Bengt Nolting Protein Folding Kinetics Biophysic PDF

Protein Folding Kinetics
Bengt Nlting
ProteinFoldingKinetics
Biophysical Methods
Second Edition
With 170 Figures, 12 in Color and 15 Tables
123
Dr. Bengt Nlting

Prussian Private Institute
of Technology at Berlin
Am Schlosspark 30
D-13187 Berlin
Germany
[email protected]
Library of Congress Control Number: 2005929411
ISBN-10 3-540-27277-1 2nd Edition Springer Berlin Heidelberg New York

ISBN-13 978-3-540-27277-9 2nd Edition Springer Berlin Heidelberg New York
2nd edition 2006. Revised and extended.
ISBN 3-540-65743-6 1st Edition Springer Berlin Heidelberg New York
This work is subject to copyright. All rights reserved, whether the whole or part of the
material is concerned, specically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microlm or in any other way, and storage in data
banks. Duplication of this publication or parts thereof is permitted only under the provisions
of the German Copyright Law of September 9, 1965, in its current version, and permission
for use must always be obtained from Springer. Violations are liable for prosecution under
the German Copyright Law.
Springer is a part of Springer Science+Business Media
springeronline.com
Springer-Verlag Berlin Heidelberg 1999, 2006
Printed in Germany
The use of general descriptive names, registered names, trademarks, etc. in this publication
does not imply, even in the absence of a specic statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
Product liability: The publisher cannot guarantee the accuracy of any information about
dosage and application contained in this book. In every individual case the user must check
such information by consulting the relevant literature.
The instructions given for carrying out practical experiments do not absolve the reader from
being responsible for safety precautions. Liability is not accepted by the authors.
Safety considerations: Anyone carrying out these methods will encounter pathogenic and
infectious biological agents, toxic chemicals, radioactive substances, high voltage and intense
light radiation which are hazardous or potentially hazardous materials or matter. It is required
that these materials and matter be used in strict accordance with all local and national
regulations and laws. Users must proceed with the prudence and precaution associated with
good laboratory practice, under the supervision of personnel responsible for implementing
laboratory safety programs at their institutions.
Typesetting: By the Author
Production: LE-TEX, Jelonek, Schmidt & Vckler GbR, Leipzig
Coverdesign: design&production, Heidelberg
Printed on acid-free paper 2/YL 5 4 3 2 1 0
To my parents
Faust
Then shall I see, with vision clear,
How secret elements cohere,
And what the universe engirds,
And give up huckstering with words.
Johann Wolfgang von Goethe
Preface
This second edition contains three new chapters covering (a) the high resolution
of the folding pathways of six proteins by using the powerful method of -value
analysis (Chap. 11; Nlting and Andert, 2000), (b) the structural determinants of
protein folding kinetics (Chap. 12; Nlting 2003; Nlting et al., 2003), and finally
(c) presenting a novel method called evolutionary computer programming
(Chap. 13; Nlting et al., 2004). The latter method involves the self-evolution of
computer programs that can lead to highly advanced programs which are able to
calculate protein folding and structure with unprecedented efficiency. The scope
of such self-evolving computer programs is far beyond protein folding and
biophysics. Section 13.3 outlines some possible future applications of selfevolving computer programs which can yield systems smarter than humans in
fulfilling certain technological tasks. For further information on biophysics
methods in general, the reader may refer also to the textbook Methods in Modern
Biophysics.
Mai 2005
Bengt Nlting
Preface to the first edition
The study of fast protein folding reactions has significantly advanced, following
the recent development of new biophysical methods which enable not only kinetic
resolution in the submillisecond time scale but also higher structural resolution.
The pathways and structures of early folding events and the transition state
structures of fast folding proteins can now be studied in far more detail. The
validity of different models of protein folding for those events may now be
elucidated and the high speed of complicated folding reactions far better
understood.
This book, which is based to a high degree on several publications by the
author and coworkers (e.g., Nlting, 1991, 1995, 1996, 1998a, b, 1999; Nlting et
al., 1992, 1993, 1995, 1997a, b; Nlting and Sligar, 1993; Pfeil et al., 1993a, b), is
particularly dedicated to students of biophysics, biochemistry, biotechnology, and
medicine as a practical introduction to the modern biophysical methods of high
kinetic (Chaps. 3 6, 810) and structural (Chaps. 23, 710) resolution of
reactions that involve proteins with emphasis on protein folding reactions. Many
methods are of truly interdisciplinary nature, ranging from mathematics to
biophysics to molecular biology and can hardly be found in other textbooks. Since
there is a rapid ongoing progress in the development and application of these
methods, in particular in protein engineering, ultrafast mixing, temperaturejumping, optical triggers of folding, and -value analysis, a large amount of
essential information concerning the equipment and experimental details is
included.
Chapter 10 reports the first high resolution of the folding pathway of a protein
from microseconds to seconds (Nlting et al., 1995, 1997a, b; Nlting, 1998a).
Requisite for this work was the development of a new method for the initiation
and study of rapid folding which involves temperature-jumping of a set of suitably
engineered mutants from the cold-unfolded to the folded state (Nlting et al.,
1995, 1997a; Nlting, 1996). This new method allows fast processes that would
normally be hidden in kinetic studies to be revealed.
Of course, the range of applicability of fast kinetic methods is far wider than
that presented. Thus, everybody working in the fields of fast chemical reactions
and physical changes, such as conformational isomerizations, enzyme kinetics and
enzyme mechanisms, might see the book as a useful introduction.
The framework that is provided for the readers is the notion that the
quantitation of kinetic rate constants and the visualization of protein structures
Preface to the first edition
along the folding pathway will lead to an understanding of function and

mechanism and will aid the understanding of important biological processes and
disease states through detailed mechanistic knowledge. Numerous figures provide
useful information not easily found elsewhere, and the book includes copious
references to original research papers, relevant reviews and monographs.
My work at Cambridge University and the Medical Research Council was
supported by a European Union Human Capital and Mobility Fellowship and a
Medical Research Council Fellowship. I gratefully acknowledge Prof. Dr. Alan R.
Fersht for the interest in our work on fast folding reactions. NMR measurements
on peptides of barstar were done by Dr. Jos L. Neira and Dr. Andrs S. SolerGonzlez.
The work at the University of Illinois at Urbana-Champaign was supported by
NIH grant GM31756. Prof. Dr. Steven G. Sligar is particularly acknowledged for
his support of acoustic relaxation experiments and many fruitful discussions.
Prof. Dr. Martin Gruebele kindly presented a LASER T-jump spectrometer with
real-time fluorescence detection in the nanosecond time scale.
Dr. Robert Clegg is acknowledged for the demonstration of an ultrafast mixing
device. Prof. Dr. Manfred Eigen and Dr. Dietmar Porschke kindly demonstrated a
T-jump and electric field-jump apparatus. I am indebted to Dr. Min Jiang and
Dr. Gisbert Berger for proof-reading the manuscript, and to Dr. Marion Hertel and
Ms. Janet Sterritt-Brunner for processing the manuscript within Springer-Verlag.
Legal remarks: A number of methods mentioned in this book are covered by
patents. Nothing in this publication should be construed as an authorization or
implicit license to practice methods covered by any patents.
January 1999
Bengt Nlting
Contents
Introduction
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Structures of proteins . . . . . . . . . . . . . . . . . . . . . . . .
2.1
2.2
2.3
3
Physical interactions that determine

the properties of proteins
3.1
3.2
3.3
3.4
4
Primary structure . . . . . . . . . . . . . . . . . . . . . . . . .
Secondary structure . . . . . . . . . . . . . . . . . . . . . . .
Tertiary structure . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . .
Electrostatic interactions . . . . . . . . . . . . . . . . . .
3.1.1 Point charges . . . . . . . . . . . . . . . . . . .
3.1.2 Point chargedipole and dipoledipole interactions
Quantum-mechanical short-range repulsion . . . . . . . .
Hydrogen bonding . . . . . . . . . . . . . . . . . . . . .
Hydrophobic interaction . . . . . . . . . . . . . . . . . .
Calculation of the kinetic rate constants

4.1
4.2
4.3
4.4
4.5
17
17
19
19
21
22
. . . . . . . . . . . .
27
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
17
.
.
.
.
.
.
Transition state theory . . . . . . . . . . . . . . . .

Two-state transitions
. . . . . . . . . . . . . . . .
4.2.1 Reversible two-state transition . . . . . . .
4.2.2 Irreversible two-state transition
. . . . . .
Three-state transitions . . . . . . . . . . . . . . . .
4.3.1 Reversible three-state transitions . . . . . .
4.3.1.1 Reversible sequential three-state transition .
4.3.1.2 Reversible two-pathway three-state transition
4.3.1.3 Reversible off-pathway intermediate . . . . .
4.3.2 Irreversible three-state transitions
. . . . .
4.3.2.1 Irreversible consecutive three-state transition
4.3.2.2 Irreversible parallel decay
. . . . . . . . .
Reversible sequential four-state transition . . . . . .
Reactions with monomerdimer transitions . . . . .
4.5.1 Monomerdimer transition . . . . . . . . .
.
.
.
.
.
.
5
9
11
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
28
29
29
30
31
31
31
34
37
38
38
39
40
41
41
Contents
XII
4.5.2
4.6
4.7
5
High kinetic resolution of protein folding events

5.1
5.2
5.3
5.4
5.5
5.6
5.7
5.8
6
43
44
47
51
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
51
55
55
55
58
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
61
62
64
65
65
69
69
72
73
75
76
. . . . . . . . . . . . . . .
79
.
.
.
.
79
80
80
81
Stopped-flow nuclear magnetic resonance (NMR)

Fluorescence- and isotope-labeling . . . . . . . .
6.2.1 Folding reactions . . . . . . . . . . . .
6.2.2 Dissociation reactions . . . . . . . . . .
. . . . . . . . . .
83
Nuclear magnetic resonance . . . . . . . . . . . . . . . . . . .

Circular dichroism . . . . . . . . . . . . . . . . . . . . . . . .
83
89
High structural resolution of transient . . . . . . . . . . . . . .

protein conformations
95
Resolution of protein structures in solution

7.1
7.2
. . . . . . .
Ultrafast mixing . . . . . . . . . . . . . . . . . . . . .
Temperature-jump . . . . . . . . . . . . . . . . . . . .
5.2.1 Electrical-discharge-induced T-jump . . . . . .
5.2.1.1 T-jump apparatus . . . . . . . . . . . . . . . .
5.2.1.2 Observation of early folding events: refolding . .
from the cold-unfolded state
5.2.1.3 Observation of unfolding intermediates . . . . .
5.2.2 LASER-induced T-jump . . . . . . . . . . . .
5.2.3 Maximum time resolution in T-jump experiments
Optical triggers . . . . . . . . . . . . . . . . . . . . . .
5.3.1 LASER flash photolysis
. . . . . . . . . . . .
5.3.2 Electron-transfer-induced refolding . . . . . . .
Acoustic relaxation . . . . . . . . . . . . . . . . . . . .
Pressure-jump . . . . . . . . . . . . . . . . . . . . . .
Dielectric relaxation and electric-field-jump . . . . . . .
NMR line broadening . . . . . . . . . . . . . . . . . .
Summary
. . . . . . . . . . . . . . . . . . . . . . . .
Kinetic methods for slow reactions

6.1
6.2
Reversible two-state folding transition linked . . . . . .

with a monomerdimer transition
Kinetic rate constants for perturbation methods . . . . . . . . .
Summary
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.1
8.2
8.3
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
NMR detection of H/D exchange kinetics . . . . . . . . . . . .

Time-resolved circular dichroism . . . . . . . . . . . . . . . . .
-value analysis . . . . . . . . . . . . . . . . . . . . . . . . .
95
98
105
Contents
8.3.1
8.3.1.1
8.3.1.2
8.3.2
XIII
.
.
.
.
107
108
108
111
115
.
.
.
.
.
.
.
.
116
117
117
119
120
120
121
123
. . . . .
125
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
125
125
126
129
129
131
133
135
10 The folding pathway of a protein (barstar) at the . . . . . . . .

resolution of individual residues
from microseconds to seconds
137
8.3.3
8.3.3.1
8.3.3.2
8.3.3.3
8.3.3.4
8.3.4
8.3.4.1
8.3.4.2
8.3.4.3
9
Protein engineering . . . . . . . . . . . . . . . . . .
Cassette mutagenesis . . . . . . . . . . . . . . . . .
PCR mutagenesis . . . . . . . . . . . . . . . . . . .
Determination of the protein stability . . . . . . . . .
in equilibrium
Measurement of kinetic rate constants of folding . . .
and unfolding
Two-state kinetics
. . . . . . . . . . . . . . . . . .
Three-state kinetics . . . . . . . . . . . . . . . . . .
Kinetic implications of the occurrence of intermediates
Discrimination between folding and association events
Calculation and interpretation of -values . . . . . .
Two-state transition
. . . . . . . . . . . . . . . . .
Multi-state transition . . . . . . . . . . . . . . . . .
Residual structure in the unfolded state . . . . . . . .
Experimental problems of the kinetic and structural

resolution of reactions that involve proteins
9.1
9.2
9.3
9.4
10.1
10.2
10.3
10.4
10.5
Protein expression problems . . .

9.1.1 Low expression level . .
9.1.2 Expression errors . . . .
Aggregation . . . . . . . . . . .
9.2.1 Detection . . . . . . . .
9.2.2 Avoidance of aggregation
Misfolding . . . . . . . . . . . .
Unstable curve fit . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Introduction . . . . . . . . . . . . . . . . . .
Materials and methods . . . . . . . . . . . . .
Structure of native barstar . . . . . . . . . . .
Residual structure in the cold-unfolded state .
Gross features of the folding pathway of barstar
10.5.1 Equilibrium studies . . . . . . . . . .
10.5.2 Kinetic studies
. . . . . . . . . . . .
10.6 -value analysis . . . . . . . . . . . . . . . .
10.7 Inter-residue contact maps . . . . . . . . . . .
10.8 The highly resolved folding pathway of barstar .
10.8.1 Microsecond transition state . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
137
138
140
141
143
143
145
149
152
155
157
Contents
XIV
10.8.2 Intermediate
. . . . . . . . . . . . .
10.8.3 Late transition state . . . . . . . . . .
10.8.4 Directional propagation of folding . . .
10.8.5 Cistrans isomerization . . . . . . . .
10.8.6 Are there further folding events? . . . .
10.9 Structural disorder and misfolding . . . . . . .
10.10 Structure of peptides of barstar
. . . . . . . .
10.11 Nucleationcondensation mechanism of folding
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
11 Highly resolved folding pathways and mechanisms

of six proteins
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
157
157
157
158
159
160
162
162
. . . . .
165
11.1 General features of the main transition states for the . .

formation of the native structures
11.2 Nucleationcondensation . . . . . . . . . . . . . . . . .
11.3 Framework-model-like properties . . . . . . . . . . . .
11.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. . . .
165
. . . .
. . . .
. . . .
176
179
179
12 Structural determinants of the rate of protein folding . . . . .
181
12.1 Chain topology as a major structural determinant of two-state

folding
12.2 Chain topology of the transition state and implications for the
mechanism of folding
12.3 Further factors . . . . . . . . . . . . . . . . . . . . . . . .
12.4 Ultrafast folding . . . . . . . . . . . . . . . . . . . . . . . .
. .
181
. .
184
. .
. .
184
185
13 Evolutionary computer programming of protein structure . .

and folding
187
13.1 Evolution method . . . . . . . . . . . . . . . . . . . . . . . . .

13.2 Protein folding and structure predictions . . . . . . . . . . . . .
13.3 Further potential applications of the evolution method . . . . . .
188
189
194
14 Conclusions
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
195
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
197
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
215
References
Index
Symbols
Ala
Arg
Asn
Asp
bar
BdpA
BP
oC
CCD
CD
CO
CO
CTP
Cys
[D]
Da
GFU
GUF
DNA
En-HD
F
FID
Gln
Glu
Gly
GuHCl
h
His
HPLC
I
Ile
IPTG
J
kB
kcal
kDa
angstrm (1010 m)
alanine
arginine
asparagine
aspartic acid
105 Pa (= 105 N m2 )
B-domain of protein A from
Staphylococcus aureus
base pair
degree Celsius
(= degree Kelvin 273.15)
charge-coupled device
circular dichroism
carbon monoxide
contact order
chain topology parameter
cysteine
concentration of denaturant
dalton, g mol1
Gibbs free energy change upon
folding
Gibbs free energy change upon
unfolding (= GFU)
deoxyribonucleic acid
Engrailed homeodomain
folded state
free induction decay
glutamine
glutamic acid
glycine
guanidine hydrochloride
Planck constant
(6.62611034 J s)
histidine
high-performance liquid
chromatography
intermediate state
isoleucine
isopropyl-1-thio--galactoside
Joule (1 J = 1 Ws = 0.239 cal)
Boltzmann constant
(1.38071023 J K1)
kilocalorie (= 4.18 kJ)
kilodalton, kg mol1
observed rate constant, relaxation

constant, decay constant (kobs is
denoted with in Chap. 4)
liter
L
leucine
Leu
lysine
Lys
magnetic circular dichroism
MCD
mercury cadmium telluride
MCT
methionine
Met
molten globule
MG
micrometer (106 m)
m
microsecond (106 s)
s
mL
milliliter
mol
6.02211023
nm
nanometer (109 m)
NMR
nuclear magnetic resonance
NOE
nuclear Overhauser effect
ns
nanosecond (109 s)
OPO
optical parametric oscillator
PCR
polymerase chain reaction
Phe
phenylalanine
pI
isoelectric point
ppm
part per million, 106
Pro
proline
ps
picosecond (1012 s)
PVC
polyvinyl chloride
R
molar gas constant
(8.3145 J mol1 K1)
Ser
serine
SHG
second harmonic generation
TFE
trifluoroethanol
Thr
threonine
T-jump temperature-jump
TMS
tetramethylsilane
Tris-HCl tris(hydroxymethyl)
aminomethane hydrochloride
Trp
tryptophan
TY
tryptoneyeast
Tyr
tyrosine
U
unfolded state
UV
ultraviolet
Val
valine
VIS
visible
#
transition state
kobs
1 Introduction
A requisite for the further understanding of the protein folding problem is the high
structural and kinetic resolution of the folding pathway in the time scale from microseconds to seconds (Fersht et al., 1994; Chan, 1995; Nlting et al., 1995, 1997a,
2003, 2004; Shakhnovich et al., 1996; Wolynes et al., 1996; Eaton et al., 1996a;
Dill and Chan, 1997; Nlting, 1998a, 1999, 2003; Nlting and Andert, 2000).
After decades of research, the folding mystery slowly unfolds. The amazing
efficiency of the folding reaction becomes immediately obvious if one tries to
imagine the huge number of conformations in the unfolded state: Estimates of the
number of conformations in the maximally unfolded state, the so-called random
coil state, are around 10100 for a protein of 100 amino acid residues (Fig. 1.1;
Finkelstein, 1997). In the folding reaction, the unique native conformation is
attained on a time scale of typically seconds or even milliseconds for small
proteins at room temperature, if there are no complications from slowly
isomerizing amino acid residues, in particular prolines.
Fig. 1.1. Folding paradox. A protein with 100 residues may attain roughly 10100 conformations
in the maximally unfolded state, the so-called random coil state. Randomly sampling of all of
these conformations would take many millions of years even if every sampling step would take
only 1 ns. In contrast, many small proteins fold in seconds or faster. What makes the folding
reaction so rapid compared to a random walk? What is the clever mechanism that has evolved?
How are conformations so efficiently directed? What are the pathways of folding?
1 Introduction
Fig. 1.2. Gigantic number of conformations of unfolded protein: In the random coil state usually
several approximately independent conformations per amino acid residue are possible. The
scheme of an arbitrary conformation of the polypeptide AVGS is shown. For reasons of
simplicity, the hydrogen atoms are not shown.
The main reason for the gigantic number of unfolded conformations is that
often the energy differences between different rotamers are small, and thus, there
are comparable occupancies of many different orientations of the protein
backbone and sidechains (Fig. 1.2). On average, the number of independent
conformations per amino acid residue is about 10 (Finkelstein, 1997). Levinthal
realized already in 1968 that folding cannot proceed via a random sampling of
conformations since, even if one assumes one nanosecond per sampling step, the
time of folding would be far greater than the measured time. Consequently there
must be folding pathways which allow folding to proceed far more efficiently than
on a random walk (Levinthal, 1968).
Several reasons have contributed to the paramount current and still increasing
theoretical and experimental interest in protein folding: 1. Protein misfolding,
aggregation and fibrillogenesis is connected with a number of diseases, such as
prion-, Huntingtons-, and Alzheimers diseases (Bychkova and Ptitsyn, 1995;
Eigen, 1996; Booth et al., 1997; Masters and Beyreuther, 1997). 2. There is a
significant interest in the overexpression of recombinant proteins with the correct
fold for industrial and research applications. 3. Enzymatic activity under severe
conditions, such as in organic co-solvent solutions, is seen as a potentially new
method for chemical synthesis (Klibanov, 1989, 1997; Griebenow and Klibanov,
1997; Kunugi et al., 1997; Wangikar et al., 1997). 4. Further, the folding problem
is connected with the significant mathematical problem of finding global minima
in highly complex energy-potential surfaces (Fig. 1.3) in high-dimensional spaces
(Stouten et al., 1993; Luthardt and Frmmel, 1994; Cvijovic and Klinowski, 1995;
Scheraga, 1996; Becker and Karplus, 1997).
Computer simulations suggest that the energy landscape along the folding
pathway of proteins is often not perfectly smooth and that stable or unstable
intermediates may be passed through (Itzhaki et al., 1994; Ptitsyn, 1994; Sosnick
et al., 1994; Abkevich et al., 1994a; Bryngelson et al., 1995; Karplus and Sali,
1995; Onuchic et al., 1995; Baldwin, 1996; Privalov, 1996; Roder and Colon,
1 Introduction
1997; Nath and Udgaonkar, 1997a). Especially, proteins in which a single, very
deep global energy minimum is absent may display poor foldability and
complicated pathways with a number of early intermediates (Fersht, 1995c;
Abkevich et al., 1996; Shakhnovich, 1997). In particular, so-called molten globule
intermediates have found significant attention (Dolgikh et al., 1981; Nlting et al.,
1993; Ptitsyn, 1994, 1995; Chalikian et al., 1995; Fink, 1995; Gussakovsky and
Haas, 1995; Kuwajima, 1996; Fink et al., 1998). Another source of the occurrence
of intermediates is the existence of co-factors which often have dramatic
contributions to protein stability (Pfeil, 1981, 1993; Pfeil et al., 1991, 1993a, b;
Elve et al., 1994; Burova et al., 1995).
Fig. 1.3. Energy landscape of a protein. Only two reaction coordinates can be drawn. In reality
the energy landscape represents an n-dimensional hyper-surface in the (n+1)-dimensional space,
where n is the degree of freedom of conformational movement of the molecule.
On the other hand and surprisingly, small proteins have been discovered which
may complete the whole folding reaction in the submillisecond time scale
(Khorasanizadeh et al., 1993; Schindler et al., 1995; Robinson and Sauer, 1996;
Sosnick et al., 1996; Chan et al., 1997; Takahashi et al., 1997). Fast folding
sequences are found far easier if the structure of the protein is symmetric
(Wolynes et al., 1995; Wolynes, 1996). The maximum rate for protein folding is
estimated to be of the order of 1 s1 (Hagen et al., 1996, 1997) !
The occurrence of rapid events in the submillisecond time scale has been
detected indirectly with slow methods by the observation of burst-phases, i.e.,
changes of the signal within the dead time of the method (Fig. 1.4). However, the
precise and comprehensive analysis of early events requires a direct kinetic
resolution. In the past years we have seen a remarkable progress in the develop-
1 Introduction
Fig. 1.4. Burst-phase observed when refolding the 10 kDa protein C40A/C82A/P27A barstar at
5oC. Jumps are with different concentrations of urea as indicated. In 3 M urea the protein is
more than 80% unfolded, and it is more than 95% folded in 0 M urea. The circular dichroism
(CD) at 270 nm mainly reflects the structure consolidation in the vicinity of the 8 aromatic
amino acid residues.
ment of new methods which enable us to access the submillisecond, microsecond,

and even nanosecond time scale of protein folding (Fig. 1.5). The high kinetic and
structural resolution has profoundly altered the picture of folding reactions and
enhanced the understanding of the tremendous speed and efficiency of protein
folding (Nlting et al., 1995, 1997, 2003; Plaxco and Dobson, 1996; Wolynes et
al., 1996; Eaton et al., 1996a, 1997; Nlting and Andert, 2000). This book focuses
on the biophysical principles of the kinetic methods and the high structural
resolution of folding.
Fig. 1.5. Typical time scale of folding events under standard conditions, 25oC, pH 7.
2 Structures of proteins
2.1
Primary structure
The main building blocks of proteins are residues of 20 natural -amino acids.
With the exception of proline, their structure is:
(2.1)
In proline the sidechain, R, is bridged to the nitrogen atom of the amino group.
The structures of the sidechains and the properties of the amino acids are given in
Tables 2.1 and 2.2, respectively. The amino acid sidechains can be grouped in the
categories given in Table 2.2 or, alternatively, in the categories non-polar (glycine,
alanine, valine, leucine, isoleucine, proline, phenylalanine), polar (serine,
threonine, cysteine, methionine, asparagine, glutamine, tyrosine, tryptophan), and
charged (aspartic acid, glutamic acid, lysine, arginine, histidine).
With the exception of glycine, which is not chiral, all acids of natural occurring
proteins are L-isomers and are optically active. Tryptophan, tyrosine and phenylalanine absorb light at wavelengths below 310 nm, 300 nm and 270 nm, respectively. The first absorption maximum is around 280 nm for tryptophan and tyrosine and around 260 nm for phenylalanine. At 280 nm, the absorption of tyrosine
is 4 times lower than that of tryptophan at pH 6. Phenylalanine absorption at
260 nm is 6 times lower than tyrosine absorption at 280 nm (Wetlaufer, 1962).
In proteins, the amino acids are linked together by the peptide bond (Eq. 2.2),
which is formed upon condensation of two amino acids (Creighton, 1993).
(2.2)
Table 2.1. Structure of the sidechains, R, of natural amino acids. For proline, the backbone
nitrogen and CH-group are included.
2.1 Primary structure
Table 2.2. Properties of amino acid residues commonly found in proteins.

