highest activity [49], Mn 2⫹, Zn 2⫹, or other divalent metal ions also fulfill the
requirement [50,51]. At high concentrations, a metal cation apparently binds at
a third site and inhibits the enzyme [52,53]. The overall catalytic cycle includes
a minimum of eight steps [53], though the following discussion pertains only to
the deoxygenation steps. Isotope exchange experiments [54–57] and analogue
TM

Figure 5 Reaction catalyzed by enolase. Though there is disagreement on the identity
of the base, isotopic exchange experiments support the intermediacy of the aci-carboxylate
(see text for details). 2-PGA, 2-phospho-d-glycerate.
studies [57–59] have supported a mechanism (Fig. 5) in which an active-site

base abstracts the α-proton, then the enolate intermediate (depicted as the aci-
carboxylate 9 in Fig. 5) undergoes vinylogous antielimination [60] of the 3-hy-
droxyl to give the product 10. Although this general, stepwise mechanism is
broadly accepted, some details such as the identities of the active-site base and
the metal ligands remain disputed.
One of the three contrasting proposals on the catalytic machinery of enolase
is derived from the analysis of crystals of the yeast enzyme soaked with the
natural substrate 2-PGA [61] and with the inhibitor phosphonoacetohydroxamate
(PhAH) [62]. In their proposal, Lebioda and Stec contend that a water molecule
serves as the catalytic base in conjunction with Glu168 and Glu211, and subse-
quent studies on the E211Q mutant supported this ‘‘charge shuttle’’ mechanism
[63]. These studies also indicated that the 3-OH of the substrate is ligated to
metal I. Janin and coworkers, judging from the x-ray structure obtained from
lobster enolase-Mg 2⫹ crystals soaked with the inhibitor phosphoglycolate, pro-
posed a similar role for metal I in the lobster enzyme [64]. However, the lobster
enolase data suggested that His157, instead of water, is likely the catalytic base.
Rayment, Reed, and their coworkers presented a third proposal, which is based
on x-ray analysis of yeast enolase cocrystallized with Mg 2⫹ and PhAH [65]. In
this proposal, Lys345 is the active-site base and metal I coordinates to the C-1
carboxylate. Recently, they cocrystallized yeast enolase with the natural substrate
2-PGA and confirmed their early conclusions regarding the metal coordination
(octahedral) and active-site base assignment (Lys345) [66]. More importantly,
both metal sites were resolved in this work, and they were found to be separated
by a distance of ⬃4.2 Å, in contrast to a distance of 6–9 Å estimated by nuclear
magnetic resonance (NMR) spectroscopy [67]. Furthermore, the 3-OH of 2-PGA
appears to interact with Glu211 instead of with metal I, as in the other two propos-
als. Studies with K345A, E168Q, and E211Q mutants provided additional support
for the role of Lys345 as the catalytic base and for the involvement of Glu211
TM

CEO Bond
in 3-OH activation [56]. Notably, the relative positions of Lys345 and Glu211
in the Reed and Rayment proposal are consistent with the antistereochemistry
of the elimination reaction.
The x-ray crystal structure obtained by Larsen et al. [66] also revealed
that both essential divalent cations interact with the carboxylate of the substrate/
product. Together with Lys396, they appear to promote the redistribution of the
negative charge developed during the formation of the aci-carboxylate intermedi-
ate. Thus, a relatively weak active-site base, most likely assisted by significant
substrate orientation by the enzyme [68], can abstract a nonacidic (pK a ⬃ 28–32)
proton and expel a poor leaving group. Another example of this class of hydrolase
is galactonate dehydratase, which depends on Mg 2⫹ [69] and appears to utilize
His285 as the active-site base [70]. The Mn 2⫹-dependent imidazole glycerol phos-
phate dehydratase [71] may also be related to this mechanistic class of hydrolase.
The properties of the latter enzymes remain to be further characterized.
C. Reactions Catalyzed by Enzymes Containing

Iron–Sulfur Clusters
1. Aconitase
Iron–sulfur clusters are common redox cofactors in biological systems. However,
cases are known in which this metal center serves as a Lewis acid in enzymic
reactions. For example, aconitase, otherwise known as citrate(isocitrate) hy-
drolase (EC 4.2.1.3), catalyzes the interconversion of citrate (11) and isocitrate
(13) in the Krebs cycle through the dehydrated intermediate, cis-aconitate (12)
(Fig. 6). This mitochondrial enzyme (83 kDa) contains a [4Fe-4S] center [72–
74], which is in the inactive [3Fe-4S]⫹ form when aconitase is purified aerobically
[75,76]. Reduction in the presence of Fe2⫹ will convert the enzyme to the active
[4Fe-4S] 2⫹ state (Fig. 7) [75]. Additional reduction to the electron paramagnetic
resonance (EPR)-active [4Fe-4S] ⫹ state causes a ⬃60% drop in activity relative
to the [4Fe-4S] 2⫹ state [77], presumably due to the decreased Lewis acidity of
the reduced [4Fe-4S] ⫹ center. Mössbauer spectroscopic analyses on enzyme re-
Figure 6 General mechanism of the reaction catalyzed by aconitase.
TM

Figure 7 Interconversion between the inactive and active forms of the iron–sulfur clus-
ter of aconitase. The labile iron atom is designated as Fe a . (Source: Adapted from Ref. 72.)
constituted with 57 Fe revealed a single labile iron site, which is called Fe a [75,78].
Protein ligands of the nonlabile iron atoms are cysteine residues [79–82], and
the labile iron is ligated to a hydroxyl ion, which becomes protonated to water
when substrate is bound [81–84].
Results gathered from x-ray analysis [82], mutation experiments [85], and
inhibitor studies [86] support a general mechanism (amino acid residue number-
ing for porcine heart aconitase), that starts with substrate’s binding to the enzyme
and causes the active-site cleft to close. Then Arg452 and Arg580 act in concert
with His101 and His147 to bind and orient the α- and γ-carboxyl groups of citrate
(Fig. 8). At the same time, both the β-carboxylate and β-hydroxyl of citrate coor-
dinate to Fe a, thus expanding the coordination sphere to octahedral, and the iron-
bound hydroxyl group becomes protonated to form water. The proton to be ab-
stracted is aligned with serine (Ser642), which is in an oxyanion hole (Arg644)
and consequently deprotonated as a result of a significant lowering of pK a. An-
ionic Ser642 accepts the proton from the substrate and eliminates the hydroxide
via a carbanion intermediate [86] to form cis-aconitate. The positive charge on
Figure 8 Binding of citrate to the [4Fe-4S] cluster of aconitase and partial active-site
structure. (Source: Adapted from Ref. 74.)
TM

CEO Bond
Fe a also assists in the dissociation of the hydroxyl group from the substrate. As
shown in Fig. 8, cis-aconitate, with its γ-carboxyl bound to Arg580, then flips
180° as a result of a conformational change [87]. Such a flip switches the positions
of the α- and β-carbons with respect to Fe a and allows addition of water onto
the α-carbon. The overall reaction is readily reversible.
2. Fumarate Hydratase
Fumarate hydratase (fumarase, EC 4.2.1.2) catalyzes the reversible hydration of
fumarate (14) to (S)-malate (15) in the citric acid cycle (Fig. 9). The mammalian
enzyme is extremely efficient and catalyzes the reaction with an apparent second-
order rate constant (k cat /K m ⫽ 2.4 ⫻ 10 8) approaching the diffusion limit [88,89].
Unlike fumarases of mammals, yeast, and Bacillis subtilis (class II), fumarase A
from E. coli (class I) is an iron–sulfur-containing hydrolase [90]. The active
fumarase A is a homodimer (60 kDa per monomer), and each subunit contains
a [4Fe-4S] cluster that can be oxidized to the inactive [3Fe-4S] state in the same
fashion as shown in Fig. 7 for aconitase [90]. The reduced [4Fe-4S] form of
fumarase A is much less active (10- to 50-fold) than the oxidized [4Fe-4S] 2⫹
form, and substrate binding dramatically affects the [4Fe-4S] ⫹ EPR signal. These
results are used as evidence to support a catalytic mechanism for fumarase that
parallels the one for aconitase [90]. Other class I fumarases may include E. coli
fumarase B [91], Euglena gracilis fumarase [92], and Rhodobacter capsulatus
fumarase [93]. It should be noted that class II fumarases bear little homology
to fumarase A and fumarase B from E. coli [94,95], and they have no cofactor
requirements, so these other enzymes do not function by the same mechanism.
Figure 9 Postulated mechanism of the reaction catalyzed by fumarase. There is no evi-

dence for a sixth ligand, but a water molecule is bound as a sixth ligand in the related
enzyme aconitase [82]. Thus, a question mark designates a possible bound water molecule
in fumarase. (Source: Adapted from Ref. 90.)
TM

3. Dihydroxy-Acid Dehydratase
Dihydroxy-acid dehydratase (DHAD, EC 4.2.1.9) catalyzes the dehydration of
2,3-dihydroxycarboxylic acids 16 to the corresponding 2-keto acids 17, a key
step in the branched-chain amino acid synthesis of valine and isoleucine (Fig.
10). The E. coli enzyme is a 125-kDa homodimeric protein containing a [4Fe-
4S] cluster per monomer [96]. This DHAD is sensitive to both O 2 and superoxide
[96–98]. However, in marked contrast to that in aconitase, the in vitro activity
of the inactivated enzyme could not be restored by incubation with Fe 2⫹ and
reducing agent [96], apparently because the Fe-S cluster completely degrades
upon oxidation. Conversely, in vivo studies demonstrated that the inactivation
was reversible [99]. Resonance Raman spectroscopy established the presence of
a single iron, which lacks cysteinyl coordination in the [4Fe-4S] cluster [96].
This iron is postulated to be directly involved in the reaction as with aconitase.
As opposed to the E. coli enzyme, DHAD from spinach leaves (homodimer,
63 kDa per monomer) is not inactivated by O2, and it has a [2Fe-2S] center instead
of a [4Fe-4S] center [100]. The EPR spectra of this unique Fe-S dehydratase
(g ave ⫽ 1.90) are characteristic of Rieske Fe-S proteins, even though the redox
potential of the [2Fe-2S] center (⫺470 mV) more closely resembles that of plant-
type ferredoxins. This dual nature of the spinach DHAD [2Fe-2S] center is re-
flected in its reaction with substrate. When substrate is added to the reduced
protein, the EPR signal transiently shifts so the g ave approaches that of other plant-
type ferredoxins (gave ⫽ 1.95). After the substrate is consumed, the EPR spectrum
reverts to the one similar to Rieske proteins. Reduction of spinach DHAD de-
creases the activity 6-fold. These observations indicate the direct involvement of
the Fe-S center in the reaction, probably as a Lewis acid, to facilitate the departure
of the β-hydroxyl group. Consistent with this suggestion, substrate analogue stud-
Figure 10 Schematic for the reaction catalyzed by dihydroxyacid dehydratase. Note

that a single enzyme is involved in the synthesis of two different amino acids, depending
on the identity of the R group.
TM

CEO Bond
ies support the development of a partial positive charge at β-C [101]. The sub-
strate must have the R-configuration at α-C, but DHAD surprisingly accommo-
dates either the 2R,3R or 2R,3S isomer with nearly equal efficiency. Solvent
hydrogen incorporation at the β-position proceeds with retention of configuration
[102]. Because DHAD from all sources examined thus far also requires a divalent
metal ion for activity [101], it would be interesting to find out whether the mecha-
nism of DHAD bears any similarity to those discussed in Section II.B as well
as to that of aconitase.
4. L-Serine Dehydratase
Recently, an l-serine dehydratase (EC 4.2.1.13) involved in both the biosynthesis
and metabolism of serine was purified from Peptostreptococcus asaccharolyticus
and shown to be a heterodimer of 55 kDa (α, 30 kDa; β, 25 kDa) containing a
[3Fe-4S] center [103,104]. The enzyme could be inactivated by exposure to air
and reactivated by incubation with Fe 2⫹ under anaerobic conditions, presumably
by regenerating an oxidized [4Fe-4S] 2⫹ center. Though EPR spectroscopic exper-
iments confirmed the presence of an oxidized [3Fe-4S] ⫹ center, direct evidence
for a [4Fe-4S] center is still lacking [104]. Instead, the purported [4Fe-4S] 2⫹
center was inferred by titration of the anaerobically purified enzyme with
K 3Fe(CN) 6 and observation of a successively larger [3Fe-4S] ⫹ EPR signal, which
correlated with a decrease in activity. Interaction of the substrate with the Fe-S
center was also inferred from the observation that l-serine could both shield the
[3Fe-4S] ⫹ center from buffer-dependent changes in the EPR signal and prevent
oxidative loss of activity when the active enzyme is exposed to air. This evidence
has been used to propose a mechanism in which the labile iron of the [4Fe-4S] 2⫹
center binds the hydroxyl and carboxyl groups of serine [104]. This iron atom
would thus serve as a Lewis acid and facilitate loss of the hydroxyl group, an
otherwise poor leaving group (Fig. 11). Such a mechanism is vastly different
from that for most l-serine dehydratases from nonbacterial sources, which utilize
PLP as the cofactor in the irreversible deamination of serine [105]. In the latter
case, the α-H and not the β-hydroxyl group is activated during catalysis.
D. Reactions Catalyzed by Enzymes Containing

Organic Cofactors
1. Pyridoxal 5′-Phosphate–Dependent Dehydratases:
L-Serine Dehydratase and L-Threonine Dehydratase
Many enzymes that catalyze β-elimination reactions are pyridoxal 5′-phosphate–

(PLP)-dependent [106,107]. Two such PLP-dependent enzymes that catalyze the
β-elimination of a hydroxyl group are l-serine dehydratase (EC 4.2.1.13) and
l-threonine dehydratase (EC 4.2.1.16), which irreversibly convert l-serine and
TM

Figure 11 Postulated mechanism for the reaction catalyzed by the iron–sulfur-con-
taining l-serine dehydratase from Peptostreptococcus asaccharolyticus. (Source: Adapted
from Ref. 104.)
l-threonine to pyruvate and α-ketobutyrate, respectively. Although all the l-

threonine dehydratases from different sources have thus far proved to be PLP-
dependent, l-serine dehydratases from Peptostreptococcus asaccharolyticus
[103] and Clostridium propionicum [108] depend instead on an iron–sulfur center
(discussed in Section II.C.4). The role of the PLP coenzyme in these hydrolases
is to facilitate the C-H bond cleavage at α-C of the substrate by delocalizing the
resulting anion throughout the pyridine ring. These electrons are then used to
drive the subsequent elimination of the β-hydroxyl group.
As depicted in Fig. 12, the reaction is initiated by the formation of Schiff
base 18 between PLP and the amino group of the substrate. Abstraction of the
α-proton generates resonance-stabilized intermediate 19. Electron pushing in this
transient quinonoid species expels the β-hydroxyl group and leads to the forma-
tion of aminoacrylate intermediate 20. The ⑀-amino group of the active-site lysine
then adds to the C-4′ position of the PLP moiety (21), followed by the release
of the corresponding imine. Hydrolysis of the resulting imine yields the 2-keto
acid product 22 and completes the catalytic cycle. The large body of evidence
supporting this mechanism has been summarized by Miles [106]. Tryptophan
synthase (EC 4.2.1.20) utilizes a similar mechanism [106,109], though the reac-
tion is a β-replacement of the hydroxyl group with an indole moiety instead of
a hydrogen.
2. Nicotinamide-Adenine-Dinucleotide-Dependent
Dehydratases: TDP-Glucose 4,6-Dehydratase
Temporarily converting a hydroxyl group to a keto function via intramolecular
oxidation–reduction is another strategy adopted by biological systems for activat-
ing an adjacent CEH bond in CEO bond cleavage reactions. For example, this
TM

CEO Bond
Figure 12 Scheme for the dehydration of l-serine (R ⫽ H) and l-threonine (R ⫽ CH 3)

catalyzed by PLP-dependent l-serine and l-threonine dehydratase. Note that these en-
zymes from most sources will act on both l-serine and l-threonine with varying degrees
of efficiency. PLP, pyridoxal 5′-phosphate. (Source: Adapted from Ref. 106.)
interesting deoxygenation method is utilized by nucleotidyl diphosphohexose

4,6-dehydratases, which transform nucleotidyl diphosphohexoses into the corre-
sponding 4-keto-6-deoxy sugar derivatives that are the precursors of many un-
usual sugars [110,111]. The best-studied enzyme within this family is TDP-d-
glucose 4,6-dehydratase (80-kDa homodimer, EC 4.2.1.46) isolated from E. coli,
and this enzyme contains 1 NAD ⫹ per dimer [112]. Notably, recent studies of
CDP-d-glucose 4,6-dehydratase, a closely related enzyme from Yersinia pseudo-
tuberculosis, have shown that it has a higher affinity for NADH than for NAD ⫹
[113]. Since two cofactor binding sites were found in all these enzymes, yet only
one NAD ⫹ per dimer was generally detected, it is possible that the other coen-
zyme binding site is occupied by NADH.
TM

As illustrated in Fig. 13, the reaction consists of three catalytically discrete
steps: step I, oxidation of nucleotidyl 5′-diphosphohexose 23 to the corresponding
4-ketohexose 24; step II, C5/C6 dehydration to a 4-keto-∆5,6-glucoseen intermedi-
ate 25; and step III, reduction at C-6 to give the resulting 4-keto-6-deoxyhexose
product 26. This net intramolecular oxidation–reduction has been shown to in-
volve an internal hydrogen transfer from C-4 of the substrate 23 to C-6 of the
product 26, and the enzyme-bound NAD ⫹ apparently functions as a hydride car-
rier in this catalysis [114,115]. Thus, there are actually two C-H cleavage events
involved in this catalysis. The first CEH bond rupture forms the keto group,
which activates the hydrogen at C-5 for the second C-H bond cleavage, triggering
the elimination of the 6-OH. The displacement of the hydroxyl group at C-6 by
the hydrogen from C-4 occurs with net inversion, the hydrogen transfer to NAD ⫹
is ‘‘si-face’’-specific, and the dehydration from C-5 and C-6 is a syn elimination
[116–119]. These results indicate that sugar 4,6-dehydratases belong to a small
group of enzymes in which the pyridine nucleotide coenzyme is a catalytic pros-
thetic group, contrary to most other nicotinamide-dependent enzymes, in which
NAD ⫹ or NADP ⫹ functions as a cosubstrate [110]. In addition to the sugar 4,6-
Figure 13 Catalytic cycle for the dehydration of nucleotidyl diphosphohexoses. The

E od label is meant as a general reference to all enzymes that catalyze this type of reaction.
The NDP at C-1 reflects the alternate nucleotidyl sugar substrates that are utilized by the
different dehydratases, NDP, nucleotidyl 5′-diphospho-. (Source: Adapted from Ref. 113.)
TM

CEO Bond
dehydratases, this unique group of catalysts includes a few other enzymes, such
as 3-dehydroquinate synthase (EC 4.6.1.3) [120,121] and 2-deoxy-scyllo-inosose
synthase [122].
3. Flavin Dependent Dehydratases

a. (R)-2-Hydroxyglutaryl-Coenzyme A Dehydratase. Another interesting
class of enzymes that catalyze the elimination of water using redox chemical
processes is the hydroxy acyl-CoA dehydratases from Clostridium and related
bacteria. These oxygen-sensitive enzymes are able to dehydrate a wide range of
hydroxyacyl-CoA derivatives in the process of anaerobic fermentation. One such
enzyme, isolated from Acidaminococcus fermentans, catalyzes the reversible syn
dehydration of (R)-2-hydroxyglutaryl-CoA (27) to (E )-glutaconyl-CoA (28) in
the glutamate fermentation pathway (Fig. 14) [123]. This reaction is of consider-
able interest since it involves the cleavage of a nonactivated C-H bond at C-3.
The active heterodimer (α, 54 kDa; β, 42 kDa) is oxygen-sensitive, and it contains
4 mol nonheme iron, 4 mol inorganic sulfur, 0.3 mol reduced riboflavin, and 1
mol reduced flavin mononucleotide (FMN) [124]. An additional adenosine tri-
phosphatic– (ATP)- and Mg 2⫹-dependent Fe-S protein (homodimer, 54 kDa per
monomer) is required in catalytic amounts to activate the dehydratase reductively.
The in vitro assays used Ti(III) citrate as the reductant, though NADH is suspected
to be the electron source in vivo, possibly via an electron-relay pathway involving
a diaphoraselike enzyme that can transfer single electrons [123]. Titanium (III)
citrate delivers a low-potential electron (E°′ ⫽ ⫺600 mV) that is apparently recy-
cled for many turnovers, suggesting a radical mechanism. The observation that
oxidants, including 2-nitrophenol, 3-nitrophenol, 4-nitrophenol, 4-nitrobenzoate,
and chloramphenicol, temporarily inhibit the dehydration reaction until these com-
pounds are consumed also supports the idea of a radical mechanism [124].
As shown in Fig. 14a, the mechanism may include a ketyl intermediate 29
that eliminates the hydroxyl group with the assistance of a [4Fe-4S] center. It
has been demonstrated that α-hydroxyketones undergo single-electron chemical
reactions in the presence of a strong one-electron donor to eliminate hydroxyl
groups [125]. The resulting enoxy radical 30 could be reduced to give 31, which
then releases the 3-H as a hydride to regenerate the reduced flavin coenzyme and
concomitantly afford the product 28. However, it is more likely that enoxy radical
intermediate 30 releases the 3-H as a proton to give 32, followed by an electron
loss to form glutaconyl-CoA 28 (Fig. 14b).
The proposed recycling of a single, low-potential electron is mechanisti-
cally appealing and may be involved in the catalysis of other related enzymes.
Two possible examples are (R)-2-hydroxyglutaryl-CoA dehydratase from Fuso-
bacterium nucleatum [126] and (R)-lactyl-CoA dehydratase (EC 4.2.1.54) from
Clostridium propionicum [127], both of which are iron–sulfur flavoenzymes in-
TM

TM

CEO Bond
volved in the anaerobic fermentation of glutamate and alanine, respectively. Stud-

ies with EPR spectroscopy have characterized a [4Fe-4S] and a [3Fe-3/4S] center
in the latter enzyme [128], and the dehydration using an alternate substrate, (R)-
2-hydroxybutyryl-CoA, was established to be a syn elimination [129]. Further
studies may reveal that these enzymes operate through a common mechanism,
in which a C-H bond is broken after the C-O bond cleavage has occurred, as
opposed to the mechanism of 4-hydroxybutyryl-CoA dehydratase discussed later
and most other enzymes reviewed here.
b. 4-Hydroxybutyryl-Coenzyme A Dehydratase. This enzyme is 4-hydroxybu-

tyryl-CoA dehydratase from C. aminobutyricum (homotetramer, 56-kDa per sub-
unit), which converts 4-hydroxybutyryl-CoA (33) to crotonyl-CoA [34] in the
metabolism of γ-aminobutyrate (Fig. 15) [130,131]. Stoichiometric quantities of
iron and sulfur as well as flavin adenine dinucleotide (FAD) have been found in
the enzyme [132], and EPR spectroscopic studies indicated the presence of an
oxygen-sensitive [4Fe-4S] cluster that is required for activity [133]. Interestingly,
this [4Fe-4S] cluster decomposes to a species other than a [3Fe-4S] center upon air
inactivation, in contrast to the cluster in aconitase (Section II.C.1). Thus, the [4Fe-
4S] cluster may serve in a structural capacity instead of, or in addition to, acting as
a Lewis acid. Other studies have revealed that the reduced enzyme is inactive [132],
so the flavin cofactor must be oxidized before catalysis can be initiated.
These data support a mechanism consisting of three steps: dehydrogenation,
dehydration, and isomerization. As shown in Fig. 15a, the reaction could be initi-
ated by an α-proton abstraction followed by a β-hydride transfer to reduce the
active-site bound FAD and give 4-hydroxycrotonyl-CoA (35). The reducing
equivalents would be subsequently used in a nucleophilic attack at the α-carbon,
and the 4-hydroxyl could be eliminated possibly with assistance from the [4Fe-
4S] center, as is the case with aconitase. The resulting vinylacetyl-CoA (36) is
then isomerized to crotonyl-CoA (34).
Alternatively, a mechanism that includes an oxidized ketyl intermediate
could be involved [134]. In this proposed radical mechanism, depicted in Fig.
15b, the intermediate enolate 37 could be oxidized by the FAD to enoxy radical
38, which would deprotonate to a ketyl radical anion 39, expel the 4-OH (with
or without assistance from a [4Fe-4S] cluster), and go through dienoxy radical
Figure 14 Proposed mechanisms for the reversible dehydration of 2-hydroxyglutaryl-CoA

involving loss of either (a) a hydride or (b) a proton from the substrate. The participation of
the activase is illustrated in (b), though it would play a similar role in (a) as well. The possible
coordination of the substrate to a [4Fe-4S] center is represented by ‘‘Fe’’ in (a), and it is shown
with more detail in (b). CoA, coenzyme A; FMN, flavin mononucleotide; ATP, adenosine
triphosphate; ADP, adenosine diphosphate. (Source: Adapted from Ref. 124.)
TM

Figure 15 Scheme for the possible (a) anionic and (b) radical mechanisms for the re-
versible dehydration of 4-hydroxybutyryl-CoA. To reduce congestion, the [4Fe-4S] cluster
is represented as ‘‘Fe’’ in (b). FAD, flavin adenine dinucleotide.
40 before regenerating the FAD cofactor and completing the catalysis. Support
for the latter hypothesis derives from two primary sources: (1) There is a close
relationship with the (R)-2-hydroxyglutaryl-CoA dehydratase, as discussed in the
previous section, and (2) the FAD of 4-hydroxyglutaryl-CoA dehydratase is
readily reduced to the neutral semiquinone whereas reduction to the hydroqui-
none appears to be kinetically hindered [133]. Though the details of this mecha-
TM

CEO Bond
nism still need to be elaborated and verified, it seems clear that the flavin coen-
zyme directly participates in the deoxygenation reaction, and it needs to be
oxidized to do so.
c. Chorismate Synthase. Another deoxygenation in which the cleavage of a
C-O bond may occur prior to the rupture of a C-H bond is catalyzed by chorismate
synthase (EC 4.6.1.4) in the seventh step of the shikimate pathway (also see
Section II.A.3). The enzyme from all known sources depends on a reduced flavin
coenzyme in the conversion of 5-enolpyruvylshikimate 3-phosphate (EPSP) (41)
to chorismate (42) [135–139], as shown in Fig. 16. Although this reduced flavin
is often generated by an alternative pathway, the enzyme from some sources is
bifunctional and can provide the reduced flavin via an additional diaphoraselike
activity using NADPH as the cosubstrate [28]. After the catalytic cycle, the flavin
cofactor remains reduced, so there is no net change in the oxidation state of the
enzyme.
Though the precise role of the flavin is unknown, studies using flavin and
Figure 16 Possible nonconcerted pathways for the chorismate synthase reaction. Cur-
rent data cannot distinguish between a cationic (top) and a radical (bottom) mechanism.
(Source: Adapted from Ref. 158.)
TM

substrate analogues [140,141] and ultraviolet–visible (UV-Vis) spectrophotome-
try [142,143] indicated that this prosthetic group may be chemically, not just
structurally, involved in turnover. The net 1,4-elimination of the 3-phosphate and
the 6-H R proceeds with antistereochemistry [144–146]. Model system studies
[147,148] and molecular orbital calculations [149,150] suggested a preference
for syn stereochemistry in concerted 1,4-eliminations. Thus, it has been proposed
that the chorismate synthase reaction is nonconcerted [151,152], and kinetic iso-
tope effect studies are consistent with this proposal [153]. However, evidence
supporting a nonconcerted mechanism is not conclusive [152,154–158], so other
possibilities must still be considered. Interestingly, the observation of a neutral
flavin semiquinone upon incubation with the competitive inhibitor (6R)-6-fluoro-
EPSP suggested the likely involvement of radical intermediates [159]. Although
single-turnover kinetics did not detect the neutral semiquinone radical, a different
flavin intermediate was found to form and decay in a kinetically competent fash-
ion [156–158]. The spectral features of this flavin intermediate may be attribut-
able in part to an anionic semiquinone radical [158]. Thus, these biophysical
studies support a nonconcerted mechanism that includes an allylic radical inter-
mediate, though the current data are unable to distinguish this mechanism from
other nonconcerted models (Fig. 16). Clearly, a more detailed investigation is
necessary to characterize further the nature of the flavin intermediate, determine
its role in the catalysis, and discriminate between an ionic and a radical mecha-
nism.
E. Protein-Radical-Dependent Dehydratases
1. Diol Dehydrases
Though the aforementioned flavin-dependent enzymes may include a free-radical
intermediate in the reaction pathway, none of those reactions requires the forma-
tion of a protein radical as part of the catalysis. However, there are a few systems
in which the generation of a protein radical appears to be a prerequisite to abstract
a hydrogen atom from the substrate before deoxygenation can occur. The reac-
tions mediated by diol dehydrases and glycerol dehydrases are two likely exam-
ples [160]. These multimeric enzymes (⬃230 kDa) catalyze the irreversible dehy-
dration of vicinal diols into corresponding aldehydes or ketones [161] in the
anaerobic fermentation of glycerol and other glycols in several bacterial organ-
isms [Table I of Ref. 162]. Adenosylcobalamin (AdoCbl) is a required cofactor
of these enzymes [163], and the adenosyl radical [164] that is generated by the
homolytic cleavage of the reactive Co-C bond of AdoCbl [165] is the initiator
of the reaction. A monovalent cation, such as K ⫹, NH 4⫹, Tl ⫹, or Rb ⫹, is also
required to assist in the binding of AdoCbl to the enzyme [166,167]. In the cases
with methylcobalamin-dependent methionine synthase [168] and methylmalonyl-
TM

CEO Bond
CoA mutase [169], the axial dimethylbenzimidazole (DMB) ligand of AdoCbl

is displaced by a histidine residue of the enzyme, and this ligand substitution
initiates the formation of a charge-relay system that seems to be important for
modulating the Co-C bond strength [170]. However, analysis of the deduced
amino acid sequence of diol dehydrase fails to locate the fingerprint sequences
characteristic of histidine ligation to cobalt [170], so the mechanism whereby
this protein affects the Co-C bond strength is not yet clear.
The AdoCbl-catalyzed dehydration of vicinal diols is mechanistically quite
similar to most other AdoCbl-dependent rearrangements [161,171]. The overall
mechanism is depicted in Fig. 17, and the reaction is initiated by the substrate-
induced homolytic cleavage of the Co-C bond of AdoCbl into a 5′-deoxyadenosyl
radical and cob(II)alamin. The adenosyl radical abstracts a hydrogen atom from
a protein residue (shown in Fig. 17 as a cysteine), and the resulting protein radical
abstracts a C-1 hydrogen atom (H a) from the substrate 43 to generate 44. Substrate
radical 44 rearranges to the product radical 45, which is converted to gem-diol
46 by receiving H a from the protein residue. Subsequent dehydration leads to the
formation of product 47. Release of the product shifts the equilibrium so the
AdoCbl cofactor returns to the associated form, thus completing the catalytic
cycle.
Signals arising from low-spin Co(II) (g ⫽ 2.2–2.4) and an organic radical
(g ⫽ 1.95 and 2.04) have been detected by EPR [172–174], and both species
were later shown to be formed at a kinetically competent rate [175]. The fact
that hydrogen isotope exchange occurs readily between C-5′ of the adenosyl moi-
ety and C-1 of the substrate corroborates the intermediacy of the adenosyl radical
[176–180]. Both the hydrogen migration and the hydroxyl elimination are stereo-
specific [181]. The stereochemistry at C-2 dictates which hydrogen atom at C-1
undergoes migration, with the 1-H S being abstracted from (S)-1,2-propanediol
and the 1-H R from the (R)-substrate [182]. The subsequent hydrogen atom re-
bound to replace the migrating hydroxyl group proceeds with inversion of the
C-2 configuration [181,182]. During the migration of the hydrogen atom, there
is no exchange with the solvent [183].
The involvement of a protein radical in diol dehydratase is implicated
mainly by an observed isotope effect of 125 for the transfer of 3 H from cofactor
to product [163]. In order to explain the large k H /k T isotope effect, an additional
step to generate a protein radical intermediate appeared to be necessary [184].
In fact, the intermediacy of a protein radical has been suggested for many vitamin
B 12 –dependent reactions. The observations of a critical thiol residue in diol dehy-
drase [185], of a possible protein-based radical in AdoCbl-dependent ethanol-
amine ammonia lyase [186,187], and of a kinetically competent thiyl radical in-
termediate in the AdoCbl-dependent ribonucleotide reductase [188] all support
the assignment of this putative protein radical as a thiyl radical, and this proposal
is reflected in Fig. 17. Although the rearrangement mechanisms have not been
TM