Amino acid
Alanine
Arginine
Asparagine
Aspartic acid
Cysteine
Glutamic acid
Glutamine
Glycine
Histidine
Isoleucine
Leucine
Lysine
Methionine
Phenylalanine
Proline
Serine
Threonine
Tryptophan
Tyrosine
Valine
Residue
massa
(daltons)
Van der Waals

volumeb
(3)
Frequency
in proteinsc
(%)
Chemical type
of sidechain
71.08
156.19
114.10
115.09
103.14
129.12
128.13
57.05
137.14
113.16
113.16
128.17
131.20
147.18
97.12
87.08
101.11
186.21
163.18
99.13
67
148
96
91
86
109
114
48
118
124
124
135
124
135
90
73
93
163
141
105
8
6
4
5
2
6
4
7
2
5
9
6
2
4
5
7
6
1
3
7
inactive, aliphatic
basic
amide
acidic
sulfur-containing
acidic
amide
inactive
basic
inactive, aliphatic
inactive, aliphatic
basic
sulfur-containing
inactive, aromatic
inactive
hydroxyl
hydroxyl
indole, aromatic
hydroxyl, aromatic
inactive, aliphatic
For the calculation of the molecular weight of proteins or peptides, 18 daltons have to be added
for the H and OH at N- and C-termini, respectively. 2 daltons are to be subtracted per disulfide
bridge. The sidechains labeled with inactive react only under extreme conditions.
a For neutralized sidechains (Lide, 1993; Creighton, 1993; Coligan et al., 1996).
b (Richards, 1974; Creighton, 1993).
c (McCaldon and Argos, 1988; Creighton, 1993; Coligan et al., 1996).
The short distance of the peptide bond, CN (Eq. 2.2; Fig. 2.1), of only about
1.311.34 , compared with about 1.44 1.48 for the nonpeptide CN bonds,
reflects its partial double-bond character.
Fig. 2.1. The peptide

bond. Distances are
averages observed in
several protein NMR
and crystal structures
(Brookhaven National
Laboratory
Protein
Data Bank; Abola et
al., 1987, 1997).
Due to the partial double-bond character of the peptide bond (Eq. 2.3), the
adjacent atom groups have a strong tendency to be coplanar.
(2.3)
(2.4)
Two configurations of the peptide bond are possible, the trans- and the cisform (Eq. 2.4). Because of steric hindrance between adjacent sidechains in the cisform, the trans-form is usually energetically favored by more than 17 kJ mol1
(4 kcal mol1), corresponding to less than 0.1% occupancy of the cis-isomer. For
the prolyl-peptidyl bond, the energy difference is significantly reduced: in small
peptides it is typically only about 23 kJ mol1 (0.50.7 kcal mol1), corresponding to about 7080% population of the trans-configuration (Fersht, 1985;
Schreiber, 1993b; Creighton, 1993).
The pKa's found in single amino acids change upon incorporation into the
protein due to the change of environment (Table 2.3). The acidic residues of
aspartic acid and glutamic acid are negatively charged and the basic residues of
lysine and arginine have a positive charge at pH 7. Histidine, which has a pKa
6 7, is a strong base at neutral pH and is involved in many enzymatic reactions
that involve a proton transfer.
Table 2.3. Observed pKa's of ionizable groups, found for single amino acids and for amino acid
residues in proteins.
Ionizable group
-Carboxyl
-Carboxyl (aspartic acid)
-Carboxyl (glutamic acid)
Imidazole (histidine)
-Amino
Thiol (cysteine)
Phenolic hydroxyl (tyrosine)
-Amino (lysine)
-Guanido (arginine)
a
b
pKa of amino acidsa
pKa of amino acid

residues in proteinsb
1.8 2.4
3.8
4.1
6.0
8.8 10.6
8.3
10.1
10.7
12.5
3.5 4.3
3.9 4.0
4.3 4.5
6.0 7.0
6.8 8.0
9.0 9.5
10.0 10.3
10.4 11.1
12.0
(Dawson et al., 1969; Fersht, 1985; Zubay, 1993; Lide, 1993).

(Bundi and Wthrich, 1979; Matthew, 1985; Creighton, 1993; Coligan et al., 1996).
2.2 Secondary structure
2.2
Secondary structure
Three main elements of well-defined secondary structure may be distinguished: the
-helix, the -sheet, and turns. These structural elements may be connected with
each other by loops. Helices are the most abundant form of secondary structure in
globular proteins, followed by sheets, and in the third place turns. Secondary
structure formation provides an efficient mechanism of pairing polar groups of the
polypeptide backbone by hydrogen bonds. Uncoupling of only a single pair of
polar groups in the protein interior may cause an energy cost of 40 kJ mol1
(9.6 kcal mol1) (Privalov and Makhatadze, 1993). Protein secondary structure can
directly be observed by atomic force microscopy (McMaster et al., 1996).
Secondary structure elements may associate through sidechain interactions to form
super-secondary structure, so-called motifs.
Different probabilities are observed for the incorporation of amino acid residues
into different types of secondary structure (O'Neil and DeGrado, 1990; Creighton,
1993; Coligan et al., 1996; Hubbard et al., 1996). Using X-ray crystallographic
data of a large set of proteins, Chou and Fasman (Chou and Fasman, 1977, 1978a,
1978b; Chou, 1989) calculated statistical conformational preference parameters
which were based on the occurrence of a specific amino acid type in a specific
type of secondary structure, on the relative frequency of that amino acid type in
the databases, and on the relative number of amino acid residues occurring in each
type of secondary structure (Table 2.4). For example, prolines and glycines are
considered as helix-breakers since their preference parameters for helices are less
than half of that of alanine, a so-called helix-former. Further progress has been
made with the recognition that the conformational preference depends on the
relative position in the secondary structure element (Presta and Rose, 1988;
Richardson and Richardson, 1988; Harper and Rose, 1993). For example, glycine,
serine, and threonine often constitute the amino-terminal residues (N-cap) in helices. Glycine and asparagine are frequently found at the carboxyl-terminal
position (C-cap) of -helices. They are referred to as being N-cap and C-cap
stabilizers, respectively.
The -helix (Fig. 2.2) is stabilized by hydrogen bonds between the carbonyl
oxygen of the amino acid residue at the position n in the polypeptide chain with
the amide group, NH, of the residue n + 4.
Hydrogen bonds between carbonyl oxygens and amide groups of adjacent
strands stabilize -sheets (Fig. 2.3). The occurrence of -sheets is often correlated
with high hydrophobicities (see Sect. 3.4) of the involved amino acid residues:
Isoleucine, valine, tyrosine, and phenylalanine prefer -sheet structure, but aspartic
acid and glutamic acid have an aversion to incorporation into -sheets (Table 2.4).
Turns (Fig. 2.4) involve a 180o change in direction of the polypeptide chain and
are stabilized by a hydrogen bond between the carbonyl oxygen of the residue at
the position n with the amide group, NH, of the residue n + 3 (Fersht, 1985).
Less common elements of secondary structure are also 310-helices and hairpins.
10
Fig. 2.2. Right-handed helix consisting of 8 amino

acid residues. -Helices are
stabilized by hydrogen bonds
between the carbonyl oxygen
atom of amino acid residue
number n and the amide
group, NH, of residue number
n + 4 in the polypeptide chain,
as indicated by dashed lines
(for simplicity, only 4 of the
hydrogen atoms are shown).
Fig. 2.3. Antiparallel -sheet consisting of 10 amino acid residues. The -sheet is stabilized by
hydrogen bonds between carbonyl oxygen atoms and amide groups, NH, of adjacent strands, as
indicated by dashed lines (for simplicity, only 4 of the hydrogen atoms are shown).
Fig. 2.4. Type I turn. Turns

are stabilized by a hydrogen
bond between amino acid
residues n and n + 3, as indicated by a dashed line (for
simplicity, the other hydrogen
atoms are not displayed).
2.3 Tertiary structure
11
Table 2.4. Preferences of amino acids for different types of secondary structure.
Conformational preference parametera
Amino acid
-Helix
-Strand
Turn
Alanine
Arginine
Asparagine
Aspartic acid
Cysteine
Glutamic acid
Glutamine
Glycine
Histidine
Isoleucine
Leucine
Lysine
Methionine
Phenylalanine
Proline
Serine
Threonine
Tryptophan
Tyrosine
Valine
1.3 1.5
0.9 1.4
0.8 0.9
0.9 1.1
0.9 1.0
1.4
1.1 1.4
0.4 0.6
1.0 1.2
1.0 1.1
1.3
1.1 1.2
1.3 1.4
1.0 1.1
0.5 0.6
0.7 0.8
0.7 0.8
1.0
0.7 0.9
0.9 1.0
0.8 0.9
0.7 1.0
0.6 0.7
0.5 0.7
0.8 1.2
0.5 0.8
0.8 1.0
0.6 0.9
0.8 1.1
1.5 1.8
1.0 1.2
0.7 0.9
1.0 1.3
1.2 1.4
0.4 0.6
0.9 1.0
1.2 1.3
1.2
1.2 1.5
1.5 1.7
0.7
0.9 1.0
1.3 1.6
1.4 1.5
0.9 1.2
0.7 1.0
1.0
1.6
0.7 1.0
0.5
0.6
1.0
0.4 0.6
0.6
1.5 1.9
1.3 1.4
1.0
0.8 1.0
1.1
0.5
(Chou and Fasman, 1977, 1978a, b; Chou, 1989; Creighton, 1993; Thornton et al., 1995).
2.3
Tertiary structure
The polypeptide chain of natural proteins is unbranched but may contain disulfide
bridges. Some proteins can incorporate cofactors, e.g., heme or chlorophyll.
Protein cores are packed as tightly as organic solids, or slightly tighter. Also the
compressibility of the interior of globular proteins, (14 3) 1011 Pa1, is
comparable with that of soft organic solids, and is 3 times less than the
compressibility of water. Partial volumes of globular proteins in aqueous solution
typically are 0.70 0.75 L kg1 (Sarvazyan, 1991; Kharakoz and Sarvazyan,
1993; Nlting and Sligar, 1993).
For most proteins the tertiary structure is very well defined, and many of the
sidechain rotation-, segmental flexibility-, and molecular breathing motions are on
a scale of less than 2 . Sidechains located in the interior of the protein molecule
usually rotate or perform 180o flips with frequencies of 102107 Hz. Rotation of
buried tryptophan sidechains is usually so infrequent that they are considered as
almost immobile. However, structural transitions which involve large conformational changes can play a crucial role in enzymatic and binding reactions. In the
crystal structures of some proteins whole domains are statically disordered, i.e.,
different conformations are occupied.
12
13
Fig. 2.5. Examples for classes of folds of soluble proteins (Poulos et al., 1986; Abola et al., 1987,
1997; Billeter et al., 1990; Kim et al., 1990; Katayanagi et al., 1992; Tilton et al., 1992; Korolev
et al., 1995; Murzin et al., 1995; Kumaraswamy et al., 1996; Qi et al., 1996; Riek et al., 1996;
Zhu et al., 1996; Chothia et al., 1997). The protein backbones are symbolized by ribbons. The
heme cofactors of cytochrome c and cytochrome P-450cam are shown as wireframes. All alpha
(or almost all alpha): A (cytochrome c), B (cytochrome P-450cam). All beta: C (-amylase
inhibitor), D (-crystallin). Alpha and beta (parallel or antiparallel -sheets; non-segregated or
segregated - and -regions): E (ribonuclease H), F (restriction endonuclease EcoRI bound to
DNA), G (prion protein domain), H (ribonuclease A). Multi domain: I (substrate-binding domain
of the chaperone DnaK), J (fragment of Thermus aquaticus DNA polymerase). The figure was
drawn using the program MOLSCRIPT (Kraulis, 1991).
The unique tertiary structure of each protein is determined by its amino acid
sequence. Usually, the native structure of protein is at the minimum free energy
(see Fig. 1.3). Exceptions to this basic tenet of protein folding are very rare (Sohl
et al., 1998).
14
C
Fig. 2.6. Two examples for the structures of membrane proteins (Abola et al., 1987, 1997;
Cowan et al., 1992; Murzin et al., 1995; Chothia et al., 1997; Prince et al., 1997).
A, B: phosphoporin, C: fragment of the purple bacteria light-harvesting complex LH2. The
backbones are symbolized by ribbons and the cofactors chlorophyll and carotenoid of the lightharvesting protein are shown as wireframes. The figure was drawn using the program
MOLSCRIPT (Kraulis, 1991).
Nature has evolved a gigantic variety of protein three-dimensional structures,

so-called protein folds. For the thousands of coordinate files deposited in protein
databases, most notably the Brookhaven National Laboratory Protein Data Bank
(Abola et al., 1987, 1997) which may be accessed via the World Wide Web with
an entry point http://www.rcsb.org/pdb/, 7 classes of folds with more than 270
15
folds may be distinguished (Table 2.5; Murzin et al., 1995). Examples for the huge
variety of protein folds are presented in Figs. 2.5 and 2.6.
Together with the chemical diversity of the sidechains, this structural variety
contributes to an astonishing functional diversity of proteins, ranging from
catalysis or inhibition of various chemical or physical reactions, to the transport of
other molecules, electrons, protons, or excitons, to the stabilization of the
architecture within cells. The examples in Fig. 2.5 are soluble proteins. Many
important proteins incorporate cofactors that are covalently (e.g., heme in
cytochrome c; Fig. 2.5) or non-covalently bound (e.g., heme in cytochrome
P-450cam; Fig. 2.5). Membrane proteins are exemplified by phosphoporin and a
fragment of the purple bacteria light-harvesting complex LH2 (Fig. 2.6). The
protein matrix of the LH2 fragment embeds 9 chlorophyll- and at least
3 carotenoid molecules. Complicated structures like this have evolved over many
millions of years and contribute to an amazing efficiency of light-harvesting
complexes of bacteria and higher plants, that cannot be reproduced in vitro when
using a simple solution of chlorophyll molecules. In some higher plants, up to
98% of the absorbed photons are transmitted to the reaction centers via an
excitonexciton transfer mechanism.
The visualization of protein structures in the folded state and along the folding
pathway, and the quantitation of kinetic rate constants is seen to be of paramount
importance for an understanding of protein function and mechanism, and will aid
the understanding of important biological processes and disease states through
detailed mechanistic knowledge. The next chapters are devoted to the mathematical, biophysical, chemical, and molecular biological methods of high kinetic
and structural resolution of chemical and biophysical reactions of proteins with
emphasis on folding reactions.
Table 2.5. Classes of folds found in the protein databases (Murzin et al., 1995; Chothia et al.,
1997).
Class of protein fold
Relative abundance
All alpha
20 30%
All beta
10 20%
Alpha and beta with mainly parallel -sheets (/)
15 25%
Alpha and beta with mainly antiparallel

-sheets with segregated - and -regions (+)
20 30%
Multi domain (alpha and beta)

Membrane and cell surface proteins
Small proteins (dominated by cofactors or
disulfide bridges)
< 10%
< 10%
5 15%
3 Physical interactions that determine the

properties of proteins
This chapter gives an introduction into the physical forces that determine, together
with covalent interactions, the conformations along the folding pathway, including
the folded and unfolded structures. These forces also dominate the non-covalent
mutual interactions between (a) two protein molecules, (b) proteins and other
macromolecules, and (c) proteins and solvent. Further information may be found
in Cantor and Schimmel, 1980; Fersht, 1985; Creighton, 1993; Makhatadze and
Privalov, 1993; Privalov and Makhatadze, 1993.
Electrostatic interactions of point charges (Sect. 3.1.1) crucially affect most
long-range interactions of proteins with proteins and other charged macromolecules. Van der Waals interactions (Sects. 3.1.2 and 3.2) are considered the main
contributors to the stabilization of globular proteins, followed by hydrogen bonds
(Sect. 3.3), and in the third place hydrophobic interactions (Sect. 3.4) of non-polar
residues (Privalov and Makhatadze, 1993). In order to produce a stable folded
protein conformation, these contributions have to overcompensate the destabilizing contributions from the hydration of polar residues (see Sect. 3.4) and the gain
in configurational entropy upon unfolding. The magnitudes of stabilizing and
destabilizing contributions to the overall protein stability typically are several
1000 kJ mol1. A delicate balance between stabilizing and destabilizing contributions causes a stability of most globular proteins in water in the range of only
10 70 kJ mol1 (Privalov, 1979; Privalov and Makhatadze, 1993).
3.1
Electrostatic interactions
3.1.1
Point charges
Coulomb's law provides the force, F, between two charged species, Z1 and Z2
F=
1
Z1Z 2
,
4 o r d12, 2
(3.1)
where d1,2 is the distance between Z1 and Z2 , o = 8.8541012 C V1 m1 is the

permittivity of vacuum, and r is the relative permittivity. The change of energy,
E , as function of the distance separation is obtained by integration:
d2
E = Fdd1, 2 =
d1
Z1Z 2 1
1
( ).
4 o r d1 d 2
(3.2)
18
3 Physical interactions that determine the properties of proteins
r reflects the polarizability of the medium between the charges. It largely differs
between different solvents (Table 3.1). For example, the energy necessary to
separate positive and negative elementary charges, e = 1.6021019 C, from a
distance of 10 to infinity in vacuum is 139 kJ mol1 (33.3 kcal mol1). In water
the energy is decreased by a factor of roughly r = 78.
Solvents that correspond chemically to the interior of proteins have a relative
permittivity, r , which is roughly one order of magnitude lower than that of water.
Thus, Coulomb interactions of charges in the interior of proteins are typically one
order of magnitude stronger than at the surface of proteins in aqueous solution.
For example, surface charge mutations often change the protein stability by less
than 4 kJ mol1 (1 kcal mol1), while changes of more than 4 kJ mol1 are not
unusual for buried charge mutations.
Point charges have a wide range of interaction. In the folding reaction,
Coulomb interactions can effectively steer one structural element towards another
distant structural element. Coulomb interactions can also steer one protein
molecule towards another. For example, the positively charged active site of the
ribonuclease barnase steers the negatively charged inhibitor barstar into the
optimal position for binding (Schreiber et al., 1994). Strong electrostatic protein
protein interactions can result in a very strong association (Schreiber et al., 1994).
Proteinprotein complexes that are stabilized mainly by electrostatic interactions, rapidly become weakened with increasing salt concentration because
protein charges become neutralized by counter ions. Proteins with large net
charges may often be stabilized by salts that suppress the intramolecular charge
repulsion.
Table 3.1. Properties of solvents.
Solvent
Molecular
formula
Water
Methanol
Ethanol
2-Propanol
2-Propanone
2-Butanone
2-Pentanone
2-Hexanone
Phenol
Benzene
Hexane
H2O
CH4O
C2H6O
C3H8O
C3H6O
C4H8O
C5H10O
C6H12O
C6H6O
C6H6
C6H14
Relative
permittivity, ra
78.4
32.7
24.3
18.3
20.7
18.5
15.5
14.6
9.8
2.3
1.9
Hydrophilicityb
(kJ g1)
0.67
0.46
0.33
0.28
0.21
0.17
0.14
0.29
0.05
0.12
(Lide, 1993).
Gibbs free energies of transfer form the gaseous phase into water (Cabani et al., 1981; Privalov
and Makhatadze, 1993); 1 kJ g1 = 0.24 kcal g1.
b
3.2 Quantum-mechanical short-range repulsion
19
3.1.2
Point charge dipole and dipole dipole interactions
The energy of interaction of a point charge with an induced dipole (for example,
of interaction of a polarizable molecule with an ion) falls off as d 4, where d is the
distance of separation between charge and dipole (Fersht, 1985).
The energies of interaction between (a) randomly orientated permanent dipoles,
(b) a permanent dipole and a dipole induced by it, and (c) mutually induced
dipoles fall off approximately as d 5 to d 6 (Fersht, 1985; Creighton, 1993). These
types of interactions are the main origin of the attractive component of the van
der Waals forces (see Figs. 3.13.3 in Sect. 3.2). Type (c) occurs between all
atoms and is also known as the dispersion forces or London forces.
3.2
Quantum-mechanical short-range repulsion
The repulsive component of the van der Waals interaction (Figs. 3.13.2) falls off
approximately as d 12 to e d, where d is the distance of separation. Its main origin
is the quantum-mechanical Pauli exclusion principle. Note that historically only
the attractive forces (a) and (c) in Sect. 3.1.2 were called van der Waals forces.
Fig. 3.1. Van der Waals potential as function of the distance separation for the interaction of two
carbon atoms with C6 = 6103 6 kJ mol1, C12 = 1.1107 12 kJ mol1 (Warshel and Levitt,
1976; Creighton, 1993). The van der Waals potential contains an attractive component that
mainly originates from mutually induced dipole dipole interactions and falls off with the sixth
power of distance separation, and a repulsive component that mainly originates from the Pauliexclusion and falls off with the twelfth power of the distance separation. Van der Waals
interactions have a short range of only a few (1 = 1010 m). The energies of van der Waals
interactions of the atoms commonly found in proteins are small and of the order of only 0.1
2 kJ mol1, compared with the energy of 10 60 kJ mol1 per hydrogen bond, and the energy of
up to several 10 kJ mol1 per buried salt-bridge (1 kJ mol1 = 0.24 kcal mol1).
20
Fig. 3.2. Van der Waals potential as function of distance separation for the interaction of two
carbon atoms with C6 = 6103 6 kJ mol1, and C12 = 1.1107 12 kJ mol1, compared with the
Coulomb interaction in vacuum of two elementary charges, e = 1.6021019 As (1 kJ mol1 =
0.24 kcal mol1). The force is repulsive for the same sign of the charges, otherwise it is
attractive. Compared with Coulomb forces of point changes, van der Waals interactions are
intrinsically weak and have a short range of interaction. However, cooperation of a large number
of van der Waals interactions can produce a stable conformation (Creighton, 1993).
Fig. 3.3. Approximate van der Waals potentials as function of distance separation for the
interaction of two hydrogen atoms, two tetrahedral carbon atoms, and two carboxyl carbon
atoms, respectively, calculated using data from (Warshel and Levitt, 1976; Fersht, 1985). The
van der Waals potentials of hydrogen, carbon, nitrogen, oxygen, and sulfur atoms display a
shallow attractive energy minimum at distances of about 2.6 4.4 , corresponding to radii of
1.3 2.2 , and a strong repulsion at shorter distances. Van der Waals interactions of hydrogen
atoms are intrinsically weaker than those of carbon atoms. Usually, carboxyl carbon atoms have
a stronger interaction and a shorter van der Waals radius than tetrahedral carbon atoms.
3.3 Hydrogen bonding
21
The different components of the van der Waals interaction are often approximated by the LennardJones 6,12 potential:
E=
C12 C6
d 12 d 6
C12 = C12, i C12, j
(3.3)
C6 = C6, i C6, j ,
where C6, i , C12, i , C6, j , C12, j are parameters for the atoms i , and j , respectively,
which usually are derived for the potential between two atoms of the same type.
Even though the van der Waals interactions are weak, in proteins they accumulate to a significant amount. The strength of the individual interaction depends on
the types of interacting atoms, and varies with the chemical environment of the
atoms involved. For example, for carboxyl carbon atoms the interactions are
usually stronger, and van der Waals distances are usually shorter than for
tetrahedral carbon atoms (Fig. 3.3).
3.3
Hydrogen bonding
A hydrogen bond contains both positive (H-donor) and negative (H-acceptor)
partial charges. It represents a combination of covalent and electrostatic
interactions, but the main component is the electrostatic attraction between
hydrogen donor and acceptor. The magnitude of reduction of the van der Waals
distance is indicative of the strength of the hydrogen bond (Table 3.2). The Gibbs
free energy contributions per hydrogen bond in the interior of proteins are
estimated to be 10 60 kJ mol1 (214 kcal mol1) (Hagler et al., 1979; Dauber
and Hagler, 1980; Privalov and Makhatadze, 1993). Consider, for example, the
hydroxyl carbonyl bond which is one of the strongest hydrogen bonds in
proteins:
+
OH ... O=C<
(3.4)
The electronegativity of the hydroxyl oxygen atom causes a positive partial charge
of the hydroxyl hydrogen atom, the H-donor. Similarly, the carbonyl oxygen atom
has a negative partial charge which attracts the hydroxyl hydrogen atom.
Table 3.2. Properties of hydroxyl hydroxyl and amide carbonyl hydrogen bonds found in
proteins.
Type of
hydrogen bond
hydroxyl hydroxyl
amide carbonyl
Molecular
formula
Typical
H ... O distance
()
Typical reduction
of van der Waals
distance
OH...OH
>NH...O=C<
1.9 2.3
1.8 2.2
20 25%
20 30%
22
Hydrogen bonding of proteins in aqueous solution is profoundly altered by

addition of co-solvents. Hydrophobic co-solvents, for example, phenol and
benzene, may form significantly fewer hydrogen bonds with polar and charged
groups at the surface of proteins than water, and can destabilize most native
proteins (see Sect. 3.4). Trifluoroethanol (TFE) stabilizes helices by strengthening
their hydrogen bonds but destabilizes most native proteins by weakening the
hydrophobic interaction in the core of the protein (Luo and Baldwin, 1998).
3.4
Hydrophobic interaction
Table 3.3. Properties of amino acids.
Amino acid
Alanine
Arginine
Asparagine
Aspartic acid
Cysteine
Glutamic acid
Glutamine
Glycine
Histidine
Isoleucine
Leucine
Lysine
Methionine
Phenylalanine
Proline
Serine
Threonine
Tryptophan
Tyrosine
Valine
a
Codes
Ala
Arg
Asn
Asp
Cys
Glu
Gln
Gly
His
Ile
Leu
Lys
Met
Phe
Pro
Ser
Thr
Trp
Tyr
Val
A
R
N
D
C
E
Q
G
H
I
L
K
M
F
P
S
T
W
Y
V
Accessible
surface area
of residuea
Relative hydrophilicity
of amino acid residuesb
Gcyclohexanewater
(2)
(kJ g1)
115
225
160
150
135
190
180
75
195
175
170
200
185
210
145
115
140
255
230
155
0.05
0.42
0.28
0.35
0.01
0.25
0.21
0.00
0.17
0.15
0.15
0.21
0.04
0.06
0.21
0.15
0.03
0.03
0.13
Estimated with the rolling ball method.