Figure 17 Proposed mechanism of the diol dehydrase reaction. Note that the AdoCbl
is predominantly in the associated form until the substrate binds. Though 1,2-propanediol
is illustrated as the substrate in this scheme, several other glycols are recognized by this
enzyme (Table I of Ref. 162). The stereochemistry as depicted may be altered by the
configuration at the C-2 position (see text for details). Ado, adenosyl moiety; [Co], cobal-
amin.
fully elucidated and the protein-based radical has not been directly detected
in diol dehydrase, the participation of the adenosyl radical in the reaction is con-
clusive. However, a reactive protein radical may actually be responsible for
abstracting a relatively inert C-H bond of the substrate to initiate the deoxygen-
ation.
TM

CEO Bond
2. Ribonucleotide Reductase
Of the many deoxygenation reactions that have been characterized, the conver-
sion of ribonucleotides to 2′-deoxyribonucleotides is perhaps the most prominent
because of its role in deoxyribonucleic acid (DNA) biosynthesis. Because of the
sheer importance of the reaction it catalyzes, ribonucleotide reductase (RNR)
has been studied extensively over the years, and the mechanism is now well
characterized. Ribonucleotide reductase utilizes a stable protein radical to cata-
lyze the irreversible, reductive replacement of the 2′-hydroxyl group with hydro-
gen in ribonucleotides. Interestingly, at least three classes (I–III) of RNR have
been discovered, and each employs different cofactors and protein radical–gener-
ating mechanisms, though the subsequent catalytic steps are the same. In all
known forms of RNR, radical-mediated abstraction of a nonacidic hydrogen atom
precedes C-O bond cleavage, similar to most anionic deoxygenation mechanisms.
Several reviews summarizing the work that has been done on this enzyme are
available [189–195], and information pertaining to cofactor requirements and the
mechanism is briefly presented.
Class I ribonucleotide reductases are characterized by a binuclear high-spin
Fe(III) complex and a tyrosyl radical, and these enzymes aerobically deoxygenate
ribonucleotide diphosphates. The 258-kDa (α 2β 2: α 2 usually known as R1 and
β 2 as R2) protein from Escherichia coli is the best studied member of this class
[196–202]. Crystal data of R2 confirmed the presence of a binuclear µ-oxo
bridged iron cluster that is close enough to stabilize the tyrosine–122 (Tyr122)
radical (Fig. 18) [203]. Scavenging of this Tyr122 radical by hydroxyurea [204]
or by the substrate analogue 2′-deoxy-2′-mercaptouridine 5′-diphosphate [205]
leads to irreversible inactivation of the enzyme. Extensive study of the spontane-
ous assembly of the iron cluster and the Tyr122 radical in the presence of Fe(II)
and O 2 revealed the involvement of a diferric Fe(III)/Fe(IV) species and a neutral
tryptophan–48 (Trp48) radical [206–211]. The Trp48 occupies a site in a hy-
drophobic region of R2 that is believed to interact with R1 [212]. It may also
be part of an uncharacterized electron relay, involving His118, Asp237, and
Trp48, that transfers the radical from Tyr122 in R2 to a critical protein residue
in R1 (a distance of 35 Å) where the substrate binds and reacts. This protein
residue is proposed to be Cys439 on the basis of mutagenesis studies [213] and
of sequence comparisons with AdoCbl-dependent RNR from Lactobacillus leich-
mannii, in which a kinetically competent thiyl radical (Cys408) was characterized
by EPR spectroscopy [188].
As shown in Fig. 19, deoxygenation is initiated by the abstraction of a
hydrogen atom from the 3′-position of the ribose ring by the thiyl radical. The
2′-hydroxyl group is then protonated by one cysteine of a redox-active thiol pair
(Cys225 and Cys462) and is lost as water, resulting in a resonance-stabilized
cation radical intermediate. Though such a substrate-based radical has not yet
TM

Figure 18 Representation of the µ-oxo diferric iron cluster and the neighboring tyrosyl
radical in class I RNR (Escherichia coli). The bridging oxide anion derives from molecular
oxygen. The tyrosyl radical is stabilized by the iron cluster. (Source: Adapted from Ref.
195.)
been detected, a possible nucleotide-based radical was recently observed during

studies with RNR and the time-dependent inactivators (E )- and (Z )-2′-fluoro-
methylene-2′-deoxycytidine 5′-diphosphate [214]. The redox-active thiol pair re-
duces the cation and protonates C-2′, becoming a disulfide in the process. Partici-
pation of this thiol pair in the reaction has been demonstrated by mutagenesis
studies [215–217]. The more recent observations of a [XN ⋅ S CysR1] radical upon
inactivation with 2′-azido-2′-deoxyuridine 5′-diphosphate [218] and of a perthiyl
radical upon inactivation with 2′-deoxy-2′-mercaptouridine 5′-diphosphate [205]
provide initial evidence for the intriguing possibility that reduction of the sub-
strate intermediate by the redox-active thiol pair may occur via one-electron
chemistry involving thiyl radicals, as opposed to a two-electron hydride transfer.
After the cation radical intermediate is reduced, the same hydrogen atom ab-
stracted by the thiyl radical is returned to the 3′-position to form the product
deoxyribonucleotide. A biological thiol reducing agent, such as dithiothreitol,
thioredoxin, or glutaredoxin, reduces the redox-active disulfide and completes
the catalytic cycle. It is important to note that once the Cys439 thiyl radical is
generated, the remaining catalytic steps are identical for all known classes of
ribonucleotide reductases.
Class II RNR depends instead on AdoCbl for its ability to reduce ribonucle-
otide triphosphates [219]. The best characterized member of this class is from
L. leichmannii, and it is active under either aerobic or anaerobic conditions. This
enzyme (82-kDa monomer) promotes the homolytic cleavage of the inherently
TM

CEO Bond
Figure 19 Proposed mechanism of the reduction of ribonucleotides catalyzed by ribonu-

cleotide reductase. The method for generating the catalytic thiyl radical depends on the
class of RNR (see text). Amino acid numbering for the Escherichia coli class I enzyme
is shown, though enzymes from class II and III utilize the same mechanism. RNR, ribonu-
cleotide reductase. (Source: Adapted from Refs. 189 and 195.)
reactive CoEC bond of AdoCbl to form cob(II)alamin and the 5′-deoxyadenosyl

radical. As for diol dehydrase (Section II.E.1), it is not understood how the RNR
induces the homolytic cleavage of the CoEC bond [170], though a nitrogen has
been identified as a ligand to cobalt [220]. The 5′-deoxyadenosyl radical then
abstracts a hydrogen atom from Cys408 [221] to form a thiyl radical, which has
recently been detected and characterized [188,222]. Subsequent ribose reduction
steps are identical to those described previously (Fig. 19).
Unlike the preceding two classes of RNR, a ribonucleotide triphosphate
reductase (heterodimer, 160-kDa α2, 35-kDa β2) produced by E. coli grown under
anaerobic conditions has been shown to have an absolute requirement of S-adeno-
sylmethionine (AdoMet) for activity [223]. Similar to AdoMet-dependent pyr-
uvate-formate lyase [224], this class III RNR uses an iron-dependent ‘‘activase,’’
AdoMet, NADPH, flavodoxin, and flavodoxin reductase to generate a protein-
based radical under anaerobic conditions [225], though in RNR the activase is
actually part of the complete reductase complex [226]. It appears that the two
TM

subunits of the small protein are held together by a [4Fe-4S] center with one
oxygen-sensitive, labile iron [226,227], and this small dimer forms a tight com-
plex with the large dimer much as with class I RNR. On the basis of mechanistic
work with pyruvate-formate lyase [228–230], the small dimer probably interacts
with AdoMet and induces the homolytic scission to form both methionine and
a 5′-deoxyadenosyl radical, which subsequently generates the stable protein radi-
cal in the large dimer. Experiments have localized this protein radical to Gly681
[231,232], which presumably takes the place of the tyrosine radical in class I
RNR and generates the catalytic thiyl radical [233]. Thereafter, reduction of the
ribose moiety and deoxygenation proceed as described (Fig. 19), except class III
RNR utilizes electrons from formate instead of thioredoxin to regenerate the
redox-active thiol pair [234].
III. A NOVEL CEO BOND CLEAVAGE EVENT

A. Background
Sugars are most commonly associated with processes involved in the storage and
generation of the energy necessary to sustain life, such as glycogen synthesis and
glycolysis. The most notable exception to the use of sugars as metabolic vehicles
is the ribose moiety of DNA and ribonucleic acid (RNA), in which the sugar
plays a definitive structural role by providing the scaffolding for all biological
genetic material and the messages needed to translate it. The only difference
between DNA and RNA molecules is the substitution of a hydrogen for the 2′-
hydroxyl group of the ribose moiety, and this structural difference has been di-
rectly linked to the greater stability of DNA over RNA. In fact, such substitutions
generally enhance the stability of sugars, as reflected by the large number of
structural roles that have been assigned to deoxysugars [235]. Although in most
cases the precise functions of deoxysugars have not yet been fully elucidated, it
appears that incorporation of a deoxysugar as a structural component provides a
facile means of modifying the surface properties of a compound or organism.
One relevant example is the incorporation of 3,6-dideoxysugars into the lipopoly-
saccharide (LPS) of some gram-negative bacteria [236–238], where these deoxy-
sugars confer immunogenic properties to the bacteria [236,239]. Among the five
known 3,6-dideoxyhexoses—ascarylose, abequose, paratose, tyvelose, and coli-
tose—the biosynthesis of ascarylose by Yersinia pseudotuberculosis is the most
thoroughly studied [111].
The proposed biosynthetic sequence of 3,6-dideoxysugars is exemplified
by the formation of cytidine 5′-diphospho– (CDP)-ascarylose (48), as shown in
Fig. 20. Initiation of the pathway proceeds with the coupling of α-d-glucose-
1-phosphate (49) and cytidine triphosphate (CTP) by α-d-glucose-1-phosphate
TM

CEO Bond
Figure 20 Overview of the biosynthetic pathway for CDP-ascarylose. CDP, cytidine

5′-diphosphate.
cytidylyltransferase (E p) to give CDP-d-glucose (50). Subsequent steps start with

an irreversible intramolecular oxidation–reduction catalyzed by the NAD ⫹-de-
pendent CDP-d-glucose 4,6-dehydratase (Eod ), which uses a well-known mecha-
nism, as described in Section II.D.2. The resulting product, CDP-6-deoxy-
l-threo-d-glycero-4-hexulose (51), is converted to 3,6-dideoxy-d-glycero-d-
glycero-4-hexulose (54) in two consecutive steps mediated by the actions of
CDP-6-deoxy-l-threo-d-glycero-4-hexulose-3-dehydrase (E1) and E1 reductase
(E3) [240,241]. The final steps are an epimerization catalyzed by CDP-3,6-dide-
oxy-d-glycero-d-glycero-4-hexulose-5-epimerase (Eep), which inverts the con-
figuration at C-5, followed by stereospecific reduction at C-4 catalyzed by
CDP-3,6-dideoxy-d-glycero-l-glycero-4-hexulose-4-reductase (Ered ) to give
CDP-ascarylose (48). There is little doubt that the most intriguing step of this
pathway is the C-3 deoxygenation catalyzed by E1 and E 3. This CEO bond cleav-
age event is particularly interesting since this is the first example of a pyridox-
TM

amine 5′-phosphate– (PMP)-mediated dehydration, and the reaction involves the
formation of a radical intermediate, which will be discussed further along with
other mechanistic aspects in the following sections.
1. CDP-6-Deoxy-L-Threo-D-Glycero-4-hexulose-3-
Dehydrase (E1)
Enzyme E 1 is a red–brown protein consisting of two identical subunits (49 kDa
per monomer). Both radiometric [242] and fluorometric [243] quantification re-
vealed one equivalent of PMP per monomer for full activity. Analytical assays
detected stoichiometric amounts of iron and sulfur, and EPR analysis of the re-
duced enzyme confirmed the presence of an adrenodoxin/putidaredoxin-type
[2Fe-2S] center (g ⫽ 2.007, 1.950, and 1.930) [244]. It should be noted that
purified E 1 is a mixture of apo- and holo-enzyme, so exogenous PMP, iron, and
sulfur are required to reconstitute its activity fully. Incubation of the enzyme with
substrate revealed that the PMP of E 1 forms a Schiff base with the C-4 keto group
of the substrate [245]. Then, His220 abstracts the pro-S hydrogen from the C-4′-
position of the PMP in the Schiff base complex and triggers the expulsion of the
C-3 hydroxyl group of the substrate to give PMP-∆ 3,4-glucoseen complex 52 (Fig.
20) [243]. Although intermediate 52 has never been isolated, it is reversibly hy-
drated on the si face of the conjugated imine, as demonstrated by isotopic incorpo-
ration with [18 O]H 2O [245]. Also, replacement of the C-3 hydroxyl group by
a solvent hydrogen in the E 1 –E 3 coupled reaction proceeds with retention of
configuration [246]. Together, these observations imply that the overall catalysis
in the active site of E 1 is most likely a suprafacial process occurring on the si
face of the PMP–substrate complex, as is the case with most other coenzyme
B 6 –dependent enzymes [247]. In fact, sequence alignments demonstrated a clear
relationship between E 1 and other PLP/PMP enzymes [243,248]. One significant
difference highlighted by the comparisons is the replacement of a highly con-
served lysine by a histidine at position 220 in E 1. This single replacement may
be responsible for converting a PLP-dependent transaminase to a PMP-dependent
dehydrase.
The preceding results clearly showed that E 1 behaves as a normal coenzyme
B 6-dependent enzyme that catalyzes a dehydration reaction. However, the C-3
deoxygenation product is formed only after NADH and E 3 are added to the reac-
tion mixture. Other reductases, such as diaphorase and methane monooxygenase
(MMO) reductase, are also able to generate small amounts of product [245].
Removal of the [2Fe-2S] center of E 1 impaired its ability to promote the final
product formation, though E 1(apoFeS) could still abstract the C-4′ hydrogen of
PMP [244]. This requirement of the [2Fe-2S] center, which is an obligatory one-
electron carrier, for product formation strongly implicated the intermediacy of
free radicals in the reaction. An EPR analysis of the chemically reduced E 1 sub-
TM

CEO Bond
strate complex confirmed the presence of a Lorentzian-type absorption (g ⬇ 2),

which displayed no detectable hyperfine splitting [249]. The participation of the
PMP cofactor in a deoxygenation is unique, but the involvement of a radical
mechanism truly places E1 in a class by itself.
2. CDP-6-Deoxy-L-Threo-D-Glycero-4-hexulose-3-Dehydrase
Reductase (E 3)
CDP-6-Deoxy-l-threo-d-glycero-4-hexulose-3-dehydrase reductase (E 3) is a mo-
nomeric (36-kDa) protein of a red–brown color, and it contains 1 mol of FAD
as revealed by high-performance liquid chromatography (HPLC) and UV-Vis
analysis [250]. Iron and sulfur quantification combined with UV-Vis and EPR
spectroscopy indicated the presence of a plant-type [2Fe-2S] center. Sequence
alignments established a close relationship between E 3 and other iron–sulfur fla-
voproteins in the ferredoxin-NADP ⫹ reductase (FNR) family [251]. Like other
members in its family, E 3 will transfer reducing equivalents from NADH to a
variety of one-electron acceptors, including O 2, with varying degrees of efficiency
[252]. Removal of the [2Fe-2S] center impairs the ability of E 3 to catalyze final
product formation in the presence of E 1 and substrate, but the FAD of E 3(apoFeS)
remains functional as a two-electron/one-electron switch in the reduction of O 2
to H 2O 2 [250]. Therefore, the FAD can operate independently of the iron–sulfur
center when E 3 acts as a NADH oxidase, but the electron-transfer relay from
NADH to substrate during sugar reduction clearly includes both FAD and the
[2Fe-2S] center.
Stopped-flow spectroscopic experiments confirmed the independent nature
of each cofactor by showing that FAD is reduced by NADH at a similar rate in
both E 3(apoFeS) and holo-E 3 [253]. After the FAD is reduced to the hydroqui-
none form, the subsequent electron transfer to the [2Fe-2S] center was found to
be pH-dependent. At pH 7, the equilibrium favors the hydroquinone and oxidized
[2Fe-2S] 2⫹ state of the two-electron reduced enzyme, whereas at pH 10 the flavin
semiquinone radical and reduced [2Fe-2S] ⫹ state are favored. Spectroelectro-
chemical studies substantiated this pH dependence by showing that the redox
potentials of both the FAD and the [2Fe-2S] center change with respect to pH
[254]. Importantly, the midpoint potential of the FAD was dramatically altered
by pH, changing from ⫺212 mV at pH 7.5 to ⫺273 mV at pH 8.4 versus a shift
of ⫺257 to ⫺279 mV for the [2Fe-2S] center under the same conditions. The
ionizable group responsible for the pH dependence is estimated to have a pK a
of 7.3 by the stopped-flow studies [253], and N-1 of the flavin is likely the protic
position in terms of work with other flavoenzymes [255]. Overall, the proposed
electron transport sequence is consistent with the role of E 3 as a 2e ⫺ /1e ⫺ switch,
and the evidence provides compelling support for a radical mechanism in the
C-3 deoxygenation of 51 (Fig. 20). In light of the fact that PMP–glucoseen adduct
TM

52 is the ultimate acceptor receiving electrons from E 3, the catalytic role of E 3
in the biosynthesis of ascarylose clearly constitutes a novel example of biological
deoxygenation.
B. Mechanism of the Coupled Deoxygenation Reaction

On the basis of the aforementioned evidence, a proposed mechanism of electron
transfer in the E 1 –E 3 coupled reaction is presented in Fig. 21. First, NADH binds
to E 3 and forms a charge-transfer complex with the oxidized FAD (Int-I) [253].
Chemical modification studies have identified Cys296 as a possible residue that
participates in the stabilization of this charge-transfer complex [251]. The bound
NADH then transfers a hydride and reduces the FAD to the hydroquinone form
(Int-II). Subsequently, electrons are shuttled one at a time from the hydroquinone
through the [2Fe-2S] centers of both E 3 and E 1 to reduce the PMP–glucoseen
intermediate bound in the E 1 active site as a Schiff base (Int-III to Int-VI). Such
Figure 21 Proposed electron-transfer pathway in the E 1 –E 3 coupled reaction. FAD,

flavin adenine dinucleotide; NAD⫹, nicotinamide-adenine dinucleotide; NADH, reduced
NAD⫹; Int, intermediate. (Source: Adapted from Ref. 256.)
TM

CEO Bond
a one-electron transfer process can be expected to produce transient radicals, and

two kinetically competent organic radicals formed during this process have been
observed by using both stopped-flow spectrophotometry and freeze-quench EPR
spectroscopy [253,256]. The first radical is clearly the flavin semiquinone dis-
cussed in the previous section; the nature of the second radical is less well charac-
terized. Nonetheless, the available evidence supports the assignment of the sec-
ond radical as a phenoxyl radical having its unpaired spin localized mainly on
the C-3 oxygen of the PMP in the PMP–substrate Schiff base [256]. After the
second electron transfer, the resulting two-electron reduced Schiff base is hy-
drolyzed to release the product and end one catalytic cycle (Int-VI to Int-VII).
It should be pointed out that during the first turnover two additional NADH mole-
cules are needed to prime the enzyme complex for steady-state catalysis (cycling
among Int-V, Int-VI, and Int-VII). Such priming is common for redox systems
like P-450 enzymes and dioxygenases.
C. Comparative Analysis of the Catalytic Role

of Each Cofactor
1. The FAD Cofactor
In the FNR family of reductases, the flavin cofactor generally serves as a switch
between one- and two-electron chemistry [257]. As a member of this family, E 3
acts in the same manner by shuttling electrons from NADH, a known two-elec-
tron donor, to the E 1 and E 3 [2Fe-2S] centers, which are single-electron acceptors.
However, the flavin in other enzyme systems involved with deoxygenation reac-
tions mentioned in this review appears to act directly as a catalytic redox center
rather than as part of an electron shuttle. For example, the enzyme 4-hydroxybu-
tyryl-CoA dehydratase (Section II.D.3.a) utilizes an FAD cofactor to store either
a hydride or a single electron from the substrate transiently, but the cofactor is
regenerated during the catalytic cycle with no net change in oxidation state. Simi-
larly, other enzymes in anaerobic fermentation (Section II.D.3.b) and chorismate
synthase (Section II.D.3.c) all leave the flavin cofactor in the same oxidation
state at the end of each turnover as in the beginning. On the other hand, one
equivalent of NADH must be consumed by E 3 to reduce the FAD for another
round of turnover, as the reducing equivalents are not catalytically recycled. In
this role, the function of E 3 more closely resembles that of the thioredoxin reduc-
tase and/or glutaredoxin reductase that participates in the regeneration of the
redox-active thiol pair in RNR at the end of each catalytic cycle. However, E 3
mediates an electron transfer to reduce the PMP–glucoseen intermediate 52 to
give the product 54, not to regenerate E 1 after 54 has formed (Fig. 20). Thus,
there is sufficient evidence to indicate that the roles of the flavin cofactor in E 3
and other flavin-dependent hydrolases are markedly different.
TM

2. The [2Fe-2S] Center Cofactors
Primarily, the metal cofactors in deoxygenation reactions catalyzed by enzymes
like carbonic anhydrase and enolase (Section II.B) serve as Lewis acids to polar-
ize the C-O bond and facilitate the departure of the hydroxyl group. The labile
iron atom in the [4Fe-4S] center of the aconitase family of enzymes (Section
II.C) acts in an identical manner. Whether the [4Fe-4S] clusters of the bacterial
anaerobic fermentation enzymes (Section II.D.3) participate as a Lewis acid or
as an important structural element (or both) remains to be established. Nonethe-
less, it is clear that these cofactors are not involved with electron transfer in
the reaction. A notable exception to this group of oxygen-sensitive iron–sulfur
enzymes is the [2Fe-2S] center, containing spinach dihydroxyacid dehydratase
(Section II.C.2), which can function in the presence of oxygen. However, the
catalytic role of this [2Fe-2S] center seems similar to that of the iron–sulfur
center of aconitase. On the contrary, the [2Fe-2S] centers of E 1 and E 3 transfer
electrons in a redox reaction (Section III.A). Removing this cofactor does not
significantly affect the rate at which E 1 forms the dehydrated glucoseen intermedi-
ate 52 (Fig. 20), so the [2Fe-2S] center does not appear to act as a Lewis acid
to facilitate loss of the hydroxyl group. Although the function of the iron–sulfur
clusters of E 1 and E 3 as part of an electron relay is well precedented, the synergetic
involvement of the [2Fe-2S] centers and PMP in redox chemical processes distin-
guishes this deoxygenation from other C-O bond cleavage events mediated by
iron–sulfur centers.
3. The Pyridoxamine 5′-Phosphate Cofactor

The B 6 coenzymes play an essential role in a wide array of biological transforma-
tions that include deamination, transamination, decarboxylation, racemization, β-
and γ-elimination/substitution, transsulfurization, among others. In these reac-
tions, the coenzyme exists in the PLP form, except transaminases, which also
utilize PMP as a result of the interconversion between these two forms of the
coenzyme during catalysis. Significantly, E 1 is the best characterized enzyme
that catalyzes a reaction using PMP without oscillating with PLP. Though the
mechanistic characterization is in the early stages, a PMP-dependent dehydratase
in the biosynthetic pathway of the nucleoside antibiotic blasticidin S may also
be a member of this class [258]. Other hydrolases that employ coenzyme B 6 as
a cofactor, such as serine and threonine dehydratase (Section II.D.1), are all PLP-
dependent, and their reactions involve the formation and stabilization of an α-
anion typical for β-eliminations catalyzed by this class of enzymes. The fact that
the PMP cofactor in E 1 directly participates not only in the β-elimination but
also in the subsequent electron transfer reduction via a radical mechanism clearly
places E 1 into a class of its own.
TM

CEO Bond
It is noteworthy that the mechanism of lysine-2,3-aminomutase, another

coenzyme B 6 –dependent enzyme, also includes an organic radical intermediate
[259–261]. In this reaction (Fig. 22), homolytic cleavage of AdoMet generates
a 5′-deoxyadenosyl radical that abstracts a hydrogen atom from the lysine sub-
strate (55) and initiates the radical rearrangment via the PLP cofactor to give 56.
Though this reaction demonstrates that a B 6 coenzyme can participate in radical
reactions, it is quite different from the reaction catalyzed by E 1. Rather than being
generated by a free radical–mediated hydrogen atom abstraction of the substrate,
the radical in E 1 –E 3 catalysis is generated by direct single-electron reduction
of the highly conjugated cofactor–substrate complex 52 (detailed in Fig. 23).
Presumably, the PMP–glucoseen Schiff base 52 tautomerizes to the quinone
methide species 57 before the facile reduction of this reactive quinone methide
Figure 22 Proposed mechanism for the radical-mediated rearrangement catalyzed by

lysine-2,3-aminomutase. (Source: Adapted from Ref. 259.)
TM

Figure 23 Detailed electron-transfer mechanism proposed for the E 1 –E 3 coupled reac-
tion. (Source: Adapted from Ref. 256.)
[262,263]. Some related enzymes involved in the biosynthesis of deoxysugars

other than ascarylose [111] have been speculated to take advantage of this inher-
ent redox capability to catalyze similar deoxygenations.
IV. CONCLUSIONS
In this review, we have illustrated the unique nature of the C-O bond cleavage
event catalyzed by the E 1 –E 3 coupled enzyme system. These enzymes recruit
several cofactors to achieve a common biological transformation, albeit via a
novel mechanism. The PMP cofactor is directly responsible for the reversible
dehydration catalyzed by E 1. Electrons are then transferred from the biological
reducing agent NADH through an FAD cofactor and two [2Fe-2S] centers of
E 1 and E 3 to drive the deoxygenation to completion and regenerate PMP. This
TM

CEO Bond
generation of a deoxysugar via radical intermediates is reminiscent of the ribonu-

cleotide reductase reaction (Section II.E.2). However, the details of the mecha-
nisms are quite distinct. Whereas a catalytic protein radical directly oxidizes the
substrate to create a substrate radical in RNR, no radical seems to form on a
protein residue in E 1, and the transient radical is generated by single-electron
reduction instead of oxidation. Interestingly, the characteristics of the proposed
quinone methide that undergoes single-electron chemical processes is somewhat
analogous to the catalytic properties of topa quinone, tryptophan tryptophylqui-
none, and pyrroloquinoline quinone [264]. This analogy opens the possibility that
B 6 cofactors may also exhibit chemistry characteristic of these newly discovered
quinone coenzymes. With a novel mechanism and these intriguing hypotheses,
E 1 may be the prototype for a new class of coenzyme B 6 –dependent enzymes
that use an alternate chemistry to catalyze transformations not normally expected
for this cofactor.
This review also summarizes the current knowledge on characterized en-
zymes that carry out deoxygenation reactions. The timeliness of this effort is
evident in the large number of recent advances in our understanding of these
important catalysts. This transformation takes place in such diverse biological
processes as energy metabolism, DNA replication, and drug biosynthesis. Not
surprisingly, for the remarkable breadth of compounds that undergo deoxygen-
ation, several mechanistic themes have evolved to achieve this transformation.
The various themes, ranging from no cofactor requirements to protein radical
involvement, seem to be finely tuned toward the reactivity of the substrates,
though transition-state stabilization by binding in the active site unquestionably
plays an important role in enzyme efficiency [89]. Unfortunately, it is difficult
to establish a meaningful correlation between the reactivity of the substrate and
the possible mechanism of a given deoxygenation because of the variability in
active-site topology and a lack of detailed thermodynamic information on all the
reactions. This difficulty is highlighted by the fact that even enzymes that catalyze
the same reaction, such as deoxynucleotide biosynthesis (Section II.E.2), may
exhibit subtle and sometimes glaring mechanistic differences. These differences
are a testament to the evolutionary diversity of biological CEO bond cleavage
events, and they will continue to challenge and inspire the ingenuity of enzymolo-
gists well into the future. As more enzyme systems become characterized, it will
be interesting to see what other novel deoxygenations may be discovered.
REFERENCES
1. P. Gonzalez-Porqué, in Vitamin B 6 Pyridoxal Phosphate, Chemical, Biochemical,

and Medical Aspects, Vol. I, Part B (D. Dolphin, R. Poulson, and O. Agramovic,
eds.) Wiley Interscience, New York, p. 391 (1986).
TM

2. G. Müller-Newen and W. Stoffel, Biol. Chem. Hoppe Seyler, 372: 613 (1991).
3. J. R. Stern, A. Del Campillo, and J. Raw, J. Am. Chem. Soc., 75: 2277 (1953).
4. J. B. Bahnson and V. E. Anderson, Biochemistry, 30: 5894 (1991).
5. P. Willadsen and H. Eggerer, Eur. J. Biochem., 54: 247 (1975).
6. J. A. Gerlt and P. G. Gassman, J. Am. Chem. Soc., 114: 5928 (1992).
7. J. A. Gerlt and P. G. Gassman, J. Am. Chem. Soc., 115: 11943 (1993).
8. R. L. D’Ordine, B. J. Bahnson, P. J. Tonge, and V. E. Anderson, Biochemistry,
33: 14733 (1994).
9. R. L. D’Oridine, P. J. Tonge, P. R. Carey, and V. E. Anderson, Biochemistry, 33:
12635 (1994).
10. G. Müller-Newen, U. Janssen, and W. Stoffel, Eur. J. Biochem., 228: 68 (1995).
11. S.-Y. Yang, and H. Schulz, J. Biol. Chem., 258: 9780 (1983).
12. S.-Y. Yang, J. Li, X.-Y. He, S. D. Cosloy, and H. Schulz, J. Bacteriol., 170: 2543
(1988).
13. X.-Y. H. Yang, H. Schulz, M. Elzinga, and S.-Y. Yang, Biochemistry, 30: 6788
(1991).
14. S.-Y. Yang and M. Elzinga, J. Biol. Chem., 268: 6588 (1993).
15. S.-Y. Yang, X.-Y. He, and H. Schulz, Biochemistry, 34: 6441 (1995).
16. K. Bloch, in The Enzymes, 3rd ed., Vol. 5 (P. D. Boyer, ed.) Acacemic Press, New
York, p. 441 (1971).
17. J. M. Schwab, A. Habib, and J. B. Klassen, J. Am. Chem. Soc., 108: 5304 (1986).
18. J. M. Schwab and J. B. Klassen, J. Am. Chem. Soc., 106: 7217 (1984).
19. K. R. Hanson and I. A. Rose, Acc. Chem. Res., 8: 1 (1975).
20. D. J. H. Brock, L. R. Kass, and K. Bloch, J. Biol. Chem., 242: 4432 (1967).
21. J. M. Schwab, C.-K. Ho, W.-b. Li, C. A. Townsend, and G. M. Salituro, J. Am.
Chem. Soc., 108: 5309 (1986).
22. J. E. Cronan, Jr., W.-B. Li, R. Coleman, M. Narasimhan, D. de Mendoza, and J. M.
Schwab, J. Biol. Chem., 263: 4641 (1988).
23. R. R. Annand, J. F. Kozlowski, V. J. Davisson, and J. M. Schwab, J. Am. Chem.
Soc., 115: 1088 (1993).
24. M. Leesong, B. S. Henderson, J. R. Gillig, J. M. Schwab, and J. L. Smith, Structure,
4: 253 (1996).
25. A. K. Joshi and S. Smith, J. Biol. Chem., 268: 22508 (1993).
26. A. K. Joshi and S. Smith, Biochem. J., 296: 143 (1993).
27. R. Bentley, Crit. Rev. Biochem. Mol. Biol., 25: 307 (1990).
28. E. Haslam, in Shikimic Acid: Metabolism and Metabolites, John Wiley & Sons,
Chichester, N. Y. (1993).
29. C. Kleanthous, R. Deka, K. Davis, S. M. Kelly, A. Cooper, S. E. Harding, N. C.
Price, A. R. Hawkins, and J. R. Coggins, Biochem. J., 282: 687 (1992).
30. C. W. G. Boys, S. M. Fawcett, L. Sawyer, J. D. Moore, I. G. Charles, A. R.
Hawkins, R. Deka, C. Kleanthous, and J. R. Coggins, J. Mol. Biol., 227: 352 (1992).
31. D. G. Gourley, J. R. Coggins, N. W. Isaacs, J. D. Moore, I. G. Charles, and A. R.
Hawkins, J. Mol. Biol., 241: 488 (1994).
32. J. Harris, C. Kleanthous, J. R. Coggins, A. R. Hawkins, and C. Abell, J. Chem.
Soc. Chem. Commun., 1080 (1993).
TM

CEO Bond
33. S. Chaudhuri, K. Duncan, L. D. Graham, and J. R. Coggins, Biochem. J., 275: 1