Hydrophilicity at 25oC is relative to glycine, and is based on the partitioning of a sidechain
analogue between the two states (1 kJ g1 = 0.24 kcal g1). The Gibbs free energy of transfer is
given by Gcyclohexanewater = RT ln(cwater / ccyclohexane) , where R is the universal gas constant,
T is the absolute temperature, and cwater and ccyclohexane are the molar concentrations of
sidechain analogues in the different phases (Radzicka and Wolfenden, 1988; Creighton, 1993;
Privalov and Makhatadze, 1993).
b
3.4 Hydrophobic interaction
23
Fig. 3.4. Temperature dependence of the Gibbs free energy of transfer from vapor into water
(hydrophilicity; 1 kJ g1 = 0.24 kcal g1) for uncharged (neutralized) amino acid sidechains
(Privalov and Makhatadze, 1993).
The absence of hydrogen bonding between water and non-polar groups rather than
the presence of favorable interactions between the non-polar groups themselves
constitutes an important source of the protein stability in aqueous solution, the socalled hydrophobic interaction (Table 3.3; Figs. 3.43.6; Rose, 1987; Weber,
1996). Hydrophobicity and hydrophilicity usually are expressed as the Gibbs free
energies of transfer from water into the reference state, and from a reference state
into water, respectively. The transfer of the sidechains of hydrophobic amino acid
residues, for example, leucine, isoleucine, and valine, from cyclohexane into
water is energetically costly, and thus, the burial of hydrophobic sidechains in the
folding reaction is energetically favorable. In contrast, hydrophilic sidechains, for
example, that of arginine, prefer an aqueous environment over a hydrophobic
environment, and are preferentially found at the surface in folded proteins.
24
Fig. 3.5. Calculated temperature dependence of the change of the Gibbs free energy of hydration
(hydrophilicity; 1 kJ g1 = 0.24 kcal g1) of internal groups upon protein unfolding for horse
heart cytochrome c, hen egg-white lysozyme, pancreatic ribonuclease A, and sperm-whale
myoglobin, as indicated (Privalov and Makhatadze, 1993). Ghyd,FU is negative because it is
largely dominated by the contributions of polar groups (Privalov and Makhatadze, 1993).
Fig. 3.6. The formation of cages around non-polar molecules in aqueous solution at low
temperatures is connected with a decrease of entropy.
The Gibbs free energy of transfer, Ghyd , of a non-polar molecule from a

reference state, such as cyclohexane, into water (hydrophilicity) is composed of
an enthalpy, Hhyd , and entropy, TShyd , term:
Ghyd = Hhyd TShyd ,
(3.5)
3.4 Hydrophobic interaction
25
where T is the absolute temperature. At room temperature, Hhyd for the transfer
from cyclohexane into water is small and Ghyd is dominated by the entropy term
(Weber, 1996). This is mainly because the formation of ordered water cages
around non-polar compounds is an entropically costly process, i.e., is connected
with a decrease of entropy (Fig. 3.6).
Different chemical groups make vastly different contributions to the Gibbs free
energies of transfer from organic solvent into water and of transfer from the
gaseous phase into water (Fig. 3.4). For example, for non-cyclic structures, the
Gibbs free energies of transfer from vapor into water for the groups CH3 ,
CH2 , CH< , >C< , >C=O , NH2 , OH , and NH are 0.25, 0.05, 0.12,
0.41, 0.83, 1.48, 1.51, and 1.71 (all in kJ g1), respectively (Privalov and
Makhatadze, 1993).
Below 100C, the hydrophobic effect usually increases with temperature (Figs.
3.4, 3.5). At very high temperatures it does not further increase, but approaches a
maximum, mainly because the structure-forming tendency of water, i.e., the
entropic contribution to the hydrophobic effect, decreases with increasing
temperature (Rose, 1987; Makhatadze and Privalov, 1993; Privalov and
Makhatadze, 1993; Weber, 1996).
Intriguingly, studies on small organic compounds and proteins suggest that the
change of Gibbs free energy of hydration of internal groups upon protein
unfolding, Ghyd,FU , is negative for most proteins because Ghyd,FU is largely
dominated by the contributions of polar groups that prefer an aqueous over a
hydrophobic environment (Fig. 3.5; Privalov and Makhatadze, 1993).
4 Calculation of the kinetic rate constants
Protein folding reactions can proceed according to a variety of different mechanisms. This chapter presents analytical solutions for kinetic rate constants and
amplitudes for common reaction mechanisms.
The simplest case is that of a two-state transition, i.e., a reaction that proceeds
without the occurrence of intermediates directly from the unfolded state, U, to the
folded state, F (Sect. 4.2). In the transition region of the reaction U
F , both
forward and backward reaction contribute significantly to the observed rate
constant (relaxation constant, decay constant). Under conditions that strongly
favor folding (or unfolding), i.e., far outside the midpoint of equilibrium between
folded and unfolded state, the transition can be treated as an irreversible reaction
with the observed rate constant being dominated by the folding (or unfolding) rate
constant.
For reversible three-state transitions, three cases have to be distinguished:
1. The intermediate, I, is on-pathway (U
I
F), i.e., is always passed through
in the reaction from U to F (Sect. 4.3.1.1). 2. All species may interconvert, i.e., the
transition from U to F may be passed through directly and also through the
intermediate, I (Sect. 4.3.1.2). 3. I is off-pathway (I
U
F or U
F
I),
i.e., the reaction from U to F cannot proceed through I (Sect. 4.3.1.3).
Derivations of solutions for four-state transitions involve the treatment of cubic
equations (Sect. 4.4).
Occasionally, folding reactions are linked with monomermultimer transitions
(Sect. 4.5). Examples are, (a) the protein is monomeric in the unfolded state but
dimeric in the folded state, or (b) the protein aggregates in the unfolded, folded, or
an intermediate state. Since these transitions affect the observed rate constants for
folding events, solutions for a few simple cases are also presented.
Many important kinetic experiments (see Chaps. 5 and 10) involve the application of perturbation methods, such as small-amplitude temperature-jumping,
repetitive pressure perturbation, ultrasonic velocimetry, and dielectric relaxation.
These methods utilize a small perturbation of the chemical or physical equilibrium: A small change of physical or chemical conditions initiates a relaxation
process to a new equilibrium. Since the amplitude of the perturbation is small, the
mathematical treatment is tremendously simplified (Sect. 4.6).
The mathematical methods and analytical solutions presented for kinetic rate
constants and amplitudes are not limited to protein folding reactions, but may be
applied to a large variety of other chemical or physical reactions, for example, (a)
in case of unimolecular mechanisms to conformational changes of other macro-
28
molecules (peptides, carbohydrates, lipids, DNA), and (b) in case of bimolecular

mechanisms to aggregation-, enzymesubstrate binding-, and enzymeinhibitor
binding reactions.
Kinetic rate constants and amplitudes of unimolecular and bimolecular reactions are solutions of differential equations. Since no general mathematical
formalism for the analytical solution of all differential equations has been found,
the finding of a particular solution is often based on a mere guess that is confirmed
by inserting it into the equation. For the confirmation of a solution as the general
solution it is important to check whether it fulfills every possible initial condition.
Fortunately, the rate equations of unimolecular reactions are ordinary linear
differential equations which generally have solutions that are linear combinations
of exponential functions.
4.1
Transition state theory
The rate constant of the formation of a product, kif , in a step of the folding reaction (Fig. 4.1; Fersht, 1985; Matouschek et al., 1989) is, in good approximation,
kif = (kBT/h) exp(G#i /(RT)) ,
(4.1)
where kB = 1.38071023 J K1 is the Boltzmann constant, h = 6.62611034 J s is

the Planck constant, T is the absolute temperature, R = 8.3145 J mol1 K1 is the
molar gas constant, and G#i is the Gibbs free energy of activation.
Fig. 4.1. Transition state theory. The transition state is the state of highest energy along the
reaction pathway that leads from the initial state (ground state) to the final state (product). The
height of the transition state barrier determines the magnitude of the rate constant of transition
(G#i and G#f = G#i Gfi determine the rate constants of i
f and f
i, respectively).
4.2 Two-state transitions
29
The Gibbs free energy change of the reaction, Gfi , is connected with the
equilibrium constant of the reaction, Kfi , i.e., the ratio of product to reactant in
equilibrium, by the well-known relation
Gfi = Hfi TSfi = RT ln(Kfi) ,
(4.2)
where Hfi is the enthalpy change and Sfi is the entropy change of the reaction.
4.2
Two-state transitions
4.2.1
Reversible two-state transition
To derive the rate equations for a reversible two-state transition between the states
U and F
,
(4.3)
we have to consider that the quantity of the decay of reactant, U , per time unit is
proportional to the quantity of reactant itself and the quantity of the decay of
product, F , per time unit is proportional to the quantity of product:
d[F ]
= k1 [U] k1[F]
dt
d[U]
= k1[F] k1[U] ,
dt
(4.4)
where [U] , [F] , k1 , k1 , and t are the concentrations of U and F , the forward rate
constant, the backward rate constant, and the time, respectively. Taking into
account that the total concentration of species, [UF] [U] + [F] , is conserved, the
rate equation for the change of the folded state may be written as
d[F ]
= (k1 + k1 )[F] + k1 [UF]
dt
[F](0) = [Fo] ,
(4.5)
where [Fo] is the concentration of F at the start of the reaction, i.e., at t = 0. The
solution of Eq. 4.5 is easily found by using the guess that the solution is a singleexponential function:
[F](t) = C1 exp(k1t k1t) + C2
(4.6)
C1 = [Fo] [UF]k1/(k1 + k1)
C2 = [UF]k1/(k1 + k1)
[U](t) = [UF] [F](t) .
It can be shown that Eq. 4.6 fulfills every initial condition [Fo] [ 0 , [UF] ] and
represents the general solution.
30
Consequently, [U](t) and [F](t) follow single-exponential functions with an

observed rate constant, kobs = k1 + k1 (Fig. 4.2, Table 4.1 in Sect. 4.7).
Fig. 4.2. Single-exponential change of the population of the folded state in a reversible two-state
transition (U
F) under conditions that favor unfolding. The parameters chosen for this
example are: k1 = 1, k1 = 3, [Fo]/[UF] = 0.85. The observed rate constant (relaxation constant) is
kobs = k1 + k1 = 4. Inset: Energy landscape.
4.2.2
Irreversible two-state transition
Under conditions which strongly favor folding, the unfolding rate constant may be
neglected:
(4.7)
The rate equation for F in this consecutive two-state transition is
d[F ]
= k1 [U]
dt
[U] + [F] = [UF]
(4.8)
[F](0) = [Fo] ,
where [U] , [F] , k1 , and t are the concentrations of U and F , the forward rate
constant, and the time, respectively. Here the solution is (see also Table 4.1)
[F](t) = ([Fo] [UF]) exp(k1t) + [UF]
[U](t) = ([UF] [Fo]) exp(k1t) .
(4.9)
4.3 Three-state transitions
31
[U](t) and [F](t) follow single-exponential functions with an observed rate

constant, kobs = k1. In contrast to reversible reactions, here the population of U
vanishes with time (Fig. 4.3).
Analogously, for conditions that strongly favor unfolding, for example, at high
concentrations of denaturant, we find kobs = k1.
Fig. 4.3. Single exponential change of the

fraction of the folded
state, [F]/[UF], in an
irreversible
two-state
F).
transition
(U
The parameters chosen
for this example are:
k1 = 1; [Fo]/[UF] = 0.85.
4.3
Three-state transitions
4.3.1
Reversible three-state transitions
4.3.1.1
Reversible sequential three-state transition
For a reversible sequential three-state transition,
,
(4.10)
between the states U , I , and F , with the positive rate constants, k1 , k1 , k2 , and
k2 , the rate equations are
d[U]
= k1[I] k1 [U]
dt
d[F ]
= k2 [I] k2 [F]
dt
[I] = [UIF] [U] [F] ,
(4.11)
32
where [UIF] is the total concentration of the species U , I , and F , which is

conserved in the reaction. Considering for simplicity first only the changes of [U] ,
[I] , and [F] ,
d [U]
(4.12)
= k1 [I] k1[U]
dt
d [F]
= k2 [I] k 2 [F]
dt
[I] = [U] [F] ,
and assuming that the solution of Eq. 4.12 is of the form
[U](t) = C1 exp(t)
[F](t) = C2 exp(t) ,
(4.13)
[U] = k1 ([U] + [F]) k1 [U]

[F] = k2 ([U] + [F]) k2 [F] .
(4.14)
we obtain,
By substituting [U] or [F] we find an equation for ,

0 = 2 (k1 + k1 + k2 + k2) + k1k2 + k1k2 +k1k2 ,
(4.15)
which has two solutions,
1,2 = 0.5 (k1 + k1 + k2 + k2

((k1 + k1 + k2 + k2)2 4 (k1k2 + k1k2 + k1k2))1/2) .
(4.16)
For physically permissible, i.e., positive rate constants, the term under the root
cannot be negative. Thus, both solutions are real. Both = 1 and = 2 inserted
into Eq. 4.13 fulfill Eq. 4.12, and thus, represent particular solutions. The general
solution of Eq. 4.11 is a superposition of the two particular solutions:
[U](t) = C 1 exp(1t) + C3 exp(2t) + C5
[F](t) = C2 exp(1t) + C4 exp(2t) + C6 .
(4.17)
The constants, Ci , may be determined by inserting Eq. 4.17 into Eq. 4.11, and
using the conservation relationship, [UIF] = [U] + [I] + [F] , and the initial conditions, [U](0) = [Uo] , [F](0) = [Fo]:
C1 = ([Fo] 2[Uo] C6 + 2C5) / (1 2)
C2 = 1 C1
C3 = ([Fo] 1[Uo] C6 + 1C5) / (2 1)
C4 = 2 C3
C5 = [UIF]k1k2/(k1k2 + k1k2 + k1k2)
C6 = [UIF]k1k2/(k1k2 + k1k2 + k1k2)
1 = (1 k1 k1) / k1
2 = (2 k1 k1) / k1 .
Alternatively, we can use the relations
(4.18)
C1 = 3 C2
3 = (1 k2 k2) / k2
C3 = 4 C4
4 = (2 k2 k2) / k2 .
33
(4.19)
From 1 = 13 follows 1 0 and 3 0, and thus, C2 = 0 only if C1 = 0 , and C4 = 0

only if C3 = 0.
Fig. 4.4. Double-exponential change of the populations of folded, F, intermediate, I, and

unfolded, U, state in a reversible sequential three-state transition (U
I
F) under conditions
that favor folding. The parameters chosen for this example are: k1 = 4, k1 = 2, k2 = 2, k2 = 1,
[Fo]/[UIF] = 0, [Uo]/[UIF] = 1. Observed rate constants ( kobs) are 1 = 7 and 2 = 2. Inset:
Energy landscape.
Fig. 4.5. Double-exponential change of the populations of folded, F, intermediate, I, and

unfolded, U, state in a reversible sequential three-state transition (U
I
F) under conditions
that favor unfolding. The parameters chosen for this example are: k1 = 2, k1 = 4, k2 = 1, k2 = 2,
[Fo]/[UIF] = 1, [Uo]/[UIF] = 0. Observed rate constants ( kobs) are 1 = 7 and 2 = 2. Inset:
Energy landscape.
34
Summarizing, the general solutions for the reversible sequential three-state

transition (Eq. 4.10) usually are double-exponential functions, given by Eqs. 4.17
and 4.18 (see Table 4.1). Two special cases, where both [F] and [U] follow only
single-exponential functions are: C1 = C2 = 0, and C3 = C4 = 0. In case C1 = C2 =
C3 = C4 = 0, no kinetic event is macroscopically observed because the reaction is
already in equilibrium.
In the example for Fig. 4.4, one can see that the intermediate accumulates
kinetically. This is because the intermediate first becomes rapidly populated by
the fast transition from U to I, but then becomes partially trapped by the high
barrier of transition from I to F. In contrast, under conditions that favor unfolding,
the early refolding intermediate is not accumulated, i.e., it is never populated
above the level of population in equilibrium (Fig. 4.5).
After infinite time the system approaches an equilibrium,
[U]() / [I]() = KUI = k1 / k1
[I]() / [F]() = KIF = k2 / k2 ,
(4.20)
where KUI , KIF are the equilibrium constants for unfolding. Using the conservation
relation, [I] = [UIF] [U] [F] , we obtain the equilibrium concentrations:
[U]() = [UIF] k1k2 / (k1k2 + k1k2 + k1k2)
[I]() = [UIF] k1k2 / (k1k2 + k1k2 + k1k2)
[F]() = [UIF] k1k2 / (k1k2 + k1k2 + k1k2) .
Rate constants for the special case of a much faster transition U
are given in Sect. 8.3.3.2.
(4.21)
I than I
4.3.1.2
Reversible two-pathway three-state transition
(4.22)
The rate equations for the two-pathway three-state reaction (Eq. 4.22) are:
d[U]
= k1[I] k1 [U] + k3 [F ] k3[U]
dt
d[F ]
= k2 [I] k2 [F] + k3 [U] k3 [F ]
dt
[I] = [UIF] [U] [F] .
(4.23)
The free energy difference between U and F must be independent of the pathway:
k3/k3 = k1k2/(k1k2)
(4.24)
35
The method of solving Eq. 4.23 is analogous to the method for the reversible
sequential three-state transition (Sect. 4.3.1.1): Consider first for simplicity only
the changes of [U] , [I] , and [F] ,
d [U]
= k1 [I] k1[U] + k3 [F] k3 [U]
dt
d [F]
= k2 [I] k2 [F] + k3 [U] k3 [F]
dt
[I] = [U] [F] ,
(4.25)
and assume that a particular solution has the form:

[U](t) = C1 exp(t)
[F](t) = C2 exp(t) .
(4.26)
[U] = k1([U] + [F]) k1[U] + k3[F] k3[U]

[F] = k2([U] + [F]) k2[F] + k3[U] k3[F] .
(4.27)
This assumption leads to:
By simplifying Eq. 4.27 and substituting [U] or [F] , we obtain an equation for ,
0 = 2 (k1 + k1 + k2 + k2 + k3 + k3) + k1k2 + k1k2
+ k1k3 + k1k2 + k1k3 + k1k3 + k2k3 + k2k3 + k2k3 ,
(4.28)
which has two real solutions (1 corresponds to the positive sign before the root;
the corresponds to 2):
1,2 = 0.5 (k1 + k1 + k2 + k2 + k3 + k3

((k1 + k1 + k2 + k2 + k3 + k3)2 4 (k1k2 + k1k2 + k1k3 + k1k2
+ k1k3 + k1k3 + k2k3 + k2k3 + k2k3))1/2)
(4.29)
[U](t) = C1 exp(1t) + C3 exp(2t) + C5

[F](t) = C2 exp(1t) + C4 exp(2t) + C6
(4.30)
C1 =
([Fo] 2[Uo] C6 + 2C5) / (1 2)
[Uo] [UIF]k1 / (k1 + k1 + k3)
for k1 k3
for k1 k3
C2 =
C1(1 k1 k1 k3) / (k1 k3)
C1(k3 k2) / ((k2 + k2)(k1 k2))
C3 =
([Fo] 1[Uo] C6 + 1C5) / (2 1)
for k1 = k3
for k1 = k3 , k1 k2
for k1 = k3 , k1 = k2
for k1 k3
for k1 = k3
36
C4 =
C3(2 k1 k1 k3) / (k1 k3)
[Fo] C2 C6
for k1 k3
for k1 = k3
C5 = [UIF](k1k2 + k1k3 + k2k3) / (k1k2 + k1k2 +

k1k3 + k1k2 + k1k3 + k1k3 + k2k3 + k2k3 + k2k3)
C6 = [UIF](k1k2 + k1k3 + k2k3) / (k1k2 + k1k2 +
k1k3 + k1k2 + k1k3 + k1k3 + k2k3 + k2k3 + k2k3)
1 = (1 k1 k1 k3) / (k1 k3)

2 = (2 k1 k1 k3) / (k1 k3)
for k1 k3
for k1 k3 .
Similar to the sequential three-state transition (U

I
F) , [U](t) , [I](t) , and
[F](t) usually display a double-exponential behavior (Eq. 4.30) with two observed
rate constants given by Eq. 4.29 (Figs. 4.6, 4.7, Table 4.1). An important
implication is that the information of kinetic rate constants alone is generally not
sufficient to distinguish between the two different mechanisms, two-pathway or
sequential reaction. There are a few special cases in which [F](t) or [U](t) or both
follow single-exponential functions. At the limits k3 0 and k3 0, Eqs. 4.29
and 4.30 transform into the solution for the reversible sequential three-state
transition (Sect. 4.3.1.1).
Fig. 4.6. Change of the populations of folded, F, intermediate, I, and unfolded state, U, in a
reversible three-state transition with two parallel pathways (U
I
F and U
F) under
conditions that favor folding. The parameters chosen for this example are: k1 = 4, k1 = 2, k2 = 2,
k2 = 1, k3 = 2, k3 = 0.5, [Fo]/[UIF] = 0, [Uo]/[UIF] = 1. Observed rate constants ( kobs) are
1 = 8 and 2 = 3.5. Inset: Energy landscape.
37
Fig. 4.7. Change of the populations of folded, F, intermediate, I, and unfolded state, U, in a
reversible three-state transition with two parallel pathways (U
I
F and U
F) under
conditions that favor unfolding. The parameters chosen for this example are: k1 = 2, k1 = 4,
k2 = 1, k2 = 2, k3 = 0.5, k3 = 2, [Fo]/[UIF] = 1, [Uo]/[UIF] = 0. Inset: Energy landscape.
4.3.1.3
Reversible off-pathway intermediate
The third important type of reversible three-state transitions is illustrated in Eqs.
4.31 and 4.32. Strictly speaking, I is a side-product, but because it is often
spectroscopically in-between U and F, it is commonly referred to as an offpathway intermediate:
,
or
(4.31)
(4.32)
The solution for Eq. 4.31 is found by transforming it into Eq. 4.10 (Sect. 4.3.1.1)
by exchanging I and U:
[I](t) = C1 exp(1t) + C3 exp(2t) + C5
(4.33)
[F](t) = C1 1 exp(1t) + C3 2 exp(2t) + C6

[U](t) = [UIF] [I](t) [F](t)
1,2 = 0.5 (k1 + k1 + k2 + k2

((k1 + k1 + k2 + k2)2 4 (k1k2 + k1k2 +k1k2))1/2)
1 = (1 k1 k1) / k1
(4.34)
38
2 = (2 k1 k1) / k1
C1 = ([Fo] 2[Io] C6 + 2C5) / (1 2)
C3 = ([Fo] 1[Io] C6 + 1C5) / (2 1)
C5 = [UIF]k1k2/(k1k2 + k1k2 + k1k2)
C6 = [UIF]k1k2/(k1k2 + k1k2 + k1k2) ,
where [Io] = [I](t = 0) , [Fo] = [F](t = 0) , and [UIF] = [U] + [I] + [F] . Analogously,
the solution for Eq. 4.32 is obtained by cyclically exchanging U , I , and F in
Eq. 4.31 . 1 and 2 are the same as for the mechanism U
I
F (Eq. 4.16).
An important implication is that the information of kinetic rate constants alone
is generally not sufficient to distinguish between on-pathway (Sect. 4.3.1.1) or
off-pathway intermediates. A method for distinguishing these two cases is the value analysis (Sect. 8.3). Originally this method was designed for the structural
resolution of transition states and intermediates (Goldenberg et al., 1989;
Matouschek et al., 1989, 1990; Fersht et al., 1991, 1992; Matouschek and Fersht,
1991; Fersht, 1992, 1993, 1995a, b; Clarke and Fersht, 1993; Otzen et al., 1994;
Itzhaki et al., 1995b, Nlting et al., 1995, 1997a; Nlting, 1998a, 1999), but it also
provides information about the kinetic mechanism: In the -value analysis
mutants are used as reporters of structural consolidation along the folding
pathway. In the case of an off-pathway intermediate, the calculated total for the
whole reaction from U to F is not 1 for all mutants when erroneously assuming an
on-pathway mechanism (Sect. 8.3.4.2).
4.3.2
Irreversible three-state transitions
4.3.2.1
Irreversible consecutive three-state transition
(4.35)
The solution (Fig. 4.8, Table 4.1) is derived by using the methods presented in the
previous sections:
[U](t) = [Uo] exp(k1t)
(4.36)
[F](t) = C2 exp(k1t) + C4 exp(k2t) + [UIF]
k2[Uo]/(k1 k2)
k1[Uo]t
for k1 k2
C2 =
for k1 k2
C4 =
[Fo] k2[Uo]/(k1 k2) [UIF]
[Fo] [UIF]
for k1 = k2
for k1 = k2
39
[Uo] = [U](0)
[Fo] = [F](0)
[UIF] = [U] + [I] + [F] .
The definitions are as in Sect. 4.3.1. One should mention, that in case k1 = k2 , C2
is a function of time, and thus, [F](t) and [I](t) are not pure superpositions of two
exponential functions anymore.
Fig. 4.8. Change of the populations of folded, F, intermediate, I, and unfolded, U, state in an
irreversible three-state transition (U
I
F). The parameters chosen for this example are:
k1 = 4, k2 = 2, [Fo]/[UIF] = 0, [Uo]/[UIF] = 1.
4.3.2.2
Irreversible parallel decay
(4.37)
This type of reaction may occur under conditions that strongly favor folding when
two conformations (e.g., a correctly folded and a misfolded, see Sect. 9.3) are
produced. Here the populations of the species are
[U](t) = [Uo] exp( (k1 + k2 ) t)
(4.38)
[F1](t) = [Uo] exp( (k1 + k2 ) t) k1 / (k1 + k2 ) + C1

[F2](t) = [Uo] exp( (k1 + k2 ) t) k2 / (k1 + k2 ) + C2 ,
where the definitions are analogous to those in Sect. 4.3.1, and the constants C1
and C2 depend on the initial conditions.
40
4.4
Reversible sequential four-state transition
(4.39)
The solution of this case involves the treatment of a cubic equation (Eq. 4.40)
(Beyer, 1991). With the exception of a few special cases, a superposition of three
exponential functions is observed:
0 = 3 + p 2 + q + r
(4.40)
p = (k1 + k1 + k2 + k2 + k3 + k3)
q = k1k2 + k1k2 + k1k3 + k1k3 + k1k2 + k1k3 + k1k3 + k2k3 + k2k3 + k2k3
r = (k1k2k3 + k1k2k3 + k1k2k3 + k1k2k3)
1 = A + B p/3
2,3 = 0.5(A + B) 0.5(A B) 3 p/3
A = 3 0.5b + c
c=
B = 3 0.5b c
a = (3q p2)/3
b 2 a3
+
4 27
b = (2p3 9pq + 27r)/27
[U](t) = C1 exp(1t) + C3 exp(2t) + C5 exp(3t) + C7

[F](t) = C2 exp(1t) + C4 exp(2t) + C6 exp(3t) + C8 .
For example, for k1 = 4, k1 = 2, k2 = 2, k2 = 1, k3 = 2, k3 = 0.5, one obtains
c = 5.449i, A = 1.630 0.733i, B = 1.630 + 0.733i, where the imaginary number
i is defined as i 1 , and 1 = 7.093, 2 = 3.473, 3 = 0.934. Compared with the
example for three-state transitions with two parallel pathways (Fig. 4.6), the
observed rate constants, 1 and 2 , are slower, and there is an additional, even
slower, phase, 3 .
Concerning the calculation of the roots in Eq. 4.40, it should be mentioned that
a complex number z = zr + izi can be transformed to z = r(cos + i sin ), where
r = (zr2 + zi2)0.5 and = arctan (zi/zr). From de Moivres identity, (ei)n = ein, and
Eulers formula, ei = cos + i sin , follows (cos + i sin )n = cos(n) +
i sin(n). Thus,
z1/n = (r(cos + i sin ))1/n
(4.41)
= r1/n(cos(( +2k)/n) + i sin(( +2k)/n))
n = 2, 3, ... ;
z = zr + izi
k = 0, 1, ... , n 1
r = (zr2 + zi2)0.5
= arctan (zi/zr) .
For example, the three third roots of 1 are 1 and 1/2 3 /2.
4.5 Reactions with monomer dimer transitions
41
4.5
Reactions with monomer dimer transitions
4.5.1
Monomer dimer transition
(4.42)
In contrast to the cases treated in the previous sections, here we are encountering a
bimolecular reaction. The rate equation for [A] is a first-order non-linear
differential equation:
d[A]
= 2k1 [A]2 k1 [A] + k 1[A tot ]
dt
[A](0) = [Ao]
(4.43)
[A2] = 0.5 ([Atot] [A]) ,

where [Atot] is the total protein concentration in equivalents of monomers, and the
other definitions are analogous to those in Sect. 4.2. We start with the guess that
particular solutions are of the types:
[A] = B + C tanh[D(t + E)]
(4.44)
[A] = B + C coth[D(t + E)]
tanh(u )
exp( u ) exp( u)
exp(u) + exp(u)
coth(u)
exp(u) + exp(u)
.
exp(u) exp(u)
By inserting Eq. 4.44 into Eq. 4.43 and using the relations:
d[tanh(u)]/dt = (1 tanh2(u)) du/dt
d[coth(u)]/dt = (1
coth2(u))
(4.45)
du/dt ,
the guess is verified and the constants are calculated (Figs. 4.9, 4.10):
B = 0.25 [Atot] K
(4.46)
C = 0.25 [Atot] (K2 + 8K)0.5

D = 2 k1 C
K = k1/([Atot]k1)
E=
arctanh[([Ao] B)/C]/D
arccoth[([Ao] B)/C]/D
for ([Ao] B)/C < 1

for ([Ao] B)/C > 1
42
[A] =
B + C tanh[D (t + E)]
B + C coth[D (t + E)]
for ([Ao] B)/C < 1

for ([Ao] B)/C > 1 .
Fig. 4.9. Change of the populations of monomers and dimers in a monomerdimer transition
(2A
A2). The parameters chosen for this example are: k1 = 2/[Atot], k1 = 3, [Ao]/[Atot] = 0.
Fig. 4.10. Change of the populations of monomers and dimers in a monomerdimer transition
(2A
A2), compared with single-exponential fits which are shifted by 0.03 units for better
visibility. Only little differences are observed between the kinetic traces and the wrong curve
fits. Thus, the information about the shape of the kinetic traces is often not sufficient to
distinguish between bimolecular and unimolecular reactions. The parameters chosen for this
example are: k1 = 2/[Atot], k1 = 3, [Ao]/[Atot] = 1.
Alternatively, the solution of Eq. 4.43 may be represented as a linear

combination of both particular solutions (Eq. 4.44). The equilibrium
concentrations are:
4.5 Reactions with monomer dimer transitions
[A]() = 0.25 [Atot] ((K2 + 8K)0.5 K)

[A2]() = 0.5 ([Atot] [A]()) .
43
(4.47)
In contrast to the first-order reactions in Sects. 4.2 4.4, the rate constants for
reactions that involve a monomerdimer transition are always dependent on the
concentration.
4.5.2
Reversible two-state folding transition linked with
a monomer dimer transition
Another common case of a kinetic reaction is a reversible two-state folding
transition of monomeric molecules followed by the dimerization in the folded
state (Fig. 4.11):
(4.48)
The rate equations are:
d[U]
= k1[F] k1[U]
dt
d[F 2 ]
= k2 [F ]2 k2 [F 2 ]
dt
[F] = [UF] [U] 2[F2] ,
(4.49)
where [UF] is the total concentration in equivalents of monomers. The rate

equation for [U] is quite complicated:
2
k1
d [U]
d[U] 2
d[U]
d[U]
+ k1 (k1 + k1 + k2 )
) + 4k1k2 [U]
2 + 2k 2 (
dt
dt
dt
dt
(4.50)
+2k12 k 2 [U]2 + k1k2 (k1 + k1 )[U] k12 k2 [UF] = 0

Exponential functions are solutions only for a few special cases. A very crude
approximation for [U] which is valid for several non-trivial cases is given by:
n
Ai e i t
,
[U] = C + i=1
n
it
Bi e
(4.51)
i =1
where n 4 , Ai , Bi , and C are constants. By inserting Eq. 4.51 into Eq. 4.50 and
using the boundary condition for [U](0), one obtains several algebraic equations
for Ai , Bi , and C . These can be solved for the time-independent terms and the
terms which contain only low-order exponential functions. Then the error due to
the non-equality of the terms which contain high-order exponential functions is
occasionally found to be relatively small.
44
Fig. 4.11. Change of the population of the unfolded state in a two-state folding transition that is
in sequence with a monomerdimer transition (2U
2F
F2). The parameters chosen for
this example are: k1 = 3, k1 = 1, k2 = 2/[UF], k2 = 3. Two cases are shown: [Uo]/[UF] = 1, and
[Uo]/[UF] = 0.
4.6
Kinetic rate constants for perturbation methods
The mathematical treatment of bimolecular reaction kinetics is tremendously
simplified for small-amplitude perturbation methods, such as small-amplitude
temperature-jumping, pressure perturbation, ultrasonic velocimetry, and dielectric
relaxation (Table 4.2). In these methods, the changes of the populations of the
states are small because the equilibrium of the species involved is only slightly
perturbed by a small change in the physical or chemical conditions. Consider, for
example, Eq. 4.52:
(4.52)
In equilibrium, per definition, the macroscopic changes in concentration are zero:
d[U]
= k1[F] k1[U] = 0
dt
(4.53)
d[F 2 ]
= k2 [F ]2 k2 [F 2 ] = 0
dt
[F] = [UF] [U] 2[F2] ,
where [UF] is the total concentration in equivalents of monomers. Thus, the
equilibrium concentrations are:
4.6 Kinetic rate constants for perturbation methods
k1
2 2
2
( (k1 + k1 ) k2 + 8k1 k2 k2 [UF] (k1 + k1 )k2 )
2
4k1 k2
1
2
[F]eq =
( (k1 + k1 ) 2 k 2
+ 8k12 k2 k2 [UF] (k1 + k1 )k2 )
4k1 k2
[U]eq =
45
(4.54)
[F2]eq = 0.5([UF] [U]eq [F]eq) .

Upon perturbation, the equilibrium moves to a new position:
[U] = [U]eq + [u]
[F] = [F]eq + [f]
(4.55)
[F2] = [F2]eq + [f2] ,
where are [u], [f], and [f2] are the small changes of [U]eq , [F]eq , and [F2]eq ,
respectively. Thus, the relaxation of [F2] is described by
d[F 2 ]eq
dt
and so,
d[f 2 ]
= k2 ([F]eq + [f ]) 2 k2 ([F 2 ]eq + [f 2 ]) ,
dt
d[f 2 ]
= k2 (2[f ][F]eq + [f ]2 ) k2 [f 2 ] .
dt
(4.56)
(4.57)
Neglecting the second-order term, [f]2 , leads to
d[f 2 ]
= 2k2 [f ] [F]eq k2 [f 2 ] .
dt
(4.58)
Fig. 4.12. Change in the populations of the monomeric unfolded state, monomeric folded state,
and dimeric folded state upon a small perturbation of their equilibrium (2U
2F
F2).
[u], [f], and [f2] are the small changes in the equilibrium concentrations, [U]eq , [F]eq , and [F2]eq ,
for the states U, F, and F2 , respectively. Parameters chosen for this example are: k1 = 3, k1 = 1,
k2 = 2/[UF], k2 = 3, 2[uo] = [fo]. Observed rate constants ( kobs) are 1 = 8 and 2 = 3.
46
Fig. 4.13. Change in the population of the unfolded state upon a small perturbation of the
equilibrium between monomeric unfolded, monomeric folded, and dimeric folded state
(2U
2F
F2). The parameters chosen for this example are: k1 = 3, k1 = 1, k2 = 2/[UF],
k2 = 3, and [fo] = [uo], [fo] = 10 [uo], [fo] = 20 [uo], respectively, as indicated. [uo] and [fo]
depend on the magnitude and type of the perturbation and on the physical and chemical
properties of the involves species. Observed rate constants ( kobs) are 1 = 8 and 2 = 3.
Analogously,
d[u]
= k1 [f] k1[u]
dt
(4.59)
[f] + [u] + 2[f2] = 0 .

By combination of Eqs. 4.58 and 4.59 we obtain:
d 2 [u] d[u]
(k1 + k1 + k2 + 4k2 [F]eq )
2 +
dt
dt
+[u](k1k2 + k1k2 + 4k1k2 [F]eq ) = 0
[u](t) = [uo]C exp(1t) + [uo](1 C) exp(2t)
C = (k1[fo]/[uo] + 2 k1) / (2 1)
1,2 = 0.5p (0.25p2 q)0.5

p = k1 + k1 + k2 + 4k2[F]eq
q = k1k2 + k1k2 + 4k1k2[F]eq
d[u ]
+ k1 [u]) / k1
dt
= ([uo]/k1){C(k1 1)exp(1t) + (1C)(k12)exp(2t)}
[f ] = (
(4.60)
(4.61)
4.7 Summary
47
[f2] = 0.5([f] + [u]) ,

where [uo] and [fo] change with magnitude and type of the perturbation, and
depend on the physical and chemical properties of the species involved.
Depending on these parameters, convex or concave curve shapes, with or without
transient accumulation, are observed (Figs. 4.12, 4.13).
Analogously, the simple case,
,
(4.62)
is easily derived:
d[f 2 ]
= 4k1 [f 2 ][F]eq k1[f 2 ]
dt
(4.63)
[F]eq = 0.25( K 2 + 8K[F tot ] K )

[F tot ] = [F] + 2[F 2 ]
K = k1 / k1 .
Here the observed rate constant, , is
= k1 + 4k1[F]eq .
(4.64)
The rate constants of unimolecular reactions are the same as those derived in
Sects. 4.2 4.4 (see Table 4.1).
4.7
Summary
In reactions which involve solely unimolecular transitions, so-called first-order
reactions, the fractions of the species involved usually change with time according
to superpositions of exponential functions with concentration-independent rate
constants (Table 4.1).
As soon as bimolecular reactions are at least partly involved, the speed of
reaction becomes concentration-dependent (Table 4.2). However, under pseudofirst-order experimental conditions this concentration-dependence may be
undetectably small, for example, under conditions where the process is dominated
by a dissociation event.
The kinetic traces of bimolecular reactions often have non-exponential shapes.
However, in small perturbation methods, generally, exponential shapes of the
kinetic traces are observed also for bimolecular reactions.
48
Table 4.1. Rate constants for first-order reaction mechanisms. For special cases and for
magnitudes of changes of reactants and products see Sects. 4.2 4.4.
Reaction
Observed rate constants, kobsa

(relaxation constants)
= k1
= k1 + k1
1,2 = k1 , k2
1,2 = 0.5 (k1 + k1 + k2 + k2
((k1 + k1 + k2 + k2)2
4 (k1k2 + k1k2 + k1k2))1/2)
1,2 = 0.5 (k1 + k1 + k2 + k2 + k3 + k3
((k1 + k1 + k2 + k2 + k3 + k3)2
4 (k1k2 + k1k2 + k1k3 +
k1k2 + k1k3 + k1k3 + k2k3
+ k2k3 + k2k3))1/2)
1 = A + B p/3
2,3 = 0.5 (A + B)
0.5 (A B) 3 p/3
A = 3 0.5b + c B = 3 0.5b c
b 2 a3
+
4 27
a = (3q p2) / 3
c=
b = (2p3 9pq + 27r) / 27

p = (k1 + k1 + k2 + k2 + k3 + k3)
q = k1k2 + k1k2 + k1k3 + k1k3 + k1k2
+ k1k3 + k1k3 + k2k3 + k2k3 + k2k3
r = (k1k2k3 + k1k2k3
+ k1k2k3 + k1k2k3)
a
In special cases, some of the given rate constants do not apply.
4.7 Summary
49
Table 4.2. Rate constants for second-order reaction mechanisms in small-amplitude perturbation
methods (see Sect. 4.6).
Reaction
Observed rate constants, kobsa

(relaxation constants)
= k1 + 4k1[F]eq
= k1 + k1([A]eq + [B]eq)
1,2 = 0.5p (0.25p2 q)0.5
p = k1 + k1 + k2 + 4k2[F]eq
q = k1k2 + k1k2 + 4k1k2[F]eq
a The rate constants are calculated for experimental conditions under which a chemical
equilibrium is approached prior to application of a small perturbation. [F]eq , [A]eq , and [B]eq are
the equilibrium concentrations of F , A , and B , respectively. In special cases, some of the given
rate constants do not apply.
5 High kinetic resolution

of protein folding events
5.1
Ultrafast mixing
One of the oldest ways of inducing rapid protein folding is to mix a solution of
unfolded protein with buffers that favor folding (Fig. 5.1). The reaction is
followed by an optical probe, for example, ultravioletvisible or infrared
absorption, circular dichroism (CD), scattering, or fluorescence. In particular, CD
detection in combination with rapid mixing is an exquisitely sensitive probe of
conformational changes (Luchins and Beychok, 1978; Pflumm et al., 1986;
Kuwajima et al., 1987, 1993, 1996; Elve et al., 1992; Kalnin and Kuwajima,
1995; Arai and Kuwajima, 1996). Kinetic resolution of molecular dimensions
became possible by advances in X-ray scattering (Semisotnov et al., 1996) and
dynamic light scattering (Gast et al., 1997). H/D exchange kinetics is frequently
followed by nuclear magnetic resonance (NMR) spectrometry to obtain
information about local and global folding events (see Sects. 7.1 and 8.1). Mass
spectrometry detection of H/D exchange requires significantly smaller quantities
of protein (Miranker et al., 1996a, b). Real-time NMR spectroscopy with kinetic
resolution has significantly advanced into the millisecond time range (Frieden et
al., 1993; Balbach et al., 1995; Hoeltzli and Frieden, 1995).
Fig. 5.1. Stopped-flow method. Two syringes containing a solution of unfolded protein and a
buffer that favors refolding, respectively, are simultaneously pushed. Turbulence in the T-mixer,
where both liquids join together, causes a rapid mixing. After a certain amount of liquid has
passed through the mixer, the position of the stop syringe triggers a stop signal and a signal for
the detector to start recording the reaction kinetics in the sample cell.
52
5 High kinetic resolution of protein folding events
Fig. 5.2. Ultrafast continuous-flow mixing. A solution of unfolded protein and a buffer that
favors folding are gently joined together and passed through a tube with a decreasing crosssection. At the end of the tube, the laminar flow is changed into a highly turbulent flow by
passing the liquid over a sphere of only a few 10 m diameter. The mixed solution forms a continuous free jet. Each position in the jet corresponds to a certain time point in the reaction
kinetics. Kinetic traces are recorded by moving the LASER/detector system along the jet. At a
typical flow speed of 10 100 m s1, a change of the position by 100 m corresponds to a change
in time of 110 s. Because in free air the jet is stable over a few cm, reactions may be followed
from microseconds to milliseconds (Regenfuss et al., 1985; Regenfuss and Clegg, 1987).
Unfortunately, the dead time of common stopped-flow devices is usually

around 1 millisecond. Faster mixing requires stronger turbulence in the mixing
chamber. Decreasing the size of the tubes and increasing the speed of flow would
require impracticably high pressures in the common design.
In Thomas Jovin's group (Regenfuss et al., 1985; Regenfuss and Clegg, 1987)
it has been realized that mixing in microseconds can be performed by using a
number of innovations (Fig. 5.2 and Table 5.1 in Sect. 5.8): 1. In order to reduce
the flow resistance, relatively wide tubes are used for the transport of the two
liquids to the mixing chamber. Before reaching the mixing chamber, the two
streams of liquid are gently joined together so that no significant mixing occurs.
2. The mixing chamber consists of a tip in which a sphere is placed. Intense
mixing is triggered by the liquid flowing over the sphere. 3. The size of the
mixing sphere may be as small as a few micrometers. At a typical flow speed of
10 100 m s1, a dead length of 100 m corresponds to a dead time of 110 s.
4. Using continuous-flow rather than stopped-flow avoids pressure waves at high
flow speeds.
These modifications led to a 100-fold reduction of the mixing time down to
10 s. According to theoretical considerations, further improvement to 1 s
should be possible by decreasing the size of the mixer (Regenfuss et al., 1985).
Rousseau et al. (Takahashi et al., 1997; Yeh et al., 1997; Yeh and Rousseau,
1998) modified the classic design (Fig. 5.1) by choosing a narrow cross-section
for the mixer, 100 m 25 m, compared with 250 m 250 m for the
5.1 Ultrafast mixing
53
Fig. 5.3. Ultrafast mixing using a continuous-flow cell. In order to reduce the flow resistance,
relatively wide tubes of several 100 m inner diameter connect the syringes with the T-mixer. At
the position of the mixer, the cross-section of the tubes is decreased down to 100 m 25 m.
Different positions of the LASER/detector system correspond to different time points of the
reaction kinetics triggered by the mixing of the two liquids. Using a continuous-flow cell instead
of a free jet in air (see Fig. 5.2) improves the optical stability but slightly increases the necessary
pressure (Takahashi et al., 1997; Yeh et al., 1997; Yeh and Rousseau, 1998). See also Lin et al.
(2003); Cherepanov and De Vries (2004).
Fig. 5.4. Ultrafast continuous-flow mixing head (Shastry et al., 1988; Sauder et al., 1996; Park et
al., 1997; Shastry and Roder, 1998). Mixing over a sphere of 200 m diameter enhances
turbulence and enables mixing in the 15-s time scale and a dead time of about 50 s. A 50-m
platinum wire serves for the adjustment of the platinum mixing sphere. Complete kinetic traces
within a certain time range are recorded by a multichannel detector that contains a chargecoupled device (CCD). See also Teilum et al. (2002); Zhu et al. (2003).
54
continuous-flow observation cell. Using a flow cell (Fig. 5.3) instead of a free jet
in air improves the optical stability, and the earliest observable time point in this
simple design was still 100 50 s.
The continuous-flow capillary mixer with a 50-s dead time (Fig. 5.4) utilized
in Roder's group (Shastry et al., 1988; Sauder et al., 1996; Park et al., 1997;
Shastry and Roder, 1998) combines the advantages of rapid mixing via a sphere
and the good stability of a continuous-flow mixing cell. A multichannel chargecoupled device (CCD) detector enables the real-time recording of complete kinetic
fluorescence traces without moving the detector. Different channels of the CCD
correspond to different time points. Simultaneous detection of the complete
kinetic traces within a certain time range of typically microseconds to milliseconds
significantly speeds up the measurement and improves optical stability.
Ultrafast continuous mixing has been combined with resonance Raman
spectroscopy (Takahashi et al., 1997). Resonance Raman spectra of the heme
group of cytochrome c suggest that in the folding reaction the protein is trapped in
a misfolded conformation with two histidines ligated to the heme iron (Takahashi
et al., 1997). The fraction of misfolded molecules has been significantly reduced
by decreasing the pH from 5.9 to 4.5 which causes the protonation of the
misligated histidines (Yeh et al., 1997). At pH 4.8 a rate constant for the main
folding pathway of 1400 s1 at 40oC has been observed. Further progress in the
resolution of the folding kinetics of cytochrome c, using ultrafast mixing, has
been made in William Eaton's and James Hofrichter's groups. Cytochrome c
refolding has been studied in the previously inaccessible time range from 80 s to
3 ms (Chan et al., 1997). Adding imidazole to the protein solution prevents
misligation of histidines and dramatically speeds up folding (Chan et al., 1997).
Fig. 5.5 shows an ultrafast continuous-flow double-jump mixing device, made
by the author, which is designed to extend the time scale of H/D exchange
experiments monitored by NMR (see Sects. 7.1 and 8.1). The mixing tubes and
the delay tube are made from grooves carved into a plate of acrylic glass which is
sealed with a plate made from stainless steel.
Fig. 5.5. Ultrafast continuous-flow double-jump mixing head with delay times of 200 500 s.
Two T-mixers are made from grooves carved into a plate of acrylic glass sealed on top with a
plate made from stainless steel (not shown). The three inlet tubes and the outlet tube have
diameters of about 0.5 mm. The device is attached to a common quenched-flow apparatus.
5.2 Temperature-jump
55
5.2
Temperature-jump
5.2.1
Electrical-discharge-induced T-jump
5.2.1.1
T-jump apparatus
Fig. 5.6. Electrical-discharge-induced temperature-jump (T-jump) method. A capacitor is

charged by a power supply up to a specific voltage and then rapidly discharged through the
sample cell that contains the protein in a buffer with a certain electrical conductivity, for example, 50 mM phosphate buffer with 100 mM KCl. The electrical discharge causes Joule heating
by 120oC with rise times of typically 500 ns 10 s, depending on the instrument settings, in
particular on resistance of the protein solution and capacitance. When starting from the
(partially) cold-unfolded state, increase of temperature causes refolding, otherwise fast unfolding
reactions may be studied. The reaction kinetics is followed by absorption or fluorescence
detection. The electrical signals are digitized by analog-to-digital converters (ADC) which are
part of an Nicolet (Madison, WI) model Pro 90 storage oscilloscope and further processed on a
computer. Large sample cells with 1 mL volume may be used which enables a high light
throughput and thus an excellent signal-to-noise-ratio (see Fig. 5.8). Minimization of photolysis
in the sample by the intense light of a 200-W mercuryxenon lamp is achieved with the help of
an optical shutter that opens only during the measurement. Prior to digitalization of the
fluorescence signal by an ADC, a constant voltage is subtracted (see Fig. 5.9). Therefore, a
12-bit ADC is usually sufficient to resolve changes of only 0.01% in the nanosecond time scale.
When using a well-stabilized power supply for the lamp, usually no reference channel is needed.
Otherwise the signal of a reference detector that is located between monochromator and sample
cell (not shown) may be used. The simplicity of the device that does not contain mechanically
moving parts makes the handling very easy and causes an exquisite reproducibility.
56
Fig. 5.7. Optics of an electrical-discharge-induced T-jump apparatus (Fig. 5.6). Photomultiplier

tube 1 (PMT1) is used for absorption measurements. For fluorescence detection, the electrical
signals of photomultiplier tubes 2 and 3 (PMT2 and PMT3) are added. To decrease the photon
shot noise, the optics is optimized for a large light throughput (e.g., relatively wide bandwidths
of fluorescence excitation and emission are used for the experiments presented in Chap. 10) and
a high aperture of fluorescence detection (DIA-LOG, Dsseldorf, Germany; Eigen and
deMaeyer, 1963; French and Hammes, 1969; Nlting et al., 1995, 1997a).
Fig. 5.8. Sample cell for an

electrical discharge T-jump
apparatus. In order to avoid
pressure due to thermal
expansion upon T-jump, the
top of the cell is not sealed.
Fluorescence detection is
perpendicular to the excitation
beam
(DIA-LOG,
Dsseldorf, Germany; Eigen
and deMaeyer, 1963; French
and Hammes, 1969; Nlting
et al., 1995, 1997a).
Protein solutions that contain electrolytes may have sufficient electrical conductivity to enable them to be heated by a rapid electrical discharge through the sample
cell (Figs. 5.6 5.10). With the simple design (Eigen and deMaeyer, 1963; French
and Hammes, 1969) Joule heating with rise times of 1 s or faster can easily be
achieved when using a buffer with 100 mM KCl. The large size of the sample cell
of about 1 mL volume enables a large light flow and thereby leads to a low photon
shot noise. Noise levels of < 0.01% root mean square (rms) of the fluorescence
signal have been achieved at a 5-s response time of the electronics (Nlting et al.,
1995). The amplitude of temperature-jump (T-jump), T, is given by
CU 2
,
Vcp
57
(5.1)
where , V , cp , C , and U represent density, volume and specific heat capacity of

the sample, electrical capacitance, and applied voltage, respectively. The exponential rise time, , of temperature is approximately
= RC ,
(5.2)
where the resistance, R , and capacitance, C , typically are around 50 200 and
10 50 nF, respectively (Fig. 5.10).
Fig. 5.9. Improvement of the detection of small changes of fluorescence in T-jump experiments:
Prior to the T-jump, the signal is adjusted to zero by subtracting a constant voltage. Thereby
faster analog-to-digital converters (ADC) with lower digital resolution may be used.
Fig. 5.10. Electrical-discharge-induced T-jumps with different instrument settings. The

measured electrical signal was derived from the instant change of protein fluorescence upon
temperature change. When adjusting the resistance of the sample to 50 and using a 50-nF
capacitor, the rise time, , is 2.5 s. With a 20-nF capacitor, the rise time can be shortened to
1 s. However, at the same voltage, the amplitude of the T-jump is 2.5 times smaller. The
electrical signal has a slightly different shape due to the response time of the electronics, in this
example 500 ns.
58
In general, a T-jump on a protein perturbs the equilibrium constant of the

folding reaction, K (Eigen and deMaeyer, 1963; French and Hammes, 1969):
ln K
H
,
=
T P RT 2
(5.3)
where H is the enthalpy change of the reaction, R = 8.314 J mol1 K1 is the

molar gas constant and T the absolute temperature. For small-amplitude jumps,
T, follows:
H
K
= T
.
(5.4)
K P
RT 2
5.2.1.2
Observation of early folding events: refolding from the cold-unfolded
state
Even though temperature-jumping (T-jumping) has been used for a long time for
the observation of spin-relaxations of heme-proteins (Sligar, 1976; Fisher and
Sligar, 1987), of fast protein unfolding (Tsong et al., 1971; Lin and Cheung,
1992), and of fast conformational relaxations in proteins (Eigen et al., 1960;
Cathou and Hammes, 1964, 1965; Wang et al., 1975; Jentoft et al., 1977; Feltch
and Stuehr, 1979; Tsong, 1982; Steinhoff et al., 1989; Walz, 1992; Narasimhulu,
1993), the application to protein refolding was complicated by the fact that
temperature increase usually favors loss of structure, i.e., unfolding (Phillips et al.,
1995).
Fig. 5.11. Observation of early (re)folding

intermediates. Energy landscape of a protein that contains a high (#2) and a low
(#1) transition state, and an intermediate
(I) on the pathway between unfolded (U)
and folded (F) state. A: In unfolding experiments the early (re)folding intermediate
cannot easily be observed since its
population is always low. This is because
the intermediate state slowly becomes
populated and rapidly depopulated. B: Refolding experiments allow us to observe
intermediates that are located on the
reaction coordinate between the main
transition state, #2, and the unfolded state:
The amplitudes of both kinetic events are
high.
59
However, early folding events can hardly be resolved in unfolding experiments