(1991).
34. A. Shneier, C. Kleanthous, R. Deka, J. R. Coggins, and C. Abell, J. Am. Chem.
Soc., 113: 9416 (1991).
35. R. K. Deka, C. Kleanthous, and J. R. Coggins, J. Biol. Chem., 267: 22237 (1992).
36. A. P. Leech, R. James, J. R. Coggins, and C. Kleanthous, J. Biol. Chem., 270:
25827 (1995).
37. A. Shneier, J. Harris, C. Kleanthous, J. R. Coggins, A. R. Hawkins, and C. Abell,
Bioorg. Med. Chem. Lett., 3: 1399 (1993).
38. T. Krell, A. R. Pitt, and J. R. Coggins, FEBS Lett., 360: 93 (1995).
39. I. Bertini, C. Luchinat, W. Maret, and M. Zeppezauer, eds., in Zinc Enzymes, Birk-
hauser, Boston (1986).
40. W. S. Sly and P. Y. Hu, Annu. Rev. Biochem., 64: 375 (1995).
41. A. E. Eriksson and A. Liljas, in The Carbonic Anhydrases: Cellular Physiology
and Molecular Genetics (S. J. Dodgson, R. E. Tashian, G. Gros, and N. D. Carter,
eds.) Plenum Press, New York, p. 33 (1991).
42. D. N. Silverman and S. Lindskog, Acc. Chem. Res., 21: 30 (1988).
43. K. M. Merz, Jr., J. Mol. Biol., 214: 799 (1990).
44. Z. W. Liang, Y. F. Xue, G. Behravan, B. H. Jonsson, and S. Lindskog, Eur. J.
Biochem., 211: 821 (1993).
45. C. K. Tu, D. N. Silverman, C. Forsman, B. H. Jonsson, and S. Lindskog, Biochemis-
try, 28: 7913 (1989).
46. Y. Pocker and N. Janjic, J. Am. Chem. Soc., 111: 731 (1989).
47. R. S. Rowlett, M. R. Chance, M. D. Wirt, D. E. Sidelinger, J. R. Royal, M.
Woodroffe, Y.-F. A. Wang, R. P. Saha, and M. G. Lam, Biochemistry, 33: 13967
(1994).
48. F. Wold, in The Enzymes, 3rd ed., Vol. 5 (P. D. Boyer, ed.) Academic Press, New
York, p. 499 (1971).
49. J. M. Brewer, CRC Crit. Rev. Biochem., 11: 209 (1981).
50. F. Wold and C. E. Ballou, J. Biol. Chem., 227: 313 (1957).
51. L. D. Faller, B. M. Baroudy, A. M. Johnson, and R. X. Ewall, Biochemistry, 16:
3864 (1977).
52. B. H. Lee and T. Nowak, Biochemistry, 31: 2165 (1992).
53. M. J. Kornblatt, Arch. Biochem. Biophys., 330: 12 (1996).
54. E. C. Dinovo and P. D. Boyer, J. Biol. Chem., 246: 4586 (1971).
55. S. R. Anderson, V. E. Anderson, and J. R. Knowles, Biochemistry, 33: 10545
(1994).
56. R. R. Poyner, L. T. Laughlin, G. A. Sowa, and G. H. Reed, Biochemistry, 35: 1692
(1996).
57. J. Stubbe and R. H. Abeles, Biochemistry, 19: 5505 (1980).
58. V. E. Anderson, P. M. Weiss, and W. W. Cleland, Biochemistry, 23: 2779 (1984).
59. V. E. Anderson and W. W. Cleland, Biochemistry, 29: 10498 (1990).
60. M. Cohen, J. E. Pearson, E. L. O’Connell, and I. A. Rose, J. Am. Chem. Soc., 92:
4095 (1970).
61. L. Lebioda and B. Stec, Biochemistry, 30: 2817 (1991).
TM

62. E. Zhang, M. Hatada, J. M. Brewer, and L. Lebioda, Biochemistry, 33: 6295 (1994).
63. V. S. Sangadala, C. V. C. Glover, R. L. Robson, M. J. Holland, L. Lebioda, and
J. M. Brewer, Biochim. Biophys. Acta, 1251: 23 (1995).
64. S. Duquerroy, C. Camus, and J. Janin, Biochemistry, 34: 12513 (1995).
65. J. E. Wedekind, R. R. Poyner, G. H. Reed, and I. Rayment, Biochemistry, 33: 9333
(1994).
66. T. M. Larsen, J. E. Wedekind, I. Rayment, and G. H. Reed, Biochemistry, 35: 4349
(1996).
67. M. E. Lee and T. Nowak, Arch. Biochem. Biophys., 293: 264 (1992).
68. S. H. Hilal, J. M. Brewer, L. Lebioda, and L. A. Carreira, Biochem. Biophys. Res.
Comm., 211: 607 (1995).
69. J. Deacon and R. A. Cooper, FEBS Lett., 77: 201 (1977).
70. P. C. Babbitt, G. T. Mrachko, M. S. Hasson, G. W. Huisman, R. Kolter, D. Ringe,
G. A. Petsko, G. L. Kenyon, and J. A. Gerlt, Science, 267: 1159 (1995).
71. A. R. Parker, J. A. Moore, J. M. Schwab, and V. J. Davisson, J. Am. Chem. Soc.,
117: 10605 (1995).
72. H. Beinert and M. C. Kennedy, Eur. J. Biochem., 186: 5 (1989).
73. M. C. Kennedy and C. D. Stout, Adv. Inorg. Chem., 38: 323 (1992).
74. H. Beinert and M. C. Kennedy, FASEB J., 7: 1442 (1993).
75. M. C. Kennedy, M. H. Emptage, J.-L. Dreyer, and H. Beinert, J. Biol. Chem., 258:
11098 (1983).
76. H. Beinert, M. H. Emptage, J.-L. Dreyer, R. A. Scott, J. E. Hahn, K. O. Hodgson,
and A. J. Thomson, Proc. Natl. Acad. Sci. USA, 80: 393 (1983).
77. M. H. Emptage, J.-L. Dreyer, M. C. Kennedy, and H. Beinert, J. Biol. Chem., 258:
11106 (1983).
78. M. H. Emptage, T. A. Kent, M. C. Kennedy, H. Beinert, and E. Münck, Proc. Natl.
Acad. Sci. USA, 80: 4674 (1983).
79. D. W. Plank, M. C. Kennedy, H. Beinert, and J. B. Howard, J. Biol. Chem., 264:
20385 (1989).
80. D. W. Plank and J. B. Howard, J. Biol. Chem., 263: 8184 (1988).
81. A. H. Robbins and C. D. Stout, Proteins Struct. Funct. Genet., 5: 289 (1989).
82. H. Lauble, M. C. Kennedy, H. Beinert, and C. D. Stout, Biochemistry, 31: 2735
(1992).
83. M. M. Werst, M. C. Kennedy, H. Beinert, and B. M. Hoffman, Biochemistry, 29:
10526 (1990).
84. M. M. Werst, M. C. Kennedy, A. L. P. Houseman, H. Beinert, and B. M. Hoffman,
Biochemistry, 29: 10533 (1990).
85. L. Zheng, M. C. Kennedy, H. Beinert, and H. Zalkin, J. Biol. Chem., 267: 7895
(1992).
86. J. V. Schloss, D. J. T. Porter, J. J. Bright, and W. W. Cleland, Biochemistry, 19:
2358 (1980).
87. H. Lauble and C. D. Stout, Proteins, 22: 1 (1995).
88. D. A. Brant, L. B. Barnett, and R. A. Alberty, J. Am. Chem. Soc., 85: 2204 (1963).
89. S. L. Bearne and R. Wolfenden, J. Am. Chem. Soc., 117: 9588 (1995).
90. D. H. Flint, M. H. Emptage, and J. R. Guest, Biochemistry, 31: 10331 (1992).
TM

CEO Bond
91. S. A. Woods, S. D. Schwartzbach, and J. R. Guest, Biochim. Biophys. Acta, 954:

14 (1988).
92. H. Shibata, W. E. Gardiner, and S. D. Schwartzbach, J. Bacteriol., 164: 762 (1985).
93. M. M. Luque-Romero and F. Castillo, Arch. Microbiol., 155: 149 (1991).
94. J. S. Miles and J. R. Guest, Nucleic Acids Res., 12: 3631 (1984).
95. P. J. Bell, S. C. Andrews, M. N. Sivak, and J. R. Guest, J. Bacteriol., 171: 5928
(1989).
96. D. H. Flint, M. H. Emptage, M. G. Finnegan, W. Fu, and M. K. Johnson, J. Biol.
Chem., 268: 14732 (1993).
97. O. R. Brown and F. Yein, Biochem. Biophys. Res. Commun., 85: 1219 (1978).
98. O. R. Brown, E. Smyk-Randall, B. Draczynska-Lusiak, and J. A. Fee, Arch. Bio-
chem. Biophys., 319: 10 (1995).
99. D. H. Flint, E. Smyk-Randall, J. F. Tuminello, B. Draczynska-Lusiak, and O. R.
Brown, J. Biol. Chem., 268: 25547 (1993).
100. D. H. Flint and M. H. Emptage, J. Biol. Chem., 263: 3558 (1988).
101. M. C. Pirrung, C. P. Holmes, D. M. Horowitz, and D. S. Nunn, J. Am. Chem. Soc.,
113: 1020 (1991).
102. R. K. Hill, S. Yan, and S. M. Arfin, J. Am. Chem. Soc., 95: 7857 (1973).
103. R. Grabowski, and W. Bückel, Eur. J. Biochem., 199: 89 (1991).
104. A. E. M. Hofmeister, S. P. J. Albracht, and W. Bückel, FEBS Lett., 351: 416 (1994).
105. R. Grabowski, A. E. M. Hofmeister, and W. Bückel, Trends Biochem. Sci., 18:
297 (1993).
106. E. W. Miles, in Vitamin B 6 Pyridoxal Phosphate, Chemical, Biochemical, and Med-
ical Aspects, Vol. I, Part B (D. Dolphin, R. Poulson, and O. Agramovic, eds.) Wiley
Interscience, New York, p. 253 (1986).
107. H. Hayashi, J. Biochem., 118: 463 (1995).
108. A. E. M. Hofmeister, R. Grabowski, D. Linder, and W. Bückel, Eur. J. Biochem.,
215: 341 (1993).
109. E. W. Miles, Subcell. Biochem., 24: 207 (1995).
110. P. A. Frey, in Coenzymes and Cofactors: Pyridine Nucleotide Coenzymes, Vol. II,
Part B (D. Dolphin, R. Poulson, and O. Avramovic, eds.) Wiley Interscience, New
York, p. 461 (1987).
111. H.-w. Liu and J. S. Thorson, Annu. Rev. Microbiol., 48: 223 (1994).
112. H. Zarkowsky and L. Glaser, J. Biol. Chem., 244: 4750 (1970).
113. X. He, J. S. Thorson, and H.-w. Liu, Biochemistry, 35: 4721 (1996).
114. A. Melo, W. H. Elliott, and L. Glaser, J. Biol. Chem., 243: 1467 (1968).
115. O. Gabriel and L. C. Lindquist, J. Biol. Chem., 243: 1479 (1968).
116. S.-F. Wang and O. Gabriel, J. Biol. Chem., 245: 8 (1970).
117. C. E. Snipes, G.-U. Brillinger, L. Sellers, L. Mascaro, and H. G. Floss, J. Biol.
Chem., 252: 8113 (1977).
118. P. J. Oths, R. M. Mayer, and H. G. Floss, Carbohydr. Res., 198: 91 (1990).
119. Y. Yu, R. N. Russell, J. S. Thorson, L.-d. Liu, and H.-w. Liu, J. Biol. Chem., 267:
5868 (1992).
120. U. S. Maitra and D. B. Sprinson, J. Biol. Chem., 253: 5426 (1978).
121. S. L. Bender, T. Widlanski, and J. R. Knowles, Biochemistry, 28: 7560 (1989).
TM

122. N. Yamauchi and K. Kakinuma, J. Org. Chem., 60: 5614 (1995).
123. W. Bückel, Eur. J. Biochem., 106: 439 (1980).
124. U. Müller and W. Bückel, Eur. J. Biochem., 230: 698 (1995).
125. G. A. Molander and G. Hahn, J. Org. Chem., 51: 1135 (1986).
126. A.-G. Klees, D. Linder, and W. Bückel, Arch. Microbiol., 158: 294 (1992).
127. R. D. Kuchta and R. H. Abeles, J. Biol. Chem., 260: 13181 (1985).
128. R. D. Kuchta, G. R. Hanson, B. Holmquist, and R. H. Abeles, Biochemistry, 25:
7301 (1986).
129. A. E. M. Hofmeister, and W. Bückel, Eur. J. Biochem., 206: 547 (1992).
130. P. Willadsen and W. Bückel, FEMS Microbiol. Lett., 70: 187 (1990).
131. W. Bückel, FEMS Microbiol. Rev., 88: 211 (1992).
132. U. Scherf and W. Bückel, Eur. J. Biochem., 215: 421 (1993).
133. U. Müh, I. Çinkaya, S. P. J. Albracht, and W. Bückel, Biochemistry, 35: 11710
(1996).
134. W. Bückel and R. Keese, Angew. Chem. Int. Ed. Engl., 34: 1502 (1995).
135. H. Morell, M. J. Clark, P. F. Knowles, and D. B. Sprinson, J. Biol. Chem., 242:
82 (1967).
136. G. R. Welch, K. W. Cole, and F. H. Gaertner, Arch. Biochem. Biophys., 165: 505
(1974).
137. P. J. White, G. Millar, and J. R. Coggins, Biochem. J., 251: 313 (1988).
138. M. K. Ramjee, J. R. Coggins, T. R. Hawkes, D. J. Lowe, and R. N. F. Thorneley,
Bioorg. Med. Chem. Lett., 3: 1409 (1993).
139. A. Schaller, M. van Afferden, V. Windhofer, S. Bülow, G. Abel, J. Schmid, and
N. Amrhein, Plant Physiol., 97: 1271 (1991).
140. C. T. Lauhon and P. A. Bartlett, Biochemistry, 33: 14100 (1994).
141. S. Bornemann, J. R. Coggins, D. J. Lowe, and R. N. F. Thorneley, in Perspectives
on Protein Engineering and Complementary Technologies (M. J. Geisow, ed.),
Mayflower Worldwide, Birmingham, p. 134 (1995).
142. M. N. Ramjee, J. R. Coggins, T. R. Hawkes, D. J. Lowe, and R. N. F. Thorneley,
J. Am. Chem. Soc., 113: 8566 (1991).
143. S. Bornemann, M. N. Ramjee, D. J. Lowe, R. N. F. Thorneley, J. R. Coggins, C.
Abell, S. Balasubramanian, T. R. Hawkes, W. W. Nichols, and G. M. Davies, in
Flavins and Flavoproteins 1993 (K. Yagi, ed.), Water de Gruyter, Berlin, p. 843
(1994).
144. R. K. Hill and G. R. Newkome, J. Am. Chem. Soc., 91: 5893 (1969).
145. D. K. Onderka and H. G. Floss, J. Am. Chem. Soc., 91: 5894 (1969).
146. H. G. Floss, D. K. Onderka, and M. Carroll, J. Biol. Chem., 247: 736 (1972).
147. R. K. Hill and M. G. Bock, J. Am. Chem. Soc., 100: 637 (1978).
148. E. Toromanoff, C. R. Seances Acad. Sci. Ser. C, 290: 81 (1980).
149. K. Fukui, Tetrahedron Lett., 2427 (1965).
150. N. T. Anh, J. Chem. Soc. Chem. Commun., 1089 (1968).
151. C. T. Walsh, J. Liu, F. Rusnak, and M. Sakaitani, Chem. Rev., 90: 1105 (1990).
152. S. Bornemann, D. J. Lowe, and R. N. F. Thorneley, Biochem. Soc. Trans., 24: 84
(1996).
153. S. Balasubramanian, J. R. Coggins, and C. Abell, Biochemistry, 34: 341 (1995).
TM

CEO Bond
154. T. R. Hawkes, T. Lewis, J. R. Coggins, D. M. Mousdale, D. J. Lowe, and R. N. F.

Thorneley, Biochem. J., 265: 899 (1990).
155. P. A. Bartlett, U. Maitra, and P. M. Chouinard, J. Am. Chem. Soc., 108: 8068
(1986).
156. S. Bornemann, S. Balasubramanian, J. R. Coggins, C. Abell, D. J. Lowe, and
R. N. F. Thorneley, Biochem. J., 305: 707 (1995).
157. S. Bornemann, M. K. Ramjee, S. Balasubramanian, C. Abell, J. R. Coggins, D. J.
Lowe, and R. N. F. Thorneley, J. Biol. Chem., 270: 22811 (1995).
158. S. Bornemann, D. J. Lowe, and R. N. F. Thorneley, Biochemistry, 35: 9907 (1996).
159. M. N. Ramjee, S. Balasubramanian, C. Abell, J. R. Coggins, G. M. Davies, T. R.
Hawkes, D. J. Lowe, and R. N. F. Thorneley, J. Am. Chem. Soc., 114: 3151 (1992).
160. J. M. Pratt, Pure Appl. Chem., 65: 1513 (1993).
161. R. H. Abeles and D. Dolphin, Acc. Chem. Res., 9: 114 (1976).
162. T. Toraya, in Metal Ions in Biological Systems, Vol. 30 (H. Sigel and A. Sigel,
eds.) Marcel Dekker, New York, p. 217 (1994).
163. M. K. Essenberg, P. A. Frey, and R. H. Abeles, J. Am. Chem. Soc., 93: 1242 (1971).
164. J. Halpern, Science, 227: 869 (1985).
165. B. P. Hay and R. G. Finke, J. Am. Chem. Soc., 108: 4820 (1986).
166. H. A. Lee, Jr. and R. H. Abeles, J. Biol. Chem., 238: 2367 (1963).
167. T. Toraya, Y. Sugimoto, Y. Tamao, S. Shimizu, and S. Fukui, Biochemistry, 10:
3475 (1971).
168. C. L. Drennen, S. Huang, J. T. Drummond, R. G. Matthews, and M. L. Ludwig,
Science, 266: 1669 (1994).
169. F. Mancia, N. H. Keep, A. Nakagawa, P. F. Leadlay, S. McSweeney, B. Rasmussen,
P. Bösecke, O. Diat, and P. R. Evans, Structure, 4: 339 (1996).
170. M. L. Ludwig, C. L. Drennan, and R. G. Matthews, Structure, 4: 505 (1996).
171. B. M. Babior, BioFactors., 1: 21 (1988).
172. S. A. Cockle, H. A. O. Hill, R. J. P. Williams, S. P. Davies, and M. A. Foster, J.
Am. Chem. Soc., 94: 275 (1972).
173. T. H. Finlay, J. Valinsky, A. S. Mildvan, and R. H. Abeles, J. Biol. Chem., 248:
1285 (1973).
174. J. E. Valinsky, R. H. Abeles, and A. S. Mildvan, J. Biol. Chem., 249: 2751 (1974).
175. J. E. Valinsky, R. H. Abeles, and J. A. Fee, J. Am. Chem. Soc., 96: 4709 (1974).
176. R. H. Abeles and B. Zagalak, J. Biol. Chem., 241: 1245 (1966).
177. P. A. Frey, M. K. Essenberg, and R. H. Abeles, J. Biol. Chem., 242: 5369 (1967).
178. P. A. Frey and R. H. Abeles, J. Biol. Chem., 241: 2732 (1966).
179. P. A. Frey, S. S. Kerwar, and R. H. Abeles, Biochem. Biophys. Res. Commun., 29:
873 (1967).
180. P. A. Frey, M. K. Essenberg, R. H. Abeles, and S. S. Kerwar, J. Am. Chem. Soc.,
92: 4488 (1970).
181. J. Rétey, A. Umani-Rouchi, J. Seibl, and D. Arigoni, Experientia, 22: 502 (1966).
182. B. Zagalak, P. A. Frey, G. L. Karabatsos, and R. H. Abeles, J. Biol. Chem., 241:
3028 (1966).
183. A. M. Brownstein and R. H. Abeles, J. Biol. Chem., 236: 1199 (1961).
184. W. W. Cleland, Crit. Rev. Biochem., 13: 385 (1982).
TM

185. S. Kuno, T. Toraya, and S. Fukui, Arch. Biochem. Biophys., 210: 474 (1981).
186. S. L. Tan, M. G. Kopozynski, W. W. Bachovchin, W. H. Orme-Johnson, and
B. M. Babior, J. Biol. Chem., 261: 3483 (1986).
187. R. J. O’Brien, J. A. Fox, M. G. Kopczynski, B. M. Babior, J. Biol. Chem., 260:
16131 (1985).
188. S. Licht, G. J. Gerfen, and J. Stubbe, Science, 271: 477 (1996).
189. J. Stubbe, Annu. Rev. Biochem., 58: 257 (1989).
190. J. Stubbe, Adv. Enzymol. Relat. Areas Mol. Biol., 63: 349 (1990).
191. P. Reichard, J. Biol. Chem., 268: 8383 (1993).
192. P. Reichard, Science, 260: 1773 (1993).
193. M. Fontecave, P. Nordlund, H. Eklund, and P. Reichard, Adv. Enzymol. Relat. Areas
Mol. Biol., 65: 147 (1992).
194. S. Booker, J. Broderick, and J. Stubbe, Biochem. Soc. Trans., 21: 727 (1993).
195. E. N. G. Marsh, BioEssays, 17: 431 (1995).
196. C. L. Atkin, L. Thelander, P. Reichard, and G. Lang, J. Biol. Chem., 248: 7464
(1973).
197. L. Petersson, A. Gräslund, A. Ehrenberg, B.-M. Sjöberg, and P. Reichard, J. Biol.
Chem., 255: 6706 (1980).
198. B.-M. Sjöberg, T. M. Loehr, and J. Sanders-Loehr, Biochemistry, 21: 96 (1982).
199. R. C. Scarrow, M. J. Maroney, S. M. Palmer, L. Que, A. L. Roe, S. P. Salowe,
and J. Stubbe, J. Am. Chem. Soc., 109: 7857 (1987).
200. G. Bunker, L. Petersson, B.-M. Sjöberg, M. Sahlin, M. Chance, B. Chance, and
A. Ehrenberg, Biochemistry, 26: 4708 (1987).
201. M. Sahlin, A. Gräslund, L. Petersson, A. Ehrenberg, and B.-M. Sjöberg, Biochemis-
try, 28: 2618 (1989).
202. J. B. Lynch, C. Juarez-Garcia, E. Münck, and L. Que, J. Biol. Chem., 264: 8091
(1989).
203. P. Nordlund, B.-M. Sjöberg, and H. Eklünd, Nature, 345: 593 (1990).
204. G. Lassmann, L. Thelander, and A. Gräslund, Biochem. Biophys. Res. Commun.,
188: 879 (1992).
205. J. Covès, L. Le Hir de Fallois, L. Le Pape, J.-L. Décout, and M. Fontecave, Bio-
chemistry, 35: 8595 (1996).
206. J. M. Bollinger, Jr., D. E. Edmondson, B. H. Huynh, J. Filley, J. R. Norton, and
J. Stubbe, Science, 253: 292 (1991).
207. N. Ravi, J. M. Bollinger, Jr., B. H. Huynh, D. E. Edmondson, and J. Stubbe, J.
Am. Chem. Soc., 116: 8007 (1994).
208. J. M. Bollinger, Jr., W. H. Tong, N. Ravi, B. H. Huynh, D. E. Edmondson, and J.
Stubbe, J. Am. Chem. Soc., 116: 8015 (1994).
209. J. M. Bollinger, Jr., W. H. Tong, N. Ravi, B. H. Huynh, D. E. Edmondson, and J.
Stubbe, J. Am. Chem. Soc., 116: 8024 (1994).
210. B. E. Sturgeon, D. Burdi, S. Chen, B.-H. Huynh, D. E. Edmondson, J. Stubbe, and
B. M. Hoffman, J. Am. Chem. Soc., 118: 7551 (1996).
211. F. Lendzian, M. Sahlin, F. MacMillan, R. Bittl, R. Fiege, S. Pötsch, B.-M. Sjöberg,
A. Gräslund, W. Lubitz, and G. Lassmann, J. Am. Chem. Soc., 118: 8111 (1996).
212. U. Uhlin and H. Eklünd, Nature, 370: 533 (1994).
213. S. S. Mao, G. X. Yu, D. Chalfoun, and J. Stubbe, Biochemistry, 31: 9752 (1992).
TM

CEO Bond
214. W. A. van der Donk, G. Yu, D. J. Silva, J. Stubbe, J. R. McCarthy, E. T. Jarvi,

D. P. Matthews, R. J. Resvick, and E. Wagner, Biochemistry, 35: 8381 (1996).
215. S. S. Mao, M. I. Johnston, J. M. Bollinger, and J. Stubbe, Proc. Natl. Acad. Sci.
USA, 86: 1485 (1989).
216. S. S. Mao, T. P. Holler, G. X. Yu, J. M. Bollinger, Jr., S. Booker, M. I. Johnston,
and J. Stubbe, Biochemistry, 31: 9733 (1992).
217. S. S. Mao, T. P. Holler, J. M. Bollinger, Jr., G. X. Yu, M. I. Johnston, and J. Stubbe,
Biochemistry, 31: 9744 (1992).
218. W. A. van der Donk, J. Stubbe, G. J. Gerfen, B. F. Bellew, and R. G. Griffin, J.
Am. Chem. Soc., 117: 8908 (1995).
219. R. L. Blakley and H. A. Barker, Biochim. Biophys. Res. Commun., 16: 301 (1964).
220. R. L. Blakley, in B12 , Vol. 2 (D. Dolphin, ed.), John Wiley & Sons, New York,
p. 381 (1982).
221. S. Booker, S. Licht, J. Broderick, and J. Stubbe, Biochemistry, 33: 12676 (1994).
222. G. J. Gerfen, S. Licht, J.-P. Willems, B. M. Hoffman, and J. Stubbe, J. Am. Chem.
Soc., 118: 8192 (1996).
223. R. Eliasson, M. Fontecave, H. Jörnvall, M. Krook, E. Pontis, and P. Reichard, Proc.
Natl. Acad. Sci. U.S.A., 87: 3314 (1990).
224. K. K. Wong, B. W. Murray, S. A. Lewisch, M. K. Baxter, T. W. Ridky, L. Ulissi-
DeMario, and J. W. Kozarich, Biochemistry, 32: 14102 (1993).
225. X. Sun, R. Eliasson, E. Pontis, J. Andersson, G. Buist, B.-M. Sjöberg, and P. Rei-
chard, J. Biol. Chem., 270: 2443 (1995).
226. S. Ollagnier, E. Mulliez, J. Gaillard, R. Eliasson, M. Fontecave, and P. Reichard,
J. Biol. Chem., 271: 9410 (1996).
227. E. Mulliez, M. Fontecave, J. Gaillard, and P. Reichard, J. Biol. Chem., 268: 2296
(1993).
228. M. Frey, M. Rothe, A. F. V. Wagner, and J. Knappe, J. Biol. Chem., 269: 12432
(1994).
229. C. V. Parast, K. K. Wong, S. A. Lewisch, and J. W. Kozarich, Biochemistry, 34:
2393 (1995).
230. C. V. Parast, K. K. Wong, J. W. Kozarich, J. Peisach, and R. S. Magliozzo, Bio-
chemistry, 34: 5712 (1995).
231. X. Sun, J. Harder, M. Krook, H. Jörnvall, B.-M. Sjöberg, and P. Reichard, Proc.
Natl. Acad. Sci. USA, 90: 577 (1993).
232. X. Sun, S. Ollagnier, P. P. Schmidt, M. Atta, E. Mulliez, L. Lepape, R. Eliasson,
A. Gräslund, M. Fontecave, P. Reichard, and B.-M. Sjöberg, J. Biol. Chem., 271:
6827 (1996).
233. R. Eliasson, P. Reichard, E. Mulliez, S. Ollagnier, M. Fontecave, E. Liepinsh, and
G. Otting, Biochem. Biophys. Res. Commun., 214: 28 (1995).
234. E. Mulliez, S. Ollagnier, M. Fontcave, R. Eliasson, P. Reichard, Proc. Natl. Acad.
Sci. USA, 92: 8759 (1995).
235. N. R. Williams and J. D. Wander, in The Carbohydrates: Chemistry and Biochemis-
try, Vol. 1B (W. Pigman and D. Horton, eds.), Academic Press, New York, p. 761
(1980).
236. C. T. Bishop and J. J. Jennings, in The Polysaccharides, Vol. 1 (G. O. Aspinall,
ed.) Academic Press, Orlando, Fla., p. 291 (1982).
TM

237. L. Kenne and B. Lindberg, in The Polysaccharides, Vol 2. (G. O. Aspinall, ed.),
Academic Press, New York, p. 287 (1983).
238. C. R. H. Raetz, Annu. Rev. Biochem., 59: 129 (1990).
239. O. Lüderitz, A. M. Staub, and O. Westphal, Bacteriol. Rev., 30: 192 (1966).
240. P. A. Rubenstein and J. L. Strominger, J. Biol. Chem., 249: 3776 (1974).
241. P. A. Rubenstein and J. L. Strominger, J. Biol. Chem., 249: 3782 (1974).
242. T. M. Weigel, L.-d. Liu, and H.-w. Liu, Biochemistry, 31: 2129 (1992).
243. Y. Lei, O. Ploux, and H.-w. Liu, Biochemistry, 34: 4643 (1995).
244. J. S. Thorson and H.-w. Liu, J. Am. Chem. Soc., 115: 7539 (1993).
245. T. M. Weigel, V. P. Miller, and H.-w. Liu, Biochemistry, 31: 2140 (1992).
246. P. A. Pieper, Z. Guo, and H.-w. Liu, J. Am. Chem. Soc., 117: 5158 (1995).
247. M. M. Palcic and H. G. Floss, in Vitamin B6 Pyridoxal Phosphate, Chemical, Bio-
chemical, and Medical Aspects, Vol. I, Part A (D. Dolphin, R. Poulson, and O.
Avramovic, eds.) Wiley Intersciece, New York, p. 25 (1986).
248. J. S. Thorson, S. F. Lo, H.-w. Liu, and C. R. Hutchinson, J. Am. Chem. Soc., 115:
6993 (1993).
249. J. S. Thorson, and H.-w. Liu, J. Am. Chem. Soc., 115: 12177 (1993).
250. V. P. Miller, J. S. Thorson, O. Ploux, S. F. Lo, and H.-w. Liu, Biochemistry, 32:
11934 (1993).
251. O. Ploux, Y. Lei, K. Vatanen, and H.-w. Liu, Biochemistry, 34: 4159 (1995).
252. S. F. Lo, V. P. Miller, Y. Lei, J. S. Thorson, H.-w. Liu, and J. L. Schottel, J.
Bacteriol., 176: 460 (1994).
253. G. T. Gassner, D. A. Johnson, H.-w. Liu, and D. P. Ballou, Biochemistry, 35: 7752
(1996).
254. K. D. Kim, P. A. Pieper, H.-w. Liu, and M. T. Stankovich, Biochemistry, 35: 7879
(1996).
255. F. Müller, J. Vervoort, C. P. M. van Mierlo, S. G. Mayhew, W. J. H. van Berkel,
and A. Bacher, in Flavins and Flavoproteins (D. E. Edmundson, and D. B. McCor-
mick, eds.), Walter de Gruyter, Berlin, p. 261 (1987).
256. D. A. Johnson, G. T. Gassner, V. Bandarian, F. J. Ruzicka, D. P. Ballou, G. H.
Reed, and H.-w. Liu, Biochemistry, 35: 15846 (1996).
257. P. A. Karplus, M. J. Daniels, and J. R. Herriott, Science, 251: 60 (1991).
258. S. J. Gould and J. Guo, J. Am. Chem. Soc., 114: 10176 (1992).
259. M. D. Ballinger, P. A. Frey, and G. H. Reed, Biochemistry, 31: 10782 (1992).
260. M. D. Ballinger, P. A. Frey, and G. H. Reed, Biochemistry, 34: 10086 (1995).
261. W. Wu, K. W. Lieder, G. H. Reed, and P. A. Frey, Biochemistry, 34: 10532 (1995).
262. H.-U. Wagner and R. Gompper, in The Chemistry of the Quinonoid Compounds
(S. Patai, ed.) John Wiley & Sons, New York, p. 1145 (1974).
263. K. Karabelas and H. W. Moore, J. Am. Chem. Soc., 112: 5372 (1990).
264. J. P. Klinman and D. Mu, Annu. Rev. Biochem., 63: 299 (1994).
TM

14
The Molecular Mechanism of
Amyloidosis in Alzheimer’s Disease
Michael G. Zagorski
Case Western Reserve University, Cleveland, Ohio
I. INTRODUCTION
Amyloidosis is the process in which normally innocuous, soluble proteins deposit

as insoluble amyloid fibrils, resulting in organ dysfunction and death [1]. This
condition is generally a disease of aging and accompanies many widespread med-
ical ailments such as cancer, rheumatoid arthritis, Alzheimer’s disease (AD),
chronic renal dialysis, familial amyloid polyneuropathy, spongiform encephalop-
athies, maturity-onset diabetes, and systemic amyloidosis. The amyloid deposits
are characterized by distinct tinctorial properties and fibrils of 7–10 nm in diame-
ter, in which the primary components are aggregated proteins with antiparallel
cross-β-pleated sheet structures [2,3].
In AD, the major protein constituent of the amyloid deposits is the ‘‘β-
peptide,’’ which varies in length and consists of 39–43 amino acids. There are
also other proteins (presently 15 known) that share the ability to form insoluble
amyloid fibrils, which include the amyloid A fragment, amylin, cystatin C, trans-
thyretin, prion, and related fragments. The amyloid proteins and their deposits
tend to be organ- and disease-specific. For example, the protein amylin is local-
ized on the pancreas of patients with maturity-onset diabetes whereas the β-pep-
tide deposits mainly in the brain of AD patients. Except for their ability to aggre-
gate into β-sheet amyloid fibrils, the majority of these proteins do not share any
similarities in their primary amino acid sequence or their presumed biological
functions. Most experts agree that the production of amyloid is harmful and may
play an important role in the disease processes. During amyloidosis, monomeric
proteins bind together, by way of hydrogen or electrostatic bonding interactions,
TM

producing dimeric, trimeric, tetrameric, and larger oligomeric species. Eventu-
ally, a critical aggregated size that ultimately precipitates as amyloid is reached.
It is also thought that the mechanism of amyloidosis for all amyloid proteins
involves a common pathway [4,5]. One idea, which was formulated by us, in-
vokes the conversion of soluble, nontoxic α-helical structures into toxic, oligo-
meric β-sheet structures. These transformations occur with the prion protein [6]
and the amyloid β-peptide of AD [7,8].
The β-peptide results from the processing of a larger amyloid precursor
protein (APP) [9–11] (Fig. 1). The biological functions of both the APP and the
β-peptide are currently unknown, although they are believed to play roles in
neuronal homeostasis, cell adhesion, G-protein coupling, and/or oxidative stress.
In humans, soluble β-peptide is ubiquitous in biological fluids [12–15], and AD
patients also have larger quantities of the insoluble β-peptide as amyloid [16–
18]. Extensive genetic and biochemical studies with early-onset AD cases have
identified mutations in the APP gene [11,19], and the genes for other proteins
Figure 1 Overview of the formation of β-peptide from the amyloid precursor protein
(APP) that contains 695 residues. The amino acid sequences of the β-(1–28) and β-
(1–42) peptides are shown. Depending upon conditions, in solution the β-peptide exists
in distinct conformations, whereas in the amyloid deposit only the oligomeric β-pleated
sheet structure is present.
TM

called the presenilins, which are larger proteins (greater than 400 residues) con-
taining six to nine transmembrane domains [11,20–22]. All of the mutations sup-
port a role for amyloid production in AD, particularly the ‘‘longer’’ 42-residue β-
(1–42) peptide rather than the ‘‘shorter’’ 40-residue β-(1–40) peptide. Selective
processing in familial AD patients with APP and presenilin mutations results in
the favored production and deposition of the β-(1–42) rather than the more solu-
ble, shorter 40-residue β-(1–40) peptide [23–26]. Additional research supports
the release of the β-(1–42) by a specific protease [27]. Together, these results,
along with other biophysical studies [7,28,29], indicate that less soluble β-(1–
42) may be the actual culprit in initiating amyloid formation in AD.
The levels of β-peptide aggregation correlate with the neurotoxicity to cor-
tical and neuronal cell cultures [30–32], and there is a direct association between
the accumulation of β-peptide into amyloid and the severity of dementia in AD
[33–35]. In AD, a major goal of research is to uncover a therapeutic procedure
that keeps β-peptide in solution long enough to be degraded/excreted and thus
prevents it from precipitating into amyloid. Once deposited as dense amyloid
plaque cores, the surrounding nerve cells are dystrophic or dead, and the β-pep-
tide becomes highly resistant to further proteolysis [36–38]. Thus, a basic under-
standing of the molecular mechanisms of β-amyloidosis, particularly in regard
to unraveling the structures of the early-folded and soluble intermediates, may
be critical to the development of an efficacious therapeutic procedure to prevent
AD.
The purpose of this review is to highlight certain aspects of our research
work with the amyloid β-peptide of AD. This project was started in early 1990
during my appointment at the Suntory Institute of Bioorganic Research
(SUNBOR) in Osaka, Japan, and is now being continued in my research group
at Case Western Reserve University (CWRU). At the time of its inception, very
little was known about the amyloidosis of the β-peptide. The lack of information
mainly resulted from the difficulty in studying the β-peptide, which is due to its
high insolubility and rapid aggregational properties. Once produced in the brain,
it was thought, the β-peptide rapidly aggregates into insoluble β-pleated sheet
structures, the probable structure in amyloid plaques [39–41].
Over the past few years, however, much has been learned about the bio-
chemical mechanisms of β-peptide production and aggregation into amyloid. De-
spite the obstacles with handling the β-peptide, the work we painstakingly began
at SUNBOR uncovered many important properties of its aggregation. This re-
search was carried out with excellent colleagues, and later at CWRU with many
outstanding graduate students and other collaborators who are acknowledged at
the end of this article. Most notably, our initial work at SUNBOR established that,
under certain solution conditions, the synthetic β-peptides can exist in monomeric
(nonaggregated) α-helical conformations. However, if the solution conditions are
appropriately changed, the β-peptide can rearrange (α-helix → β-sheet) and even-
TM

tually precipitate to produce insoluble amyloidlike deposits [7]. Our unique ap-
proach, by carefully studying the solution conformations of synthetic β-peptides,
uncovered these new properties, whereas at that time studies on natural β-peptides
isolated from amyloid plaques in other laboratories around the world failed to
detect these important features. These results raised the possibility that β-peptide
may exist in vivo in a soluble, nonaggregated state and that an understanding of
the mechanisms and the intrinsic solution conditions for the α-helix → β-sheet
conversion should be sought. In fact, in our initial report in [7] we proposed that
the β-peptide may normally exist in human biological fluids in a soluble form,
and that the longer 42-residue peptide results from an abnormal proteolysis. Sig-
nificantly, both of these propositions were later shown to be correct in AD patients
[23–26]. As mentioned before, familial mutations for early-onset AD cases have
greater amyloid production, which results from the favored production and depo-
sition of the β-(1-42) rather than the more soluble, shorter 40-residue β-(1-40)
peptide [23–26]. Thus, our chemical and structural work illustrated an important
point in biomedical research: detailed basic knowledge of the solution chemical
properties and the structures can facilitate parallel pharmacological studies.
On the basis of biophysical studies with synthetic β-peptides, x-ray fibril
diffraction established that the major conformation in the amyloid plaques is an
oligomeric, antiparallel cross-β-pleated sheet structure [39–41]. In regions of the
brain free of the amyloid plaque, synchrotron Fourier transform infrared micros-
pectroscopy showed that the β-peptide can fold into α-helical or random coil
structures [42]. In solution and before precipitation as amyloid, the β-peptide can
adopt α-helix, random coil, β-turn, and/or β-sheet structures, in relative ratios
that are strongly dependent on the solution conditions [7,40,43–46]. The α-heli-
cal and β-sheet/β-turn structures are monomeric, whereas the β-sheet structure
in solution is oligomeric, is toxic [31,32,47] and is a precursor to that found in
the amyloid plaque. Our work at SUNBOR was the first to demonstrate that the
solution conformations vary with pH and solvent [7,48], as later confirmed by
other laboratories [45,49,50]. The spectroscopic techniques we employ include
circular dichroism (CD), nuclear magnetic resonance (NMR), and infrared (IR)
spectroscopy. A major portion of our initial studies used the β-(1–28) peptide
(Fig. 1), which contains residues 1–28 and is an appropriate structural model for
the complete β-(1–42) peptide. Although the β-(1–28) peptide is incapable of
adhering to the surface of preexisting plaques [51,52], it produces soluble mono-
meric α-helical structures [7,50] as seen in the brain [42], and also plaquelike
oligomeric β-sheet structures, similar to those found in natural amyloid plaques
[40,43]. The hydrophobic 29–42 region increases the rate of aggregation and β-
sheet production [44,45,48,53] and greatly contributes to the insolubility of the
peptide in aqueous solution.
Because of space limitations, this review will highlight only specific aspects
of our research with the amyloid β-peptide. The topics were selected to provide
TM

a framework of our approach to elucidating the mechanisms of β-amyloidosis.
Our studies have focused on examining the solution-state of the β-peptide, before
it begins to precipitate as amyloid. The selected sections are summarized as fol-
lows:
1. The extraordinary sensitivity of the β-peptide solution structures (α-
helix, β-sheet, random coil) and aggregational states to the environ-
mental conditions such as pH, temperature, and solvent hydrophobicity
2. Our approach to provide reproducible starting points for biophysical
studies
3. Progress toward obtaining the three-dimensional solution structure of
the physiologically important β-(1–42) peptide
4. Using nicotine as an example, research for the design of β-amyloid
inhibitors that may eventually serve as drug targets
5. Proposed biological mechanisms for β-amyloidosis that take into ac-
count the in vitro chemical studies
II. SOLUTION CONDITIONS AND THEIR IMPACT ON

THE STRUCTURE
A. Peptide Preparation and Characterization
All peptides were either purchased from commercial sources or prepared in our
own laboratory on solid phase using conventional t-Boc or Fmoc strategies [54].
Because of their widespread use in research, the amyloid β-peptides are sold by
many commercial laboratories. We find it most economical to purchase unpuri-
fied peptide batches and purify them in our own laboratory by standard reverse-
phase high-performance liquid chromatography (HPLC) using acetonitrile, water,
and trifluoroacetic acid (TFA) solutions for elution [48,55]. By purifying the
peptides ourselves, we remove the inconsistencies with varying purity levels from
different commercial sources. Peptide identities and purity levels are always veri-
fied by mass spectrometry and NMR spectroscopy.
An important protocol with our current studies is to ensure that the β-pep-
tide adopts a well-defined, monomeric state before starting the biophysical mea-
surements. A major difficulty with solution studies of the amyloid peptides, par-
ticularly the more insoluble β-(1–42) peptide, relates to the ease with which they
aggregate and precipitate. Although it has been established that the soluble β-
sheet structure of the β-peptide is neurotoxic [31,32,56], there still exists consid-
erable disagreement about the levels of neurotoxicity [31,57–59]. In fact, several
groups still report that synthetic β-peptides show either trophic or toxic responses
on neurons in vitro. The reasons for these discrepancies are believed to arise in
part from differences in the aggregation states and the structures of the β-peptides.
TM

Different commercially prepared batches of high-performance liquid chromatog-
raphy–(HPLC)-purified β-(1–42) peptides can have different starting aggregation
states and structures [60], which will then in turn affect their solubility, aggrega-
tion rates, biological activities in solution [31,57–59], and ability to reproduce
biophysical measurements. The reproducibility of the presumably neurotoxic ag-
gregated β-sheet structure is dependent on many factors, particularly the peptide
concentration, ionic strength, and solvent polarity [29,44,48,61]. For example,
the aggregation rate is extremely rapid in aqueous acetonitrile solutions, such as
those used for HPLC purification of the peptide [62]. The longer the β-(1–42)
peptide remains in aqueous acetonitrile solution, the more likely it will become
an aggregated β-sheet structure.
To overcome the complications described, we developed a simple pretreat-
ment protocol that involves sonicating the dry peptide in neat TFA solution before
biophysical measurements [55]. The TFA breaks up the preaggregated peptides
and affords monomeric random coil structures. This method ensures that different
batches of the purified β-peptides will provide reproducible starting points for
biophysical and neurotoxicity studies.
B. Circular Dichroism Studies: Effect of Solvent and pH

Because of the structural instability of the β-peptide, before starting the more
time-consuming NMR measurements, we typically first conduct thorough CD
studies. This enables us to explore rapidly the influences of solvent, temperature,
pH, and peptide concentration on the secondary structures. The CD technique is
fairly reliable for measuring peptide secondary structure [63,64], and, unlike
NMR, it is easy to use, very sensitive, and capable of working with peptide con-
centrations down to the micromolar (µM ) range.
A remarkable feature of the β-peptides is that the relative ratios of second-
ary structures are greatly influenced by the solvent hydrophobicity and the pH
[7,8,29,44,45,48,49,62,65,66]. In earlier work we showed that, in aqueous triflu-
oroethanol (TFE) solutions, the β-(1–28), β-(1–39), and β-(1–42) peptides all
adopt monomeric α-helical structures at low and high pH, as shown by the in-
tense, negative CD bands at 208 and 222 nm [7,48]. However, at intermediate
pH (4–7) an oligomeric β-structure (the probable structure in plaques) predomi-
nates, as shown by the weak, broad negative bands at about 214 nm. The CD
spectra were analyzed using standard poly-l-lysine curves [63,64] and a graphical
summary of the results for the β-(1–28) peptide in 60% TFE is shown in Fig.
2A. These experiments were performed by recording CD spectra for two identical
peptide solutions at high and low pH, followed by a stepwise increase or decrease
of the pH. To prevent errors that can arise from secondary structural changes
over time, CD spectra were recorded within 10 min after each pH alteration.
TM

Figure 2 Graphs showing the variation of secondary structures of the β-(1–28) peptide
with pH in 60% TFE (a) and water alone (b) from CD data; the β-sheet structure is favored
at midrange pH (4–7), particularly in 60% TFE solution. TFE, trifluoroethyl; CD, circular
dichroism; —䊐—, % helix; —䉫—, % sheet; —䊊—, % random coil.
Trifluoroethanol is a structure-promoting solvent that encourages intramolecular

hydrogen bonding [67–69]. In solutions containing up to 80% TFE, a synthetic
peptide that comprises residues 29–42 [β-(29–42)] remained in a β-sheet confor-
mation, indicating that this segment may direct folding of the complete β-(1–
42) peptide to produce the β-pleated sheet structure found in amyloid plaques.
As expected, the secondary structures for the completely hydrophobic β-(29–42)
peptide were not altered with pH. Under equivalent conditions, we found that an
α-helical structure is more easily formed in the shorter β-(1–39) than in β-(1–
42), indicating that β-(1–42) forms a more stable β-sheet structure than does β-
(1–39).
The β-(1–28) peptide at pH 2.8 and 7.4 in 60% TFE solution is 58% and
47% α-helix, respectively, with the remainder being mostly random coil (Fig.
2A). The conformation was independent of peptide concentration throughout the
1.7- to 3000-µM range; this finding, together with those of other NMR studies
[70], establishes that the α-helical structure is monomeric. These solutions were
stable over time, and no changes in the relative amounts of α-helix and random
coil were observed over a 2-month period. Within the pH 4–7 range, however,
mixtures of β-sheet, α-helix, and random coil coexist in solution. Moreover, for
peptide solutions kept at pH 4–7, the relative proportions of α-helix, β-sheet,
and random coil structures are both time- and concentration-dependent. At pH
4–7, the amount of β-sheet structure increases over time and is accompanied by
TM

precipitation as an amyloidlike β-pleated sheet deposit. The relative proportion
of β-sheet structure increases at higher peptide concentrations, consistent with
its highly aggregated state in the amyloid plaques of AD brain.
In water with or without TFE, the β-sheet structure is preferred at pH 4–
7, whereas the random coil and α-helical conformations are favored at pH 1–4
and 7–10. A summary of the secondary structures observed for the β-(1–28)
peptide in water solution as a function of pH without any cosolvents is shown
in Fig. 2B. Analysis of the CD spectra using standard poly-l-lysine curves
showed that, for all pH values, 80%–90% of the major conformation was random
coil, except at pH 6.0, where approximately 52% β-sheet and 42% random coil
conformations coexist (Fig. 2B). The presence of an isodichroic point in the CD
data indicates that only two conformations (random coil and β-sheet) are in
equilibrium. By contrast, there are two isodichroic points for the CD data in
60% TFE [48]. At pH 4–7 in 60% TFE, the α-helical content drops abruptly
and the β-sheet content rises, whereas the random coil content remains rela-
tively unchanged. These data suggest that, in TFE solution at midrange pH, an
α-helix → β-sheet conversion occurs directly without a random coil intermediate.
However, in TFE solution at low and high pH little, if any, β-sheet structure
exists, and instead α-helix and random coil structures coexist. If an α-helix →
random coil → β-sheet transformation occurred at pH 4–7, then an increase
in random coil structure should occur. Related α-helix (soluble, membrane-
bound) → β-sheet (insoluble, aqueous solution) conversions of other peptides
and proteins, including the scrapie prion proteins, are well known [6,71].
C. Circular Dichroism Studies: Effect of Micelles and pH

To examine the effect of a lipid environment on the solution structures of the β-
peptide, additional CD studies were undertaken in micellular solution. Micelles
adequately mimic a membranelike or bilayer environment and are frequently em-
ployed in the structural studies of peptides and proteins [66,72,73]. In earlier
work we proposed that the β-peptide adopts an α-helical structure in lipid solu-
tions [48,74], which may be pertinent to the native structure when bound to lipo-
proteins and albumin in human plasma [75]. The present studies were undertaken
to explore the effect of pH and the surface charge of micelles on the solution
structures of the β-(1–28) peptide.
Shown in Fig. 3 are graphs summarizing the percentage secondary struc-
tures with negatively charged, positively charged, and neutral micelles at different
pH values. To prevent possible structural perturbations over time, the experiments
were performed by recording CD spectra for two identical peptide solutions at
high and low pH, followed by a stepwise increase or decrease of the pH. The
results establish that the α-helical structure is produced in the physiological pH
range with either the negatively charged sodium dodecyl sulfate (SDS) or the
TM

positively charged dodecyltrimethylammonium chloride (DTAC) micelle. The
chemical structures of these and other micelles are presented in Fig. 4. The zwit-
terionic, neutral dodecylphosphocholine (DPC) micelle does not produce signifi-
cant α-helical structure above pH 4. For the CD spectra obtained in all three
micelles, there are mixtures of two conformations, random coil and α-helix. This
interpretation is consistent with the presence of isodichroic points, together with
α-helical and random coil bands observed in the CD spectra (data not shown).
In 20-mM SDS solution, which is above the critical micelle concentration of 8
mM, the peptide adopts predominantly α-helical structure at low to weakly basic
pH (Fig. 3), as shown in the CD spectra by a strong positive band at 195 nm
and negative bands at 208 and 222 nm. The α-helical content remains relatively
constant within two ranges, pH 2–4 and 7–9, which is approximately 80% and
50% α-helix, respectively, with the remaining structure being random coil. At
pH 10 and 11 the peptide adopts mostly random coil structures, as demonstrated
by the presence of the intense negative band at 198 nm. No β-sheet structure
exists in any of the micelle solutions, regardless of pH. These data are in stark
contrast to the results in TFE–water solution, especially at pH 4–7, where an α-
helix → β-sheet conversion occurs. One possibility is that the histidine side chains
of the β-peptide are shielded from the solvent by the micelles, as these residues
are critical to stabilizing the β-sheet structure [70,76]. These studies establish
that membranelike media with a charged surface stabilize the soluble α-helical
conformation of the β-peptide in the physiological pH range.
D. Nuclear Magnetic Resonance Studies

Another analytical technique we use with the amyloid β-peptides is NMR spec-
troscopy [77,78]. In comparison to the more qualitative techniques such as CD
and IR spectroscopy, the NMR approach has the advantage of being able to deter-
mine specifically where a particular secondary structure is located within the
primary sequence. Furthermore, for drug design the NMR method can provide
valuable information about protein dynamics and specific data about the individ-
ual amino acid side chains that bind to a particular ligand [79,80]. With sufficient
distance and dihedral angle constraints, the NMR approach is also capable of
rendering a complete three-dimensional structure. However, for solution NMR
applications, a general requirement is that the protein predominantly adopts one
well defined and sufficiently stable structure usually for a period of several days
and at relatively high concentrations (up to several milli-molar). This stability is
mandatory when long, multidimensional data acquisitions may be run over a
period of several days. These necessities have greatly restricted NMR studies of
the amyloid β-peptides. In water solution, freshly dissolved monomeric β-(1–
42) peptide adopts a mixture of rapidly interconverting α-helix, random coil, and
aggregating β-sheet structures. This rapidly equilibrating mixture causes severe
TM

TM

Figure 4 Chemical structures of the sodium dodecyl sulfate (SDS), dodecyl phospho-
choline (DPC), hexadecyl-N-methylpiperidium (HMP), myristyltrimethylammonium
(MTMA), dodecyltrimethylammonium chloride (DTAC), N-tetradecyl-N, N-dimethyl-3-
ammonio-1-propanesulfonate (Z 3–14), Congo red (CR), (S)-(⫺)-nicotine, and (S)-(⫺)-
cotinine. Depending on whether the compound has a net positive or negative charge, the
counterions are either sodium or bromide.
line broadening and prevents the detection of well-resolved NMR signals. Addi-
tional serious problems are the subsequent aggregation and precipitation, particu-
larly with the longer β-(1–42) peptide, that readily occur at the higher peptide
concentrations required for NMR spectroscopy. It is possible, however, to per-
form NMR studies of fragments that are less likely to aggregate such as the β-
(1–28), β-(12–28), β-(10–35), and β-(25–35) in water solution, or with the com-
plete β-(1–42) peptide in water solution containing TFE or micelles [61,70,74,
81–83].
One of the most useful NMR parameters is the nuclear Overhauser effect
Figure 3 Graphs depicting the variation of random coil and α-helical secondary struc-
tures of the β-(1–28) peptide with pH in SDS, DPC, and DTAC micelle solutions from
CD data. The chemical structures of the micelles are shown in Fig. 4. The charged SDS
and DTAC micelles stabilize the α-helix at physiological pH and β-sheet structure was
not observed in any micelle solution. SDS, sodium dodecyl sulfate; DPC, dodecyl phos-
phocholinc; DTAC, dodecyl trimethylammonium chloride; DPC, dodecyl phosphocholine;
CD, circular dichroism.
TM

(NOE), which provides through-space information between protons that are
within 5 Å of each other and are dipolar-coupled [84]. The pattern of interresidue
NOEs and the magnitude of another parameter, the amide (NH) to αH J-coupling
constant, together permit an accurate representation of the secondary structure
of proteins [77,78,85,86]. Another useful marker for secondary structures relies
on the αH chemical shifts, with α-helices showing upfield shifts relative to β-
sheets [87,88]. The major secondary structural elements in proteins are α-helices,
turns, parallel β-sheets, and antiparallel β-sheets. The NOE patterns, backbone
J-coupling constants, and αH chemical shifts are distinct for these structures. The
α-helix has smaller NH-αH J-coupling constants (⬍6.0 Hz), relative to β-sheets
(⬎7.0 Hz). In addition, the α-helix can be identified by strong NN(i, i ⫹ 1),
αN(i, i ⫹ 3), and αβ(i, i ⫹ 3) NOEs, whereas β-sheets are characterized by very
strong αN(i, i ⫹ 1) NOEs and the absence of any other NOEs among the back-
bone protons. The antiparallel β-sheet exhibits NOEs between αH protons on
adjacent strands whereas these are replaced by NOEs between NH and αH pro-
tons on adjacent stands for the parallel β-sheet. There are also differences in NH
exchange rates, in which the first four residues in helices, the first three residues
in a 310 helix, and every second residue in the peripheral strands of β-sheets have
greater exchange rates than neighboring NHs. Similarly, there are characteristic
patterns of NOEs, spin–spin couplings, and amide proton exchange rates for turns
of types I, II, I′, and II′, and half-turns derived from type II tight turns.
The first NMR studies were performed with the β-(1–28) peptide and these
involved two-dimensional (2D) NMR studies below pH 3 in 60% TFE-d3, [7,70].
According to earlier CD work, under these conditions the β-(1–28) peptide
should fold into a largely α-helical structure (Fig. 2) that is stable and amenable
to long-term 2D NMR data acquisitions. This initial work represented the first
high-resolution NMR data obtained with an amyloid peptide in solution. Subse-
quent studies done in our laboratory used perdeuterated-DPC (DPC-d38) and per-
deuterated-SDS (SDS-d25) solutions [74] under conditions that also maintained
stable α-helices (Section III.C). Analysis of the 2D NOESY data showed numer-
ous NN(i, i ⫹ 1), αN(i, i ⫹ 3), and αβ(i, i ⫹ 3) NOEs that established a predomi-
nantly α-helical structure, in accordance with the CD data (Fig. 2 and 3). As
shown in Fig. 5, the three-dimensional NMR structure in 60% TFE-d3 contained
right-handed α-helices (residues 2–11 and 13–27), connected by a bend-centered
at Val12 [74]. The bend is in agreement with the observed periodic nature of
the αH chemical shifts [87,88] and also with the NMR-derived NH temperature
coefficients. At 25°C and below pH 3, nearly identical NOE data were seen in
aqueous solutions containing TFE-d3, DPC-d38, or SDS-d25, suggesting that simi-
lar tertiary α-helical structures are present under these conditions. However, there
are differences in the stabilities of the two α-helical segments (residues 2–11 and
13–27) with temperature. The NH temperature coefficients in SDS-d25 solution
showed that both α-helical segments are stable up to 50°C, whereas in 60% TFE-
TM

Figure 5 The final 95 structures for the β-(1–28) peptide superpositioned using the
backbone atoms (pH 3, 60% TFE solution). The N-terminus is shown at the top and the
C-terminus is shown at the bottom. The side chains of Val12 and Phe20 are labeled to
illustrate their location on the same face of the helix TFE, trifluoroethyl. (Source: From
Ref. 74.)
d3 and DPC-d38 solutions, the smaller α-helix (residues 2–11) rapidly unfolds
above 25°C with the other helix (residues 13–27) unfolding above 30°C. Both
α-helices unfold to random coil structures and refold back to α-helices with cool-
ing. The NH temperature coefficients were determined by recording NOESY data
at 10, 25, 35, 40, and 50°C. The random pattern of coefficients suggests that the
peptide is not buried within the hydrophobic interior of the SDS-d25 micelle,
but instead is located at the surface near the lipid/water interface. These studies
demonstrate that the negatively charged SDS micelle provides an environment
most conducive to α-helix formation and that this stability is not due to the loca-
tion of the peptide within the hydrophobic interior of the SDS micelle.
To evaluate further the effect of pH on the loss of α-helical and the forma-
tion of β-sheet structure, we are currently determining the dissociation constants
(pKa) of the aspartic acid (Asp), glutamic acid (Glu), histidine (His), and tyrosine
(Tyr) residues under different solution conditions. We have used the β-(1–28)
TM

peptide as a model for the longer β-(1–42) peptide, since the remaining 29–42
region is devoid of polar and charged amino acid residues. The respective chemi-
cal shifts for the 2H, 4H; 3H, 5H; γCH2s, and βCH2 signals of the His; Tyr; Glu;
and Asp residues at different pH values are being determined by a combination
of 1D and 2D NMR spectroscopy. Preliminary analysis demonstrates that the
Glu and His residues have larger pKa values in SDS solution, compared with
60% TFE, DPC, and water alone. This result may partly account for the greater
stability of the α-helix in SDS solution. Complete loss of the helix occurs above
pH 9, which corresponds to the pKa of two His residues. Below pH 9 the His
side chains are predominantly positively charged, that charge should bind them
more strongly with the negatively charged SDS surface. A complete summary
of these results is presented in a forthcoming paper [89].
With a relatively clear understanding of the effects of solvent, pH, tempera-
ture, and lipids on the structures and aggregational properties of the β-(1–28)
peptide, we next examined the more difficult, longer, and naturally occurring 42-
residue β-(1–42) peptide. It was felt that the 1–28 region should function in an
analogous manner in both peptides, since the remaining 29–42 region is com-
pletely hydrophobic and can be considered a distinct subregion. As it happens,
our studies with the β-(1–28) did indeed provide sufficient background knowl-
edge that enabled us to examine the more important β-(1–42) peptide. The β-
(1–42) peptide can be extremely insoluble and act as a seed that nucleates amy-
loid formation with the other shorter amyloid peptides [23,28,29]. In addition,
the β-(1–42) peptide is the predominant protein component in cerebrovascular
amyloid and plaque core deposits [90,91], and it is thought that the longer peptide
may be the actual culprit behind amyloid deposition in AD [24,90–92]. Unfortu-
nately, despite its biological importance to AD, biophysical and biological studies
have primarily used the shorter peptides since the β-(1–42) is difficult to dissolve
and aggregates/precipitates too rapidly.
Before beginning NMR studies with the longer peptide, we first conducted
CD studies. However, unlike the β-(1–28) peptide, the β-(1–42) peptide in water
solution above pH 7 forms an aggregated β-sheet structure and rapidly precipi-
tates at concentrations above 100 µM. Therefore, NMR studies in water solution
at physiological pH are clearly precluded. When the CD studies were repeated
with 1 molar-equivalent of SDS pH 7.2, the CD spectra obtained at 0, 24, 48,
96, and 120 h were nearly identical, consistent with a predominantly α-helical
structure, and significantly, no precipitation was seen. These results suggest that
the SDS micelle promotes formation of a predominantly α-helical structure,
which is probably similar to that observed for the β-(1–28) peptide. The stability
and lack of precipitation of the β-(1–42) peptide provided by the SDS micelle
suggest an amenable system for detailed structural analysis by high-resolution
NMR methods. This ability of SDS to prevent aggregation is analogous to other
recent work, in which the micelle, hexadecyl-N-methylpiperidinium bromide
TM

(HMP), inhibited aggregation of the β-(1–40) peptide [93]. The chemical struc-
tures of HMP, plus other micelles and compounds that are β-amyloid inhibitors,
are shown in Fig. 4. These include myristyltrimethylammonium (MTMA), N-
tetradecyl-N,N-dimethyl-3-ammonio-1-propanesulfonate (Z 3–14), Congo red
(CR) [94], and (S)-(⫺)-nicotine [95]. The first three compounds were selected
from a screening of 4300 compounds as β-amyloidosis inhibitors [93]; both HMP
and MTMA inhibited β-(1–40) aggregation whereas Z 3–14 showed weak inhibi-
tion.
The NMR characterization of the β-(1–42) peptide in SDS-d25 solution has
proceeded very well. In fact, aqueous solutions of the β-(1–42) at relatively high
concentrations (up to 2 mM ) are stable for several months. This is important for
multiple 2D NMR data acquisitions that can require several days. The SDS-d25
concentrations were significantly higher than those for the β-(1–28) peptide. Typ-
ically, the SDS-d25 concentrations are 80–150 mM, so that the peptide/micelle
ratio is always 1 :1; with no salt, the average SDS aggregation number is 62 and
the critical micelle concentration is 8 mM [73]. With the exception of minor
chemical shift differences for some residues at the N-terminus, 1D NMR spectra
obtained at protein concentrations of 1.4, 0.94, 0.56, and 0.16 mM were virtually
identical. Therefore, it appears that aggregation does not proceed above the level
of a monomer within the concentration range 1.4–0.16 mM. Additional studies
at lower peptide concentrations are currently being explored in our laboratory.
A contour plot of the complete 2D NOESY spectrum of the β-(1–42) pep-
tide in SDS-d25-H2O solution at pH 7.2 is shown in Fig. 6. The NOESY spectrum
reveals through-space connectivity among protons separated by less than 5 Å.
The numerous well-resolved cross peaks demonstrate that the β-(1–42) peptide
is folded, and numerous sequential NN(i, i ⫹ 1) NOEs between adjacent amide
protons (NHs) are present (Fig. 7). The NOESY spectra obtained with suppres-
sion of the H2O signal by presaturation and with the ‘‘Jump and Return’’ pulse
sequence [96] were almost identical in the NH–NH region. This demonstrates
that NH signals are not lost by presaturation of the water signal, which can lead
to saturation transfer. In particular, strong NN(i, i ⫹ 1) NOEs are observed within
the Tyr10–Val24 and Lys28–Ala42 segments. These and other medium-range
backbone NOEs, together with the upfield chemical shift positions for the αH
signals, are consistent with α-helical structure within these regions [77,87].
The complete proton NMR spectrum has been assigned and further efforts
for building the three-dimensional structure are currently under way. Shown in
Fig. 6 is a hypothetical model for the association of the β-(1–42) peptide with
the SDS micelle. Although not shown, side-chain binding interactions between
the positively charged side chains of the β-(1–42) and the negatively charged
SDS surface are most likely occurring. These interactions stabilize the α-helix
and prevent formation of the aggregated β-sheet, which is the predominant struc-
ture in water without SDS. The location of the peptide at the micelle surface,
TM