(i.e., under conditions that favor unfolding). This is illustrated in Fig. 5.11, which
displays the energy landscape of a protein that has an early folding intermediate, I,
which is located on the reaction coordinate in-between the unfolded, U, and the
main transition state, #2. One can see that in unfolding experiments (Fig. 5.11, A)
the molecules slowly pass through the high transition state barrier, #2, and thus
the intermediate state slowly becomes populated. Then the molecules rapidly pass
through the low-energy transition state, #1, thereby rapidly de-populating the
intermediate state. Thus, the occupancy of the intermediate state is always low,
and the fast transition from I to U may not easily be monitored since it has a very
small amplitude. In contrast, in refolding experiments (Fig. 5.11, B), after passing
through the fast transition, the molecules become trapped by the high transition
state barrier which can only slowly be overcome. Thus, both phases have a high
amplitude and may be measured (see also Sect. 4.3.1.1).
Fortunately, globular proteins usually display the property of cold-unfolding
(cold-denaturation), i.e., a decrease of stability with decreasing temperature
(Fig. 5.12; Privalov, 1990; Griko and Privalov, 1992; Damaschun et al., 1993;
Gast et al., 1993, 1995; Nishi et al., 1994). The temperature-dependence of the
Gibbs free energy change upon folding, GFU , is
GFU = HFU, g (1 T/Tg) + Cp(T Tg T ln(T/Tg)) ,
(5.5)
where T , Tg , HFU, g , and Cp are the absolute temperature, the absolute

temperature at the midpoint of the heat-unfolding transition, the change of
enthalpy at Tg , and the change of molar heat capacity, respectively (Pfeil, 1988).
Fig. 5.12. Thermodynamic stability, GUF = GFU , of P27A/C40A/C82A barstar in 2 M

urea as function of temperature, calculated from data in (Nlting et al., 1995). Qualitatively most
other globular proteins show a similar behavior, where the stability increases upon cooling at
high temperatures, but upon heating at low temperatures.
60
At first glance, this surprising thermodynamic phenomenon might be counterintuitive since one might expect a higher degree of disorder with increasing
temperature but not the opposite. The reason for the existence of cold-induced
unfolding of protein is seen in the fact that the protein is dissolved in water, and
the overall entropy of the system represents a delicate balance of the contributions
from water and protein (Makhatadze and Privalov, 1993; Weber, 1996): Upon
lowering the temperature, the structure-forming propensity of water increases; the
hydrophobic stabilization of the protein core decreases, and the preference of
polar groups in the core for hydration increases (see Sect. 3.4). Thus, it has been
realized that rapid refolding may be triggered in T-jump experiments when
starting from the cold-unfolded state (Nlting et al., 1995; Nlting, 1996).
Probably all globular proteins cold-unfold at a sufficiently low temperature, but
for many proteins in the absence of denaturants, a significant degree of coldunfolding can only be achieved at temperatures below the freezing point of water.
In these cases, the experiments are performed at different denaturant concentrations, and the kinetic rate constants are extrapolated to zero concentration of
denaturant. Thus, the T-jump method for the rapid initiation of refolding is
generally applicable to a wide range of proteins.
A protein of enormous interest has been barstar, the 10 kDa inhibitor of the
ribonuclease barnase, for which the occurrence of an early intermediate has been
predicted from burst-phase analysis (Schreiber and Fersht, 1993b; Shastry and
Udgaonkar, 1995). NMR and CD spectroscopy have shown that the colddenatured state of barstar is a highly unfolded state with only a small amount of
residual structure, especially in helix1 and helix2 (Wong et al., 1996; Nlting et al.,
1997b). T-jumping of cold-unfolded barstar has enabled the first kinetic and
structural resolution of the folding pathway of a protein at the level of individual
amino acid residues from microseconds to seconds (Chap. 10; Nlting et al., 1995,
1997a). Early on the folding pathway, barstar passes through an intermediate with
a rate constant of formation of 2300 s1 at 10oC and rate constant of decay of
800 s1. The activation enthalpy of the fast event is positive and its fluorescence
change accounts for roughly 40% of the total for the transition from the unfolded
to the folded state in the absence of denaturants. -value analysis (see Sect. 8.3)
shows that a significant degree of secondary and tertiary structure, mainly located
in strand1 , helix1 , and helix4 , is formed on a time scale of 1 ms. At this stage,
large parts of the molecule are still in a molten-globule-like state. The high
structural resolution of the folding pathway of barstar from microseconds to
seconds (Nlting et al., 1995, 1997a; Nlting, 1998a) reveals that the folding
mechanism is consistent with a nucleationcondensation model (Abkevich et al.,
1994b; Itzhaki et al., 1995a; Fersht, 1995c, 1997; Freund et al., 1996; Ptitsyn,
1998; Nlting, 1999). Similar to the growth of a crystal, a part of the molecule,
the so-called nucleus, forms early. The diffuse nucleus becomes increasingly
stabilized as further structure condenses around it in the course of the folding
reaction. Later folding events are characterized by a hierarchical assembly of
structure (Chaps. 10 and 11; Nlting et al., 1995, 1997a; Nlting, 1998a, 2003).
61
5.2.1.3
Observation of unfolding intermediates
By using double-jumps, the T-jump method may be extended to the study of
intermediates which are located on the reaction coordinate between the folded
state and the main transition state (Nlting, 1996), so-called unfolding
intermediates (Kiefhaber and Baldwin, 1995; Kiefhaber et al., 1995; Wallenhorst
et al., 1997). Lactoglobulin which contains several prolines, unfolds on a time
scale of hours in 4 M urea at temperatures around 4oC. When T-jumping after a
short incubation time under conditions of partial cold-unfolding, lactoglobulin
displays a rapid kinetic event with increase in fluorescence intensity (Nlting,
1996). The amplitude of this kinetic event disappears with the same rate constant
as the protein approaches the fully unfolded state, suggesting the occurrence of an
unfolding intermediate (Figs. 5.13, 5.14; Nlting, 1996).
Fig. 5.13. Refolding of

-lactoglobulin from the
partially cold-unfolded
state: The amplitude of
the fast reaction disappears after prolonged
incubation under conditions of cold-unfolding
at 4oC in 4.5 M urea
(Nlting, 1996).
Fig. 5.14. Free energy

diagram of -lactoglobulin at low temperatures. U, I, F, unfolded,
intermediate,
and folded states; #2
slow transition state; #1
fast transition state
(Nlting, 1996).
62
5.2.2
LASER-induced T-jump
When the electrical discharge was replaced by a LASER pulse, temperature rises
in the nanosecond time scale were obtained (Fig. 5.15; Thompson, 1997).
At wavelengths around 1.5 2 micrometers the absorption of H2O and D2O is
not too large to ensure a relatively uniform heating in a sample cell of about
50 micrometers path length. Nanosecond pulses of several mJ at this wavelength
may be generated using optical parametric oscillators (OPO; e.g., from
B.M.Industries, Lisses, France; Elliot Scientific, Harpenden, U.K.; Continuum,
Santa Clara, CA; see Fig. 5.16), or even stronger pulses with up to about 200 mJ
using Raman shifters (e.g., Edinburgh Instruments, Edinburgh, U.K.; Lambda
Photometrics, Harpenden, U.K.; Light Age, Somerset, NJ; see Fig. 5.17; Ballew et
al., 1996b). Temperatures around 4oC where the coefficient of thermal expansion
of water approaches zero, small sample cell volumes, and small amplitudes of the
T-jumps are used to avoid significant pressure effects upon rapid heating.
Fig. 5.15. LASER T-jump and optical triggers. One LASER beam serves as a pump beam to
excite the desired chemical or physical changes which are measured by using a probe beam.
Fig. 5.16. Optical parametric oscillator. The nonlinear polarizability of the crystal medium is
used to generate two low frequency photons (the signal and idler waves) from a single, high
frequency photon (pump wave): 1/pump wavelength = 1/signal wavelength + 1/idler wavelength.
63
Fig. 5.17. Raman shifter. The pump line with the wavenumber o (1/wavelength) is converted
(scattered) into the Stokes line o R. The magnitude of R depends on the gas. For example,
R is about 4155 cm1, 2991 cm1, 2331 cm1, and 2915 cm1, for H2 , D2 , N2 , and CH4 ,
respectively. Optimal conversion efficiencies at pressures of several 10 bar (several MPa) are
typically 10 50% at the first Stokes wavelength. Anti-Stokes lines and higher-order Stokes lines
have lower intensities. For better visibility, the optical axes of the two beams are shifted against
each other.
Fig. 5.18. Nanosecond T-jump machine with real-time fluorescence detection. 1.54-m light
pulses with energies of 200 mJ are generated by passing the 1.064-m output of a Q-switched
Nd:YAG LASER through a Raman shifter (see Fig. 5.17). Temperature-jumps in the sample cell
are achieved by utilizing water absorption at 1.54 m. The frequency-tripled output of a
titaniumsapphire LASER serves for the excitation of protein fluorescence which is used to
monitor the reaction kinetics initiated by the temperature change. When T-jumping the protein
from the cold-unfolded to the folded state, similar to that described in Sect. 5.2.1.2, early folding
events in the nanosecond time scale may be detected (Service, 1996; Ballew et al., 1996a, b, c).
The folding reaction proceeds under essentially isothermal conditions, since the
temperature of a typical protein molecule equilibrates with the solvent within
nanoseconds or faster depending on the molecular weight (see Sect. 5.2.3;
Nlting, 1995; Ballew et al., 1996a; Nlting, 1998b).
Several groups have applied the method (Nlting et al., 1995) of T-jumping of
protein from the (partially) cold-unfolded state. Using a T-jump machine with
real-time fluorescence detection in the nanosecond time scale (Figs. 5.18, 5.19),
Martin Gruebele's group demonstrated the partial formation of a subunit of
myoglobin, comprising the A, G, and H helix of the molecule, in an initial folding
64
event of only about 7 s duration (Service, 1996; Ballew et al., 1996a, b, c;

Sabelko et al., 1998). Since methionine 131 is a significant quencher of
fluorescence in the folded state, folding at this position is connected with a
decrease of the fluorescence signal. More work on myoglobin folding was also
performed by other groups (Gilmanshin et al., 1997a, b, 1998; Pappu and Weaver,
1998).
The detection method used in the LASER T-jump set-up has been extended to
far-infrared absorption at wavenumbers around 1500 1700 cm1 (5.88 6.67 m
wavelength), which is a very sensitive probe of the protein secondary structure
content. Helixcoil transitions of a 21-amino acid residue peptide with a folding
rate constant of 6 107 s1 at 28oC have been observed directly in the time domain
(Williams et al., 1996; Thompson et al., 1997). The folding transition of a hairpin
consisting of 16 amino acid residues is surprisingly slow, 6 s at room temperature (Muoz et al., 1997, 1998).
Fig. 5.19. Second-harmonic generation in a non-linear crystal which is contained in a ringcavity. Short wavelengths useful for the excitation of fluorescence or for the measurement of
protein circular dichroism are generated by doubling the frequency of a pump LASER.
5.2.3
Maximum time resolution in T-jump experiments
When heating the buffer via electrical discharge or LASER, the maximum speed
of initiation of rapid folding is usually limited by the speed of the propagation of
heat in the sample. An approximate analytical solution for the heat diffusion in
and around a spherical molecule that is contained in a constant energy flow has
been given, originally in the context of sound velocity measurements (Nlting,
1995), and later extended to T-jumping (Nlting, 1998b). Fig. 5.20 shows
different temperature profiles in and around the protein molecule as a function of
the parameter = (2t1cp1ro2)0.5, where , , cp , ro and t are heat conductivity,
density, specific heat capacity, radius of the molecule and time, respectively
5.3 Optical triggers
65
(Nlting, 1995, 1998b). For 1, corresponding to a small radius of the molecule

and a large time, there is a good temperature equilibration between protein
molecule and bulk solvent.
This evaluation is in good agreement with LASER T-jump experiments in
which the buffer is heated via excitation of a dye (Phillips et al., 1995). From
experiment and theory it is estimated that the initiation of protein folding via
T-jumping may be extended down to about 20 ps for a 10-kDa protein (Nlting,
1998b).
Fig. 5.20. Temperature profile for a spherical molecule in a constant energy field as a function of
the parameter , where , , cp , ro , t , and T are heat conductivity, density, specific heat
capacity, radius of the molecule, time, and temperature change, respectively. r is the radius under
consideration, and is the temperature change which would be obtained in the absence of heat
conduction (Nlting, 1998b).
5.3
Optical triggers
In devices for optical triggers, a LASER pump beam usually causes a desired
chemical change and a probe beam tests the induced changes (Fig. 5.15). The
wavelength of the pump beam is selected to excite specific chromophores of the
protein or of attachments to the protein, rather than water in T-jumping.
5.3.1
LASER flash photolysis
Flash photolysis was already being used in 1993 in groundbreaking experiments
on cytochrome c and the structural resolution of refolding in the nanosecond time
scale was obtained (Jones et al., 1993). Since the CO-bound cytochrome c is less
66
stable than cytochrome c, refolding of the protein may be initiated by flash

photolysis of CO. LASER flash photolysis has been used for a long time in
protein unfolding experiments and for the observation of small conformational
relaxations in proteins (Dyer et al., 1989; Hofrichter et al., 1991; Xie and Simon,
1991; Causgrove and Dyer, 1993; Hu et al., 1996). But the method developed in
Eaton's group enabled the first ultra-rapid refolding. Photodissociation takes less
than 100 femtoseconds (Boxer and Anfinrud, 1994; Eaton et al., 1996b; Figs.
5.21, 5.22). A thorough and precise analysis of the early folding events shows a
complicated refolding behavior where rebinding of native and non-native heme
ligands occurs. Rebinding of carbon monoxide limits the applicability of the
method to the fast time scale.
Fig. 5.21.
LASER flash
photolysis of CO bound to
heme.
Fig. 5.22. LASER flash-photolysis-induced refolding of cytochrome c. COcytochrome c is

less stable than cytochrome c. Flash photolysis of COcytochrome c in moderate concentrations of denaturant initiates rapid refolding (Jones et al., 1993).
5.3 Optical triggers
67
Infrared detection of conformational changes following proteinCO dissociation with a time resolution of a few nanoseconds or less became possible due to
the availability of infrared LASERs (e.g., diode LASERs manufactured by Laser
Components, Olching, Germany) and fast mercury cadmium telluride (MCT)
detectors (e.g., made by Kolmar Technologies, Conyers, GA; EG&G Judson,
Montgomeryville, PA; Fermionics, Simi Valley, CA) with high sensitivity around
6 m wavelength (Fig. 5.23; Dyer et al., 1994; Yuzawa et al., 1994).
Fig. 5.23. An example for infrared detection of protein refolding induced by CO flash
photolysis. The output of a Nd:YAG LASER (1064 nm) is frequency-doubled (532 nm) by using
second-harmonic generation (SHG) and converted into a wavelength of 572 633 nm with a dye
LASER. In a first mixer made from a potassium titanyl phosphate crystal (KTP), the difference
frequency between the 1064 nm and the output of the dye LASER is generated. The resulting
difference frequency is mixed with 1064 nm in a second mixer made from silver thiogallate
(AgGaS2). This generates a pulse which is tunable from about 3333 to 7600 nm
(3000 1300 cm1). Mercury cadmium telluride (MCT) detectors are used for the detection of
the infrared radiation. Strong pump and probe pulses of a few picoseconds duration may be
obtained when using additional optical amplifiers. In this measurement principle, the time point
of detection and the time resolution are determined by the position of the delay line and the
duration of the LASER pulses, respectively. Thus, picosecond time resolution may be obtained
even with slower MCT detectors (Dyer et al., 1989, 1994; Causgrove and Dyer, 1993).
68
Subpicosecond resonance Raman detection has been utilized for the detection
of CO-flash-photolysis-induced conformational changes in hemoglobin and myoglobin (Fig. 5.24; Petrich et al., 1987; Varotsis and Babcock, 1993; Franzen et al.,
1995). This measurement principle could be used also for protein refolding
experiments in a similar way to that shown in Fig. 5.22.
Fig. 5.24. Subpicosecond resonance Raman spectroscopy. A combination of a colliding-pulse

LASER, continuum generation (H2O), optical filters, second-harmonic generation (SHG), and
pumped dye amplifiers is used to generate subpicosecond pulses of a pump beam at 575 nm and
a probe beam at 435 nm (Petrich et al., 1987; Varotsis and Babcock, 1993; Franzen et al.,
1995).
A further major breakthrough has been the extension of circular dichroism

(CD) spectroscopy to the nanosecond time scale (Lewis et al., 1985; Lewis et al.,
1992; Goldbeck and Kliger, 1993; Bjrling et al., 1996; Chen et al., 1998) by
using a new type of CD spectrometer (see Sect. 8.2) in combination with LASER
flash photolysis for the initiation of conformational changes in proteins. In this
new spectrometer, the CD signal is optically amplified prior to the conversion into
an electrical signal. Studies have been performed on kinetic processes effecting
changes in the chiral structures of hemoglobin (Lewis et al., 1985; Bjrling et al.,
1996), myoglobin (Lewis et al., 1985; Milder et al., 1988), and phytochrome A
(Chen et al., 1997). Kliger et al. even succeeded in the measurement of the far-UV
CD in the time scale from nanoseconds to seconds (Zhang et al., 1993; Chen et
al., 1997; Chen et al., 1998).
5.4 Acoustic relaxation
69
A similar measurement principle with magnetic circular dichroism (MCD)

detection (Milder et al., 1988) has been applied to conformational relaxations in
cytochrome oxydase (Goldbeck et al., 1991; Woodruff et al., 1991), cytochrome ba3 (Goldbeck et al., 1992), and cytochrome c3 (O'Connor et al., 1993).
MCD is sensitive to structural features of the protein molecule that directly affect
the energy levels of the MCD chromophores (Goldbeck, 1988).
In a study by Chen et al. (1998) the refolding of cytochrome c, induced by
flash photolysis of CO-bound cytochrome c in moderate concentrations of
guanidine hydrochloride, was followed by CD in the far-UV ( 215 nm in this
study), near-UV, and Soret region (about 400 460 nm). Conformational
relaxations which have been observed in these experiments in the time scales of
500 ns and 2 s are not connected with a significant formation of secondary
structure (Chen et al., 1998). The method of time-resolved circular dichroism
spectroscopy is presented in more detail in Sect. 8.2.
5.3.2
Electron-transfer-induced refolding
Rapid refolding of redox-active proteins for which there is a significant difference
in stability between the oxidized and reduced forms has been induced by electron
transfer (Chan et al., 1996; Mines et al., 1996; Pascher et al., 1996). The redoxactive cofactor is covalently bound to the protein to prevent the effect of
bimolecular reactions on the observed rate constant. Reduction of unfolded ferri
cytochrome c within less than 1 s initiates rapid refolding.
5.4
Acoustic relaxation
The frequency-dependence of sound absorption and sound velocity of protein is
related to conformational transitions (Hammes and Roberts, 1969; Sarvazyan,
1991). Acoustic relaxation, i.e., sound absorption or sound velocity, can be
measured using acoustic resonators (Fig. 5.25, 5.26; Eggers and Kustin, 1969;
Gavish et al., 1983a, b; Sarvazyan, 1991; Chalikian et al., 1994, 1996) or by
determining the time of flight of sound pulses (Sarvazyan, 1991). Electrical
transducers which are used in acoustic resonators (Fig. 5.26) contain piezoelectric
material that expands or contracts upon application of a voltage and can generate
sound over the wide range of frequencies from 1 kHz to 1 GHz, corresponding
to folding relaxations from 1 millisecond to 1 nanosecond.
Relaxations at about the frequency of the sound wave and at higher frequencies
contribute to the frequency-dependent component of the sound velocity. The total
observed signal is affected also by the properties of the bulk solvent, the
properties of the hydration shell of the molecule, and the intrinsic properties of the
molecule.
70
Fig. 5.25. Resonator sound velocity meter (Gavish et al., 1983a, b; Nlting et al., 1993; Nlting
and Sligar, 1993). The resonator is composed of a sample cell that contains an emitting and a
receiving transducer. A positive feedback-loop maintains a standing sound wave in the
resonator. Since there is a number of resonance frequencies (higher harmonics), a frequency
generator is used to select a specific oscillation. At a given number of nods of the standing wave,
the resonance frequency is a linear function of the sound velocity. By precisely measuring the
resonance frequency, the difference of sound velocity of the protein solution relative to buffer is
determined with a precision of 0.2 cm s1, compared with the absolute sound velocity of water
of 1500 m s1 at 25oC ! The protein concentration is typically 110 g L1.
Fig. 5.26. Simple and robust design of a sample cell for a resonator sound velocity meter. For
the precise coplanar adjustment of the transducers, the two sides of the cell are squeezed
together with the help of three screws (not shown). The cell is temperature-controlled with a
temperature drift of less than 0.005oC h1.
Significant acoustic relaxation is observed if there is a large volume change

involved in the chemical or physical processes that are excited by the sound wave
(see also Sect. 5.5). Computer dynamic simulations of the tiny volume
fluctuations of the protein molecule are complicated because there are different
5.4 Acoustic relaxation
71
definitions for the volume which is considered as the hydration shell, and so, the
calculation of the volume of the protein molecule is ambiguous. Fortunately, the
relaxational part of the acoustic signal can be identified by measuring at different
frequencies with no need to separate hydration shell and intrinsic contributions.
Sound velocity, u , density, , and adiabatic compressibility, , of a solution are
related to each other by the Laplace equation:
1 = u2 .
(5.6)
For the compressibility measurement of protein in solution by using Eq. 5.6, a

high precision of density measurement is necessary, depending on the
concentration of protein, which is typically 110 g L1 (for further information see
Gavish et al., 1983a; Sarvazyan, 1991). Density meters for the measurement of
small density differences with a precision of a few mg L1 are commercially
available (e.g., MettlerToledo, Greifensee, Switzerland).
Ultrasonic compressibility measurements have significantly expanded our
understanding of the physical properties of molten globule states (MG). The MG
of apo-cytochrome b562 at neutral pH displays only a small increase of relaxations
at 2 MHz relative to the native conformation (Nlting and Sligar, 1993). Partially,
this may be the case because in this MG three of its four helices have a relatively
defined native-like secondary and tertiary structure. The sound velocity of the
acidic MG of -lactalbumin displays a large relaxational part at 2 MHz (Fig. 5.27;
Nlting et al., 1993). Half of the sound absorption at 7 MHz is accounted for by
relaxation (Kharakoz and Bychkova, 1997). This shows that a large amount of
relaxations in the fluid-like MG state proceeds exceedingly fast, in the
submicrosecond time scale.
Fig. 5.27. Relative sound velocity changes urel = (u uo) /uo , where uo is measured at
pH 7, and u at the indicated pH. Closed circles, -lactalbumin; open circles, lysozyme which
has a large structural and sequential homology to -lactalbumin (Nlting et al., 1993).
72
5.5
Pressure-jump
Pressure causes a perturbation of the physical or chemical equilibrium (Takahashi
and Alberty, 1969). The relaxation of the system to a new equilibrium can be
observed by an optical detection system (Fig. 5.28). Protein folding is generally
accompanied by a change in volume. According to the principle of LeChatelier,
an increase in pressure favors transitions to the state with the smaller volume
(Weber, 1993; Jung et al., 1995, 1996; Topchieva et al., 1996), which is usually
the unfolded state at sufficiently high pressure (Gross and Jaenicke, 1994; Foguel
and Weber, 1995; Tamura and Gekko, 1995; Bismuto et al., 1996; Michels et al.,
1996; Mozhaev et al., 1996; Tanaka and Kunugi, 1996).
Fig. 5.28. Repetitive pressure-perturbation method (Pryse et al., 1992). A stack of transducers
generates a pressure of 10 50 bar (15 MPa) which causes (partial) unfolding of the protein.
After rapid release of the pressure by changing the voltage, the refolding kinetics of the protein
is monitored by using absorption or fluorescence spectroscopy. Usually the stack of transducers
is separated from the sample volume by a membrane (not shown).
Fig. 5.29. Repetitive pressureperturbation method (Pryse et

al., 1992). Pressures of several
100 bar (several 10 MPa) may
be generated using a smaller
cross-section for the sample cell
than for the transducers and a
piston to propagate the force
from the transducers to the
sample cell.
5.6 Dielectric relaxation and electric-field-jump
73
The repetitive pressure-perturbation method (Figs. 5.28, 5.29) repeatedly uses