Figure 6 The complete 2D NOESY spectrum (600 MHz, mixing time 270 ms) for the
β-(1–42) peptide (1.4 mM ) in H2O :D2O (9:1) containing SDS-d25 (80 mM ) at pH 7.2
and 20°C. The H2O signal was suppressed by presaturation. The proposed secondary struc-
ture of the β-(1–42) peptide based on the NMR chemical shift, NOE, and NH-temperature
coefficient data is shown. The monomeric peptide appears to remain on the surface at the
lipid–water interface of the SDS micelle and does not become embedded into the hy-
drophobic interior. NOESY, nuclear overhauser effect; SDS, sodium dodecyl sulfate;
NMR, nuclear magnetic resonance.
rather than the hydrophobic interior, is consistent with the amide-NH temperature
coefficients and the rapid deuterium exchange rates. The irregular pattern of the
NH-temperature coefficients and the rapid NH-exchange rates are consistent with
the peptide being located on the surface of the SDS micelle. If residues were
buried within the hydrophobic interior of the micelle, then significantly lower
temperature coefficients and NH-exchange rates should be observed. Overall, the
secondary structure is defined by an extended structure (residues 1–9), α-helix
(residues 10–24), loop (residues 25–27), and α-helix (residues 28–42). The pres-
ence of a loop within the Gly25–Ser26–Asn27 segment is supported by the ab-
sence of any major backbone NOEs and the downfield chemical shift of the
Asn27 αH. Interestingly, the location of this loop is nearly identical to a reverse
turn suggested for the β-sheet structure of a β-(10–43) peptide, in which the
turn was proposed to be within the Ser26–Asn27–Lys28–Gly29 sequence [44].
Recent NMR studies of a β-(10–35)-CONH2 fragment suggests a turn–strand–
turn motif between His13 and Val24 in aqueous solution [51], whereas the β-
(1–40) peptide forms two α-helical segments between Gln15–Asp23 and lle31–
Met35 in 30% TFE-d3 solution [81].
TM

Figure 7 An expanded NOESY region for the β-(1–42) showing the NH-NH connectiv-
ities, as well as other peaks for NOE interactions between the aromatic side chains and
the backbone NH. NOESY, nuclear overhauser effect.
III. NICOTINE INHIBITION OF ␤-PEPTIDE

FIBRIL FORMATION
An intense area of research in AD involves identifying potential inhibitors that

either slow down or prevent the precipitation of the β-peptide into amyloid. The
recent reports of an inverse relationship between the risk of AD and cigarette
smoking [97–102] inspired us to investigate the effects of nicotine and cotinine
(structures in Fig. 4) on the solution conformations and aggregational properties
of the β-peptide. Nicotine is a major component of cigarette smoke that is metabo-
TM

lized and excreted as cotinine. Using CD and ultraviolet spectroscopic techniques,
we recently showed that nicotine inhibits amyloid formation by the β-(1–42)
peptide [95]. Cotinine slows down amyloid formation, but to a lesser extent than
nicotine. Although it is not yet known whether a similar nicotine inhibition to
amyloidosis occurs in the human brain, these studies may facilitate the design
of less toxic and more efficient amyloid inhibitors that are structurally related to
nicotine. In this review, we summarize unpublished molecular modeling studies
that were not included in the original manuscript [95]. The molecular modeling
studies provide a more detailed representation of the nicotine inhibition to β-
amyloidosis.
Our previous NMR studies demonstrated that nicotine binds to the 1–28
peptide region when folded in an α-helical conformation [95]. The β-(1–28) pep-
tide is an appropriate structural model for the complete β-(1–42) peptide, since
it produces soluble monomeric α-helical structures [7,50], as well as plaquelike
oligomeric β-sheet structures, similar to those found in natural amyloid plaques
[40,43]. The hydrophobic 29–42 region increases the rate of aggregation and β-
sheet production [44,45,48,53] but should not affect the ability of nicotine to
bind to the β-peptide. Nicotine is a weak base with pKa of 7.9 [103], and at blood
pH of 7.4, nicotine is about 69% ionized (protonated) and 31% nonprotonated.
Therefore, if binding does indeed occur, nicotine would be more likely to bind
to the polar, hydrophilic 1–28 peptide region, rather than the completely hy-
drophobic 29–42 region (Fig. 1).
On the basis of chemical shift data, the binding between nicotine and the
β-(1–28) peptide primarily involves the N-CH3 and 5′CH2 pyrrolidine moieties
of nicotine and the His residues of the peptide. The binding is in fairly rapid
exchange, as shown by single averaged NMR peaks. A summary of the bound
chemical shifts and intermolecular NOEs between nicotine and the peptide region
is shown in Table 1.
The modeling studies used the NMR data, molecular dynamics, and energy
minimization protocols. On the basis of the bound NMR chemical shift data (Ta-
ble 1), we envisioned a molecule of nicotine bound to the α-helix of the β-peptide,
as depicted in Fig. 8. In the α-helical structure, the side chains of His13 and
Lys16 are proximate [74] and could conceivably accommodate nicotine. The pos-
sibility that the Lys16–ζNH⫹3 is tied up with an intramolecular hydrogen bond
to the N3 of His13 was ruled out on the basis of previously published NMR data
[70], which showed that the Lys16–ζNH⫹3 is solvent exposed. Besides the bound
chemical shift data that support this model, another attractive feature for this
proposed binding is the complementary charge distribution between nicotine and
the peptide. At pH 7.2 the nicotine pyrrole nitrogen (N1′) will be predominantly
protonated (pKa 7.9), whereas the pyridine N1 is not protonated (pKa 3.12) [104].
For the β-peptide, at pH 7.2 the side chain N3 of His13 is not protonated (pKa
6.2–7.0) whereas the Lys–ζNH⫹3 remains protonated up to pH 10–11 [105].
TM

Table 1 Chemical Shifts, NOEs, and Distances in the Complex of Nicotine and the
β-(1–28) Peptide
Bound Starting Final

chemical Minimization and distance distance
NMR resonance shift (ppm)a distance restraints (Å)b (Å)c (Å)c
Peptide
His13-2H ⫺0.22 6–10 Above the pyridine 10.07 6.84
ring
His13-4H ⫺0.06 12.66 9.94
Lys16-ζNH⫹3 0.00 12.10 8.90
Nicotine
2H 0.10
6H 0.07
4H 0.07
5H 0.03
2H′ 0.07
5-α-CH′ 0.24 6–10 In plane of His13 11.64 8.69
ring
5-β-CH′ 0.31 6–10 In plane of His13 11.64 7.96
ring
3CH′2 0.02
4CH′2 0.05
N-CH3 ⫺0.50
0.19 6–10 In plane of His13 8.97 6.05
ring
Intermolecular NOEs
His13-4H to 4CH′2 3–5 Å
Tyr10-2, 6H to 5-β-CH′ 3–5 Å
Va118-βH to 2H′ 3–5 Å
Tyr10-βH to 5-β-CH′ 3–5 Å
Va112-γCH3 to 5-α-CH′ 3–5 Å
a
Obtained by subtracting the chemical shifts in nicotine and the peptide from those seen in the com-
plex. The negative shifts are upfield.
b
For the upfield or downfield shifts, distance restraints refer to a proton located either above or in
the plane, respectively, of the nicotine–pyridine and peptide–His13 aromatic rings. Pseudoatoms
were used to represent the three equivalent NH⫹3 and CH3 protons.
c
The starting and final distances, before and after molecular dynamics plus several cycles of energy
minimization. For the 2H of His13, the distances refer to its location relative to the center of the
nicotine–pyridine aromatic ring and for the Lys16-ζNH⫹3 relative to the N1 of nicotine. NOE, nuclear
overhauser effect; NMR, nuclear magnetic resonance.
TM

One molecule of nicotine was annealed and energy minimized to its lowest
energy conformation in water solution at pH 7.2. The nicotine–peptide complex
was generated by manually docking the minimized (S)-nicotine molecule to the
NMR-derived averaged, energy-minimized β-(1–28) α-helical structure [74].
Pseudoatoms for the five atoms of the imidazole ring of His13 and for the six
atoms of the pyridine ring of nicotine were created. The docking was performed
in a manner consistent with the diagram in Fig. 8 and the bound chemical shift
data (Table 1), which were interpreted as resulting from ring current effects [106].
The N-CH3 of nicotine displays two NMR signals when mixed with peptide: one
signal appears downfield from the original unbound shift by 0.19 ppm, and the
other is upfield by 0.50 ppm in an approximate 2 :1 ratio, respectively. The two
signals probably represent two different orientations of the pyrrole ring when
bound to the His residues of the peptide. For the purpose of modeling, we used
the more abundant downfield signal.
Figure 8 Proposed structure of the (S)-(⫺)-nicotine/β-(1–28) peptide complex, when

the peptide folds as an α-helix, with the N-terminus at the top and the C-terminus at the
bottom. The schematic diagram proposes the His13-N3 and Lys16-ζNH⫹3 binding to the
pyrrole H-N1′⫹ and N1 of nicotine, respectively.
TM

The upfield shift of the His 13-2H results from its location above or below
the nicotine–pyridine ring; the downfield shifts of the N-CH3, and 5CH′2 protons
of nicotine were explained by their location in the His13 imidazole ring plane.
Accordingly, the N-CH3 and 5CH′2 protons were positioned to the periphery of
the His13 ring, and the His13-2H proton was placed above the pyridine ring
(Table 1). The possibility that other aromatic rings in the peptide, such as those
of the phenylalanine and tyrosine residues, contribute to the binding and ring
current shifts was ruled out, since the NMR peaks for these rings do not undergo
chemical shift changes in the mixture. A restraint file of 6–10 Å was created for
the preceding protons on the basis of well-known methods that use ring current
shifts to relate the position of a proton relative to an aromatic ring [106,107].
For bound chemical shifts less than 0.5 ppm, the distances from the ring center
are larger (6–10 Å) and all bound chemical shifts are less than 0.31 ppm (Table
1). These restraints kept the N-CH3, and 5CH′2 protons within the deshielding
plane of the His13 ring throughout the course of the dynamics and minimizations.
However, to prevent movement of the His-2H away from the shielding region
of the nicotine–pyridine ring, individual 6–10 Å restraints between the 2H and
each heavy atom of the pyridine ring were required.
To provide a more realistic binding environment and stabilize the charges,
the SOAK Assembly operation in INSIGHT II (Biosym, Inc.) was utilized to
solvate the nicotine–peptide complex. This involved building a two-layer solvent
sphere, which contained 336 and 186 water molecules in the inner and outer
spheres, respectively. The diameter of the complete sphere was 21 Å with the
inner sphere taking up 15 Å To prevent evaporation of the water during the
dynamics, the water molecules in the outer sphere were tethered, whereas the
water molecules in the inner sphere were allowed to move freely. Thus, the outer
layer provided a boundary and prevented escape of the water molecules in the
inner sphere. The starting nicotine–peptide complex was placed in the center of
the inner sphere, so that the Val12–Leu17 region completely occupied the sphere
(Fig. 9A). The SOAK operation automatically repositioned those water molecules
that overlapped with the atoms of the complex. To reduce computation time, a
cutoff distance was set to 25 Å The cutoff distance defines the minimum range
for atoms that will be considered in the calculation. The atoms that were outside
this region (amino acid residues 1–11 and 18–28) were kept fixed. Within the
Val12–Leu17 segment, the backbone atoms were tethered and the side chains
were completely free to move throughout the dynamics and minimizations. The
only distance restrictions within the Val12–Leu17 region involved the 6– 10–
to Å restraints placed on four hydrogen atoms: His13-2H, 5-α-CH′, 5-β-CH′, and
the pseudoatom for the N-CH3, as shown in Table 1. These restraints were made
on the basis of the bound chemical shifts and ring current effects (discussed
previously). The computations began with 100 steps of steepest descent energy
minimization, followed by molecular dynamics at 300 K for 1 ps, and finally 6000
TM

(a)
(b)
TM

steps of conjugate gradient energy minimization. A globally enforced dielectric
constant of 1.0, cvff force field, and a template force with a force constant of
1000 kcal/Å2 were employed. A simple harmonic potential was used for the bond
energies, and the cross terms were omitted from the energy expression. Distances
are provided for selected protons in nicotine and the peptide in Table 1. The final
total energy of the solvated nicotine–peptide complex was ⫺5219 kcal/mole.
The main adverse factor that contributed to the total energy were the nonbonded
repulsion energies from the fixed atoms in amino acid residues 1–11 and 18–
28. Longer molecular dynamic runs did not produce significantly lower total ener-
gies, demonstrating that the final nicotine–peptide complex was not trapped in
a local energy minimum. Other annealing methods are currently being explored
for obtaining lower energies.
An expanded view of the β-peptide/nicotine complex is shown in Fig. 9B,
and distances between selected protons are provided in Table 1. As shown, all
final distances are substantially shorter, indicating that nicotine and the peptide
move closer in space. If an unfavorable interaction were present, then greater
distance would be seen. In particular, both the His13-2H and 5-β-CH′ have sig-
nificantly shorter distances than their initial distances, and these NMR signals
likewise show the largest changes in NMR chemical shifts with binding. The
distance between the nicotine-N1 and the Lys16–ζNH⫹3 is reduced by 3.20 Å;
additionally, the labile nicotine H-N1′ hydrogen is less than 3 Å from the N3 of
His13. These results support the proposed binding mode shown in Fig. 8, in
which nicotine binds to the Lys16-ζNH3⫹ and the His13-N3 by hydrogen bonding.
A similar nicotine binding motif may occur with other proteins. In human
plasma, 5% nicotine becomes bound to albumin and several serum lipoproteins
[108], and in vitro nicotine binds to peptide segments of the nicotinic acetylcho-
line receptor (AChR). The AChR is a well-characterized protein that produces
rapid ligand-gated ion channels [109]. Although the structure of the binding sites
for the lipoproteins, albumin, and AChR are not completely characterized, several
His and Lys residues are believed to occupy important locations. The His and
Lys residues occupy α-helical sections in the α-bungarotoxin binding site of
AChR, where nicotine also becomes bound. Chemical modifications or replace-
Figure 9 The three-dimensional model of the nicotine–peptide complex (a and b) after

molecular dynamics and energy minimizations. (a) The NMR-derived averaged, energy-
minimized β-(1–28) α-helical structure is shown with the His13, His14, and Lys16 side
chains, and the backbone as a ribbon. The water molecules that solvate the complex are
shown in dark gray (a). (b) Expanded view that corresponds to the region within the box
in (a) shows the His-2H positioned above the nicotine–pyridine ring, consistent with the
NMR data. The modeling studies also suggest that the Lys16-ζNH3⫹ may bind to the nico-
tine-N1. NMR, nuclear magnetic resonance. (Source: From Ref. 74.)
TM

ment of the His residues results in ion channel impairment and a decreased affin-
ity for nicotine binding [110–112]. Substitution of Lys185 in a peptide segment
corresponding to residues 181–200 of the Torpedo alpha 1 AChR protein reduced
the binding of nicotine to the peptide [113]. Thus, the model of nicotine binding
shown in Figs. 8 and 9 may be general with other proteins.
The model shown in Figs. 8 and 9 may assist in the development of nicotine
analogues to retard amyloidosis by preventing an α-helix → β-sheet conforma-
tional transformation that is important in the pathogenesis of Alzheimer’s disease.
Further studies with additional nicotine analogues are required to support this
binding mode. Potential analogues could be synthesized with electron donating
substituents in the 4 position of the pyridine ring of nicotine, since this would
increase the basicity of the N-1 nitrogen and perhaps strengthen the interaction
with Lys 16-ζNH⫹3 . Interestingly, an analogous structural motif may be involved
in the binding of the β-peptide to transthyretin, a normal protein component of
plasma [114]. A structural model for the complex of transthyretin and the α-
helical structure of the β-peptide showed that the side chain of Lys16 forms part
of a positive potential binding surface that makes contact with a negative potential
surface of transthyretin.
IV. MECHANISM OF AMYLOIDOSIS OF THE ␤-PEPTIDE
Despite the importance to AD, the underlying mechanisms for the accumulation
of the β-peptide into insoluble amyloid remain largely unknown. A popular con-
cept involves an ‘‘amyloid-initiated-cascade’’ phenomenon, in which altered pro-
duction, removal, and aggregation of the amyloid β-peptide initiate a sequence
of events that leads to neuronal death [11,38,115]. It should be kept in mind that
other lesions such as the neurofibrillary tangles are also abundant in AD brains,
and it may happen that the tangles are more important in the pathogenesis of the
disease [116]. The tangles are intracellular deposits consisting of twisted fila-
ments of the cytoskeletal tau protein. However, since the majority of genetic and
biological data support a critical role for amyloid formation in AD, the ‘‘amyloid-
initiated-cascade’’ hypothesis appears to be the most promising model for drug
discovery [11,117].
A hypothetical model for β-amyloidosis is outlined in Scheme 1. This
mechanism attempts to take into account the results of our work, as well as work
from other laboratories. The various pathways shown in the scheme connect the
in vitro biophysical studies to a natural situation that may exist in the brain. The
model emphasizes a conformationally driven mechanism, in which the three ma-
jor solution structures of the β-peptide coexist in equilibrium: random coil (mono-
meric), α-helix (monomeric), and β-sheet (oligomeric). The aggregated β-sheet
TM

Scheme 1
is the predominant structural motif in the amyloid plaques. When the aggregation
reaches a critical mass, the β-sheet structure precipitates as amyloid, and recon-
version back to soluble random coil or α-helical structures is no longer possible
[48,62,118]. Recent data suggest that soluble β-sheet dimers are toxic [47], with
the larger amyloid plaques that form later further disrupting neuronal circuitry.
During β-sheet formation, the β-peptide-induced neurotoxicity increases [31,32],
suggesting that the processes involved in amyloid plaque formation are critical
in the pathogenesis of AD. The mechanisms whereby the β-sheet structure exerts
its neurotoxicity remain unknown [119].
Once released by proteolysis of APP, the β-peptide would be expected
initially to adopt a monomeric, random coil conformation. This interpretation is
based on the known physiological concentrations of the β-peptide in cerebrospi-
nal fluid, which are in the nano- to picomolar range [91,120]. At these low con-
centrations, the production of aggregated β-peptide as amyloid is very unlikely
[28,29,48,66] since the threshold concentrations required for generation of amy-
loidlike oligomeric β-sheet structure are much higher 50–350 µM. Presumably,
the random coil structure is biologically inactive and would not be involved in
specific binding processes to other macromolecules. This suggests that, in order
TM

to produce amyloid deposits, localized regions of the brain must have much
higher peptide concentrations, or alternatively some change in the brain microen-
vironment must occur to promote β-aggregation.
Under different environmental conditions, the β-peptide can fold into an-
other structure dominated by β-sheet and β-turns [51,52,121]. Since this structure
adheres to the surface of preexisting plaques, it is probably the immediate precur-
sor to the aggregated β-sheet structure. Conversion of this presumably monomeric
structure to aggregated β-sheet structure can occur with reductions in brain pH
[122] or by the attachment of the β-peptide to larger macromolecules such as
ApoE [123–127], metals, or small amounts of preaggregated peptide that can
act as a seed for precipitation [28]. Some of the other components that induce
aggregation include glycosaminoglycans (GAGs) [128,129], zinc [130,131],
Apo-E [123,125,132,133], glycation [134,135], or (non-Aβ protein component
precursor (NACP) [136]. These processes occur during normal aging that can be
accompanied by changes of brain microenvironments, including β-peptide con-
centrations, pH, and membrane integrity.
The binding between heparan sulfate proteoglycans and the β-peptide is
extremely rapid below pH 7 [128,129,137], and analysis of the precipitates of
β-peptide containing sulfate ions or heparan showed that the sulfate moieties of
the GAG chains promote extensive lateral aggregation of β-peptides [138]. The
heparan sulfate proteoglycans are components in the amyloid plaques. It was
proposed that the interactions of GAGs and the β-peptide may stabilize one of
the secondary structures and/or accelerate the aggregation properties of the β-
peptide. Unpublished data from our laboratory established that the β-sheet struc-
ture reacts with heparin to produce an amyloidlike precipitate. Related studies
with other peptides have shown that different GAGs can markedly influence their
solution conformations and folding properties [139–141], in that heparin pro-
moted formation of a β-sheet structure rather than an α-helical structure. Molecu-
lar modeling and NMR studies showed that the His14 and Lys16 side chains,
which both reside within the presumed heparan sulfate binding motif [138] are
critical to the binding interaction of the GAGs and the β-peptide. At physiological
pH, both the α-helix and random coil structures do not bind to heparin (Scheme
1). The general consensus is that amyloid fibril formation is promoted by the
binding of components to the β-sheet structure. The components that promote
or retard aggregation and toxicity of the β-peptide have recently been summarized
[18].
If the β-peptide encounters a hydrophobic environment, similar to TFE or
micelles, then an α-helical structure should form. This situation may occur in
human plasma when the β-peptide becomes attached to lipoproteins, albumin,
ApoJ, or transthyretin [75,114,142]. It is thought that the β-peptide injures cells
by a surface membrane effect [9,143] and membrane-bound β-peptide that is
closely associated with GM1 ganglioside forms a catalytic seed for amyloid for-
TM

mation [144,145]. Moreover, a recent report using synchrotron Fourier transform
infrared (FTIR) microspectroscopy showed that significant quantities of α-helical
structure of the β-peptide are present in the brain gray matter [42]. In all probabil-
ity, the α-helical structure seen in the brain should be identical with the NMR
structure seen with the synthetic β-(1–42) peptide in SDS. As mentioned before,
the SDS micelle adequately mimics a biologically relevant, membranelike envi-
ronment and is extensively employed in biophysical investigations of peptide and
protein structures [72,73]. The NMR structure of the β-(1–42) is defined by an
extended structure (residues 1–9), α-helix (residues 10–24), loop (residues 25–
27), and α-helix (residues 28–42). Most likely, this structure is nontoxic, since
it is monomeric, ordered, and very soluble [146]. The binding interactions of the
α-helical structure with other macromolecules such as lipoproteins are probably
desirable, since they prevent the β-peptide from aggregating into a β-sheet struc-
ture that is toxic and precipitates as amyloid.
We believe that the α-helical structure is an appropriate target for the design
of amyloid inhibitors. The α-helical structure is monomeric and stable, thus suit-
able for modern structure-based drug design methods [147]. As shown with our
work on the nicotine inhibition to amyloidosis [95], the NMR work established
that nicotine binds to the His residues of the α-helical structure, which provides
a rationale for the loss of β-sheet production and precipitation. It should be
pointed out that evidence to support such a binding interaction in the brain is
not yet available. However, if the binding occurs, then a reduction in the α-
helix → β-sheet conversion should occur, which would prevent or slow down
amyloidosis. Related α-helix → β-sheet conversions for other peptides and pro-
teins [148,149], including peptide segments of the prion proteins [150,151], are
well documented. In water solution, freshly prepared, monomeric β-(1–42) pep-
tide adopts mixtures of α-helix, random coil, and β-sheet structures that are in
equilibrium. The precipitation drives the equilibrium toward β-sheet structure,
so once precipitation has begun the α-helical structure will rearrange to the oligo-
meric β-sheet structure [α-helix → β-sheet (solution) → β-sheet (precipitate)].
Once plaque formation begins, the oligomeric β-pleated sheet structure that forms
is resistant to further proteolysis and turnover [37,152]. It may happen that the
monomeric β-sheet/β-turn and/or the early soluble β-sheet structures that form
before the plaque are likewise resistant to proteolysis, similar to soluble peptide
segments of the amyloid forming prion proteins [151].
V. CONCLUSIONS
There are currently no effective treatments for AD, and the current drug develop-
ment process is often cumbersome, taking an average 9 years for approval [153].
The methods discussed in this review, notably the solution NMR technique, can
TM

provide critical experimental data to confirm that a newly designed compound
binds to a protein receptor [79,80], in a manner consistent with pure computa-
tional approaches [147,154]. The NMR data may be an important prerequisite
to the further development of additional compounds. In closing, we believe that
the approaches discussed in this review, which involve combined molecular mod-
eling and NMR, will assist in the design of less toxic analogues to prevent β-
amyloidosis in AD.
ACKNOWLEDGMENTS
I would like to thank the many excellent colleagues who completed the studies
shown in this review. These include Colin Barrow, Takashi Iwashita, and Vassil
Vassilev from the Suntory Institute for Bioorganic Research, Osaka, Japan, and
likewise Keith Marcinowski, Joseph Talafous, Haiyan Shao, Kan Ma, Arthur
Salomon, Shu-chuan Jao, Charlene Keane, Erin Clancy, Jing Yang and Rob
Friedland from Case Western Reserve University, Cleveland, Ohio. The work
was supported in part by grants from the American Health Assistance Foundation,
the Suntory Institute for Bioorganic Research, the National Institutes of Aging
(AG-08992-06 and AG-14363-01), American Federation of Aging Research,
Philip Morris, Inc., Gliatech, Inc., Pfizer, Inc., Smokeless Tobacco Agency, and
a Faculty Scholars Award from the Alzheimer’s Association (FSA-94-040). The
600-MHz NMR spectrometer was purchased with funds provided by the National
Science Foundation, the National Institutes of Health, and the state of Ohio. I
would also like to thank Anita Hong (Anaspec, Inc.), Larry Sayre, and Witold
Surewicz for helpful discussions. Finally, I would like especially to thank Koji
Nakanishi for his sound advice, particularly his support and encouragement
throughout my stay in his groups in both New York and Japan.
REFERENCES
1. J. Sipe, Annu. Rev. Biochem., 61: 947 (1992).

2. E. D. Eanes and G. G. Glenner, J. Histochem. Cytochem., 16: 673 (1968).
3. H. Inouye, P. E. Fraser, and D. A. Kirschner, Biophys. J., 64: 502 (1993).
4. J. W. Kelly and P. T. Lansbury, Amyloid, 1: 186 (1994).
5. J. W. Kelly, Structure, 5: 595 (1997).
6. K.-M. Pan, M. Baldwin, J. Nguyen, M. Gasset, A. Serban, D. Groth, I. Mehlhorn,
Z. Huang, R. J. Gletterick, F. E. Cohen, and S. B. Prusiner, Proc, Natl. Acad. Sci.
USA, 90: 10962 (1993).
7. C. J. Barrow and M. G. Zagorski, Science, 253: 179 (1991).
TM

8. C. Soto, E. M. Castano, B. Frangione, and N. C. Inestrosa, J. Biol. Chem., 270:
3063 (1995).
9. D. J. Selkoe, Annu. Rev. Neurosci., 17: 489 (1994).
10. L. Hendriks and C. Van Broeckhoven, Eur. J. Biochem., 237: 6 (1996).
11. J. Hardy, Trends Neurosci., 20: 154 (1997).
12. C. Haass, M. G. Schlossmacher, A. Y. Hung, C. Vigo-Pelfrey, A. Mellon, B. L.
Ostaszewski, I. Lieberburg, E. H. Koo, D. Schenk, D. B. Teplow, and D. J. Selkoe,
Nature, 359: 322 (1992).
13. M. Shoji, T. E. Golde, T. T. Cheung, J. Ghiso, S. Estus, and L. M. Shaffer, Science,
258: 126 (1992).
14. P. Seubert, C. Vigo-Pelfrey, F. Esch, M. Lee, H. Dovey, and D. Davis, Nature,
359: 325 (1992).
15. J. Busciglio, D. H. Gabuzda, P. Matsudaira, and B. A. Yanner, Proc. Natl. Acad.
Sci. USA, 90: 2092 (1993).
16. G. G. Glenner and C. W. Wong, Biochem. Biophys. Res. Commun., 120: 885
(1984).
17. C. L. Masters, G. Simms, N. A. Weinman, G. Multhaup, B. L. McDonald, and K.
Beyreuther, Proc. Natl. Acad. Sci. USA, 82: 4245 (1985).
18. L. L. Iversen, R. J. Mortishire-Smith, S. J. Pollack, and M. S. Shearman, Biochem.
J., 311: 1 (1995).
19. A. Goate, M.-C. Chartier-Harlin, and M. Mullan, Nature, 349: 704 (1991).
20. E. Levy-Lahad, W. Wasco, P. Poorkaj, D. M. Romano, J. Oshima, W. H. Pettingell,
C.-e. Yu, P. D. Jondro, S. D. Schmidt, K. Wang, A. C. Crowley, Y.-H. Fu, S. Y.
Guenette, D. Galas, E. Nemens, E. M. Wijsman, T. D. Bird, G. D. Schellenberg,
and R. E. Tanzi, Science, 269: 973 (1995).
21. E. Levy-Lahad, E. M. Wijsman, E. Nemens, L. Anderson, K. A. B. Goddard,
J. L. Weber, T. D. Bird, and G. D. Schellenberg, Science, 269: 970 (1995).
22. R. Sherrington, E. I. Rogaev, Y. Liang, E. A. Rogaeva, and G. Levesque, Nature,
375: 754 (1995).
23. A. Tamaoka, T. Kondo, A. Odaka, N. Sahara, N. Sawamura, K. Ozawa, N. Suzuki,
S. Shoji, and H. Mori, Biochem. Biophys. Res. Commun., 205: 834 (1994).
24. N. Suzuki, T. T. Cheung, X.-D. Cai, A. Odaka, L. Otvos, C. Eckman, T. E. Golde,
and S. G. Younkin, Science, 264: 1336 (1994).
25. K. Duff, C. Eckman, C. Zehr, X. Yu, C.-M. Prada, J. Perez-tur, M. Hutton, L.
Buee, Y. Harigaya, D. Yager, D. Morgan, M. N. Gordon, L. Holcomb, L. Refolo,
B. Zenk, J. Hardy, and S. Younkin, Nature, 383: 710 (1996).
26. D. Scheuner, C. Eckman, M. Jensen, X. Song, N. Suzuki, T. D. Bird, J. Hardy, M.
Hutton, W. Kukull, E. Larson, E. Levy-Lahad, M. Viitanen, E. Peskind, P. Poorkaj,
G. Schellenberg, R. Tanzi, W. Wasco, L. Lannfelt, D. Selkoe, and S. Younkin,
Nat. Med., 2: 864 (1996).
27. H. W. Klafki, D. Abramowski, R. Swoboda, P. A. Paganetti, and M. Staufenbiel,
J. Biol. Chem., 271: 28655 (1996).
28. J. T. Jarrett and P. T. Lansbury, Cell, 73: 1055 (1993).
29. S. W. Snyder, U. S. Ladror, W. S. Wade, G. T. Wang, L. W. Barrett, E. D. Matay-
oshi, H. J. Huffaker, G. A. Krafft, and T. F. Holzman, Biophys, J., 67: 1216 (1994).
30. C. Pike, A. Walencewicz, C. Glabe, and C. Cotman, Brain Res., 563: 311 (1991).
TM