pressure-jumps and accumulations of the kinetic traces (Pryse et al., 1992).
Combined with signal averaging, the repetition of pressure-perturbations of about
10 bar (1 MPa) causes a dramatic improvement of signal-to-noise ratio and
reproducibility, and a dead time of 1 ms has been obtained. Also in this method a
further improvement down to a dead time of 100 s should be possible, for
example with smaller sample cells.
Single pressure-jumps of up to 700 bar (70 MPa) have been applied and
relaxed within 60 s using a mechanical valve (Fig. 5.30). Fluorescence detection
revealed the slow refolding kinetics of barnase (Oliveberg and Fersht, 1996b).
Fig. 5.30.
A pressure-jump
apparatus with a mechanical
valve. After release of the spring,
the bar is pushed which opens
the valve. Fast pressure release
enables jumps from 700 bar
(70 MPa) to atmospheric pressure
(Oliveberg and Fersht, 1996b).
5.6
Dielectric relaxation and electric-field-jump
Charge interactions are predicted to pull a protein apart if a sufficiently strong
electric field is applied (DeMaeyer, 1969; Oliveberg and Fersht, 1996a). Early
work on peptides using dielectric relaxation techniques (Fig. 5.31) has
significantly contributed to our knowledge about the high speed of helixcoil
transitions. It has been discovered that helixcoil transitions of small peptides
may occur in the nanosecond time scale (Schwarz and Seelig, 1968).
The design of the field-jump set-up (Fig. 5.32) with a time resolution of 40
nanoseconds, which has been developed in the group of Manfred Eigen (Porschke
and Obst, 1991; Porschke, 1996), resembles that of the electrical discharge
T-jump apparatus illustrated in Figs. 5.6 5.8. The capacitor used in the T-jump
apparatus is replaced by a coaxial cable. If the impedance of the cable is matched
with that of the cell, the time of discharge is limited by the speed of the
propagation of electromagnetic waves in the cable and by the length of the cable.
74
For example, the time of discharge of a 10-m cable is about 100 ns. Since the total
released energy is comparatively small, the temperature increases are far smaller
than in T-jumping. A low electrical conductivity of the protein solution is chosen,
but the electrodes of the sample cell are not completely isolated from the protein
solution because otherwise the electric field would rapidly be neutralized by
counter ions from the sample. Similar to T-jumping but different from dielectric
relaxation, the folding kinetics is monitored by fluorescence or absorption
detection which enables the use of lower protein concentrations than those
typically used in dielectric relaxation studies.
Fig. 5.31. Electrical wiring for the measurement of dielectric relaxation. An electrical bridge
contains four resistors; one of them (Z2) is adjustable; another (Z1) represents the sample cell. In
the measurement, Z2 is adjusted to zero voltage at the meter indicated. Then Z1 is given by:
Z1 = Z2Z3/Z4. In the illustrated example of this measurement principle, Z2 is a complex resistor
that is constructed, e.g., by parallel connection of an adjustable capacitor with an adjustable
Ohm resistor. Frequencies between 1 Hz and 1 GHz are relatively easy to generate and to apply.
Fig. 5.32. Electric-field-jump machine. A coaxial cable is charged to a high voltage and rapidly
discharged through the sample cell that contains the protein solution. The optical system (not
shown) used for the detection of the fast kinetics is similar to that of the T-jump machine, see
Figs. 5.6 5.8.
5.7 NMR line broadening
75
5.7
NMR line broadening
The principle of nuclear magnetic resonance (NMR) has been reviewed and
discussed widely (see Sects. 7.1 and 8.1; Wthrich, 1986; Williams and Fleming,
1995). Briefly, the effect of NMR originates from the absorption of
electromagnetic radiation by the atomic nuclei which have a nuclear spin, such as
1H, 13C, and 15N, when a magnetic field is applied. Nuclei with different
xxxxxxxxx
Fig. 5.33. NMR line broadening. An equilibrium between folded, F, and unfolded, U,
conformations is considered, where k is the rate constant of exchange between U and F. F and
U are the NMR frequencies for a particular atomic nucleus in the folded and unfolded
conformation, respectively. Top: Folded and unfolded conformations are in slow exchange,
k |F U |. Separate lines are observed for the folded and unfolded states. Middle:
Intermediate frequency of exchange, i.e., k |F U |. A complicated line shape, which varies
with the magnitude of k, is observed and from this k may be calculated. Bottom: Folded and
unfolded conformation are in rapid exchange, k |F U |. The atomic nucleus feels only an
averaged environment, and thus only one line is observed.
76
orientations relative to the externally applied magnetic field differ in their energy
and in their occupancy. Transitions between these orientations may be excited by
electromagnetic radiation.
Usually, an NMR line changes its position by ( = |F U | ) upon transition
from the folded, F, to the unfolded state, U, because of a change of the chemical
environment of the nuclei. Under conditions in which F and U are in equilibrium,
three cases may be distinguished (Fig. 5.33; Williams and Fleming, 1995):
1. The rate constant, k, of the transition between F and U is low relative to :
Two lines are observed. The two amplitudes correspond to the populations of F
and U, respectively (Fig. 5.33, top).
2. The rate constant of the transition between F and U is high relative to : Only
a single line is observed, since the nucleus feels only an averaged environment, in-between that of F and U (Fig. 5.33, bottom).
3. The rate constant of the transition between F and U is comparable to : A
complicated lineshape, which strongly varies with the magnitude of k, is
observed. From this shape, the rate constant of exchange can be calculated for
the specific nucleus (Fig. 5.33, middle).
With high-field NMR spectrometers, rate constants of exchange between F and
U in the submillisecond time scale may be estimated using line broadening
measurements of suitable 1H lines with changes of chemical shifts upon unfolding
by a few ppm (106). NMR has the highest structural resolution among the
methods mentioned in this chapter, but it requires a significant population of the
species involved, and thus, is difficult to apply to early intermediates of low
occupancy in equilibrium.
The aromatic 1H NMR spectra of a truncated form of the N-terminal domain of
phage -repressor have been measured at various concentrations of urea from 1.3
to 3.1 M (Huang and Oas, 1995). The extrapolated folding rate constant for the
absence of denaturants is (3600 400) s1 at 37oC !
5.8
Summary
Currently, the highest, i.e., atomic, resolution of protein folding events may be
achieved by NMR-related methods (Table 5.1). Usually, sample concentrations of
0.1 to 5 mM are required to obtain a sufficient NMR signal, and thus systems
which are prone to strong aggregation may not be investigated. Currently, the
time resolution of NMR when combined with H/D exchange (Sect. 8.1) initiated
by ultrafast mixing (Sect. 5.1) is roughly 200 s.
Significantly higher time resolution at low protein concentration has become
possible along with the development of a number of fast methods for the initiation
of folding, including temperature-jumping, optical triggers, acoustic relaxation,
pressure-perturbation, and dielectric relaxation. With the exception of acoustic
and dielectric relaxation, these methods may be combined with the -value
5.8 Summary
77
analysis which enables structural resolution at the level of individual amino acid
residues, and is the only existent method to characterize structurally the transition
states of folding (see Sect. 8.3).
Table 5.1. Methods for the study of fast folding events. For important results on ultrafast-folding
proteins see Sect. 12.4.
Method
Mixing
Electrical discharge T-jump
LASER T-jump
LASER flash photolysis
Electron-transfer-induced refolding
Acoustic relaxation
Pressure-jump
Dielectric relaxation
NMR line broadening
Approximate time range
Structural resolution
10 s
100 ns 10 sb
1 nsd 100 msb
100 fse 1 ms
1 s 1 ms
1 ns 1 ms
60 s 1 s
1 ns 1 s
100 s 100 ms
Atomica
Individual residuesc
Global properties
Global properties
Atomic
In combination with NMR, on a time scale of roughly 200 s .

The resolution of slow kinetics is limited by the re-cooling of the sample after application of
the T-jump, which depends on the size and heat conductivity of the sample cell. For example,
the re-cooling rate constant of the sample cell with 1 mL volume shown in Fig. 5.8 is 0.03 s1.
c Individual amino acid residues; in combination with protein engineering as described in
Sect. 8.3 and Chap. 10.
d According to theoretical considerations, further extension to about 20 ps should be possible.
e The propagation of the conformational relaxation through the molecule may be slower.
a
b
6 Kinetic methods for slow reactions
6.1
Stopped-flow nuclear magnetic resonance (NMR)
One-dimensional NMR (see Sects. 5.7, 7.1, and 8.1) of small proteins with 5 to 20
kDa molecular weight may be performed on a time scale of a few 100 ms. This is
fast enough to use stopped-flow techniques in combination with NMR to monitor
folding reactions with rate constants of up to a few s1 (Balbach et al., 1995;
Hoeltzli and Frieden, 1995, 1996; Dyson and Wright, 1996). Fig. 6.1 displays a
typical stopped-flow NMR design. The stop syringe used in common stoppedflow devices (Fig. 5.1) is replaced by stop bars. After a sample volume of a few
100 L has passed through the mixer, the syringes are stopped and the acquisition
of the NMR signal is started (Fig. 6.2).
One of the main applications of stopped-flow NMR is the structural resolution
of folding intermediates. For higher sensitivity of stopped-flow NMR detection,
the 5 tryptophan residues of Escherichia coli dihydrofolate reductase have been
replaced by 6-19F-tryptophan (Hoeltzli and Frieden, 1995, 1996, 1998). Using
site-directed mutagenesis (see Sect. 8.3.1) the resonances have been assigned to
individual tryptophan residues.
Fig. 6.1. Set-up for stoppedflow nuclear magnetic resonance. The stop syringe in the
common stopped-flow apparatus (Fig. 5.1) is replaced by a
stop-bar.
80
Fig. 6.2. Change of NMR

lines in stopped-flow-induced folding- or unfolding
experiments (see text). Rate
constants of the processes
involved are determined by
fitting the appropriate equations (see Chap. 4) to the
intensities of the NMR lines
as functions of time.
The folding pathway of dihydrofolate reductase has been found to involve the
cooperative formation of one or more intermediates (Hoeltzli and Frieden, 1995,
1996, 1998). This direct NMR evidence for the cooperativity of folding represents
a significant support of optical-spectroscopical and equilibrium-thermodynamical
methods for the study of proteins in which the cooperativity of folding is often an
important assumption for the interpretation of the data.
Complications in stopped-flow NMR experiments occasionally are transient
and permanent aggregation (see Sect. 9.2) because the protein concentration
needed is typically 100 M to 5 mM. Permanent aggregation may easily be
detected by using light scattering or ultracentrifugation methods (Sect. 9.2).
However, transient aggregation, i.e., aggregation that occurs only for a short
period of time after initiation of the reaction kinetics, may easily remain
undetected, and may affect the properties of the molecules involved, or may even
be confused with the formation of folding intermediates. However, NMR is not
very sensitive to small populations of aggregates. Usually populations of less than
5% of aggregated or otherwise modified species are not detected.
6.2
Fluorescence- and isotope-labeling
Measurements of very slow folding kinetics or enzymeinhibitor dissociation
reactions may be problematical: Direct spectroscopic detection of the involved
species may be too insensitive or not stable enough over weeks of data acquisition.
In particular, fluorescence detection over a long period of time is complicated by
photolysis of aromatic amino acid residues. In these cases the competition between two chemically similar but physically slightly different labels might be used
for the observation of the reaction kinetics.
6.2.1
Folding reactions
1. A reaction mixture of protein and label1 that will produce a high yield of
proteinlabel1 complex is made. Label1 is an active label, i.e., it is detectable
by optical spectroscopy, NMR, or radioactive methods.
6.2 Fluorescence- and isotope-labeling
81
2. The proteinlabel1 complex is incubated under conditions of competition of

label1 with a large excess of the non-active label2 that is chemically similar to
label1 but may be distinguished by physical methods. The conditions are
chosen in such a way that label1 may rapidly dissociate in the case of a partial
opening or unfolding of the protein molecule (Fig. 6.3).
3. The concentration of the proteinlabel1 complex is measured for different time
points, in real-time or indirectly. Under suitable conditions, the rate constant of
decrease of label1 in the complex reflects the rate constant of partial opening or
unfolding of the protein.
An important variant of this method is the quenched-flow technique (Sect. 8.1)
which is used for the NMR- and mass-spectroscopical characterization of folding
intermediates: Protein is labeled with 2H that is later chased off by 1H from the
unprotected parts of intermediates. Under conditions of fast exchange, the rate
constant of protection against label exchange reflects the rate constant of structure
formation. 2H (D, deuterium) is not radioactive and differs only moderately from
1H in many properties, and thus the exchange often causes only a small
perturbation of the sample. Usually the H/D exchange measurement is not realtime, and thus, may also be performed on a fast time scale.
Fig. 6.3. Observation of very slow kinetics with the help of the method of competition between
an active (i.e., detectable) and non-active label. Suitable labels may be fluorescent labels or
isotopes. The active label is chased off with an excess of non-active label that is chemically
similar. The rate constant for the decay of the concentration of the active complex is measured,
e.g., by using optical methods, NMR, or scintillation counting. Under suitable conditions, the
decay rate constant is equal to the rate constant of the global or local folding event.
6.2.2
Dissociation reactions
1. For the observation of the slow dissociation kinetics of a strong enzyme
inhibitor complex, enzyme (inhibitorlabel1) complex and inhibitorlabel2 are
made. Label1 is detectable by optical spectroscopy, NMR, or radioactive
methods. Label2 is chemically similar but physically distinguishable from
label1 , and may be a part of the inhibitor itself, for example, 1H.
2. The enzyme (inhibitorlabel1) complex is incubated under conditions of
competition of the inhibitorlabel1 with a large excess of inhibitorlabel2.
82
3. Provided that the labels do not introduce significant perturbations of the reaction, the rate constant for the decrease in concentration of label1 in the complex
reflects the dissociation rate constant of the enzymeinhibitor complex. This
measurement under pseudo-first-order conditions is not significantly affected
by the association rate because essentially every association event leads to the
binding of the silent inhibitorlabel2 (Fig. 6.4).
With this method, the dissociation rate constant of the barnasebarstar complex
has been determined using tritium (3H) isotope labeling (Fig. 6.4; Schreiber and
Fersht, 1993a, 1995). The barnasebarstar interaction is one of the strongest
proteinprotein interactions known. That is why the dissociation rate constant of
this complex can hardly be measured by common mixing methods in which the
rapid association interferes with the measurement of the dissociation rate constant.
In the presence of a large excess of native barstar, the concentration of 3Hlabeled barstar in complex with barnase decays single-exponentially, and the
observed rate constant is equal to the dissociation rate constant of the complex.
For the measurement of the disappearance of 3H-labeled barstar from the complex,
fractions of the sample were analyzed at different time points. For this purpose,
the barnasebarstar complex was separated from the fraction of sample by using
chromatography, and the concentration of 3H in the complex was determined by
using a scintillation counter (Schreiber and Fersht, 1993a, 1995).
Caution: 3H is radioactive and must be used in strict accordance with all local
and national regulations and laws.
Fig. 6.4. Observation of very slow dissociation rate constants of an enzymeinhibitor complex
with the help of a competition method. Inhibitorlabel1 is chased off with an excess of
inhibitorlabel2 , and then removed, for example by using chromatography. For non-perturbing
labels under suitable experimental conditions, the rate constant for the change of the
concentration of enzyme(inhibitorlabel1) complex reflects the dissociation rate constant of the
enzymeinhibitor complex (Schreiber and Fersht, 1993a, 1995). This method is analogously
applicable to enzymesubstrate complexes.
7 Resolution of protein structures in solution
7.1
Nuclear magnetic resonance
Nuclear magnetic resonance (NMR) has emerged as an important tool for the
study of protein structures, and it is the method with the highest structural resolution for proteins in solution that is currently available. A large number of excellent textbooks on protein NMR has been published (Atta-ur-Rahman, 1986;
Wthrich, 1986; Williams and Fleming, 1995). Extensive practical hints have
been given by Croasmun and Carlson (1994). Here a brief introduction is
presented to support the understanding of the kinetic applications of NMR in
Sects. 5.7, 6.1, and 8.1.
NMR spectroscopy is based on the measurement of the absorption of electromagnetic radiation by the nuclei of atoms in the radio frequency range of several
MHz up to several 100 MHz. The most important observable nuclei that occur in
proteins or can be incorporated into proteins are 1H , 13C , 15N , 19F , 31P. Their natural abundances are 99.985%, 1.10%, 0.367%, 100%, and 100%, respectively
(Lide, 1993). Due to their low natural abundance, 13C and 15N often require
isotope enrichment for a sensitive detection. For 1H, 13C, 15N, 19F, and 31P the
nucleic spin is 1/2, and thus, in a magnetic field the nucleus takes up two possible
orientations that differ in the energy by E (Williams and Fleming, 1995) ,
E = B h /(2) ,
(7.1)
where , B , and h are the magnetogyric ratio, the applied magnetic field strength,
and Planck's constant, respectively. According to Boltzmann's distribution, the
ratio of occupancy of the two orientations, N and N , is
N /N = exp(E /(kBT)) ,
(7.2)
where kB and T are Boltzmann's constant and absolute temperature, respectively.

Electromagnetic radiation is absorbed when its frequency, , matches the frequency at which the nuclear spins precess in the magnetic field (Wthrich, 1986) ,
= B / (2) .
(7.3)
Eqs. 7.1 and 7.2 show that the occupancy difference, and thus the intensity of
an NMR line, increases with increasing magnetic field strength, B. Furthermore,
the spectral resolution also increases with increasing B.
84
Fig. 7.1. Schematic representation of a high field Fourier transform NMR spectrometer. The
sample is contained in a strong magnetic field (18.79 Tesla for an 800 MHz spectrometer)
generated with a superconducting coil that is contained in liquid helium. Superconductivity is
the ability of a material to carry electricity without any measurable resistance. The low
temperature (typically 2 4 K) of the main coil enables a high magnetic field strength without
breakdown of superconductivity. Spinning of the sample improves the homogeneity. Smaller
coils contained in the probe serve for radiowave excitation and detection. Different pulse
sequences are executed with the help of computer, acquisition processor, synthesizer and linear
power amplifier. The free induction decay (FID) is amplified and digitized, and then further
processed in the acquisition processor and computer (BRUKER Analytik, Karlsruhe, Germany).
Currently, NMR spectrometers with B = 18.79 Tesla are available (BRUKER

Analytik, Karlsruhe, Germany). In high-field NMR spectrometers, the magnetic
field is generated by using superconducting coils (Fig. 7.1). At a magnetic field
strength of 18.79 Tesla, the protons (1H) come into resonance at 800 MHz, and it
is common to refer to an 800 MHz instrument.
The NMR approach to studies of biological macromolecules is based largely
on four types of experiments: the chemical shifts, coupling constants, nuclear
Overhauser effect (NOE), and amide proton exchange rates (Wthrich, 1986;
Williams and Fleming, 1995).
Chemical shifts. Since the magnetic field strength, B, is slightly affected by the
chemical environment of the nucleus, i.e., electrons and other nuclei, the nuclei of
the molecule come into resonance at slightly different resonance frequencies, .
Lines are shifted to a different due to their different chemical environment. See,
7.1 Nuclear magnetic resonance
85
for example, Fig. 7.2 which displays the 1H spectrum of ethanol: Resonance
frequencies are measured relative to a reference frequency and are in units of ppm
(part per million, 106). Tetramethylsilane (TMS) is chosen as the internal
standard, i.e., for the calibration of the reference frequency. One can see that the
protons of the methyl group CH3 (a) have the lowest chemical shift, i.e., lowest
resonance frequency. These nuclei are significantly shielded by the electrons, and
thus experience a lower magnetic field strength. The highest resonance frequency
is found for the methylene protons (c) that are less shielded by electrons.
Coupling constant. Nuclei that are connected by covalent bonds will exert
magnetic interactions on each other. The precession of covalently bound
neighbored nuclei in the magnetic field affect each other, and fine structures, socalled multiplets, instead of single lines, are observed. This is called coupling. A
nucleus that couples equally to n others will give rise to an (n + 1)-multiplet. For
example, the signal of the methyl group, CH3 , of ethanol (Fig. 7.2) is a triplet
caused by couplings of the methyl protons with the two methylene protons.
Furthermore, the coupling constants vary with the conformation of the molecule
(Williams and Fleming, 1995).
Nuclear Overhauser effect (NOE). NOE signals arise from spinspin interactions of nuclei through space. These interactions influence the relaxation rates of
nuclei and thereby affect the occupancies of different energy levels of the nuclei.
The observed effect of these interactions is a change in intensity of an NMR line
when another nucleus is irradiated at its resonance frequency. The NOE is only
noticeable over a distance, d , of less than about 5 and its magnitude falls off as
d 6. Thus, NOE experiments may provide distance constraints between nuclei,
which is very important information for simulating and refining the threedimensional structure of a macromolecule.
Proton exchange rates. Proton exchange quenched-flow experiments are an
important method for studying the rapidly changing conformations of macromolecules, such as protein folding intermediates (Sect. 8.1).
Fig. 7.2.
1H NMR
spectrum
of ethanol.
86
Fig. 7.3. Fourier-transform principle for NMR. The magnetization, M , of the sample is tipped
through an angle, , by a powerful radio frequency pulse. In this example = 90o (/2 - pulse).
After the pulse has been applied, the magnetization starts to precess around the z-axis. The
resulting radio frequency is detected and Fourier transformed (see Fig. 7.4). Relaxation of the
magnetization causes an exponential decrease in the radio signal with time and return of the
system to the original state.
Fig. 7.4. Fourier transforms of free induction decays (FID's). Top: The Fourier transform of a
single-sinusoidal function is a single NMR line. Bottom: A superposition of two sinusoidal
functions in the time domain corresponds to two peaks in the frequency domain.
Precise data from mainly the first three of these four types of experiments serve
for deducing the chemical environments of nuclei, for the assignment of the nuclei
to specific amino acid residues, and for the simulation and the refinement of native
protein structures. NMR spectroscopy on small stable proteins in solution can
enable atomic resolution with errors of less than a few .
7.1 Nuclear magnetic resonance
87
Fig. 7.5. Sequence of two-dimensional (2D) NMR experiments (see text).
Fig. 7.6. Sequence for the heteronuclear two-dimensional nuclear Overhauser effect (HOESY).
Top: Pulse sequences for protons and carbon nuclei. Bottom: Change of proton magnetization:
The first 90o pulse tips the 1H z-magnetization into the xy-plane. The 1H magnetization vectors
fan out during the first half of the evolution period (t1/2). A 180o 13C pulse serves to interchange
the spin labels and to refocus the 1H magnetization vectors in the subsequent t1/2 period.
Another 90o 1H pulse tips the magnetization back into the longitudinal direction. 1H and 13C
spins interact during the mixing period (Atta-ur-Rahman, 1986).
To obtain the NMR spectrum, the first spectrometers, so-called continuous

wave spectrometers, used to measure the absorption while steadily changing the
88
radio frequency in the desired frequency range. Nowadays, most NMR spectrometers use a Fourier-transform principle: The radio frequency is applied as a
single pulse of typically a few s duration. This pulse tips the magnetization of
the sample by a certain angle, (Fig. 7.3). After application of the pulse, the
magnetization of the sample starts to precess around the z-axis. The difference
between the radio frequency caused by the precession and a reference frequency
is detected and Fourier transformed (Fig. 7.4). This mathematical operation transforms the signal from the time domain into the frequency domain. For example,
the Fourier transform of a sine function is a single line (Fig. 7.4 top). The superposition of two sine functions gives rise to a complicated interference in the time
domain (Fig. 7.4 bottom) but is easily recognized in the frequency domain.
A major breakthrough in the field has been the introduction of two-dimensional
(2D) NMR spectroscopy (Ernst and Anderson, 1966). Usually, the 2D NMR
measurement involves four phases that are cyclically passed through several times
(Figs. 7.5 and 7.6):
1. Thermal equilibration of the system is achieved during the preparation period.
Similar to 1D NMR, a strong radio frequency pulse is applied which tips the
magnetization by an angle , for example /2.
2. In contrast to 1D NMR, detection is not started immediately after this pulse, but
the system is allowed to evolve, i.e., different nuclei are given time to interact.
After a time of evolution of the system, t1 , a further radio frequency pulse is
applied.
3. During the mixing period, tm , the system evolves further.
4. Prior to the detection of the free induction decay (FID), a further pulse is
applied in some experiments. The time of detection of the FID is t2. Fourier
transformation of the two time domains, t1 and t2 , results in two frequency
domains, 1 and 2 , respectively.
Fig. 7.7. Two-dimensional COSY NMR spectra of alanine (Ala) and threonine (Thr).
2D NMR enables us to determine the interactions between different nuclei and

thereby to estimate intramolecular distances and angles. To solve a protein structure, the first step is usually the assignment of the NMR peaks to the individual
7.2 Circular dichroism
89
amino acid residues. Different amino acid residues display various connectivities
between the nuclei, and thus may be identified by using their different patterns of
2D NMR spectra, see, for example, Fig. 7.7. Further, different amino acid
residues display different patterns of fine structure of the NMR peaks. After
assignment of the NMR peaks to the individual amino acid residues, the
information of the chemical shifts, the information from the connectivities, and
most importantly, the information from through-space energy transfer experiments
is used to establish distance constraints. Using the distance constraints and also
energy potentials, the structure is simulated and refined. 2D and multidimensional
NMR techniques provide not only information on the structures of folded proteins
but also on folding intermediates, on unfolded states, and on the dynamics of the
different conformations (e.g., Dyson and Wright, 1996; Eliezer et al., 1998). For
further information see (Atta-ur-Rahman, 1986; Wthrich, 1986; Croasmun and
Carlson, 1994; Williams and Fleming, 1995).
7.2
Circular dichroism
Circular dichroism (CD) is exquisitely sensitive to changes of protein conformations (Johnson, 1990; Nakanishi et al., 1994; Duddeck, 1995). Far-UV CD, i.e.,
CD in the wavelength region of about 170 260 nm, is sensitive to the secondary
structure, whereas the near-UV CD signal from about 260 nm to 330 nm is diagnostic of the chiral environment of aromatic residues and of the tertiary structure
of proteins. CD signals above 330 nm may be sensitive to the conformation of
cofactors, for example, heme. Although the resolution is not as high as with NMR
(see Sect. 7.1), usually CD studies are much simpler and may be performed more
cheaply and, most importantly, on a faster time scale (see Sect. 8.2).
Fig. 7.8. Circular dichroism is the difference between the absorptions for left, Al , and right, Ar ,
circularly polarized light. Alternately left and right circularly polarized light with the initial
intensity Io is passed through the sample cell. In most CD spectrometers, the difference between
Al and Ar is calculated using solely the final intensities, Il and Ir (see Fig. 7.9; Eq. 7.6; Velluz et
al., 1965).
90
Protein CD is usually expressed in units of or R ,

= l r
R = /nR ,
(7.4)
where l and r are the molar extinction coefficients for left and right circular
polarized light, respectively, and nR is the number of amino acid residues of the
protein (Fig. 7.8). and R are called molar ellipiticity and mean residual
ellipticity, respectively. The difference in absorption, A = Al Ar , for left and
right circularly polarized light is given by
A = c d ,
(7.5)
where c and d are concentration of the sample and pathlength of the sample cell,
respectively. Sometimes one still encounters the unit deg cm2 dmol1 ([]) that is
obtained by multiplying with a factor of 3298.
Fig. 7.9. Scanning CD spectroscopy. The light of a xenon lamp is passed through a polarizing
double monochromator. Then the linearly polarized light is circularly polarized with the help of
a modulator, passed through the sample cell, and converted into an electrical signal by a
photomultiplier. The signal of the photomultiplier consists of an alternative current (AC) and a
direct current (DC) component. A positive feedback loop ensures the constancy of the DC
component by adjusting the high voltage supply of the photomultiplier (HT). The AC
component with a frequency equal to the frequency of the modulator arises from the difference
between the light absorptions with left and right circular polarization. Its amplitude is a measure
of the circular dichroism. After phase-sensitive detection and rectification of the AC component
by the lock-in amplifier (LIA), it is digitized by the analog-to-digital converter (ADC) and
further processed in the computer. The CD spectrum is obtained by scanning the wavelength
region of interest with simultaneous variation of the amplitude of modulation.
91
CD spectrometers are available for a wavelength range of about 1651100 nm

(e.g., from JASCO, Tokyo, Japan). In the widely used principle for scanning CD
spectroscopy, a double monochromator ensures the selection of the desired wavelength and an excellent suppression of scattered light (Fig. 7.9). Circularly polarized light is generated by conversion of linearly polarized light with the help of a
photoelastic or electro-optical modulator that acts as a retardation element with a
variable phase delay (Fig. 7.10; Sect. 8.2). Photoelastic modulators made from
SiO2 are transparent to about 155 nm. For shorter wavelengths, CaF2 , MgF2 , or
LiF2 with a cut-off at about 130 nm, 125 nm, and 110 nm, respectively, are used
(Nlting, 1991). For technical reasons, during the modulation cycles, most of the
time the polarization is not exactly circular but elliptical. The CD is measured by
passing alternately left and right elliptically polarized light through the sample and
measuring the transmitted intensity. The difference in absorption, A = Al Ar , is
calculated from the difference of the light intensities falling onto the detector,
2 log(e) Il Ir
,
Il + Ir
f
(7.6)
where Il and Ir are the intensities of left and right elliptically polarized light,
respectively, 2 log(e) 0.87, and f is correction factor for the deviation of the
polarization from exactly circular polarization. For the sinusoidal modulation in
most CD spectrometers, f is about 0.79. The CD spectrum is obtained by scanning
the wavelength region of interest. For this purpose, the prisms of the double
monochromator are rotated with the help of computer-controlled step motors. For
the measurement principle of CD and for the construction of CD spectrometers see
also Velluz et al. (1965), Nlting (1991), and Sect. 8.2.
Fig. 7.10. Generation of circularly polarized light by passing linearly polarized light through a
quarter-wavelength (/4) crystal phase plate. The angle of the optical axis of the phase plate
relative to the plane of polarization of the incident beam is 45o. At this angle the crystal splits the
incident beam into two components of exactly equal amplitude. These two linearly polarized
components travel through the crystal plate with different velocities. The birefringence and
thickness of the crystal are adjusted to cause a difference between the two components of
exactly /4. The superposition of the two components of equal amplitude but 90o phase
difference produces a circular polarization of the beam that exits the crystal.
92
Fig. 7.11. Far-UV CD signal for different types of secondary structure: -helix of infinite length,
-sheet, turns, and random coil, as indicated (Yang et al., 1986).
Fig. 7.12. Far-UV CD spectrum of a barstar mutant (C40A/C82A/P27A) at 25oC as function of

the concentration of urea, as indicated (Nlting et al., 1997b). The protein unfolds upon
increasing the urea concentration, [urea], which causes characteristic changes of the amplitude
and shape of the CD spectrum around 220 nm.
Different types of secondary structure, -helix, -sheet, and turns,

corresponding to different chiral environments of the amide chromophores, give
rise to markedly distinct far-UV CD spectra (Fig. 7.11). The typical spectrum of
barstar which is rich in -helix is shown in Fig. 7.12. The precision of the
determination of the secondary structure content by decomposition of CD spectra
improves with increasing range of wavelengths used. However, below 180 nm the
measurement of CD spectra is very complicated because of the absorption of
protein, salts, and water, and the information content increases only slightly,
93
considering that the methods of secondary structure determination by using

decomposition of CD spectra have a considerable systematic error.
The sensitivity of the near-UV CD of proteins to changes of the tertiary
structure content is illustrated in Fig. 7.13. Unfolding of barstar by urea, heat, or
low temperature causes a weakening of the near-UV CD signal at 260290 nm,
and a change of sign, due to a loss of chiral environment of the aromatic
sidechains. The signal at 295310 nm that arises mainly from buried tryptophan
sidechains vanishes almost completely upon heat-, cold-, or urea-induced
unfolding (Nlting et al., 1997b).
Fig. 7.13. Near-UV CD signal of a barstar mutant (C40A/C82A/P27A) under different

conditions (Nlting et al., 1997b). Top: CD at 25oC as function of the concentration of urea, as
indicated. Middle and bottom: CD in 2.2 M urea as function of temperature, as indicated.
8 High structural resolution of transient

protein conformations
8.1
NMR detection of H/D exchange kinetics
Nuclear magnetic resonance (NMR) techniques not only monitor individual backbone amide protons, but also discriminate deuterons (D, 2H) and protons (H, 1H)
(see also Sects. 5.7 and 7.1; Roder et al., 1988; Udgaonkar and Baldwin, 1988;
Bycroft, 1990; Wthrich, 1986; Williams and Fleming, 1995). Most of the amide
protons of unfolded protein rapidly exchange with the protons of the solvent.
Proton exchange is both acid- and base-catalyzed. For random chain polyalanine,
the pD of slowest exchange is about 3 (Fig. 8.1; Englander and Mayne, 1992). For
the fully folded protein the exchange rates of most of the amide protons are
dramatically slowed down, in some cases by more than a factor of 109
(Udgaonkar and Baldwin, 1988). The protection against exchange of the amide
protons originates from the burial of the amide hydrogens and from hydrogen
bonding (Roder et al., 1988). Because folding intermediates usually already have
protected parts in their structures, they may be characterized by using H/D
exchange techniques in combination with NMR.
A typical quenched-flow H/D exchange pulse labeling experiment (Figs. 8.2
8.4) for the structural characterization of folding intermediates proceeds as follows
(Roder et al., 1988; Englander and Mayne, 1992; Elve et al., 1994; Dyson and
Wright, 1996): Initially, the protein is unfolded in a denaturantD2O solution
xxxxxx
Fig. 8.1. Kinetic rate constant, k , of H/D exchange for amide protons of random chain
polyalanine as function of pD (Englander and Mayne, 1992).
96
8 High structural resolution of transient protein conformations
where exchangeable NH sites become deuterated. Then the unfolded protein is

allowed to partially refold by mixing it with a deuterated buffer that favors
refolding. After a variable time, the partially refolded protein is exposed to a H2O
labeling pulse of 550 ms duration at pH 9 where exchange of free hydrogens
takes about 1 ms, so that amides at the unstructured parts of the protein become
fully protonated but sufficiently protected sites stay deuterated. Then the pH is
lowered to provide slow-exchange conditions and the protein is allowed to refold
completely. Prior to the NMR measurement, the protein solution is concentrated
by ultrafiltration.
Using the individual proton occupancies as function of the refolding time and
the assignment of the resonances, the protected parts of early folding intermediates
are mapped out. Kinetic information on a time scale of about 1 ms and longer can
be obtained. Slightly higher time resolution, 200 s, can be achieved for proteins
which are stable at high pH using the double-jump mixing head illustrated in Fig.
5.5 (Sect. 5.1) and a labeling buffer at about pH 10.
H/D exchange kinetics is the currently available method with the highest
structural resolution for monitoring early folding intermediates (Bycroft, 1990).
Unfortunately, some proteins are prone to aggregation at low or high pH. Great
care has to be taken not to confuse folding intermediates with aggregation.
Fig. 8.2. Double-jump mixing for H/D exchange kinetic experiments (quenched-flow). H/D
exchange of solvent-exposed amides is both acid and base catalyzed (see Fig. 8.1). The protein
was stored in D2O with a high denaturant concentration so that all exchangeable protons became
replaced by deuterons. Refolding is initiated in the first mixer. After a certain period of time,
determined by the velocity of the solution and the length of the delay loop, the partially refolded
protein is H-labeled with buffer at pH about 9. Refolding of the labeled protein is completed in
quench buffer at low pH where exchange is very slow.
8.1 NMR detection of H/D exchange kinetics
97
Fig. 8.3. Kinetic H/D exchange experiment. Prior to the refolding reaction, all exchangeable
protons were replaced by deuterons in the unfolded state. Refolding is initiated with deuterated
buffer. After partial refolding, the protein solution is mixed with H2O. Parts of the protein which
are protected at this time stay deuterated. Then H/D exchange is quenched with buffer of low pH
and refolding is completed prior to NMR measurement.
Fig. 8.4. pH profile

of a typical kinetic
H/D exchange experiment (quenchedflow).
98
8.2
Time-resolved circular dichroism
The kinetic resolution of single-wavelength detection of commercial CD spectrometers in combination with stopped-flow accessories is typically only a few
milliseconds, and the measurement of complete spectra is even slower.
Complete CD spectra of up to about 100 nm width may rapidly be recorded by
using a real-time multichannel spectrometer (Fig. 8.5; Nlting, 1991; Brandl et al.,
1991; Nlting et al., 1992). In this spectrometer, a polychromator instead of the
monochromator in scanning spectrometers (Fig. 7.9), and a multichannel detector
instead of a photomultiplier are used. A charged-coupled device (CCD) that has a
quantum efficiency of 60% at 600 nm wavelength, and still 20% at 200 nm,
serves as a detector with 512 channels.
Fig. 8.5. Multichannel CD spectrometer (Nlting et al., 1992). The light of a deuterium- or
xenon lamp is first linearly polarized by the polarizer, and then circularly polarized by the
modulator. The modulator generates alternately left and right circularly polarized light. After
passing through the flow cell, the different wavelengths are separated by a holographic grating.
The spectrum is recorded by a charged-coupled device (CCD). The operation of the CCD is
synchronized with that of the modulator, so that in every cycle of the CCD the transmitted
intensity for only one of the two polarizations, left- or right-circular, is recorded. CD amplitudes
are calculated for every channel from the intensities of transmitted left and right circularly
polarized light. The absorption of an optimal signal-to-noise ratio for this measurement principle
is about 0.80.9, and sharply falls off with increasing absorption above 1.5 (see Fig. 8.7).
8.2 Time-resolved circular dichroism
99
In the same way as for most scanning CD spectrometers (Sect. 7.2), the difference in absorption between left and right circularly polarized light, A = Al Ar ,
is calculated using Eq. 8.1:
2 log( e ) Il Ir ,
(8.1)
A
Il + Ir
f
where Il and Ir are the intensities of left and right circularly polarized light,
respectively, 2 log(e) 0.87, and f (here 0.9) is a correction factor for the degree
of circular polarization. For a typical measurement on proteins, A 105 103.
Because all wavelengths are recorded simultaneously within a certain
wavelength region, , for a given photon shot noise, the speed for recording a
spectrum is increased by a factor of f:
f = / ,
(8.2)
where is the spectral resolution. For example, at a spectral resolution of 1 nm,

a wavelength region of 100 nm may be measured 100 times faster than with a
scanning CD spectrometer, at a given photon shot noise. Further, since the
instrument does not need moving parts for the change of wavelength within a
certain wavelength region, its operation is very stable and its construction is very
compact. A short length of the optical path in this spectrometer is of advantage at
short wavelengths where CD instruments have to be purged with nitrogen to
remove absorbing oxygen, ozone, and water vapor.
For a signal, S , of photons converted into electrons per time unit by the CCD,
the noise, S , is approximately (Nlting, 1991; Stark et al., 1992)
S2 = S + R2 ,
(8.3)
where R is the signal-independent noise of the detector (Fig. 8.6).
Fig. 8.6. Standard deviation, S , as function of the magnitude, S , of the electrical signal per
pixel for the CCD detector used for the multichannel CD spectrometer (Fig. 8.5). Only at low
amplitudes is the signal dominated by the signal-independent noise. For this CCD, at signals
>105 photons per pixel, the measurement is photon-shot-noise-limited, i.e., the noise is about
equal to the square root of the number of detected photons (Nlting, 1991; Stark et al., 1992).
100
Thus, a good signal-to-noise ratio in fast kinetic measurements requires a large

light throughput. That is why the arrangement of the optics of the multichannel
CD spectrometer is specifically optimized for a very high light throughput: The
light is focused onto the CCD device by a holographic grating with an aperture
angle of 16o. The small number of optical elements and the high aperture of the
grating ensures that for a non-absorbing sample about 50% of the light emitted by
the lamp is transmitted to the CCD detector.
It should be pointed out, that in CD spectrometers the sensitivity of the measurement strongly depends also on the sample absorption (Fig. 8.7). Obviously, a
small concentration of the sample causes a small CD amplitude and thus a low
signal-to-noise ratio. On the other hand, a very large concentration causes a large
CD amplitude, but also a large photon shot noise since the sample absorbs too
much light. The optimal signal-to-noise ratio of a photon-shot-noise-limited CD
measurement is achieved at an absorption of 0.87. Furthermore, at a very high
absorption, CD spectrometers show a non-linear response. Thus, when measuring
a CD spectrum, the sample should be adjusted to an extinction not outside the
range of 0.11.5 for the whole wavelength region measured, if possible.
The polarization modulator (Fig. 8.8) acts as a retardation element and the
modulation cycles convert linearly polarized light into alternately left and right
circularly polarized light (Velluz et al., 1965; Nlting, 1991). In contrast to the
sinusoidal modulation in commercial CD spectrometers, in this instrument the
electro-optical modulator is operated with a nearly rectangular voltage profile
(Fig. 8.9). This causes a smaller deviation of the polarization state of the light
from the exactly circularly polarized state and thus, a further improvement in the
signal-to-noise ratio. However, the suppression of artifacts caused by optical
rotation is more complicated than in the classic design. The electro-optical
modulator shown in Fig. 8.8 is transparent down to about 190 nm. For shorter
wavelengths, photoelastic modulators are used.
Fig. 8.7. Signal-to-noise ratio of a photon-shot-noise-limited CD measurement as function of the

absorption of the sample (Nlting, 1991).
101
Fig. 8.8. Electro-optical modulator (Pockels cell) for the conversion of linearly into circularly
polarized light (Nlting, 1991). Transparent electrodes are made by using glycerol which
contains a few percent of water. The diameter of the z-cut ammonium dihydrogen phosphate
(ADP) crystal is 40 mm, its thickness is 2 mm, and that of the glycerol layer is about 0.01 mm.
In the absence of a voltage, the optical axis of the ADP crystal is in line to the incident linearly
polarized light beam. Upon application of a voltage, the optical axis splits up into two axes
which are tilted relative to the direction of the incident beam (see right side of the figure). The
optical effect of these two axes is approximately the same as the effect of two optical axes, of
which one is orientated parallel, and the other perpendicular, to the direction of the incident light
beam. Then, the axis that is perpendicular to the incident beam has an orientation of 45o relative
to the plane of linear polarization of the incident beam (see Fig. 7.10).
Fig. 8.9. Nearly rectangular voltage profile applied to the electro-optical modulator (Fig. 8.8) of
the multichannel CD spectrometer (Fig. 8.5).
The time resolution of the real-time multichannel CD spectrometer (Fig. 8.5) is

limited by the frequency of the read-out of the CCD detector and by the photon
shot noise. Unfortunately, even the use of a significantly stronger light source
would not enable a high-precision real-time measurement in the s- or ns-time
scale because available CCD devices and diode arrays can convert only relatively
low photon currents.
102
Further extension of the time resolution is obtained by repetitive methods: The

reaction is repeated many times and the signals are accumulated (Xie and Simon,
1989, 1991). Fig. 8.10 shows a repetitive single wavelength CD spectrometer with
a 50-ps time resolution, limited by the lengths of the LASER pulses. To obtain
one time point for one wavelength, the reaction is carried out several times for the
left and right circular polarizations and the signals are averaged. An optical delay
line serves for a time delay between pump and probe beam: In air, light travels 30
cm in 1 ns.
Fig. 8.10. Repetitive pump and probe beam CD spectroscopy with a time resolution of 50 ps. A
pump beam is generated by frequency-doubling the 1064-nm output of a YAG LASER using a
CD*A crystal. After careful depolarization, carried out with the help of depolarizer and rotating
half-wavelength plate, the pump beam is passed onto the sample where it initiates the desired
reaction kinetics. Relaxation in the sample is monitored at 355 nm with the help of the probe
beam which is generated by a frequency-tripling stage using a KDP crystal. The probe beam is
linearly polarized with a polarizer and polarization modulated with a photoelastic modulator that
is timed to the LASER so that successive light pulses alternately pass through the modulator
when it is in the +/4 and /4 position. After passing through the sample, the transmitted probe
beam is phase-sensitively detected (Xie and Simon, 1989, 1991).
A solution to the problem of the saturation of the detector at high photon

currents has been shown by Kliger et al. (Lewis et al., 1985, 1992; Bjrling et al.,
1996; Chen et al., 1998). The groundbreaking measurement principle (Figs. 8.11
8.13) enables CD spectroscopy in the time scale from picoseconds to seconds (see
also Sect. 5.3.1). In this principle, the difference of intensity between transmitted
left and right polarized light is optically amplified prior to the conversion into an
electrical signal.
103
Fig. 8.11. CD measurement in the nanosecond time scale (Einterz et al., 1985; Lewis et al.,
1985, 1987, 1992; Milder et al., 1988). Linearly polarized light generated by flash lamp and first
polarizer is converted with the help of a strain plate into alternately left and right elliptically
polarized light. In the absence of an optically active sample, a second polarizer absorbs most of
the elliptically polarized light. However, placing a CD-active sample into the beam causes a
significant change of the transmitted intensities (see Fig. 8.12). Compared to commercial
spectrometers (Fig. 7.9), the relative difference of intensities is amplified by a factor of about
2.5 1, where is the phase difference of the strain plate. In order to avoid artifacts, usually
should not be smaller than 0.01 (0.6o) , corresponding to an amplification by a factor of 250.
Fig. 8.12. Optical amplification of the relative transmitted intensity difference in a new type of
fast CD spectrometers (Einterz et al., 1985; Lewis et al., 1985, 1987, 1992; Milder et al., 1988;
Wen et al., 1996). Highly eccentric elliptically instead of circularly polarized light is used. A
polarizer (second polarizer in Fig. 8.11) which is orientated perpendicular to the major axis of
elliptical polarization absorbs most of the light in the absence of an optically active sample
(top). In the presence of an optically active sample, the eccentricity of the light is changed which
gives rise to a significant relative difference of the two light intensities after passing through the
polarizer (bottom).
104
In contrast to the common design variants (Figs. 7.9 and 8.5), in this design,
elliptically polarized light of high eccentricity is used to measure CD, and here the
relative difference of detected light intensities is
Il Ir
1 A ,
Il + Ir log(e)
(8.4)
where Il and Ir are the transmitted intensities for left and right elliptical polarization, respectively, A is the difference of absorptions between left and right
circularly polarized light, and is the retardation of the strain plate (Figs. 8.11 and
8.12; Esquerra et al., 1997). In this way, a certain CD signal, A, and also optical
rotatory dispersion (Shapiro et al., 1995), may be detected with an about 100
times larger relative difference of the light intensity falling onto the electrical
detector, corresponding to a 10.000 times faster detector-saturation- and photonshot-noise-limited measurement. Also this measurement principle has been
combined with multichannel detection (Fig. 8.13).
Fig. 8.13. Nanosecond multichannel CD spectroscopy. Pump (excitation) beam and probe beam
are obtained from two dye LASERs which are synchronously pumped by the second harmonic
of a mode-locked Nd:YAG LASER. Transients are obtained by scanning the delay line and
taking several shots at each step. A change in the difference between the two optical paths in air
by 0.3 mm corresponds to a 1-ps change of the delay between pump and probe beam. Optical
activity in the sample is probed by using alternately left and right elliptically polarized light of
high eccentricity that is generated with the first polarizer and a strain plate. A second polarizer
which is orientated perpendicular to the first polarizer serves for a highly sensitive analysis of
the polarization state of the probe beam after passing through the sample. Artifacts due to
transient birefringence caused by the pump beam are diminished with the help of a depolarizer
(Einterz et al., 1985; Lewis et al., 1985, 1987, 1992; Milder et al., 1988; Wen et al., 1996).
8.3 -value analysis
105
8.3
-value analysis
All of the aforementioned kinetic methods in which rate constants of folding are
obtained may be combined with a protein engineering approach (Goldenberg et
al., 1989; Matouschek et al., 1989, 1990; Matouschek and Fersht, 1991, 1992,
1993, 1995a, b; Fersht et al., 1991, 1992; Otzen et al., 1994; Itzhaki et al., 1995b,
Nlting et al., 1995, 1997a; Nlting and Andert, 2000; Nlting, 1998a, 1999,
2003). This method, the so-called -value analysis, uses the build-up of interaction energies in the protein molecule along the folding pathway as a measure of
structure formation, and is of paramount importance for the high structural
resolution of folding transition states and early folding intermediates (Fig. 8.14).
Fig. 8.14. -value analysis. The case of a protein that folds via a two-state mechanism is illustrated. A mutation causes a change of stability, GFU , of the folded state (F) compared with
the unfolded state (U). G#U is the energy difference between mutant and wild-type in the
transition state, #. The fraction of energy difference, # = G#U /GFU , depends on the
amount of structure that has built up in # at the position of the mutation. = 0 corresponds to no
structure formation; for complete formation of structure at the position of the mutation, = 1. A
set of single mutants, strategically distributed over the molecule is used to map out the structure
of the transition state at the resolution of single amino acid residues. For the assumptions made
for this analysis see (Fersht et al., 1992, 1994). This approach may analogously be applied on
proteins with more complicated energy landscapes which contain intermediates on the reaction
pathway, and on proteins which have residual structure in their unfolded states.
The basic idea is to create a mutation in the protein molecule which causes a
difference in stability between mutant and wild-type, GFU . This difference in
stability builds up in the course of the folding reaction (Figs. 8.14, 8.15). Early in
the folding reaction, when no structure has yet formed at the position of the
mutation, there is no difference in Gibbs free energy, G, between mutant and
106
wild-type. When, in the course of the folding reaction, structure has completely
formed at the position of the mutation, the difference in stability between wildtype and mutant is as large as in the folded state, i.e., G = GFU . Using the
energy as a structurally sensitive probe, and a set of non-disruptive mutants, strategically distributed over the molecule, the structures of folding intermediates can
be mapped out for the same time scale for which kinetic methods are available.
Most importantly, this is the only existent method which can be used to
determine the structures of transition states at the resolution of individual amino
acid residues. The knowledge of the transition state structures is of paramount
importance for understanding the protein folding problem since transition states
represent, per definition, the highest points in the energy landscape along the
folding pathway (see Sect. 4.1). Knowledge of these structures can give insight
into the processes that determine the speed of folding and may be used to
re-engineer the speed of folding in a rational way.
Fig. 8.15. -value analysis with a mutation that deletes an interaction in the protein molecule: In
the course of the folding reaction of wild-type protein, an interaction builds up which causes a
change of energy. In the mutant, this interaction is deleted. By comparing the energies of mutant
and wild-type protein in all individual steps of the folding reaction one can determine at which
stage of the folding reaction this interaction has formed. If at a certain (early) time point the
interaction is not formed yet, there is no difference in energy (and kinetics) between mutant and
wild-type: an interaction which does not exist, makes no physical difference. If, in contrast
structure is completely formed at the position of the mutation at a later time point, then at this
time point the difference in stability between mutant and wild-type is as large as in the folded
state. Thus, the deletion of interactions serves as a probe of structure consolidation; the Gibbs
free energy difference is used as a measure of the degree of structure consolidation (see also
Figs. 8.14, 8.29, and 8.30). A set of mutants serves for mapping out the structural changes along
the folding pathway throughout the whole molecule.
8.3 -value analysis
107
8.3.1
Protein engineering
The -value analysis requires the availability of a set of non-disruptive mutants,
i.e., mutants in which the structure is not significantly altered outside the position
of the mutation (Matouschek et al., 1989, 1990; Matouschek and Fersht, 1991). In
hydrophobic cores of the protein molecule, preferentially the mutations delete only
one or two methyl-groups, for example ValAla, LeuAla, IleAla, AlaGly.
In helices mutations may be used which delete interactions of neighboring amino
acid residues, for example GlnAla, GlnGly, AlaGly, and SerAla.
There is an ongoing progress in the development of simpler and faster methods
for protein engineering (see, e.g., Ausubel et al., 1992). For reasons of brevity,
here only the principles of two very robust methods of site-directed mutagenesis
are explained, the PCR-mediated mutagenesis (PCR = polymerase chain reaction),
and the cassette mutagenesis. Currently, the net-time effort for preparing,
expressing and purifying a set of 60 point mutants may be less than 2 months for
one person when using PCR-mediated mutagenesis.
The initial step for protein genetic engineering is to clone the wild-type protein.
Usually a plasmid, which is a circular piece of DNA (deoxyribonucleic acid) of
typically a few 1000 base pairs in length (Fig. 8.16), is used as a cloning vector.
The gene of the protein can be obtained by isolation from the host or by synthesis.
Insertion of the gene into the vector may similarly be done as in cassette
mutagenesis (Sect. 8.3.1.1). A plasmid which contains the DNA that encodes for
the protein is called a clone. Clones for numerous proteins may be obtained from
several laboratories of large-scale sequencing projects. For clones which are
published in the literature there is usually an honorary obligation for the authors to
make these available to other laboratories for non-commercial research purposes.
Fig. 8.16. Example for a plasmid cloning vector with a promoter and the origin of replication
(ori). DNA which encodes for a protein (the gene) may be inserted into the polylinker site that
contains 6 restriction sites in this example. In order to simplify the selection of transformants,
this plasmid is engineered to carry the gene for antibiotic resistance (amp).
108
Four steps are necessary for the site-directed mutagenesis presented here
(Ausubel et al., 1992; Creighton, 1993):
1. Isolate the plasmid containing the gene of wild-type protein from the host cells
via a miniprep.
2. Mutagenesis reactions (see Sects. 8.3.1.1 and 8.3.1.2).
3. Insert the mutant plasmid into the host cells and let the transformed cells grow.
4. Isolate the plasmid with the mutant gene from the host cells via a miniprep, and
sequence the gene to verify the mutation.
8.3.1.1
Cassette mutagenesis
A fragment of the DNA encoding for the wild-type protein, a so-called cassette, is
cut out with the help of specific restriction enzymes at the unique restriction sites
A and B (Fig. 8.17). The cloning vector obtained is then thoroughly purified by
gel electrophoresis. Two mutagenic synthetic oligonucleotides are hybridized to
form a mutagenic cassette which is inserted into the cloning vector between points
A and B.
Fig. 8.17. Cassette mutagenesis of plasmid DNA (Ausubel et al., 1992).
8.3.1.2
PCR mutagenesis
The availability of restriction sites limits the applicability of the cassette
mutagenesis. Since the purity of synthetic oligonucleotides rapidly decreases with
increasing length, it is impractical to use the cassette mutagenesis if there are no
suitable restriction sites nearby the position of the intended mutation. With the
technique of polymerase chain reaction (PCR), DNA of many 100 base pairs (BP)
length may easily be synthesized with less than 1 error per 10,000 BP on average.
In the PCR reaction (Figs. 8.18, 8.19), the target DNA is amplified in a cyclic
reaction containing three simple steps:
8.3 -value analysis
109
1. Specific hybridization of two oligonucleotides (so-called primers) with complementary, or nearly complementary, sequences on the template DNA.
2. Extension of the primers by a polymerase.
3. Thermal denaturation, i.e., dissociation, of the double stranded DNA, and
return to step 1.
Fig. 8.18. Polymerase chain reaction (PCR). At higher temperature, the double stranded template
DNA dissociates. After re-cooling, single stranded DNA may bind specifically to primers that
have a complementary or nearly complementary sequence. At high concentrations of primers,
compared with template DNA, the binding of template DNA to primers is favored over the
unproductive binding of two strands of template DNA. Now the polymerase may extend the
DNA (dashed lines). The procedure of (a) denaturation, (b) binding of the primers, and (c) extension is repeated, typically 20 to 40 times. This can result in a many million-fold amplification
of a specific DNA fragment. Efficiencies per cycle can exceed 70% (see also Sect. 9.1.2).
Fig. 8.19. Example for the temperature profile in a PCR with 30 cycles. During the extension
reaction, a polymerase catalyzes the growth of new DNA strands in a 53 direction.
Extension time and temperature depend on the type of polymerase used and on the length of the
DNA fragment that is to be amplified. For example, for Thermus aquaticus DNA polymerase,
the optimal temperature for polymerization is about 70 80oC, and the rate of nucleotide
incorporation under optimal reaction conditions may exceed 40 nucleotides per second.
110
Fig. 8.20. A variant

of the PCR-mediated
mutagenesis:
Introduction of a mutation
by two sequential PCR
steps (Ausubel et al.,
1992).
5'... C AGC C C T CCG GC A