31. L. K. Simmons, P. C. May, K. J. Tomaselli, R. E. Rydel, K. S. Fuson, E. F. Brig-
ham, S. Wright, I. Lieberburg, G. W. Becker, D. N. Brems, and W. Li, Mol. Phar-
macol., 45: 373 (1994).
32. C. J. Pike, A. J. Walencewicz-Wasserman, J. Kosmoski, D. H. Cribbs, C. G. Glabe,
and C. W. Cotman, J. Neurochem., 64: 253 (1995).
33. J. F. Flood, J. E. Morley, and E. Roberts, Proc. Natl. Acad. Sci. USA, 88: 3363
(1991).
34. N. W. Kowall, M. F. Beal, J. Busciglio, L. K. Duffy, and B. A. Yanker, Proc. Natl.
Acad. Sci. USA, 88: 7247 (1991).
35. H. Braak, E. Braak, J. Bohl, and R. Reintjes, Neurosci. Lett., 210: 87
(1996).
36. M. Knauer, B. Soreghan, D. Burdick, J. Kosmoski, and C. Glabe, Proc. Natl. Acad.
Sci. USA, 89: 7437 (1992).
37. C. Nordstedt, J. Näslund, L. O. Tjernberg, A. R. Karlström, J. Thyberg, and L.
Terenius, J. Biol. Chem., 269: 30773 (1994).
38. T. Wisniewski, J. Ghiso, and B. Frangione, Neurobiol. Aging, 15: 143 (1994).
39. D. A. Kirschner, C. Abraham, and D. J. Selkoe, Proc. Natl. Acad. Sci. USA, 83:
503 (1986).
40. P. D. Gorévic, E. M. Castano, R. Sarma, and B. Frangione, Biochem. Biophys. Res.
Commun., 147: 854 (1987).
41. P. T. Lansbury, Biochemistry, 31: 6865 (1992).
42. L.-P. I. Choo, D. L. Wetzel, W. C. Halliday, M. Jackson, S. M. LeVine, and
H. H. Mantsch, Biophys. J., 71: 1672 (1996).
43. D. A. Kirschner, H. Inouye, L. K. Duffy, A. Sinclair, M. Lind, and D. J. Selkoe,
44. C. Hilbich, B. Kisters-Woide, J. Reed, C. L. Masters, and K. Beyreuther, J. Mol.
Biol., 218: 149 (1991).
45. D. Burdick, B. Soreghan, M. Kwon, J. Kosmoski, M. Knauer, A. Henschen,
J. Yates, C. Cotman, and C. Glabe, J. Biol. Chem., 267: 546 (1992).
46. E. Terzi, G. Hölzemann, and J. Seelig, Biochemistry, 33: 1345 (1994).
47. A. E. Roher, M. O. Chaney, Y.-M. Kuo, S. D. Webster, W. B. Stine, L. J. Haver-
kamp, A. S. Woods, R. J. Cotter, J. M. Tuohy, G. A. Krafft, B. S. Bonnell, and
M. R. Emmerling, J. Biol. Chem., 271: 20631 (1996).
48. C. J. Barrow, A. Yasuda, P. T. M. Kenny, and M. G. Zagorski, J. Mol. Biol., 225:
1075 (1992).
49. P. E. Fraser, J. T. Nguyen, W. K. Surewicz, and D. A. Kirschner, Biophys. J., 60:
1190 (1991).
50. L. J. Otvos, G. I. Szendrei, V. M. Y. Lee, and H. H. Mantsch, Eur. J. Biochem.,
211: 249 (1993).
51. J. P. Lee, E. R. Stimson, J. R. Ghilardi, P. W. Mantyh, Y.-A. Lu, A. M. Felix, W.
Llanos, A. Behbin, M. Cummings, M. van Criekinge, W. Timms, and J. E. Maggio,
Biochemistry, 34: 5191 (1995).
52. J. E. Maggio, and P. W. Mantyh, Brain Pathol., 6: 147 (1996).
53. J. T. Jarrett, E. P. Beger, and P. T. Lansbury, Biochemistry, 32: 4693 (1993).
54. M. Bodanszky and A. Bodanszky, The Practice of Peptide Synthesis, Springer-
Verlag, Berlin, (1994).
TM

55. S.-C. Jao, K. Ma, J. Talafous, R. Orlando, and M. G. Zagorski, Amyloid, 4: 239
(1997).
56. M. R. Harrigan, D. D. Kunkel, L. B. Nguyen, and A. T. Malouf, Neurobiol. Aging,
16: 779 (1995).
57. J. Busciglio, A. Lorenzo, and B. A. Yanker, Neurobiol. Aging, 13: 609
(1992).
58. C. J. Pike, D. Burdick, A. J. Walencewicz, C. G. Glabe, and C. W. Cotman, J.
Neurosci., 13: 1676 (1993).
59. J. R. Wujek, M. D. Dority, R. C. A. Frederickson, and K. R. Brunden, Neurobiol,
Aging, 17: 107 (1996).
60. C. Soto, E. M. Castaño, R. A. Kumar, R. C. Beavis, and B. Frangione, Neurosci.
Lett., 200: 105 (1995).
61. C.-L. Shen, M. C. Fitzgerald, and R. M. Murphy, Biophys. J., 67: 1238 (1994).
62. C.-L. Shen and R. M. Murphy, Biophys. J., 69: 640 (1995).
63. N. Greenfield, and G. D. Fasman, Biochemistry, 8: 4108 (1969).
64. R. W. Woody, Circular Dichroism of Peptides and Proteins, VCH Publishers, New
York (1994).
65. K. Halverson, P. E. Fraser, D. A. Kirschner, and P. T. Lansbury, Biochemistry, 29:
2639 (1990).
66. E. Terzi, G. Hölzemann, and J. Seelig, J. Mol. Biol., 252: 633 (1995).
67. R. W. Storrs, D. Truckses, and D. E. Wemmer, Biopolymers, 32: 1695 (1992).
68. A. Jasanoff and A. R. Fersht, Biochemistry, 33: 2129 (1994).
69. A. Cammers-Goodwin, T. J. Allen, S. L. Oslick, K. F. McClure, J. H. Lee, and
D. S. Kemp, J. Am. Chem. Soc., 118: 3082 (1996).
70. M. G. Zagorski and C. J. Barrow, Biochemistry, 31: 5621 (1992).
71. M. A. Hemminga, J. C. Sanders, and R. B. Spruijt, Prog. Lipid Res., 31: 301 (1992).
72. P. A. McDonnell and S. J. Opella, J. Magn. Reson., B102: 120 (1993).
73. G. D. Henry and B. D. Sykes, Methods Enzymol., 239: 520 (1994).
74. J. Talafous, K. J. Marcinowski, G. Klopman, and M. G. Zagorski, Biochemistry,
33: 7788 (1994).
75. A. L. Biere, B. Ostaszewski, E. R. Stimson, B. T. Hyman, J. E. Maggio, and
D. J. Selkoe, J. Biol. Chem., 271: 32916 (1996).
76. P. E. Fraser, D. R. McLachlan, W. K. Surewicz, C. A. Mizzen, A. D. Snow, J. T.
Nguyen, and D. A. Kirschner, J. Mol. Biol., 244: 64 (1994).
77. K. Wüthrich, NMR of Proteins and Nucleic Acids, John Wiley, New York, (1986).
78. D. A. Case and P. E. Wright, Determination of High-Resolution NMR Structures
of Proteins, CRC, Ann Arbor, Mich. (1993).
79. S. B. Shuker, P. J. Hajduk, R. P. Meadows, and S. W. Fesik, Science, 274: 1531
(1996).
80. D. J. Craik, in Drug Design, in (W. S. Hancock ed.), CRC Series in Analytical
Biotechnology, CRC Press, Boca Raton, Fla. (1996).
81. H. Sticht, P. Bayer, D. Willbold, S. Dames, C. Hilbich, K. Beyreuther, R. W. Frank,
and P. Rösch, Eur. J. Biochem., 233: 293 (1995).
82. T. Kohno, K. Kobayashi, T. Maeda, K. Sato, and A. Takashima, Biochemistry, 35:
16094 (1996).
83. T. G. Fletcher and D. A. Keire, Protein Science, 6: 666 (1997).
TM

84. D. Neuhaus and M. P. Williamson, The Nuclear Overhauser Effect in Structural
and Conformational Analysis, VCH, New York (1989).
85. G. Wagner, S. G. Hyberts, and T. F. Havel, Annu. Rev. Biophys. Biomol. Struct.,
21: 167 (1992).
86. A. Bax and S. Grzesiek, Acc. Chem. Res., 26: 131 (1993).
87. D. S. Wishart, B. D. Sykes, and F. M. Richards, Biochemistry, 31: 1647 (1992).
88. F. J. Blanco, J. Herranz, G. Carlos, M. A. Jiménez, M. Rico, J. Santoro, and J. L.
Nieto, J. Am. Chem. Soc., 114: 9676 (1992).
89. E. L. Clancy, K. Ma, K. Wollenberg, and M. G. Zagorski, submitted.
90. A. E. Roher, J. D. Lowenson, S. Clarke, A. S. Woods, R. J. Cotter, E. Gowing,
and M. J. Ball, Proc. Natl. Acad. Sci. USA, 90: 10836 (1993).
91. S. A. Gravina, L. Ho, C. B. Eckman, K. E. Long, L. Otvos Jr., L. H. Younkin, N.
Suzuki, and S. G. Younkin, J. Biol. Chem., 270: 7013 (1995).
92. M. Citron, T. S. Diehl, G. Gordon, A. L. Biere, P. Seubert, and D. J. Selkoe, Proc.
Natl. Acad. Sci. USA, 93: 13170 (1996).
93. S. J. Wood, L. MacKenzie, B. Maleef, M. R. Hurle, and R. Wetzel, J. Biol. Chem.,
271: 4086 (1996).
94. A. Lorenzo and B. A. Yankner, Proc. Natl. Acad. Sci. USA, 91: 12243 (1994).
95. A. R. Salomon, K. J. Marcinowski, R. P. Friedland, and M. G. Zagorski, Biochemis-
try, 35: 13568 (1996).
96. P. Plateau and M. Gueron, J. Am. Chem. Soc., 104: 7310 (1982).
97. A. B. Graves, C. M. van Duijn, V. Chandra, L. Fratiglioni, A. Heyman, A. F. Jorm
et al., Int. J. Epidemiol., 20 suppl. 2: S48 (1991).
98. C. Van Duijn and A. Hofman, Br. Med. J., 302: 1491 (1991).
99. R. P. Friedland, Neurobiol. Aging, 15: 239 (1994).
100. D. Brenner, W. Kukull, G. van Belle, J. Bowen, W. McCormick, L. Teri, and E.
Larson, Neurology, 43: 293 (1993).
101. R. P. Friedland, Neurobiol. Aging, 15: 239 (1994).
102. A. B. Graves and J. A. Mortimer, J. Smoking-Related Disord., 5: 79 (1994).
103. R. T. Jones, Tobacco Dependence, Raven Press, New York (1987).
104. D. R. Lide, CRC Handbook of Chemistry and Physics, 71st ed., CRC Press, Boca
Raton, Fla. (1991).
105. C. R. Cantor and P. R. Schimmel, Biophysical Chemistry, Part 1. The Conformation
of Biological Macromolecules, W. H. Freeman, New York (1980).
106. S. J. Perkins, Applications of Ring Current Calculations to the Proton NMR of
Proteins and Transfer RNA, Plenum Press, New York, (1982).
107. C. E. Johnson and F. A. Bovey, J. Chem. Phys., 29: 1012 (1958).
108. G. A. Kyerematen and E. Vesell, S., Nicotine Metab. Rev., 23: 3 (1991).
109. A. Karlin and M. H. Akabas, Neuron, 15: 1231 (1995).
110. H. D. Lacorazza, M. S. O. D. Bengtsson, and M. B. D. J. Bonino, Neurochem. Int.,
20: 521 (1992).
111. T. L. Lentz, Biochemistry, 34: 1316 (1995).
112. C. B. Bouzat, H. D. Lacorazza, M. B. J. Bonino, and F. J. Barrantes, Pflügers Arch.,
423: 365 (1993).
113. T. L. Lentz and B. Conti-Fine, personal communication.
114. A. L. Schwarzman, L. Gregori, M. P. Vitek, S. Lyubski, W. J. Strittmatter, J. J.
TM

Enghilde, R. Bhasin, J. Silverman, K. H. Weisgraber, P. K. Coyle, M. G. Zagorski,
J. Talafous, M. Eisenberg, A. M. Saunders, A. D. Roses, and D. Goldgaber, Proc.
Natl. Acad. Sci. USA, 91: 8368 (1994).
115. D. J. Selkoe, J. NIH Res., 7: 57 (1995).
116. G. Giaccone, B. Pedrotti, A. Migheli, L. Verga, J. Perez, G. Racagni, M. A. Smith,
G. Perry, L. DeGioia, C. Selvaggini, M. Salmona, J. Ghiso, B. Frangione, K. Islam,
O. Bugiani, and F. Tagliavini, Am. J. Pathol., 148: 79 (1996).
117. J. Hardy, Proc. Natl. Acad. Sci. USA, 94: 2095 (1997).
118. S. J. Tomski and R. M. Murphy, Arch. Biochem. Biophys., 294: 630 (1992).
119. L. M. Sayre, M. G. Zagorski, W. K. Surewicz, G. A. Krafft, and G. Perry, Chem.
Res. Toxicol., 10: 518 (1997).
120. P. C. Southwick, S. K. Yamagata, C. L. Echols, G. J. Higson, S. A. Neynaber,
R. E. Parson, and W. A. Munroe, J. Neurochem., 66: 259 (1996).
121. W. P. Esler, E. R. Stimson, J. R. Ghilardi, Y.-A. Lu, A. M. Felix, H. V. Vinters,
P. W. Mantyh, J. P. Lee, and J. E. Maggio, Biochemistry, 35: 13914 (1996).
122. C. M. Yates, J. Butterworth, M. C. Tennant, and A. Gordon, J. Neurochem., 55:
1624 (1990).
123. J. Ma, A. Yee, J. H. Bryan Brewer, S. Das, and H. Potter, Nature, 372: 92 (1994).
124. P. L. Richey, S. L. Siedlak, M. A. Smith, and G. Perry, Biochem. Biophys. Res.
Commun., 208: 657 (1995).
125. K. C. Evans, E. P. Berger, C.-G. Cho, K. H. Weisgraber, and P. T. Lansbury, Proc.
Natl. Acad. Sci. USA, 92: 763 (1995).
126. S. J. Wood, W. Chan, and R. Wetzel, Biochemistry, 35: 12623 (1996).
127. A. A. Golabek, C. Soto, T. Vogel, and T. Wisniewski, J. Biol. Chem., 271: 10602
(1996).
128. A. D. Snow and T. N. Wight, Neurobiol. Aging, 10: 481 (1989).
129. K. R. Brunden, N. J. Richter-Cook, N. Chaturvedi, and R. C. A. Frederickson, J.
Neurochem., 61: 2147 (1993).
130. P. W. Mantyh, J. R. Ghilardi, S. Rogers, E. DeMaster, C. J. Allen, E. R. Stimson,
and J. E. Maggio, J. Neurochem., 61: 1171 (1993).
131. A. I. Bush, W. H. Pettingell, G. Multhaup, M. Paradis, J.-P. Vonsattel, J. F. Gusella,
K. Beyreuther, C. L. Masters, and R. E. Tanzi, Science, 265: 1464 (1994).
132. W. J. Strittmatter, K. H. Weisgraber, D. Y. Huang, L.-M. Dong, G. S. Salvesen,
M. Pericak-Vance, D. Schmechel, A. M. Saunders, D. Goldgabr, and A. D. Roses,
133. C. Soto, E. M. Castaño, F. Prelli, R. A. Kumar, and M. Baumann, FEBS Lett., 371:
110 (1995).
134. M. A. Smith, S. Taneda, P. L. Richey, S. Miyata, S.-D. Yan, D. Stern, L. M. Sayre,
V. M. Monnier, and G. Perry, Proc. Natl. Acad. Sci. USA, 91: 5710 (1994).
135. M. P. Vitek, K. Bhattacharya, J. M. Glendening, E. Stopa, H. Vlassara, R. Bucala,
K. Manogue, and A. Cerami, Proc. Natl. Acad. Sci. USA, 91: 4766 (1994).
136. M. Yoshimoto, A. Iwai, D. Kang, D. A. C. Otero, Y. Xia, and T. Saitoh, Proc.
Natl. Acad. Sci. USA, 92: 9141 (1995).
137. L. Buee, W. Ding, A. Delacourte, and H. Fillit, Brain. Res., 601: 154 (1993).
138. P. E. Fraser, J. T. Nguyen, D. T. Chin, and D. A. Kirschner, J. Neurochem., 59:
1531 (1992).
TM

139. R. A. Gelman and J. Blackwell, Arch. Biochem. Biophys., 159: 427 (1973).
140. R. A. Gelman and J. Blackwell, Biopolymers, 12: 1959 (1974).
141. A. L. Stone, Fed. Proc., 36: 101 (1977).
142. E. Matsubara, B. Frangione, and J. Ghiso, J. Biol. Chem., 270: 7563 (1995).
143. D. H. Cribbs, C. J. Pike, S. L. Weinstein, P. Velazquez, and C. W. Cotman, J. Biol.
Chem., 272: 7431 (1997).
144. K. Yanagisawa, A. Odaka, N. Suzuki, and Y. Ihara, Nat. Med., 1: 1062 (1995).
145. J. McLaurin and A. Chakrabartty, J. Biol. Chem., 271: 26482 (1996).
146. Z. S. Farhangrazi, H. Ying, G. Bu, L. L. Dugan, A. M. Fagan, D. W. Choi, and
D. M. Holtzman, NeuroReport, 8: 1127 (1997).
147. C. L. Verlinde and W. G. Hol, Structure, 2: 577 (1994).
148. P. Fan, C. Bracken, and J. Baum, Biochemistry, 32: 1573 (1993).
149. A. Graf v. Stosch, M. A. Jiménez, V. Kinzel, and J. Reed, Proteins Struct. Funct.
Genet., 23: 196 (1995).
150. J. Nguyen, M. A. Baldwin, F. E. Cohen, and S. B. Prusiner, Biochemistry, 34: 4186
(1995).
151. H. Zhang, K. Kaneko, J. T. Nguyen, T. L. Livshits, M. A. Baldwin, F. E. Cohen,
T. James, and S. B. Prusiner, J. Mol. Biol., 250: 514 (1995).
152. R. A. Crowther, Biochem. Biophys. Acta, 1096: 1 (1991).
153. N. R. Cutler, and J. J. Sramek, Drug Development Must Be Optimized in Order to
Choose the Correct Strategy, John Wiley, New York (1995).
154. J. G. Vinter, and M. Gardner, in Topics in Molecular and Structural Biology, (S.
Neidle, W. Fuller and J. Cohen eds., CRC Press, Boca Raton, Fla. (1994).
TM

15
Bacteriorhodopsin Structure/
Function Studies: Use of the
Demethyl Retinal Analogues for
Probing of the Arg82Ala Mutant
Rosalie K. Crouch, Donald R. Menick, and Yan Feng

Medical University of South Carolina, Charleston, South Carolina
Rajni Govindjee and Thomas G. Ebrey
University of Illinois at Urbana-Champaign, Urbana, Illinois
I. INTRODUCTION
Bacteriorhodopsin (bR), the only protein in the purple membrane of the archae-
bacterium Halobacterium salinarium, is a retinal-based, light-transducing pig-
ment [1]. The dark adapted form contains roughly equivalent amounts of the all-
trans and 13-cis isomers of retinal (1, Fig. 1), the aldehyde of vitamin A, as its
chromophore [2]. The protein functions as a light driven transmembrane proton
pump and is of quite some interest for its potential to harvest light energy. The
discovery that bR with its seven transmembrane α-helices has the same spatial
arrangement as the large class of biologically important G-protein receptors has
further increased interest in this protein. In addition, the protein is easily grown
and purified, making it readily available for a wide range of spectral and physio-
logical studies.
The photocycle of all-trans bR [WT (wild type)/1a] initiates with the isom-
erization of all-trans retinal to the 13-cis isomer and consists of a sequence of
at least five steps (Fig. 2). The 13-cis pigment [WT/1] is also photochemically
active and undergoes a distinct photocycle [3] in which proton translocation is
not observed [4]. The photocycle of the all-trans isomer has been studied in the
TM

Figure 1 Structures of retinal analogues: 1, R 1 ⫽ R2 ⫽ methyl, in retinal; 2, R 1 ⫽
hydrogen, R 2 ⫽ methyl, in 9-demethylretinal; 3, R 1 ⫽ methyl, R 2 ⫽ hydrogen, in 13-
demethylretinal; 4, R 1 ⫽ R 2 ⫽ hydrogen, in 9,13-demethylretinal.
Figure 2 Photocycle of bR showing the main intermediates. bR, bacteriorhodopsin.
TM

greatest detail as it is the one associated with the transport of protons. A complica-
tion of these photocycles is that the photointermediates themselves undergo pho-
toreversible reactions [5]. The exact role of the chromophore in the proton trans-
location process is still not completely defined although it is clear that the
chromophore in bR and in its various photointermediates is capable of light-
induced isomerization that results in changes in the protein conformation. Little
is currently understood of the chromophore–protein interactions inducing these
conformational changes.
There are two analogue-type approaches that can be taken to probe the
structure–function questions regarding this protein: the chromophore can be de-
rivatized or the protein can be mutated (see Fig. 3 for a general scheme). The
native chromophore of bR is retinal (1, vitamin A aldehyde), which is a small
molecule (molecular weight [MW] ⬍ 300) containing a conjugated polyene sys-
tem. This native chromophore can be easily replaced with a suitably derivatized
retinal [6], and bR has been shown to accept a wide variety of retinal analogues
(Fig. 4). The binding site is selective for the polyene chain as pigments cannot
be formed with analogues that have significant additional bulk along this chain.
However, the portion of the binding site that accommodates the methylated cyclo-
hexyl ring is quite flexible as pigments have been formed with a number of ring-
derivatized analogues. The approach of replacing the native chromophore with
retinal derivatives has been extensively utilized by Nakanishi and his group, both
at Columbia and later in his students’ own laboratories [7–9].
Figure 3 Schematic of two approaches to modification of bacteriorhodopsin for

structure/function studies.
TM

Figure 4 Schematic of structural requirements of retinal analogues for pigment forma-
tions with bR. bR, bacteriorhodopsin.
9-Demethylretinal (2) has previously been reported to form a bR pigment

that has a blue shifted absorption maximum (20 nm) [10,11]. The wild-type pig-
ment containing 2 (WT/2) appeared to function normally with the same proton
pumping action as the native pigment. Using a mutant bR in which the tryptophan
182 (Trp 182) was mutated to a phenylalanine, Yamazaki et al. [12] have shown
that there is a strong interaction of the 9-methyl group with this residue. Thus,
FTIR studies show that the L intermediate is diminished and M formation is
delayed. These data indicate that the 9-methyl is located near Trp 182 at this step
of the photocycle.
A stable bR pigment has also been reported to form with 13-demethylretinal
(WT/3). Gärtner et al. [10,13] found that 85% of the chromophore was in the
13-cis conformation in WT/3, in contrast to the pigment with 2, which had an
isomeric ratio close to that for the native bR (WT). These investigators concluded
that the 13-methyl of retinal is critical for the chromophore isomerization. This
analogue therefore provides an excellent tool for the study of systems involving
the 13-cis conformation. Proton pumping has been found to be reduced to about
20% that of the native pigment [14]. Removal of methyls from both the 9- and
13-positions generated a pigment that resembled the 2 pigment in absorption
properties [11].
Site-directed mutagenesis is a tool that has been extensively explored in the
study of bR. Numerous mutants have been made, and the study of the biophysical
properties of these mutants has led to much of our current understanding of the
TM

structure and function of bR [15–17]. The arginine at position-82 (R82) has been
found to be key to both the absorption and proton pumping properties of bR. This
amino acid has been found to stabilize the ionized state of aspartate at position 85,
the proton acceptor and part of the counterion to the Schiff base. The aspartate
85 is deprotonated in purple membrane, but decreasing the pH results in the
formation of blue membrane and protonation of D85 [18,19]. Replacing the argi-
nine with the neutral residue alanine (R82A) strongly affects the shift in the pKa
of the purple to blue transition, shifting it from 2.6 in WT to 7.2–7.5 in the
mutant [20,21]. The result is that the mutant R82A is inactive at pH ⬎ 6.
We have combined the use of site-directed mutagenesis and regeneration
with retinal derivatives to examine structure/function relationships in the re-
sulting pigments. The chromophore has been altered by deletion of the retinal
side chain methyls, one of which is known to affect the isomeric composition
of the chromophore in the pigment. The mutant studied is R82A, in which the
critical charged amino acid arginine has been mutated to a neutral residue alanine.
In each case, the analogue pigment has been compared to its corresponding pig-
ment with either the native chromophore or the native protein.
II. MATERIALS AND METHODS

A. Synthesis of Retinals
The demethyl retinals [22] (all-trans isomers) were synthesized as outlined in
Fig. 5. This synthetic route is similar to those used for the syntheses of numerous
retinal analogues. Diethyl (3-cyano-2-methylprop-2-enyl)phosphonate was pre-
pared from 1-chloro-3-cyano-2-methylpropene by heating with triethylphosphite.
Sodium bis(dimethylsilyl)amide in dry tetrahydrofuran (Aldrich) was used as the
base and diisobutylaluminumhydride (DiBal; Aldrich) was used as the reducing
agent. The products were purified by TLC (5% ethyl acetate/hexane) and/or
HPLC)(Varian 5000 gradient system, Altech Econosphere Silica SU column, 250
⫻ 6 mm, 0.9% ethylacetate, 0.1% methanol/99% hexane). The retinals were
characterized by nuclear magnetic resonance data from a Varian VXR400 spec-
trometer with deuterated chloroform as solvent. The analogues were used as soon
as possible after preparation but, when necessary, were stored at ⫺70°C under
argon. The purity was confirmed by HPLC and absorption spectra before use.
All experiments with retinals were performed under dim red light.
B. Bacterial Strains and Growth Conditions

Escherichia coli BMH71-18 mutL [23] was used for site-directed mutagenesis,
E. coli JM101 was used to propagate mutant phages, and competent E. coli Nova
Blue (Novagen, Madison, WI) was used for transformation of the pT7Blue vector
TM

(Novagen, Madison, WI) containing polymerase chain reaction–(PCR)-generated
fragments. All E. coli strains were grown in 2X YT or LB medium [24]. The
wild-type and R82A bop genes were expressed in Halobacterium salinarium IV-
8 [25] (a gift of Richard Needleman), which contains a stable ISH1 insert in the
bop gene [26].
C. Site-Directed Mutagenesis of bR
Oligonucleotide-directed, site-specific mutagenesis was performed essentially as
described by Menick [27]. The 2.7-kb BamHI-HindIII bop gene insert was re-
stricted from the H. salinarium shuttle vector pMC-1 (a generous gift of R. Nee-
dleman) and ligated into M13mp19. The Arg82 was replaced with Ala by using
the mutagenic primer 5′CTGG GCG GCG TAC GCT GAC 3′, which contained
the indicated two-nucleotide mismatch. Mutants were initially screened by colony
blot hybridization and then plaque purified, and the mutation was verified by
dideoxy sequencing [28]. The entire coding region of the bop gene was sequenced
to ensure that no additional mutations had occurred. The BamHI-HindIII bop
gene insert was restricted from the M13mp19RF and religated into pMC-1.
D. Transformation of Halobacterium salinarium IV-8

The R82A bR mutant and wild-type control were transformed into the H. sali-
narium strain IV-8 containing ISH1 insert within the bop gene as described
[21,29]. The spheroplasts were pelleted by centrifugation, resuspended in growth
medium (plus 15% sucrose) [25], and incubated at 37°C for 48 h. Aliquots were
plated on growth medium agar plates with 15% sucrose and 5 mg/mL mevinolin
(a generous gift of Merck Sharp & Dohme). Several independent clones were
picked (after 2 weeks of growth) and struck out on growth medium plates; single
colony isolates were grown in 2 L of grown medium to characterize each bR
mutant.
E. Southern and Sequence Analysis of the R82A Mutant

Southern blot analysis was carried out as described [24]. Genomic deoxyribonu-
cleic acids (DNAs) from H. salinarium cells IV-8 and IV-8 cells transformed
with wild-type or the R82A bop mutant were digested with PstI, transferred to
Genescreen Plus membranes, and hybridized with a 32 P-labeled 500-bp KpnI bop
gene fragment. Additionally, the R82A bop gene was amplified by polymerase
chain reaction from the R82A transformed IV clones. The PCR was performed
with 1 mL of sheared H. salinarium genomic DNA as template as previously
described [30]. The 1150-bp PCR product was gel purified and sequenced.
TM

F. Isolation of Purple Membrane
Purple membrane was also isolated from cultures of H. salinarium by the method
of Becker and Cassim [31]. Purple membrane was prepared from the two sources
(cultures of H. salinarium IV-8 transformed with wild-type and R82A bop) as
described [32] and purified with an additional sucrose gradient (30%–60% su-
crose, 100,000 g, 17 h).
G. Pigment Regeneration
The native and transformed pigments were bleached according to the method of
Tokunaga et al. [6]. Pigments were regenerated by additions of the all-trans reti-
nals in ethanol (⬍1% vol) to bleached membranes (50 mM Tris-HCl, pH 8.8) at
25°C. Pigment formation was measured after 24 h by absorption changes with
bleached membrane as reference.
H. Spectroscopic Methods
Absorption spectra were recorded in 5-mm quartz cuvettes thermostatted at 20°C
using a Cary-Aviv 14 spectrophotometer (Aviv Associates, Lakewood, NJ) by
digital measurement with 1-nm steps and a spectral bandwidth of 1 nm. Kinetic
measurements of dark adaptation were done at 580 nm using the kinetic mode
of the spectrophotometer. A homemade cryostat was used for low-temperature
measurements [33].
Flash-induced absorption changes were measured on a kinetic spectropho-
tometer [34,35]. The actinic pulse was from a Quanta Ray DCR11 Nd:YAG laser
(532 nm, 7 ns, 5–10 mJ/pulse; Spectra Physics, Mountain View, CA). Flash-
induced pH changes were measured at pH 7–7.2 using the pH-sensitive dye py-
ranine.
III. RESULTS AND DISCUSSION
The retinal analogues were synthesized as outlined in Fig. 5. This scheme is quite
similar to those used in other laboratories for the syntheses of various retinal
analogues and relies primarily on Horner-Emmons olefination for chain elonga-
tion. Yields of the various steps were between 40% and 65%, but the products
were seldom isolated other than for an initial characterization, because of the
lability of these polyene compounds.
Stable pigments were formed between bacteriorhodopsin and the trans iso-
mer of all three retinal analogues 2–4. The absorption maxima of these WT pig-
ments were similar to those reported by Gärtner et al. [10] (Table 1). Pigments
TM

Figure 5 Synthesis of retinal analogues. Route 1, synthesis of 9-demethylretinal (2);
route 2, synthesis of 13-demethylretinal (3); route 3, synthesis of 9,13-demethylretinal
(4).
TM

Table 1 Absorption of Pigment Analogues a
Absorption maximum (nm)
Retinal (1) 2 3 4
Native WT 558 540 569 539
Transformed WT 558 540 567 538
R82A 555 537 566 539
a
All pigments were regenerated with the all-trans isomer of the respective retinal. Pigment solutions
were dark adapted for 1 h before measurement. Spectra were recorded in 50-mM-Tris-HCl, pH 8.8
at 24°C. Values ⫾2 nm. WT, wild type.
were formed with both bR apoprotein generated from the normal purple mem-
brane purified from cultures of the halobacteria and from protein of H. salinarium
IV-8 transformed with the WT bop gene. The transformed and native pigments
had identical absorption properties. The data presented here are from pigments
regenerated with the apoprotein isolated from the native pigment. The pigments
formed with the apoprotein of R82A and the three demethyl retinals showed
absorption maxima at 537 nm (2), 566 nm (3), and 539 nm (4), respectively
(Table 1). The pigments were reasonably stable in the presence of 10-mM hy-
droxylamine, pH 7.5, with only modest degradation noted for the pigments
formed with 4 (⬍25% over 2 h). The deletion of the 9-methyl group shifted the
absorption maxima in both of the pigments, whereas the removal of the 13-methyl
and the mutation at the 82-position had little effect on the absorption properties.
The R82A pigment has a pK a for the purple to blue transition of 8.7 in
25% glycerol/water (Fig. 6). For the R82A/3, the pK a was shifted to 8.1. The
origin of this shift can be attributed to the finding that for the 3 pigment most
of the chromophore is probably in the 13-cis conformation and the pK a of the
13-cis form of the pigment is lower than that for the all-trans form [36].
The photocycle intermediates were detected by monitoring the flash-in-
duced absorption changes at 410 nm, 540 nm, and 600 nm. The WT changes
(Fig. 7a) represent predominantly M (410 nm) and O (640 nm) intermediates as
well as the recovery of the pigment (540 nm). The WT/2 bR converted to a 410-
nm M intermediate and a 600-nm species, as a result of the O intermediate, after
20 ms (Fig. 7c). The formation of the M intermediate is in agreement with the
results of Yamazaki et al. [12].
Excitation of WT/3 produced only a small amount of the 410-nm M inter-
mediate species (Fig. 7e). The half-time of decay observed for this species was
about 3 s, whereas the decay of M in the WT bR is 10 ms. The predominant
conversion of WT/3 was to a 600-nm species that is much less in WT/2 (Fig.
7c). It has been proposed that this 600-nm species, which is different from the
TM

Figure 6 Fraction of blue membrane as a function of pH in dark-adapted R82A (2) and
R82A/3 (1).
O intermediate of the trans cycle, arises from the 13-cis cycle. [7,12] This identi-
fication of the photoproduct and a small amount of M is consistent with the fact
that the 13-cis isomer accounts for at least 80% of the retinal in the binding
site of 3 pigment. Furthermore, upon the illumination of WT/4 pigment, the M
intermediate species from the trans cycle was quite small and only an unusual,
long-lived 600-nm species was found (Fig. 7g). These results of flash-induced
spectroscopic experiments are consistent with the conclusion that the 13-position
of retinal is important for the stability of the all-trans configuration of retinal in
the binding site [7].
In R82A, the light-induced absorption change at 410 nm decayed much
faster than the absorption at 580 nm recovered, and an O intermediate was not
observed (Fig. 7b). Interestingly, a rapidly forming 600-nm species is observed
in all three R82A/2–4 pigments (Fig. 7d, f, and h), indicating an increase in the
proportion of the 13-cis isomer in the pigment. The R82A/3 and R82A/4 pig-
ments have particularly long-lived 600-nm species.
The flash-induced proton changes were measured by the transient absorp-
tion changes of the pH indicator pyranine at pH 7.2. The fast (⬍1 ms) decrease
in dye absorbance observed in WT corresponds to release of protons, and
the subsequent restoration of absorbance (rise time 12 ms) corresponds to the
uptake of protons (Fig. 8a) [11]. The signals from all three WT/2–4 pigments
were ⬍20% that of WT (Fig. 8b, c, and d). The proton signal from R82A was
about 40% that of WT, and the order of proton release and uptake is reversed
TM

Figure 7 Flash-induced absorption changes: a, WT; b, R82A; c, WT/2; d, R82A/2; e,
WT/3; f, R82A/3; g, WT/4; h, R82A/4. Measuring wavelengths as marked, λ actinic 532
nm, temperature 20°C, 0.1 M NaCl, pH 7.
TM

Figure 8 Flash-induced absorption changes of the pH-sensitive dye pyranine showing
proton release/uptake: a, WT; b, WT/2; c, WT/3; d, WT/4; e, R82A; f, R82A/4. Measur-
ing wavelengths at 458 nm with other conditions as in Fig. 4.
compared to that of the WT, as observed previously (Fig. 8e) [11]. No absorption
change of the dye in R82A/3 was observed (Fig. 8f), demonstrating that the
combination of the deletion of the 13-methyl and the removal of the charged
amino acid at position-82 prevents the light-induced movement of protons for
this pigment.
IV. CONCLUSIONS
In conclusion, the use of analogues of retinal lacking the 9- and/or 13-methyl in

combination with the R82A mutation has allowed us to examine the photocycle
TM

and function of pigments in which the 13-cis to all-trans isomeric ratio is expected
to be increased. The 9-methyl is critical to maintaining the absorption maximum
of bR, and the environment of interaction that causes this shift has not been
altered in this mutant. The photocycles of the pigments are affected in that the
R82A/2 shows an inhibition of intermediate M formation, and all the R82A/2–
4 pigments show the 600-nm species. Proton pumping is eliminated in the R82A/
3 pigment. These results support previous findings on the importance of the
Arg82 residue to the function of bR and illustrate the value of the combination
of the two analogue approaches in probing the complicated chromophore–protein
interactions in the proton pumping process.
ACKNOWLEDGMENTS
This work was supported by NIH (EY04939 and GM52023), DOE (95ER20171),
and Research to Prevent Blindness, Inc. We thank Dr. R. Needleman for provid-
ing the shuttle vector containing the bop gene, pMC-1, and bR-defective H. sali-
narium strain, IV-8. We particularly thank Dr. Koji Nakanishi, without whose
initial guidance and enthusiasm this collaboration would not have existed, and
therefore these studies would not have been accomplished.
REFERENCES
1. D. Oesterhelt and W. Stoeckenius, Nature, 233NB: 149 (1971).

2. For recent reviews, (a) T. G. Ebrey, Thermodynamics of Membrane Receptors and
Channels; (M. Jackson, ed.), CRC Press, Boca Raton, Fla. pp 353–386. (1993).
(b) J. K. Lanyi and G. Varo, Israel J. Chem., 35: 365 (1996).
3. C. Gergely, C. Ganea and G. Varo, Biophys. J., 67: 855 (1994).
4. A. Fahr and E. Bamberg, FEBS Lett., 140: 251 (1982).
5. S. P. Balashov, Isr. J. Chem., 35: 429 (1996).
6. F. Tokunaga, R. Govindjee, T. G. Ebrey, and R. K. Crouch, Biophys. J., 19: 191
(1977).
7. W. Humphrey, I. Logunov, K. Schulten, and M. Sheves, Biochemistry, 33: 3668
(1994).
8. Y. Feng, D. R. Menick, B. M. Katz, C. J. Beischel, E. S. Hazard, S. Misra, T. G.
Ebrey and R. K. Crouch, Biochemistry, 33: 11624 (1994).
9. For a recent review: K. Nakanishi and R. K. Crouch, Isr. J. Chem., 35: 253 (1996).
10. W. Gärtner, P. Towner, H. Hopf and D. Oesterhelt, Biochemistry, 22: 2637 (1983).
11. M. M. Szweykowska, A. D. Broek, J. Lugtenburg, R. L. van der Bend and P. W. M.
van Dijek, Recl. Trav. Chim. Pays-Bas. 102: 42 (1983).
12. Y. Yamazaki, J. Sasaki, M. Hatanaka, H. Kandori, A. Maeda, R. Needleman, Shi-
nada, K. Yoshihara, L. S. Brown, and J. K. Lanyi, Biochemistry, 34: 577 (1995).
13. H.-W. Trissl and W. Gärtner, Biochemistry, 26: 751 (1987).
TM

14. Y. Feng, R. K. Crouch, B. Katz, and D. R. Menick, Biophys. J., 64: 213a (1993).
15. R. A. Mathies, S. W. Lin, J. B. Ames, and W. T. Pollard, Annu. Rev. Biophys. Chem.,
20: 491 (1991).
16. J. K. Lanyi, J. Bioenerg. Biomembr., 24: 169 (1992).
17. D. Oesterhelt, J. Tittor, and E. Bamberg, J. Bioenerg. Biomembr., 24: 181 (1992).
18. S. Subramaniam, T. Marti, and H. G. Khorana, Proc. Natl., Acad. Sci. USA. 87:
1013 (1990).
19. L. S. Brown, L. Bonet, R. Needleman, and J. K. Lanyi, Biophys, J., 65: 124 (1993).
20. L. Stern and H. G. Khorana, J. Biol. Chem., 264, 14202, (1989).
21. S. P. Balashov, R. Govindjee, M. Kono, E. Imasheva, E. Lukashev, T. G. Ebrey,
R. K. Crouch, D. R. Menick, and Y. Feng, Biochemistry, 32: 10331 (1993).
22. P. J. van den Tempel and H. O. Huisman, Tetrahedron, 46: 233 (1966).
23. B. Kramer, W. Kramer, and H. J. Fritz, Cell, 38: 879 (1984).
24. F. M. Ausubel, R. Brent, R. Kingston, D. D. Moore, J. G. Seidman, J. A. Smith,
and K. Strul, Current Protocols in Molecular Biology, Wiley, New York (1989).
25. R. Needleman, M. Chang, B. Ni, G. Varo, J. Fornes, S. White, and J. Lanyi, J. Biol.
Chem., 266: 11478 (1991).
26. S. DasSarma, U. L. Raj Bhandary, and H. G. Khorana, Proc. Natl. Acad. Sci. USA,
81: 125 (1984).
27. D. R. Menick, in Methods in Nucleic Acids Research (Karam, J. D., Chao, L., Warr,
G. W., eds.), CRC Press, Boca Raton Fla., pp. 45–59 (1991).
28. F. Sanger, S. Nickelen, and A. R. Coulson, Proc. Natl. Acad. Sci. USA, 74: 5463
(1977).
29. B. Ni, M. Chang, A. Duscht, J. Lanyi, and R. Needleman, Gene, 90: 169 (1990).
30. R. K. Saiki, in PCR Protocols: A Guide to Methods and Applications, Academic
Press, New York, pp. 13–20, (1990).
31. B. Becker and J. Y. Cassim, Prep. Biochem., 5: 161 (1975).
32. D. Oesterhelt and W. Stoeckenius, Methods Enzymol., 31: 667 (1974).
33. S. P. Balashov, R. Govindjee, and T. G. Ebrey, Biophys. J., 60: 475 (1991).
34. Z. Dancshazy, R. Govindjee, B. Nelson, and T. G. Ebrey, FEBS Lett. 209: 44 (1986).
35. R. Govindjee, S. P. Balashov, and T. G. Ebrey, Biophys. J., 58: 597 (1990).
36. S. P. Balashov, E. S. Imasheva, R. Govindjee, and T. G. Ebrey, Biophys. J., 70: 473
(1996).
TM

16
Autonomous Genomes
David G. Lynn
The University of Chicago, Chicago, Illinois
I. INTRODUCTION
In 1995, the first complete genomic sequence was published [1], and since then
several others have become available. Of particular importance to the bioorganic
chemist, this list includes Escherichia coli [2] and Bacillus subtilis [3], the organ-
isms whose proteins and molecular genetics have been used so extensively to
define and characterize the chemical reactions of biology. These primary se-
quence data will certainly extend and enrich our understanding of the reactions
that drive these organisms, but it also opens an entirely new set of questions. It
is likely that no scientific discipline will benefit more directly from this knowl-
edge than bioorganic chemistry and chemical biology. Just as the structure and
the synthesis of the steroid nucleus placed human reproduction and hormone
physiological processes at the atomic level, and the structure of proteins revealed
the exquisite sophistication of the mechanisms of biological catalysts, the struc-
ture of the genome, the molecule that codes for the entire organism, places the
most central questions of biology on an atomic scale.
More than just understanding the molecular machines of biology, it is now
possible to consider how chromosomes function: to wonder about the logic for
the organization and positions of genes that are functionally interdependent, the
function of intervening sequences that tie the coding regions together, the reason
for the positioning of the origin and terminus of replication, the functional sig-
nificance of G-C-rich domains; the positions, frequencies, and function of re-
peating sequences; and the organization and utility of coding redundancy. With
data on multiple genomes, strategies for organization and stability can be com-
pared and the rules governing success may be understood. Questions as to how
TM

genome organization differs across a genus, a family, a kingdom; how an entire
genome responds to major environmental changes; how stages and specific rates
of molecular evolution occur across the genome can be experimentally addressed.
The genome is a far larger structure than a steroid or an enzyme, and differ-
ent methods are needed for its analysis. In this chapter I would like to consider
several approaches that we have explored to understand genome function. These
approaches will be presented in the context of the minimal structure for an auton-
omous genome [4–9]. The small genomes were the first to be sequenced, and
these are already providing new glimpses of theoretical evolutionary genomics
and the rooting of evolutionary trees [6–9]. But the term autonomous is slippery
in a biological context. For example, Morowitz [4] suggested that the smallest
cell is likely to contain the minimal genome and, making estimates of average
macromolecule volume, proposed that Mycoplasma genitalium, with a 0.3 mm-
diameter, could encode only 600 proteins. This genome sequence was published
in 1995 and contains 469 genes [10], putting a lower limit on the number of
catalysts needed for an autonomous cellular entity. But M. genitalium is a parasite
and contains, for example, very few genes necessary for the synthesis of its amino
acids and nucleic acid bases. Its autonomy, then, depends very much on the envi-
ronment in which the genome functions. Once the environment becomes equivo-
cal, then a virus, a plasmid, or a viroid becomes an autonomous genome within
the host cell.
At the molecular level, then, the minimal structure for genome function is
context-dependent and is a chemical issue. Although it is within biology that the
term genome has evolved and must be understood, it is within the limits of the
chemical environment that the reactions necessary for autonomy must be defined.
Much insight will be gained over the next few years as genome sequences are
compared, but just as with the steroid nucleus, it is within the context of the
construction of new structures that build on the operational principles uncovered
within these genomes that we will be able to advance beyond what exists today
and to explore the limits of what is autonomous and what is chemically possible.
II. TOP-DOWN GENOME REDUCTION
Maniloff [6] argued that small autonomous genomes could have arisen along two
independent paths, which he referred to as ‘‘top-down’’ and ‘‘bottom-up.’’ The
bottom-up path is essentially an origin of life argument, to which I will return.
The top-down path requires the evolutionary development of symbiotic, parasitic,
or other interdependent associations such that genomic information can be re-
duced. The biosphere today is a tightly woven network of interdependent organ-
isms. Plants can utilize CO 2, O 2, and NO 2, inorganic materials, and photons for
energy, seemingly functioning in the absence of other organisms. There are many
TM

associations, cellular organelles, micorrhizal/plant symbioses, lichens, legume/
rhizobium symbioses, insect vectored pollination strategies, etc., in which inter-
dependence appears to have evolved. In many of these systems there is an obvious
driver for the association, but the actual evolutionary pathway and the molecular
and mechanistic changes that allowed for its development are less clear. The
parasitic angiosperms provide a situation that may allow for a molecular level
description of the traits that led to this loss of genome autonomy and of the
resulting evolutionary consequences.
Searcy and MacInnis proposed three general developmental phases for the
molecular evolution of the parasitic plants [11,12]. The first consisted of the de-
velopment of the specialized organ that forms the physiological bridge to the
host—the haustorium. Once formed, the subsequent events are of specialization
and adaptation. These events include the loss of biochemical pathways and mor-
phological structures that become redundant with host attachment, and the accrual
of more complex adaptations specific to an obligate parasite, such as host special-
ization and mechanisms to overcome host defenses.
Analyses of the nonphotosynthetic holoparasite Epifagus virginiana have
provided striking support for evolutionary reduction [13,14]. The plastid deoxyri-
bonucleic acid (DNA) (ptDNA) of this organism was found to be approximately
a third the size of a nonparasitic relative, Nicotiana tabacum, and essentially all
of the plastid encoded photosynthetic genes, the reduced nicotinamide-adenine
dinucleotide phosphate (NADPH) dehydrogenase genes, all four ribonucleic acid
(RNA) polymerase genes, 13 of the 30 plastid encoded tRNAs, and 6 of the 22
plastic encoded ribosomal protein genes were either missing completely or pres-
ent as pseudogenes [15,16]. Accelerated rates of molecular evolution of both the
remaining plastid genes and the rDNA of these parasites have also been docu-
mented and explained by molecular population genetic models of mutation, selec-
tion, and drift [17,18].
Significant biochemical evidence also exists for highly specialized pro-
cesses for host selection. A component exuded from the roots of sorghum seed-
lings was shown to induce Striga asiatica seed germination and was the first
characterized host-derived stimulant, the sorghum xenognosin for Striga spp.
germination (SXSg) [19,20]. This component, however, was not sufficient to ex-
plain the distance regulation observed for S. asiatica seed germination in culture,
and mathematical models of host commitment uncovered an additional compo-
nent, which ironically proved to be produced via the same biosynthetic pathway
as SXSg [21,22]. Several other examples of host specialization have been docu-
mented [23], and, more recently, mechanisms for overcoming host resistance
have been detected [24–27].
By Searcy’s model, the development of the host attachment organ was the
critical first step leading to the subsequent genomic adaptations. Current estimates
are that 1% of all flowering plants are parasitic [13,28], broadly distributed across
TM

at least 16 families [29], suggesting that the parasitic strategy, driven by haustorial
development, was repeatedly discovered throughout angiosperm evolution [23].
The identification of 2,6-dimethoxy-p-benzoquinone (DMBQ) as a natural sor-
ghum xenognosin for Striga haustoria (SXSh) that was both necessary and suffi-
cient for haustorial induction in several of the parasitic Scrophulariaceae [20,30]
suggested that oxidative release from host cell walls provided an important sig-
naling strategy [31]. It has now been shown that the S. asiatica root meristem
constitutively produces H 2O 2 (Fig. 1). This oxidant has been hypothesized to
serve as a cosubstrate, along with host cell wall phenols, for apoplastic localized
host peroxidases to generate the xenognostic signals [32]. The released quinones
mediate host cell surface detection (Fig. 2). This strategy would provide an effec-
tive means for the parasite to identify a viable host and to respond in a distance-
dependent manner to the host root surface [33].
H 2O 2 has been shown to be produced at the cell wall during pathogen inva-
sion [34], and two pathways are known for its production. Apoplastic peroxidases
[35,36] generate H 2O 2 via an H 2O 2-independent oxidation of a number of sub-
strates, including NADH, NADPH, thiols, and certain phenols [36–40]. At pres-
Figure 1 Light micrograph of a 2-day-old Striga asiatica seedling that has been stained
with pyrogallol. The red coloration [shown here as a darkening at the root meristem (see
arrow)], of the root meristem highlights the H 2O 2 production. (Source: Ref. 32.)
TM

Figure 2 The proposed mechanism for haustorial induction involves a four-step autocat-
alytic cycle: (a) H 2O 2 is produced at the parasite root meristem and diffuses to the host,
(b) the H 2O 2 serves as a substrate for host apoplastic peroxidases that oxidatively remove
wall bound phenolics at kpx[H 2O 2][PhOH], (c) some percentage of these phenols are oxi-
dized to quinones that diffuse to the parasite, and (d) a parasite oxidoreductase uses the
quinones to produce H 2O 2 at k et[Q]. The parasite oxidoreductase is coupled to a signal
transduction chain that induces haustorial development.
ent there is no evidence for the exudation of these more easily oxidized substrates
to the parasite cell surface. Evidence for a membrane-bound NADPH oxidase
complex, homologous to the human neutrophil oxidative burst machinery, has
also been found in plants [41–44]. In that regard, SXSh quinones were recently
shown to be detected via a benzoquinone-dependent oxidoreductase [45,46]. In
terms of the observed redox range (Fig. 3), the terminal acceptor could be O 2,
via a cytochrome-mediated reduction. It may be, therefore, that signal detection
is coupled with oxidative signal release through the production of H 2O 2 in an
autocatalytic host detection mechanism. It has been suggested that a mutation in
the oxidative burst machinery controlling H 2O 2 production in a dicotyledonous
plant has made it possible for the parasite to exploit monocot cell surfaces via
a dicot defense pathway. A good defense is converted into an offensive host
TM

Figure 3 The active haustorial inducers fall within a narrow first half-wave reductive
potential window, between 20 and ⫺280 mV, consistent with their role as redox carriers.
(Source: Ref. 46.)
detection strategy [32]. Such a gain of function mutation could have easily
evolved and played a critical role in the development of parasitism. Further analy-
sis of the oxidoreductase genes responsible for H 2O 2 production in Striga should
allow the model to be evaluated.
The study of Striga spp. makes possible a structure/activity analysis of the
parasitic genome. By defining the elements that lead to parasitism, a functional
understanding for the process of genome reduction can be obtained. Through
comparisons with other related parasites, both the rate of gene loss and the re-
sulting changes in genomic organization may be understood. Analogous ‘‘com-
pare and contrast’’ approaches have taught us much about the forces that drive
biology, and it is clear that these analyses have now moved to an atomic level.
In our case, the genomic reorganization that occurs between parasitic and nonpar-
asitic relatives provides glimpses into the structure/function analysis of their ge-
nome.
TM

III. BOTTOM-UP GENOME CONSTRUCTION
The soil bacterium Agrobacterium tumefaciens can harbor an extrachromosomal

genome, the Ti plasmid, which is responsible for coupling the bacterium and a
plant cell into a disease state known as the crown gall tumor. An elaborate meta-
bolic interface is developed with the bacterium, which results in the oncogenic
transformation of a plant cell host [47–49]. Together the two organisms are coor-
dinated to ensure the survival of the Ti plasmid.
Fig. 4, presented as a timeline, outlines the signaling events that coordinate
this interface. In the left panel, the first signal is generated from a dividing compe-
Figure 4 The timeline for the infection process of Agrobacterium tumefaciens. Across
the top, the oncogenetically transformed plant cell displays tumorigenic growth. In the
left-hand panel, the gene transfer event is induced by specific signaling molecules (repre-
sented by phenol) that originate from the plant cell. The T-DNA is transferred after activa-
tion of the vir regulon of the Ti plasmid (pTiA6). In the middle panel, octopine is produced
by the transformed plant cell and activates both occ and traR expression. In the right-
hand panel, CF, a factor whose biosynthetic genes are constitutively expressed from Ti,
together with TraR, activates conjugal transfer of the Ti plasmid.
TM

Figure 5 Comparisons of the two gene transfer events in Agrobacterium spp. The onco-
genetic transfer (a) integrates multiple inputs, of which the phenol signal is most critical
as the initiation event. Conjugal transfer (b) shows an autocatalytic amplification step to
induce the machinery for gene transfer. The genetically encoded machinery for both trans-
fer processes is homologous.
tent plant cell and recognized via the activation of the virulence regulon of the
Ti plasmid. A histidine two-component sensor kinase and response regulator are
constitutively expressed from the regulon and integrate information from mono-
saccharide, phenol, and acidic pH to control expression of the remaining genes
(Fig. 5). The virulence regulon encodes the machinery that catalyzes the excision
of the T-DNA, its transfer into the plant cell, and targeting aspects of its incorpo-
ration into the genome of the plant cell. The center panel shows the result of
incorporation of the T-DNA. Driving the prokaryotic genes for critical steps in
the biosynthesis of auxin and cytokinin behind eukaryotic promoters leads to the
uncontrolled growth of the plant cell. In addition to the oncogenes, an expressed
dehydrogenase catalyzes the reductive amination of pyruvate with arginine to
generate octopine, a substance not metabolized by the plant, but that can serve
as the sole carbon and nitrogen source for A. tumefaciens that harbor the catabolic
genes of the Ti plasmid. Octopine accumulation activates both the expression of
the octopine catabolism regulon (occ) and the regulatory protein, TraR. TraR and
a homoserine lactone adduct autoinducer, AI, constitutively biosynthesized from
TM

S-adenosyl methionine and the acyl-CoA keto ester by gene products of the Ti
[50], are required to activate the expression of the tra regulon [51]. This regulon
controls transfer of the entire Ti plasmid, as shown in the right-hand panel via
an autocatalytic amplification of the machinery for conjugal transfer (Fig. 5).
Given the homology that exists between the two gene transfer machines shown
in Fig. 5, the process has little specificity with respect to recipient cell and trans-
fers genes to both prokaryotic and eukaryotic recipients.
Thus, the Ti plasmid, the original tumor inducing principle of Armin Braun,
successfully orchestrates an interface between eukaryotic and prokaryotic cells
and, once the economic advantage of opine production is engineered, delivers
itself to any cell capable of exploiting the advantage. The only requirements for
the host cell are the replication and expression of Ti. Ti is therefore an autono-
mous genome, but its autonomy is context-dependent. The eukaryotic host plant
range is very broad, virtually for any dicotyledonous plant, and as such A. tumefa-
ciens provides a general vector for higher plant transformation [47]. The bacterial
host, however, is restricted to A. tumefaciens. In a bottom-up approach, Ti can
be explored as a scaffold on which increased capability can be added to make a
more autonomous genome.
Bacterial host range competence may be limited by several factors, but a
central one involves the initial step of establishing the interface with the plant,
activation of the VirA/VirG signal transduction chain. Two different experimen-
tal approaches have suggested that VirA and VirG alone are not sufficient for
signal transmission. First, analysis of the known inducing structures led to a pro-
posed mechanistic-model for induction [52]. From this model a series of specific
irreversible inhibitors were prepared and developed into affinity reagents. These
affinity reagents did not label VirA, but did label a series of low-molecular-weight
chromosomally encoded proteins in vivo within a time frame consistent with the
physiological inhibition. Therefore, VirA was proposed not to be the xenognostic
receptor [53,54]. More recently, A. tumefaciens mutants have been selected for
increased sensitivity to the signal molecule. These mutants show an enhanced
response to specific methoxy geometric isomers of the designed inducers. This
trait has been localized to the chromosomal background [55]. The identification
of chromosomal loci that are critical for signal perception suggests that A. tumefa-
ciens maintains some parental control over the promiscuous Ti plasmid. Defining
those loci should provide an understanding of the origin of this apparent control
and further raise the possibility of cloning these genes back into Ti to extend its
bacterial host range.
Plasmids represent mobile elements whose molecular genetic manipulation
is now highly developed, and methods are available to extend their capability.
They were among the first genetic elements to be synthetically engineered and
that synthetic potential is still growing. The advantage with Ti is the great flexi-
bility of the DNA transfer complexes, and increased replication autonomy repre-
TM

sents a biotechnological potential that will allow for the molecular entraining of
other organisms. Such synthetic approaches, which share a conceptual link with
the synthesis of alternate functional groups on the steroid nucleus, should ulti-
mately extend to the construction of larger increasingly autonomous genomes.
IV. EXTENDING THE PRINCIPLES TO OTHER SKELETONS
As the requirements of a functioning genome are understood, it becomes possible

to develop rationally, as has been the case with steroid hormone action, different
architectures capable of genome function. Conceptually, the storage and transfer
of molecular information, copying, reading, and translation, are all template-di-
rected polymerization reactions. In biological systems these reactions are medi-
ated by a complex array of catalysts, and kinetic control of both the rate and the
fidelity of the information transfer is critical. However, in a primordial bottom-
up approach to genome construction, species capable of such catalysis would
not be present and must have evolved from thermodynamically controlled self-
associations.
Several model templates have shown features of self-replication, including
nucleotide-based dimers and oligomers [56–64], peptides [65,66], and micelles
[67]. The goal of many of these studies has been to demonstrate exponential
growth of the encoded information, and this effort has seen limited success be-
cause template turnover in replication is limited by product dissociation. In ex-
isting biological systems separate catalysts evolved that use chemical energy to
drive template dissociation prior to subsequent rounds of replication and, in this
way, achieve exponential growth.
The evolution of catalysts is therefore central, and in considering possible
biopolymer catalysts, both DNA and RNA are candidates [68–73]. The range of
reactions catalyzed by the nucleic acid polymers is however, more limited than
that of proteins for at least two reasons. First, the narrower range of side chain
functional groups and the more specific associations of the existing side chains
in Watson–Crick and Hoogsteen pairing motifs limit their catalytic availability.
Second, the hydration requirements of the polyanionic phosphate backbone reduce
its ability to pack into rigid structures. Hence, binding site definition, substrate
binding and release, as well as large-scale and rapid conformational rearrange-
ments important for allostery in proteins are more limited in nucleic acids [74].
The thermodynamic stability of Watson–Crick duplexes is significant, and
we have harnessed this energy to drive reversible imine condensation [75,76].
Under reducing conditions, the imine is trapped as the amine, and the base se-
quence information encoded within a DNA heximer is irreversibly transferred
with good fidelity (Fig. 6). By defining the thermodynamic stability of the amine
and imine linkage [77,78], conditions were developed in which product inhibition
TM

Figure 6 The thermodynamic and catalytic cycles for translating DNA information into
the altered backbone product. The substrates are labeled with 15 N to allow the assignment
of the equilibrium constants, K, for each step. DNA, deoxyribonucleic acid. (Source: Ref.
79.)
of template turnover was reduced [79] such that the template efficiently catalyzed
the amplification of the encoded information. This opportunity to have multiple
turnovers on the template allows for the information to be transferred accurately
and amplified many times with catalytic amounts of template.
This ligation reaction gives a different product backbone from the template
and, in that way, has characteristics of a translation process. To extend this pro-
cess to replication would require the amine product to catalyze the construction
of the template as shown in Fig. 7. The ethyl amine linkage is more hydrophobic
than the phosphate diester, and its reduced hydration and increased flexibility
TM

Figure 7 Proposed two-step replication process in which the phosphate backbone poly-
mer catalyzes the synthesis of the amine backbone and the amine backbone polymer cata-
lyzes the synthesis of the phosphate backbone.
should enable polymers containing such linkages to fold and pack in a way that
is more like that of proteins. The basic amine in the linkage should expand the
catalytic competence of the polymer much as the 2′-OH does in RNA [68,74].
Therefore it may be possible to reduce the central dogma of biology, DNA to
RNA to protein, to a two-step process as shown in Fig. 7.
The ligation efficiency of the reaction outlined in Fig. 6 has allowed us to
consider new approaches to template-directed synthesis, even a synthetic ap-
proach that extends the architectural limits of the bottom-up construction of a
genome. Now the term autonomy becomes more relevant because this genome
would be synthetic. The extension of the ligation reaction to template-directed
polymerizations would be necessary, and eventually, for the template to be auton-
omous, it must encode the capability of catalyzing a ‘‘metabolism’’ sufficient
for monomer generation [80–82]. Therefore, the central question becomes how
the required diversity of catalysts might be produced from the template. The
principles inherent in the imine coupling reactions can be extended to other cou-
pling reactions, giving different functionality in the backbone of the translation
product. For example, DNA has recently been shown to direct the condensation of
peptide nucleic acid amides under dehydrating conditions [83,84]. These products
have new backbones and represent new translation products that could be selected
for new catalytic function. In other words, one template could be translated into
many different products in which the base sequence is the same, but the different
TM

backbone increases the diversity of catalytic function. In the future, the definition
of autonomy will be solely in the hands of the chemist who defines the environ-
ment, i.e., the access to raw materials and the energy source.
V. CONCLUSIONS
Chemistry is as much a philosophical approach to problem solving as it is a

scientific discipline. As our understanding of biology increasingly approaches an
atomic scale, chemistry’s problem solving approaches become more generally
applicable. Clearly genome structure has radically increased the convergence of
chemistry and biology, and it is the contributions of the fathers of bioorganic
chemistry, including Koji Nakanishi, to whom this volume is dedicated, who have
developed the methods and the ideas that have made this convergence possible.
At the center of chemistry is the ability to create new structures syntheti-
cally and change composition in a rational, definable way. As a result of the
phenomenal synthetic advances that have occurred, notably since the 1960s, the
current methodology boasts incredible diversity. The physical laws that drive life
must operate within the structural limits of molecular frameworks and environ-
mental conditions, and, as I have tried to develop here, this synthetic flexibility
will allow those limits to be expanded continually.
ACKNOWLEDGMENTS
The ideas presented here have grown from discussions and experiments with very
talented coworkers, most of whom are referenced here. I am grateful for a long-
standing and valuable collaboration with Andy Binns, University of Pennsylva-
nia, and support from the NIH, NSF, DOE, Novartis, and the Rockefeller Founda-
tion. Most specifically, I acknowledge Koji Nakanishi for establishing a breadth
of inquiry and an enthusiasm for learning that have so broadly impacted science,
and for the personal support that has made these investigations possible.
REFERENCES
1. R. D. Fleischmann, Science, 269: 496 (1995).

2. F. R. Blattner, G. Plunkett III, C. A. Bloch, Science, 277: 1453 (1997).
3. F. Kunst, N. Ogasawara, I. Moszer, A. M. Albertine, Nature, 390: 249 (1997).
4. H. J. Morowitz, Isr. J. Med. Sci., 20: 750 (1984).
5. M. Itaya, FEBS Lett., 362: 257 (1995).
6. J. Maniloff, Proc. Natl. Acad. Sci. USA, 93: 10004 (1996).
TM

7. A. R. Mushegian and E. V. Koonin, Proc. Natl. Acad. Sci. USA, 93: 10268 (1996).
8. E. V. Koonin and A. R. Mushegian Curr. Opin. Genet. Dev., 6: 757 (1996).
9. S. Razin, Indian J. Biochem. Biophys., 34: 124 (1997).
10. C. M. Fraser, Science, 270: 397 (1995).
11. D. G. Searcy, Evolution, 24: 207 (1970).
12. D. G. Searcy and A. J. MacInnis, Evolution, 24: 796 (1970).
13. J. Kuijt, The Biology of Parasitic Flowering Plants, University of California Press,
Berkeley (1969).
14. A. Cronquist, The Evolution and Classification of Flowering Plants, New York Bo-
tanical Garden, Bronx, N.Y. (1988).
15. C. W. dePamphilis and J. D. Palmer, Nature, 348: 337 (1990).
16. K. H. Wolfe, C. W. Morden, and J. D. Palmer, Proc. Natl. Acad. Sci. USA, 89:
10648 (1992).
17. C. W. dePamphilis, in Parasitic Plants (M. C. Press and J. D. Graves, eds.), Chap-
man & Hall, London, pp. 177–205 (1995).
18. D. L. Nickrent and E.M.C. Starr, J. Mol. Evol., 39: 62 (1994).
19. M. Chang, D. H. Netzly, L. G. Butler, and D. G. Lynn, J. Am. Chem. Soc., 108:
7858 (1986).
20. D. G. Lynn and M. Chang, Annu. Rev. Plant Physiol. Plant Mol. Biol, 41: 497
(1990).
21. G. W. Fate, M. Chang, and D. G. Lynn, Plant. Physiol., 93: 201 (1990).
22. G. W. Fate and D. G. Lynn, J. Am. Chem. Soc., 118: 11369 (1996).
23. L. J. Musselman and M. C. Press, in Parasitic Plants (M. C. Press and J. D. Graves,
eds.), Chapman & Hall, London, pp. 1–13 (1995).
24. A. R. Saunders, Dept. Agric. S. Afr. Bull., 128: 1 (1933).
25. A. Olivier, N. Benhamou, and G. D. Leroux, Can. J. Bot., 69: 1679 (1991).
26. J. H. Visser, I. Dörr, and R. Kollmann, J. Plant Physiol., 135: 737 (1990).
27. J. L. Riopel and M. P. Timko, in Parasitic Plants (M. C. Press and J. D. Graves,
eds.), Chapman & Hall, London, pp. 39–79 (1995).
28. P. R. Atsatt, in Encyclopedia of Plant Physiology (O. L. Lange, P. S. Nobel, C. B.
Osmond, and H. Ziegler, eds.), Springer-Verlag, Berlin, pp. 519–535 (1983).
29. U. Molau, in Parasitic Plants (M. C. Press and J. D. Graves, eds.), Chapman &
Hall, London, pp. 141–176 (1995).
30. M. Chang and D. G. Lynn, J. Chem. Ecol., 12: 561 (1986).
31. J. I. Yoder, Planta, 202: 407 (1997).
32. D. Kim, R. Kocz, L. Boone, W. J. Keyes, and D. G. Lynn, Chem Biol., 5: 103
(1998).
33. J. B-H. Tok, Y-L. Tzeng, K. Lee, Z. Zeng, and D. G. Lynn, in Phytochemicals for
Pest Control (P. Hedin, ed.), ACS Symposium Series 658, Washington, D.C.,
pp. 108–116 (1997).
34. C. S. Bestwick, I. R. Brown, M.H.R. Bennett, and J. W. Mansfield, Plant Cell, 9:
209 (1997).
35. R. Vera-Estrella, E. Blumwald, and V. J. Higgins, Plant Physiol., 99: 1208 (1992).
36. G. P. Bolwell, V. S. Butt, D. R. Davies, and A. Zimmerlin, Free Radic. Res. 23:
517 (1995).
37. B. Halliwell, Planta, 140: 81 (1978).
TM

38. M. Mader and V. Amberg-Fisher, Plant Physiol., 70: 1128 (1982).
39. A. Vianello and F. Macri, J. Bioenerg. Biomembr., 23: 409 (1991).
40. H. Pichorner, A. Couperus, S.A.A. Korori, and R. Ebermann, Phytochem., 31: 3371
(1992).
41. A. Levine, R. Tenhaken, R. Dixon, and C. Lamb, Cell, 79: 583 (1994).
42. C.-K. Auh and T. M. Murphy, Plant Physiol., 107: 1241 (1995).
43. N. Doke and Y. Miura, Physiol. Mol. Plant Pathol., 46: 17 (1995).
44. S. C. Dwyer, L. Legendre, P. S. Low, and T. L. Leto, Biochim. Biophys. Acta, 1289:
231 (1996).
45. C. E. Smith, M. W. Dudley, and D. G. Lynn, Plant Physiol., 93: 208 (1990).
46. C. E. Smith, T. Ruttledge, Z. Zeng, R. O’Malley, and D. G. Lynn, Proc. Natl. Acad.
Sci. USA, 93: 6986 (1996).
47. A. N. Binns and M. F. Thomashow, Annu. Rev. Microbiol., 42: 575 (1988).
48. P. C. Zambryski, Annu. Rev. Plant Physiol. Plant Mol. Biol., 43: 465 (1992).
49. C. Fuqua, S. C. Winans, and E. P. Greenberg, Annu. Rev. Microbiol., 50: 727 (1996).
50. M. I. Moré, L. D. Finger, J. L. Stryker, C. Fuqua, A. Eberhard, and S. C. Winans,
Science, 272: 1655 (1996).
51. K. R. Piper, S. Beck von Bodman, and S. K. Farrand, Nature, 362: 448 (1993).
52. K. M. Hess, M. D. Dudley, D. G. Lynn, R. D. Joerger, and A. N. Binns, Proc. Natl.
Acad. Sci. USA, 88: 7854 (1991).
53. K. Lee, K. M. Hess, M. D. Dudley, D. G. Lynn, R. D. Joerger, and A. N. Binns,
54. K. Lee, Y.-L. Tzeng, N. Liu, M. D. Dudley, and D. G. Lynn, Pure Appl. Chem.,
65: 1241 (1993).
55. A. Campbell, J. B-H Tok, Y. Wang, D. G. Lynn, and A. N. Binns, in preparation
(1998).
56. R. Naylor and P. T. Gilham, Biochemistry, 5: 2722 (1966).
57. L. E. Orgel, Nature, 358: 203 (1992).
58. J.-I. Hong, Q. Feng, V. Rotello, and J. Rebek, Science, 255: 848 (1992).
59. D. Sievers and G. von Kiedrowski, Nature, 369: 221 (1994).
60. T. Achilles and G. von Kiedrowski, Angew. Chem. Int. Ed. Engl., 32: 1198 (1993).
61. T. Li and K. C. Nicolaou, Nature, 369: 218 (1994).
62. G. Ertem and J. P. Ferris, Nature, 379: 238 (1996).
63. Orgel, L. E., Acc. Chem. Res., 28: 109 (1995).
64. D. Philp and J. F. Stoddart, Angew. Chem. Int. Ed. Engl., 35: 1155 (1996).
65. D. Lee, J. R. Granja, J. A. Martinez, K. Severin, and M. R. Ghadiri, Nature, 382:
525 (1996).
66. S. Yao, I. Ghosh, R. Zutshi, and J. Chmielewski, J. Am. Chem. Soc., 119: 10559
(1997).
67. P. A. Bachmann, P. L. Luisi, and J. Lang, Nature, 357: 57 (1992).
68. T. R. Cech, Annu. Rev. Biochem., 59: 543 (1990).
69. R. H. Symons, Annu. Rev. Biochem., 61: 641 (1992).
70. F. Michel and J.-L. Ferat, Annu. Rev. Biochem., 64: 435 (1995).
71. R. R. Breaker and G. F. Joyce, Chem. Biol., 1: 223 (1994).
72. B. Cuenoud and J. W. Szostak, Nature, 375: 611 (1995).
73. Y. Li and D. Sen, Nat. Struct. Biol., 3: 743 (1996).
TM

74. G. J. Narlikar and D. Herschlag, Annu. Rev. Biochem., 66: 19 (1997).
75. J. T. Goodwin and D. G. Lynn, J. Am. Chem. Soc., 114: 9197 (1992).
76. J. T. Goodwin, P. Luo, J. C. Leitzel, and D. G. Lynn, in Self-Production of Supramo-
lecular Structures, from Synthetic Structures to Models of Minimal Living Systems,
Kluwer Academic, Maratea, Italy, pp. 99–104 (1994).
77. P. Luo and D. G. Lynn, submitted (1998).
78. P. Luo, J. C. Leitzel, J. Z-Y. Zhan, and D. G. Lynn, J. Am. Chem. Soc., 120: 3019
(1998).
79. J. Z. Y. Zhan, and D. G. Lynn, J. Am. Chem. Soc., 114: 9197 (1997).
80. C. de Duve, Blueprint for a Cell: The Nature and Origin of Life, Neil Patterson,
New York, NY, chapter 6 (1991).
81. N. Lahav J. Theor. Biol., 151: 531 (1991).
82. S. A. Kauffman, The Origins of Order, Oxford University Press, New York, NY,
chapter 7 (1993).
83. C. Böhler, P. E. Nielsen, and L. E. Orgel, Nature, 376: 578 (1995).
84. J. G. Schmidt, P. E. Nielsen, and L. E. Orgel, J. Am. Chem. Soc., 119: 1494 (1997).
TM

17
Stereochemical Considerations of
Immunoglobulin Heavy Chain
Enhancer Activation
Barbara S. Nikolajczyk* and Ranjan Sen

Brandeis University, Waltham, Massachusetts
I. INTRODUCTION
Immunoglobulin secreting cells, or B cells, differentiate in the bone marrow from

a common blood cell precursor. This developmental process requires changes in
gene expression, one example of which is the B cell–specific expression of the
immunoglobulin µ heavy chain gene (IgH). We have focused on dissecting the
details of transcriptional activation of the IgH gene to gain insight into the mecha-
nisms responsible for B cell development.
B lymphocyte–specific expression of the IgH gene is activated by the µ
enhancer, a deoxyribonucleic acid (DNA) regulatory sequence located in the J H-
C µ intron. The µ enhancer is activated at early stages of B cell differentiation
and participates in targeting the IgH locus for V(D)J recombination close to the
time of B lineage precursor cell commitment [1,2]. The µ enhancer is hypothe-
sized to increase accessibility of the recombinase machinery to the µ locus during
V(D)J recombination. Later in B cell differentiation, after recombination has po-
sitioned a V H promoter close to the C µ locus, the enhancer is required for expres-
sion of the functionally rearranged µ gene as demonstrated by numerous transfec-
tion and transgenic assays. Furthermore, in transgenic assays, the µ enhancer is
sufficient to activate a reporter gene in lymphocytes, but not in other hematopoi-
* Current affiliation: Boston University Medical School, Boston, Massachusetts.
TM

etic cell types, indicating its autonomous cell-type specificity [3–5]. Our objec-
tive is to understand the molecular mechanisms of µ enhancer function.
We have previously defined a core region of the µ enhancer that is active
in B cells [6]. This region contains three elements—µA, µE3, and µB—all of
which are simultaneously required for transcriptional activation. The µA and µB
elements bind ETS domain transcription factors; µE3 binds several widely ex-
pressed leucine zipper-containing basic helix–loop–helix proteins (bHLH-zip).
The ETS domain proteins are a family of DNA binding factors that have amino
acid homology in the DNA binding domain, the ETS domain. Similarly, bHLH-
zip transcription factors are homologous in the bHLH-zip domains, which are
responsible for the dimerization and DNA-binding properties of this protein fam-
ily. Several ETS proteins can bind µA [6–9], and the µB site binds PU.1, an
ETS protein that is highly expressed in B cells and macrophages [10]. None of
the ets or bHLH-zip family members relevant to µ enhancer activation are re-
stricted to pre-B and B cells where IgH genes are expressed. The lack of correla-
tion between the tissue distribution of enhancer binding proteins and cells in
which the enhancer is active led us to propose that tissue specificity of the en-
hancer is achieved by a combinatorial mechanism [6]. Specifically, we proposed
that µ enhancer activation is B cell–specific because the appropriate combination
of transcription factors is present only in B cells and this combination must bind
enhancer DNA in the appropriate three-dimensional arrangement to form a tran-
scriptionally active nucleoprotein complex. Consistent with this idea, multimeri-
zation of µA, µE3, or µB does not reconstitute a regulatory element that activates
transcription in B cells [11–13]. These observations suggest that the properties
of this tripartite enhancer domain are determined by the appropriate juxtaposition
of the three sites.
II. A PROPOSED MECHANISM FOR ␮ ENHANCER

ACTIVATION IN B CELLS
A. ␮A–␮B Spacing Is Critical for Enhancer Activity in B
Cells
Our previous analysis demonstrated that the µA and µB sites are essential for
activation of the murine µ heavy chain gene enhancer [6]. Examination of the
corresponding regions of the murine and human IgH loci showed that both µA
and µB sites were highly conserved (Fig. 1). Interestingly, the spacing between
the µA and µB motifs is evolutionarily conserved; however, we found that the
µE3 site defined in the murine enhancer was absent in the human enhancer. Pres-
ervation of µA–µB spacing even in the absence of a recognizable E box suggested
that correct spatial orientation of these two sites was critical for enhancer func-
tion.
TM

Figure 1 Comparison of murine and human µ enhancer sequences. Note high conserva-
tion of µA and µB and lack of µE3 conservation. Spacing between µA and µB is identical.
To test the hypothesis that the organization of the µA and µB sites was
critical for enhancer activity in B cells, we constructed mutated µ enhancers with
either 4 or 10 base pairs inserted between the µE3 and µB sites (Fig. 2) and
tested these mutations in DNA binding and functional assays. Insertion of 4 or
10 bp rotates the µB site approximately 145° or 350°, respectively, relative to
the µA or µE3 sites, assuming a typical DNA helical periodicity of 10.5 bp.
Protein binding to the three sites was not affected in the rotation mutations. How-
ever, transcription activation in B cells was significantly impaired in both the ⫹4
Figure 2 Precise alignment of µA and µB is necessary for µ enhancer function in B

cells. Insertion of one-half ([⫹4] construct) or one full ([⫹10] construct) helical turn of
DNA between µA and µB inactivates the enhancer in B cells, despite restoration of helical
alignment in the [⫹10] construct. Activity of the control [⫹18] construct excludes trivial
effects due to changing enhancer–promoter distance.
TM

and ⫹10 mutations (Fig. 2). These observations suggest that appropriate spacing
between the µA and µB sites is critical for enhancer activity.
B. Comparison of Orientation Mutated Enhancers in B

Cells and Macrophages
The µA and µB sites that bind ETS domain proteins are nonpalindromic. This
‘‘directionality’’ is depicted in Figs. 2 and 3 as single-headed arrows. In contrast,
the intervening µE3 element is palindromic and therefore shown as a double-
headed arrow. To investigate further the spatial constraints of a functional µA/
µB motif, we tested enhancers in which the relative orientations of the µA and
µB sites had been altered. Mutated enhancers F2-5, as indicated in Fig. 3, con-
tained changes in the relative orientation of the µA, µE3, and µB sites alone or
in combination. In the wild-type enhancer (F1), the core GGAA of the µA site
is 11 bp 5′ to the core CATGTG of µE3, and the core GGAA of the µB site is
19 bp 3′ to the µE3 core binding site (spacing counts are based on centers of µA
Figure 3 Precise alignment of µA and µB is necessary for µ enhancer activity in B

cells, but not macrophages. Any change in µA–µB orientation (F2–F4) inactivates the
enhancer in B cells. The macrophage µ enhancer nucleoprotein complex can retain func-
tion despite changes in µA–µB orientation (F2–F3). Therefore, a given array of ets sites
can define a tissue-specific enhancer activated by the combination of DNA-binding pro-
teins present in a particular cell type.
TM

and µB that are between the G and A of the GGAA cores, and the center of µE3
between the middle TG). Enhancer F2 contains a 14-bp spacing between the µE3
and µB cores, and the µB GGAA is located on the noncoding DNA strand,
whereas F3 contains a 19-bp spacing between the µE3 and µB cores, and µB is
again located on the noncoding DNA strand. In F4, µA is moved to the noncoding
DNA strand and the distance between the core GGAA of µA and the core
CATGTG of µE3 is conserved at 11 bp, resulting in the helical realignment of
µA relative to µB. In F5 we moved the µE3 site further 3′ by flipping the DNA
between µA and µB. Because µE3 is partially palindromic, inversion of the site
should not affect the stereochemistry of protein binding to DNA. Enhancer F5
places the µA and µE3 cores 14 bp apart and the µE3 and µB cores 16 bp apart.
In vitro binding analyses with mutated enhancers and purified proteins indicated
that the mutations had not inadvertently affected protein binding.
To test the hypothesis of B cell–specific combinatorial activation of the µ
enhancer directly, we tested activity of F1–F5 in the hematopoietic cell types B
cells and macrophages. The wild-type enhancer F1 was active in both S194 B
cells and RAW 264.7 macrophages (Fig. 3). Functional tests of F2–F5 probed
the mechanism that activates the wild-type enhancer in the two cell types. A
change in the orientation of µA (e.g., F4) or µB (e.g., F2, F3, F4) inactivated
the enhancer in B cells. However, when the orientation and spacing between µA
and µB were maintained, the enhancer was active in B cells even if the spacing
between the µE3 and µA or µB sites was altered (F5). In contrast, the F2 and
F3 constructs, in which the orientation of the µB site is changed, were active in
RAW 264.7 cells (Fig. 3). Enhancer F4, in which the µA–µE3 distance and
relative orientations of µA and µB are altered, was inactive in macrophages. The
F5 enhancer was about 50% active in macrophages, indicating that the moderate
modification in µE3 alignment relative to µA and µB was not critical for enhancer
activity in the context of appropriately arranged µA and µB sites. Overall, we
conclude that proper juxtaposition of µA and µB is critical for µ enhancer activa-
tion in B cell lines, but not in macrophages. The results of the orientation mutated
enhancers strongly suggest that the mechanism underlying activation of the mini-
mal µ enhancer differs in B cells versus macrophage. These observations high-
light the tissue-specific regulatory possibilities that can be achieved by appropri-
ately juxtaposing a combination of ETS domain protein binding sites.
C. ␮ Enhancer Deoxyribonucleic Acid Flexure Is

Increased by Protein Binding
The studies described showed that changes in either the spacing or the orientation
of the µA and µB sites drastically reduced enhancer activity in B cells, but did
not affect protein–DNA interactions. Clearly, recruiting the proteins to the DNA
is not sufficient to activate transcription; rather, a stereochemically precise super-
TM

complex of the three proteins must be assembled appropriately on the DNA. To
examine one structural component of such a complex, we investigated whether
the proteins that bound to the µA, µE3, and µB elements affected DNA structure.
We assayed two categories of changes in DNA structure that are known to result
from DNA–protein interactions: (1) an increase in DNA flexure and (2) the induc-
tion of a directed bend in the DNA. Deoxyribonucleic acid flexure refers to a
distortion in the DNA that brings the ends of the DNA molecule closer to each
other, but not in any fixed orientation. One way to visualize it is a bend in the
DNA in which one ‘‘half ’’ of the molecule can take on several positions on part
of the surface of a cone (Fig. 4). A directed bend is one in which the ‘‘half’’ of
the molecule takes up a unique position on the surface of such a cone (bold
arrow).
Protein-induced changes in DNA flexure were determined by circular per-
mutation assays. These assays were carried out using DNA fragments where the
µ enhancer was located at different positions relative to the end of the DNA.
Increased flexure was detected as reduced mobility of the DNA/protein complex
in nondenaturing acrylamide gels. Two proteins, PU.1 and TFE3, scored positive
in this assay, increasing DNA flexure by 48° and 40°, respectively. By compari-
son, Ets-1, a µA binding protein, had no effect on DNA structure in these analy-
ses.
Figure 4 Model of how a DNA binding protein can alter DNA structure. A DNA bind-
ing protein that increases DNA flexure can result in numerous conformations indicated
by thin and dotted arrows. A DNA binding protein that induces a directed DNA bend
locks the DNA into a particular structure, represented by the bold arrow.
TM

D. Only PU.1 Protein Induces a Directed Bend in
␮ Enhancer Deoxyribonucleic Acid
Although circular permutation analysis provides evidence for protein-induced
DNA distortion, this assay more accurately reflects changes in DNA flexure that
result from protein–DNA interactions. Alternatively, anomalous migrations in
circular permutation assays may result from the nonglobular conformations of
some DNA binding proteins [14]. To distinguish between DNA distortion and
directed DNA bending, we performed phasing analysis. For phasing analysis, the
µ enhancer fragment was cloned adjacent to three poly A tracts [14,15]. Each
poly A tract confers an intrinsic DNA bend of 18° for an overall 54° minor groove
directed DNA bend. This bend serves as an internal standard against which pro-
tein-induced bending in the adjacent DNA fragment can be compared. The µ
enhancer was cloned at five different distances (designated the spacer length and
ranging from 29 to 41 bp) from the A 5 tracts to position it at various helical positions
relative to the intrinsic minor groove bend (Fig. 5A). Protein-induced bends directed
toward the minor groove reinforce the intrinsic A tract bend when the protein bind-
ing site is positioned at integral (n) helical turns from the A tracts. In contrast, major
groove bends reinforce the intrinsic bend when the protein binding site is positioned
on the opposite helical face (n ⫹ 5). If a DNA binding protein induces only a change
in flexure, the location of the binding site with respect to the A tracts does not
affect DNA bending as assayed by nucleoprotein complex migration. The EMSA
(electrophoretic mobility shift assay) analysis demonstrates bend reinforcement by
decreased mobility of the DNA–protein complex.
We performed phasing analysis with the DNA binding domain of PU.1 and
Ets-1 [ETS (PU.1) and ETS (Ets-1) respectively], and a naturally occurring splice
variant of TFE3, TFE3S (Fig. 5B). Differences in protein–DNA complex mobil-
ity were apparent for ETS(PU.1) but not for ETS(Ets-1) or TFE3S, indicating
that only ETS(PU.1) induced a directed DNA bend. Results with full-length PU.1
were similar to results with ETS(PU.1) (data not shown). Phasing analysis further
allowed us to calculate the degree of ETS(PU.1)-induced DNA bending, as de-
scribed by Thompson and Landy [16] and modified by Kerpolla and Curran
[14,15,17]. The best fit cosine curve for ETS(PU.1) (Fig. 5C) indicated that PU.1
bent µ enhancer DNA to an angle of 24° after correction for the slight bend (1°)
contributed by the probe alone.
Direction of the ETS(PU.1)-induced DNA bend can also be ascertained
from the phasing analysis once the center of the bend is determined. Although
circular permutation results indicated the center of the ETS(PU.1)-induced flex-
ure lies within the µB element, the precise bend center of this nonpalindromic
site is not discernible by this method. However, PU.1 binding to the µB site (5′-
TTCCCCAAA-3′) induces a strong deoxyribonuclease (DNAase) I–hypersensi-
TM

Figure 5A Phasing analysis DNA fragments. The helical positioning of µ enhancer
protein binding sites relative to an internal control bend (A 5 tracts) was altered by increas-
ing spacer DNA from 29 to 41 bp. Directed DNA bending was interpreted relative to the
internal control bend. DNA, deoxyribonucleic acid.
tive site in the noncoding strand between the C and A residues opposite the
T-C junction of the coding strand, suggesting protein induced distortion of the
phosphodiester backbone [18]. We approximated that this hypersensitive site is
the center of the PU.1-induced DNA bend and calculated the lengths between
µB and the poly A tract bend centers with reference to the position of the hyper-
sensitive site. The ETS(PU.1)-induced DNA bend reinforced the poly A tract
bend, resulting in the slowest mobility nucleoprotein complex when the µB site
was positioned 36 bp (3.4 helical turns) from the poly A tract. Because ETS(PU.1)
reinforced the minor groove-directed poly A tract bend when the µB element
was on the opposite side of the DNA helix, we propose ETS(PU.1) bends µ
enhancer DNA toward the major groove (Fig. 6).
TM

Figure 5B PU.1 induces a directed bend in µ enhancer DNA. EMSA analysis demon-
strates the ETS (PU.1) nucleoprotein complex migration changes as the helical alignment
(determined by spacer length) between µB and the A 5 tracts changes. ETS (Ets-1) and
TFE3S-µ enhancer complex migration is independent of spacer length. ETS (PU.1) or
ETS(Ets-1)-DNA complexes are bracketed; TFE3S-DNA complex is demarcated by an
arrow. ETS (PU.1), and ETS (Ets-1) refer to the ETS domains of the transcription factors
PU.1 and Ets-1, respectively. EMSA ⫽ electrophoretic mobility shift assay.
Figure 5C The best-fit cosine curve of ETS (PU.1)–µ enhancer nucleoprotein com-
plexes from Figure 5B. Analysis indicated ETS (PU.1) induces a 24° directed bend in the
µ enhancer DNA. ETS (PU.1) ⫽ the ETS domain of PU.1.
TM

Figure 6 Model: a functional nucleoprotein complex on the µ enhancer. We hypothesize
specific interfaces must be formed by protein–protein interactions of Ets-1, PU.1, and
TFE3 to result in B cell–specific enhancer activity. Appropriate alignment and DNA struc-
ture of µA, µE3, and µB forms a scaffold for this active complex. One possible mecha-
nism for activity is that Ets-1/PU.1 interfaces are needed to recruit transcriptional co-
activator(s).
III. DISCUSSION
To facilitate mechanistic analysis of the lymphoid-specific immunoglobulin µ

heavy chain gene enhancer, we previously described a minimal enhancer that
contains three sequence elements: µA, µE3, and µB [6]. Here we have summa-
rized the results of experiments that show that precise spatial organization of the
sites, and consequently the three-dimensional nucleoprotein structure formed
with enhancer DNA, is necessary for B cell–specific transcriptional enhancement.
Furthermore, one of the three minimal enhancer binding proteins induces a di-
rected bend in the DNA that may facilitate the formation of the functional stereo-
specific nucleoprotein complex.
Comparison of the activity of orientation mutated enhancers in B cells and
macrophages showed that the requirements differed greatly in the two cell types.
For example, altering the head-to-tail organization of the µA and µB elements
abrogates transcriptional activity in B cells but results in relatively little change
in macrophage activity (i.e., Fig. 3, F2, 3, and 4). Note that in F1 the ‘‘arrow-
head’’ of µA points toward µB, whereas in F4, the arrowhead of µA points away
from µB. Inactivity of F4 demonstrates that proper 5′-3′ organization of µA and
µB is insufficient for enhancer activity, and the further stereochemical consider-
ations constrain the combinatorial use of ETS protein binding sites in both cell
types. These results are reminiscent of the properties of hormone receptor binding
TM

sites, where the spacing of half sites has been shown to affect the transcriptional
outcome [19,20].
In our interpretation of the PU.1 phasing data, we approximated the center
of the bend to the site of a strong PU.1 induced DNAase 1–hypersensitive site
between residues 5 and 6 on the noncoding strand of the µB site, 5′-TATTTGGG-
GAAGGGAA-3′. We have found that the three adenosine residues complemen-
tary to the three contiguous thymidines in the µB site score strongly in methyla-
tion interference assays [18]. Since the methyl group of methyl adenosines fall
in the minor groove, these observations suggest that PU.1 is located over the
minor groove in the vicinity of the PU.1 induced bend. Taken together with the
phasing analysis that shows the induced bend is toward the major groove, our
observations are consistent with the DNA’s being bent away from the bound
PU.1 protein. It remains to be determined how the PU.1-induced DNA bend
contributes to the formation of the three-protein DNA complex.
CONCLUSIONS
Our data suggest that precise stereochemistry of the µ enhancer nucleoprotein

complex is critical for B cell–specific immunoglobulin heavy chain gene expres-
sion. The relative positioning of ETS domain proteins and the PU.1-induced DNA
bend determines the nature and extent of both protein–DNA and protein–protein
contacts within the multicomponent complex. We suggest that these contacts de-
fine an active enhancer only in B cells because of the coexpression of a unique
combination of transcriptional activators.
REFERENCES
1. A. Alessandrini and S. V. Desiderio, Mol. Cell. Biol., 11: 2096 (1991).

2. T. K. Blackwell, M. W. Moore, G. D. Yancopoulos, H. Suh, S. Lutzker, E. Selsing,
and F. W. Alt, Nature, 324: 585 (1986).
3. J. M. Adams, A. W. Harris, C. A. Pinkert, L. M. Corcoran, W. S. Alexander, S.
Cory, R. D. Palmiter, and R. L. Brinster, Nature, 318: 533 (1985).
4. W. Y. Langdon, A. W. Harris, S. Cory, and J. M. Adams, Cell, 47: 11 (1986).
5. W. Reik, G. Williams, S. Barton, M. Norris, M. Neuberger, and M. A. Surani, Eur.
J. Immunol., 17: 465 (1987).
6. B. Nelsen, G. Tian, B. Erman, J. Gregoire, R. Maki, B. Graves, and R. Sen, Science,
261: 82 (1993).
7. M. Lopez, P. Oettgen, Y. Akbarali, U. Dendorfer, and T. A. Libermann, Mol. Cell.
Biol., 14: 3292 (1994).
8. R. R. Rivera, M. H. Stuiver, R. Steenbergen, and C. Murre, Mol. Cell. Biol., 13:
7163 (1993).
TM

9. P. Oettgen, Y. Akbarall, J. Boltax, J. Best, C. Kunsch, and T. A. Libermann, Mol
Cell. Biol., 16: 5091 (1996).
10. M. J. Klemsz, S. R. McKercher, A. Celada, C. VanBeveren, and R. A. Maki, Cell,
61: 113 (1990).
11. B. Nelsen, T. Kadesch, and R. Sen, Mol. Cell. Biol., 10: 3145 (1990).
12. T. A. Libermann and D. Baltimore, Mol. Cell. Biol., 13: 5957 (1993).
13. H. Beckmann, L. K. Su, and T. Kadesch, Genes Dev., 4: 167 (1990).
14. T. K. Kerppola and T. Curran, Science, 254: 1210 (1991a).
15. T. K. Kerppola and T. Curran, Cell, 66: 317 (1991b).
16. J. F. Thompson and A. Landy, Nucleic Acids Res., 16: 9687 (1988).
17. T. K. Kerppola and T. Curran, Mol. Cell. Biol., 13: 5479 (1993).
18. E. Rao, W. Dang, G. Tian, and R. Sen, J. Biol. Chem., 272: 6722 (1997).
19. C. K. Glass, J. M. Holloway, O. V. Devary, and M. G. Rosenfeld, Cell, 54: 313
(1988).
20. M. Beato, Cell, 56: 335 (1989).
TM

(Raymond Cooper, John K. Snyder) Biology - Chemist (BookFi) PDF

Uploaded by

Copyright:

Available Formats

(Raymond Cooper, John K. Snyder) Biology - Chemist (BookFi) PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(Raymond Cooper, John K. Snyder) Biology - Chemist (BookFi) PDF

Uploaded by

Copyright:

Available Formats

The

Bio Iogy-C he mistry

World Wide Web

Copyright  1999 by Marcel Dekker, Inc. All Rights Reserved.

Current printing (last digit):

PRINTED IN THE UNITED STATES OF AMERICA

Copyright n 1999 by Marcel Dekker, Inc. All Rights Reserved.

Natural products science, a fascinating cornerstone of modern research, has long

Copyright n 1999 by Marcel Dekker, Inc. All Rights Reserved.

Copyright n 1999 by Marcel Dekker, Inc. All Rights Reserved.

Copyright n 1999 by Marcel Dekker, Inc. All Rights Reserved.

Copyright n 1999 by Marcel Dekker, Inc. All Rights Reserved.

1. Insect Antifeeding Limonoids from the Chinaberry Tree Melia azedarach

Copyright n 1999 by Marcel Dekker, Inc. All Rights Reserved.

Copyright n 1999 by Marcel Dekker, Inc. All Rights Reserved.

Valeria Balogh-Nair Department of Chemistry, The City College of the City

Colin J. Barrow School of Chemistry, The University of Melbourne, Parkville,

Timothy A. Blizzard Medicinal Chemistry, Merck Research Laboratories,

Rosalie K. Crouch Department of Ophthalmology, Medical University of

Thomas G. Ebrey School of Cellular and Molecular Biology, University of

Yan Feng Department of Ophthalmology, Medical University of South Caro-

Copyright n 1999 by Marcel Dekker, Inc. All Rights Reserved.

Nobuyuki Harada Institute for Chemical Reaction Science, Tohoku Univer-

Kurt Hostettmann Institut de Pharmacognosie et Phytochimie, Université de

Maryse Hostettmann Institut de Pharmacognosie et Phytochimie, Université

David A. Johnson Department of Chemistry, University of Minnesota, Minne-

Isao Kubo Department of Environmental Science, Policy, and Management,

Takenori Kusumi Faculty of Pharmaceutical Sciences, Tokushima University,

Hung-wen Liu Department of Chemistry, University of Minnesota, Minneapo-

David G. Lynn Department of Chemistry, The University of Chicago, Chi-

Donald R. Menick Departments of Medicine, and Biochemistry and Molecular

Munehiro Nakatani Department of Chemistry and Bioscience, Kagoshima

Copyright n 1999 by Marcel Dekker, Inc. All Rights Reserved.

Ikuko I. Ohtani Department of Chemistry, Biology, and Marine Science, Uni-

Sylvain Rodriguez Institut de Pharmacognosie et Phytochimie, Université de

Ranjan Sen Rosenstiel Research Center and Department of Biology, Brandeis

Steven P. Tanis Medicinal Chemistry I, Pharmacia & Upjohn, Inc., Kalama-

Jesús Trujillo Vázquez Instituto Universitario de Bio-Orgánica ‘‘Antonio

Jean-Luc Wolfender Institut de Pharmacognosie et Phytochimie, Université

Michael G. Zagorski Department of Chemistry, Case Western Reserve Uni-

Copyright n 1999 by Marcel Dekker, Inc. All Rights Reserved.

March 16, 1998

Copyright n 1999 by Marcel Dekker, Inc. All Rights Reserved.

Copyright n 1999 by Marcel Dekker, Inc. All Rights Reserved.

Merci, Koji, pour ton amitié

Professor Guy Ourisson

Copyright n 1999 by Marcel Dekker, Inc. All Rights Reserved.

Limonoids are tetranortriterpenoids derived from euphane (H-20β) or tirucallane

Copyright n 1999 by Marcel Dekker, Inc. All Rights Reserved.

Similarly, Melia azedarach is known to produce azadirachtin-type limo-

II. LIMONOIDS FROM Melia azedarach Linn.

Melia azedarach is a native of Persia, India, and China but is naturalized in a

Copyright n 1999 by Marcel Dekker, Inc. All Rights Reserved.

quently oxidatively cleaved to a carboxylic acid or lactone (C-seco limonoids).

Copyright n 1999 by Marcel Dekker, Inc. All Rights Reserved.

Figure 2 Limonoids from M. azedarach, Group 1: apo-euphol limonoids (continued ),

Copyright n 1999 by Marcel Dekker, Inc. All Rights Reserved.

Copyright n 1999 by Marcel Dekker, Inc. All Rights Reserved.

Figure 4 Limonoids from M. azedarach, Group 3: C-seco limonoids, highly oxidized