...
5' G T T ATA GCC T T C AGC C C T CCG GC G
3'... CAA TAT CGG AAG TCG GGA GGC CG C
3'... CGG AAG TCG GGA GGC CG T
ATA GCG CAT AGC C ...3'

ATA GCG CAT AGC CGT ...3'
TAT CGC GTA TCG GCA ...5'
TAT CGC GTA TCG ...5'
primer 2
primer 3
8.3 -value analysis
111
In the variant of the PCR-mediated mutagenesis illustrated in Fig. 8.20, the

whole gene of a protein is amplified (Ausubel et al., 1992):
1. Two overlapping fragments of the mutant gene for the protein including
flanking regions are prepared in two PCR reactions.
2. The product of these two reactions serves as the template for the third PCR in
which the whole mutant gene is combined and amplified.
3. The product of the third PCR is digested at the positions A and B which are
located outside the gene.
4. For the preparation of the cloning vector, the wild-type (wt) gene is removed
from the wt-plasmid with the help of specific restriction enzymes that digest
the plasmid specifically at two unique restriction sites, A and B. The cloning
vector obtained is thoroughly purified and any undigested DNA removed to
avoid wt-contamination in the later steps.
5. The purified digest of the third PCR is ligated into the vector.
8.3.2
Determination of the protein stability in equilibrium
A convenient method to determine the stability of a protein is the equilibrium
titration with denaturants: The protein is unfolded with denaturants and a
spectroscopic signal, for example, fluorescence or circular dichroism, is recorded
as function of the concentration of denaturant (see, e.g., Privalov, 1979;
Matouschek et al., 1989). From the curves obtained with sigmoidal shape
(Fig. 8.21), one can calculate the Gibbs free energy change upon folding of the
protein, GFU , as described in the following paragraphs.
Fig. 8.21. Urea-induced unfolding of P27A/C40A/C82A barstar at 25oC, monitored by the mean
residual ellipticity at 222 nm, R,222 , and fitted to Eq. 8.13 with F set to 0. K is the equilibrium
constant for unfolding (K = [U]/[F]), that is sometimes also marked as K1.
112
Fig. 8.22. Linear change of the stability of proteins with the concentration of denaturant. The
Gibbs free energy change upon folding, GFU , is negative under conditions that favor folding.
Usually GFU increases, in good approximation, linearly with the denaturant

concentration, [D] (Fig. 8.22; Matouschek et al., 1989):
GFU([D]) = GFU (0) + m [D] ,
(8.5)
where m = (GFU([D])) /[D]. This is because denaturants, for example, urea

and guanidine hydrochloride (GuHCl), are better solvents for the unfolded state
than for the folded state. Urea and GuHCl are supposed to form hydrogen bonds
to the polar groups of the polypeptide backbone, and thereby, to favor the more
solvent-exposed state, i.e., the unfolded state.
Note that for a stable protein in water GFU has a negative sign. According to
Eq. 8.5, GFU in the absence of denaturants is
GFU (0) = m [D]50% ,
(8.6)
where [D]50% is the denaturant concentration at the midpoint of the equilibrium

between folded and unfolded state, i.e., GFU([D]50%) = 0. Thus,
GFU([D]) = m ([D] [D]50%) .
(8.7)
Consider a two-state transition,

U
F,
(8.8)
where K is the equilibrium constant for unfolding, i.e., the ratio of the population
of the unfolded state, [U] , relative to that of the folded state, [F] ,
K = [U]/[F] .
(8.9)
The observed signal, S , is given by the sum of the signal of the folded state times
its population, plus the signal of the unfolded state times its population:
8.3 -value analysis
S = {F + F[D] + (U + U[D])K }/ (1 + K ) ,
113
(8.10)
where F , F , U , and U define linear baselines for the signals of the folded and
unfolded states as function of denaturant concentration, respectively. The equilibrium constant for unfolding, K , is related with the Gibbs free energy change upon
folding, GFU , by Eq. 8.11:
K = exp{GFU /(RT )} ,
(8.11)
where R and T are the universal gas constant and absolute temperature,
respectively. According to Eqs. 8.7 and 8.11, K may be expressed as follows:
K = exp{m([D] [D]50% ) /(RT )} .
(8.12)
Now Eq. 8.12 may be inserted into Eq. 8.10:

S=
F + F [D] + ( U + U [D]) exp{m([D] [D]50% ) /( RT )}

.
1 + exp{m([D] [D]50% ) /( RT )}
(8.13)
The unfolding curve, e.g., in Fig. 8.21, is fitted to Eq. 8.13, and so, m and [D]50%
are obtained which determine the change of Gibbs free energy upon folding in the
absence of denaturants according to Eq. 8.6 (GFU (0) = m [D]50% ) .
For CD detection, occasionally, F is very small and may be set to zero without
significantly altering the calculated GFU (0). This may result in a better stability
of the curve fit (see Sect. 9.4).
Eqs. 8.58.13 are derived for two-state transitions. However, Eqs. 8.58.13
may also be applied on a multi-state transition involving a number of kinetic
intermediates if the population of the intermediates in equilibrium is small.
Further, small systematic errors usually cancel out to a large degree when
comparing only differences of Gibbs free energies between different mutants as
done in the -value analysis.
Fig. 8.23. Principle of differential scanning calorimetry (DSC): Two identical sample cells, one
containing the protein solution, and the other the buffer, are heated at the same constant rate,
typically 0.11 K min1. The difference of the heat uptake between both cells is measured by
monitoring the electrical energy needed for the temperature increase.
114
Fig. 8.24. Differential calorimetric scan of cytochrome P-450cam (Pfeil et al., 1993a).
Fig. 8.25. CD T-scan of

C40A/C82A/P27A barstar in 2 M urea
(bottom), and the difference to a curve fit with
Eq. 8.14 (top).
Alternatively, GFU may be measured using calorimetry, e.g., (Privalov, 1979;

Pfeil, 1981, 1993; Pfeil et al., 1993a, b). In differential scanning calorimetry, the
protein is refolded or unfolded by changing the temperature, and the heat emission
or heat absorption during the reaction is measured (Figs. 8.23, 8.24). Relatively
large quantities of protein are needed for calorimetry, typically 110 mg protein
dissolved in 1 mL buffer, but enthalpy and heat capacity are also obtained.
8.3 -value analysis
115
The third common method for the determination of GFU is to use thermal
scans, i.e., to measure a spectroscopic signal of the protein in the transition region
as function of temperature (e.g., Fig. 8.25). Usually the curve is fitted to
Y (T ) =
K = exp(
YF + mFT + (YU + mUT ) K

1+ K
H F U (1 T / Tg ) + Cp (T Tg T ln(T / Tg ))
RT
(8.14)
) ,
where T is the absolute temperature, R the universal gas constant, Tg the

temperature of half transition, Y(T) is the signal at T, YF and YU are the signals of
folded and unfolded state at T = 0, respectively, mF and mU set the slopes of the
signals of the folded and unfolded state, respectively, HFU is the enthalpy
change upon folding at T = Tg , and Cp the heat capacity change at T = Tg. Then
the Gibbs free energy change upon folding is simply given by
GFU = RT ln (K) .
(8.15)
However, often the quality of the fit is unsatisfactory (Fig. 8.25), partially because
of aggregation at higher temperature. Even more importantly, Eq. 8.14 contains 7
free parameters which are fitted to a curve of relatively simple shape, and so the fit
is often unstable and may lead to wrong results. Sometimes, improved results are
obtained with Eq. 8.16 which takes into account that the baseline of the unfolded
state is not a straight line, but is slightly curved:
Y (T ) =
YF + mF (T To ) + (Y + (YU Y ) exp(mU (T To ))) K

,
1+ K
(8.16)
where To is a conveniently chosen fixed temperature, and here YF and YU are the
signals of folded and unfolded state at To , respectively, and Y is the signal of the
unfolded state at infinite temperature in the case of mU < 0. mU and the, often
small, mF may be determined outside the transition region and inserted into Eq.
8.16. Still, this method is only advisable for proteins with little if any aggregation,
and when a very wide range of temperature is accessible.
8.3.3
Measurement of kinetic rate constants of folding and unfolding
Measurements of kinetic rate constants have to be done with care: The occurrence
of transient aggregation has to be checked spectroscopically and by comparing the
rate constants and relative amplitudes between different concentrations of protein
(see Sects. 8.3.3.4 and 9.2). It should be noted that the concentrationindependence of the observed rate constant is not sufficient to rule out a
monomermultimer mechanism since the observed kinetics might be dominated
by the dissociation reaction which has a concentration-independent rate constant.
116
Fig. 8.26. Observed rate constants as function of the urea concentration for the two folding
transitions, #1 and #2, of the 10 kDa protein barstar (see Sect. 10.5.2).
For the -value analysis both the folding (k1 , k2 , ... ) and unfolding (k1 , k2 , ... )
rate constants have to be known. This information is extracted by decomposition
of the curves of the observed rate constants (k1obs , k2obs , ...) as functions of the
concentrations of denaturant and extrapolation to zero denaturant concentration.
The observed rate constants display a V-shaped behavior where they are
dominated by the folding rate constants at low denaturant concentrations, and by
the unfolding rate constants at high denaturant concentrations (Fig. 8.26).
8.3.3.1
Two-state kinetics
For a two-state transition between the unfolded, U , and folded, F , state,
,
(8.17)
the observed rate constant, kobs , as a function of the concentration of denaturant,

[D], is given by Eq. 8.18 (see Sect. 4.2):
kobs([D]) = k1(0) em1[D] + k1(0) em1[D] ,
(8.18)
where m1 and m1 determine the slopes of the rate constants for folding and
unfolding, respectively, as a function of [D] (see Sect. 8.3.2). Sometimes, kobs is
also called the relaxation constant. In order to obtain k1(0) and k1(0) , Eq. 8.18 is
fitted to the curve of kobs([D]) (Nlting et al., 1995; Nlting et al., 1997a). For the
-value analysis, the ratios of k1(0) and k1(0) between mutant and wild-type are
needed. In the case of only small differences, if any, in m1 and m1 between
mutant and wild-type, using the same m-values for mutant and wild-type may improve the stability of the fit, and small errors may partially cancel each other out.
8.3 -value analysis
117
8.3.3.2
Three-state kinetics
The general solutions for three-state transitions are given in Sect. 4.3. For a
consecutive three-state transition, composed of a slow and a much faster
transition,
,
(8.19)
where U , I , and F designate the unfolded, intermediate, and folded state,

respectively, the observed slow rate constant, k2obs , is approximately
k2obs k2 {K2 + 1/(1 + K1)} = k2 + k2 /(1 + K1)
(8.20)
K2 = k2 /k2
K1 = k1 /k1 ,
where k1 , k1 are the rate constants of the fast folding transition, and k2 , k2 are
the rate constants of the slow folding transition (Nlting et al., 1995, 1997a; see
also Sect. 4.3). Analogous to the exact solution for a two-state transition, the
observed fast rate constant k1obs is approximately the sum of the folding and
unfolding rate constants (Nlting et al., 1995, 1997a) ,
k1obs k1 + k1 = k1 (1 + K1) .
(8.21)
Thus, Eq. 8.18 applies analogously also for the fast transition of a three-state
folding mechanism, but Eq. 8.20 has to be applied for the extrapolation of the
slow rate constants to zero concentration of denaturant.
8.3.3.3
Kinetic implications of the occurrence of intermediates
According to Eq. 8.20, k2obs is always smaller than it would be in the absence of
an intermediate, namely k2 + k2 (see Eq. 8.18), and thus, the occurrence of an
early intermediate retards folding. This retardation of folding is often the source
of the roll-over effect (Fig. 8.27) which is a deviation of the measured curve for
the observed rate constant as function of the concentration of denaturant from the
shape expected for a two-state transition. However, depending on the stabilities of
the species involved as function of the denaturant concentration, the roll-over
effect may be quite small and easily remain undetected. On the other hand, from
the presence of the roll-over effect one cannot necessarily conclude the existence
of a folding intermediate. Transient aggregation may cause a similar effect.
There are two interesting implications of the occurrence of unfolding intermediates, i.e., of intermediates that are located on the reaction coordinate between
118
Fig. 8.27. Roll-over effect due to the presence of an early folding intermediate: At low
denaturant concentration, the observed rate constant is slowed down relative to that expected for
a two-state transition.
the folded and main transition states (see Sect. 5.2.1.3): First, the structure of the
main transition state cannot be very close to the structure of the folded state, and
second, similar to (re)folding intermediates, the observed rate constant is
decelerated: For a three-state transition,
,
(8.22)
here composed of a slow transition between unfolded state, U , and intermediate

state, I , and a much faster transition between I and folded state, F , the observed
slow rate constant, k1obs , is approximately
k1obs k1 {K1 + 1 / (1 + K2)} = k1 + k1 / (1 + K2)
(8.23)
K1 = k1/k1
K2 = k2/k2 ,
where k1 , k1 are the rate constants of the slow transition, and k2 , k2 are the rate
constants of the fast transition (derived from Eq. 8.20). For a two-state transition,
the observed slow rate constant is given by the sum of the folding and unfolding
rate constants, k1 + k1 , which is always larger than the right-hand term in
Eq. 8.23. Thus, both (re)folding and unfolding intermediates decrease the
observed rate constant for the folding reaction; both retard folding (Nlting,
1996).
8.3 -value analysis
119
8.3.3.4
Discrimination between folding and association events
In experiments with small-amplitude T-jumps or small changes of denaturant
concentration, the observed rate constant, kobs , for a monomerdimer equilibrium,
(8.24)
is
kobs = 4[A] k1 + k1 ,
(8.25)
where 4[A] k1 and k1 are the rate constants of association and dissociation,
respectively (k1 is the second-order rate constant of association).
Fig. 8.28. Relative amplitude, /( Ao) , (1000) and relative observed rate constant, kobs /k1 , of
the relaxation as function of the relative concentration, Ao /K , for T-jumping of a monomer
dimer equilibrium (Eq. 8.24), in which the forward and backward reactions have the same transition state. K = k1/k1 and Ao is the total protein concentration. At low protein concentrations,
corresponding to a small Ao /K , the observed rate constant is concentration-independent and the
relative amplitude of the relaxation increases with increasing concentration. At high protein
concentrations, corresponding to a large Ao /K , the observed rate constant increases and the
relative amplitude decreases with increasing protein concentration (see also Sect. 4.6).
At high protein concentrations, i.e., under conditions where the major fraction in
equilibrium is A2 , the observed rate constant is dominated by the concentrationdependent association rate constant (Fig. 8.28). At low concentrations, i.e., under
conditions where A is the major fraction in equilibrium, the observed rate constant
is dominated by the concentration-independent dissociation rate constant. In this
case, the relative amplitude of the relaxation is concentration-dependent:
A T-jump changes the constant K = k1/k1 = [A]2/[A2] approximately to (1 + ) K =
([A] + 2 )2/([A2] ), where and are small perturbations. Thus,
120
[A][A2]/([A] + 4[A2]) [A2] .
(8.26)
So, the relative amplitude is proportional to [A2]/Ao , in which Ao = [A] + 2 [A2]

[A]. Doubling of Ao at low concentrations leads to an approximately 4fold
increase of [A2] , and thus, to a doubling of the relative amplitude of the observed
kinetics. Thus, dimerization or decay of dimers can be ruled out as the origin of a
relaxation when both the rate constant and the relative amplitude do not change
over a wide range of concentration. Also for more complicated reactions
involving association/dissociation events, a concentration-dependence of the
relative amplitude of relaxation is predicted, which may be weaker, however.
8.3.4
Calculation and interpretation of -values
8.3.4.1
Two-state transition
For a two-state transition (Fig. 8.29; Fersht, 1985; Matouschek et al., 1989), the
Gibbs free energy difference, G#U , of the transition state (#) between mutant
and wild-type is (the unfolded state is taken as the reference state):
G# U = RT ln(
k1,mt
),
k1,wt
(8.27)
where k1,mt and k1,wt are the rate constants for folding of mutant and wild-type,
respectively (see Sect. 4.1).
In the -value analysis, G#U is compared with the difference in the Gibbs
free energy changes upon folding between mutant and wild-type GFU , i.e., with
the difference in the stabilities of the folded states between mutant and wild-type,
GF U = GF U,mt GF U,wt
# =
G# U ,
GF U
(8.28)
(8.29)
where GFU,mt and GFU,wt are the changes in Gibbs free energy upon folding for
mutant and wild-type, respectively.
Because
GF U = G# U + GF # ,
(8.30)
analogously one can determine the -value of # by using the rate constants for
unfolding:
GF #
(8.31)
# = 1
GFU
GF # = RT ln(
k1,mt
).
k1,wt
(8.32)
8.3 -value analysis
121
If the -values, # , of the transition state determined by Eqs. 8.29 and 8.31 do
not match, the protein under consideration does not fold via a two-state
mechanism, provided artifacts of measurement are excluded.
Fig. 8.29. -value analysis for a two-state transition. In the folded state, mutant and wild-type
protein differ in stability by GFU , which may be measured by equilibrium titration of the two
proteins with denaturants (see Sect. 8.3.2). This energy difference builds up in the course of the
folding reaction (the unfolded state is taken as the reference state). By using the kinetic rate
constants for wild-type and mutant, one can find out whether an energy difference already exists
in the transition state, #. The ratio, # , of the Gibbs free energy difference between mutant and
wild-type in #, G#U , relative to GFU is taken as a measure of structure formation at the
position of the mutation in # (see also Figs. 8.14 and 8.15).
8.3.4.2
Multi-state transition
For every step in the folding reaction which leads from the state i to the state
j = i + 1, the increase of the -value is (Nlting et al., 1995, 1997a)
j i =
G j i
GFU
(8.33)
where Gji is the difference in the Gibbs free energy changes for this reaction
step between mutant (mt) and wild-type (wt). Gji is calculated from the
122
folding rate constants, ki,mt and ki,wt , using Eq. 8.34 for a reaction step that is
leading from a less folded state to a transition state, and from the unfolding rate
constants, ki,mt and ki,wt , using Eq. 8.35 for a reaction step leading from a
transition state to a more folded state:
k
(8.34)
G ji = RT ln( i,mt )
ki,wt
G ji = RT ln(
ki,mt
) .
ki,wt
(8.35)
The -value of the state n is given by summing the increments of the -values,
n
n = j i ,
(8.36)
j=2
where i denotes the state previous to the state j , and the unfolded state has the
number 1.
Fig. 8.30. -value analysis of a three-state transition. For each of the 4 steps of the reaction,
#1, #1
I, I
#2, and #2
F, the kinetic rate constants are measured for mutant
U
(mt) and wild-type (wt) protein. Then, using the kinetic rate constants and the total difference in
Gibbs free energy changes upon folding between mutant and wild-type protein, GFU
(measured in equilibrium), the changes of -value for every step are calculated. The total of the
4 increments of -value must be 1 for all mutants if the protein folds according to the assumed
on-pathway three-state mechanism (see text). A lower total of the -value increments may be an
indication for the occurrence of a further folding event of significant free energy contribution.
8.3 -value analysis
123
For example, of the intermediate state, I, of a three-state folding reaction

according to Fig. 8.30 is
I =
G#1U + GI #1 GIU
=
GF U
GFU
G#1U = RT ln(
k1,mt
)
k1,wt
GI #1 = RT ln(
(8.37)
k1,mt
),
k1,wt
and for #2 we have
#2 =
G#1U + GI #1 + G# 2 I G# 2 U
GF # 2
=
= 1
GF U
GF U
GF U
G#2 I = RT ln(
k2,mt
)
k2,wt
GF # 2 = RT ln(
(8.38)
k2,mt
).
k 2,wt
If the correct reaction mechanism was assumed, the -value for the folded
state, F , determined by Eq. 8.36 must be 1: The total of the energy differences
obtained from the kinetic experiments (Gji) must be equal to the energy
difference obtained from equilibrium measurements (GFU), unless a transition
was left out in the analysis. Like other kineticthermodynamic methods, the
-value analysis cannot prove reaction mechanisms, but can only rule out
alternatives. However, if for all mutants in a large set of mutants, the total of the
-value increments (ji) for the whole reaction (= F) is 1, and the mutations
are well distributed over the protein molecule, then it is very likely that no global
folding event with significant energy contribution is missing in the analysis.
8.3.4.3
Residual structure in the unfolded state
In the previous sections the unfolded state was taken as an unstructured reference
state. However, often unfolded states of proteins differ from the random coil state
and may not be considered as fully unfolded (Nlting et al., 1997b). It can be
shown that in case of residual structure in the (partly) unfolded state, the -values
calculated as shown in the previous sections represent the degree of formation of
structure on top of the amount of residual structure in the unfolded state. In
particular = 0 indicates the same degree of structure as in the unfolded state and
= 1 is still obtained for complete structure consolidation in the state under
consideration unless there are non-native interactions.
9 Experimental problems of the kinetic and

structural resolution of reactions that
involve proteins
9.1
Protein expression problems
9.1.1
Low expression level
Low protein expression is often one of the most severe problems, and it may take
a significant effort to find improvements. Here a number of hints is given which
can help in many cases to advance the process of optimization (Table 9.1).
General. The level of expression of recombinant proteins usually decreases
with the number of cell cycles. A main reason for this effect is that expression of
non-native proteins, and also overexpression of native proteins, usually causes a
negative selection pressure, i.e., cells with lower expression levels tend to outgrow
the others. Thus, one should always avoid an unnecessarily large number of cell
cycles and strong selection pressures. It may be safer to use freshly transformed
cells for the cell culture, but in many cases not-too-old glycerol stocks, made after
a fresh transformation, are sufficient. Often, colonies of freshly transformed cells
may directly be transferred from the media plate into a flask with several liters of
pre-warmed medium. By keeping the cells at the optimal temperature of growth
without interruption during the period from transformation to harvest, one does
not only speed up the cell growth, but may occasionally improve the expression
level as well. The selection pressure towards lower expression is significantly
reduced in inducible expression systems with a low basal level of expression. In
these systems expression is usually induced only in the medium or at later stages
of the exponential growth phase of the cell culture, i.e., after a large number of
cells has grown up under conditions of very little negative selection pressure.
Protein is toxic to the host cells. Use an inducible expression system with a low
basal level of expression, and induce expression in a late stage of the growth phase
of the host cell culture. If this does not help, for example, because the protein is
highly toxic to the host cells, change the host. For different expression systems
see (Ausubel et al., 1992; Coligan et al., 1996).
Expression level differs between different cells. Colonies of freshly transformed cells sometimes differ in size or shape. Different types of colonies may
display significant differences in expression level. Thus, it is a good idea to
analyze the expression level of the different transformant colonies. Occasionally
126
9 Experimental problems of the kinetic and structural resolution ...
it is observed that the slowly growing colonies, which are visible only after a
significantly longer duration of growth, display the highest expression level.
Occasionally, this observation is made even for inducible expression systems.
Cells lose the plasmid. Reduce the number of cell growth cycles as described in
the General section or insert the gene of the protein into the genomic DNA of the
host cells.
Plasmid is damaged. The host cell culture may evade the pressure of high
expression by selecting for a damaged plasmid with lower expression level:
Subclone the gene into a new plasmid.
Protein is expressed in form of inclusion bodies. In the case of a low protein
concentration in the supernatant that forms upon centrifugation of the lysed cells,
check whether the pellet contains the recombinant protein. If the protein is
expressed in the form of inclusion bodies, dissolve the inclusion bodies with a
denaturant, e.g., urea, and slowly refold the protein by removing the denaturant,
e.g., by dialysis. To minimize aggregation and misfolding, the refolding of the
protein should proceed in a sufficient volume of solution. A method for increasing
the yield of correctly folded protein is described in Sect. 9.3 (Fig. 9.9).
Lysozyme causes co-aggregation. Under some experimental conditions, lysozyme is prone to aggregation, especially at high temperature and during thawing
after flash-freezing. When using lysozyme for the harvest, co-aggregation with
the recombinant protein may cause losses. The application of ultrasonication,
instead of lysozyme, for cell lysis may improve the protein preparation.
Table 9.1. Clues which might help to solve expression problems.
Problem
Possible solution
Protein is toxic to the host cells.
Use a different, inducible expression system with a

low basal level of expression, and induce expression
in a late stage of the cell growth phase.
Expression level differs between

different cells.
After transformation, analyze the expression level of

the different types of colonies, including that of the
slowly growing ones.
Cells lose the plasmid.

Plasmid is damaged.
Use only freshly transformed cells, or insert the gene

into the genomic DNA of the host cells.
Subclone the gene into a new plasmid.
Protein is expressed in form of

inclusion bodies.
Lysozyme causes co-aggregation.
Dissolve the inclusion bodies with denaturant, then

slowly refold the protein.
Use ultrasonication for cell lysis.
9.1.2
Expression errors
There are two main error sources which lead to the expression of a wrong recombinant protein (Table 9.2): 1. A wrong gene is used, for example, because of (a)
9.1 Protein expression problems
127
wild-type contamination in preparations of mutants, (b) a polymerization error

during DNA amplification, or (c) the use of impure oligonucleotides. 2. Incorrect
post-translational modifications of the protein (for common types of alterations of
the protein after assembly of the polypeptide chain on the ribosome see below).
To rule out both error sources it is obviously not sufficient to sequence the gene.
Wild-type contamination. The frequency of this problem depends on the
method of mutagenesis and on the care taken to ensure clean preparations of
DNA. For example, in the PCR-mediated mutagenesis presented in Sect. 8.3.1.2,
this problem is generally absent if one ensures (a) an excellent separation of
digested and undigested plasmid in the preparation of the cloning vector, for
example, by using a very clean gel chromatography (applies also for cassette
mutagenesis), and (b) a clean separation of PCR product and template DNA.
Point mutations. Methods of mutagenesis which involve the use of
polymerases may introduce additional point mutations. Error rates per nucleotide
per extension, PP , of most of the thermostable DNA polymerases are between
106 and 104 under optimal reaction conditions, corresponding to a less than 3%
chance of one or several errors occurring in one extension reaction for a gene of
300 base pairs in length. In the PCR, the error grows with the number of extension reactions, z. The chance, P, that a randomly selected DNA molecule of the
reaction product is error-free is
P (1 PP) n (z0.5) ,
(9.1)
where n is the number of base pairs of the amplified gene. Thus, in the case of a
large number of cycles of the PCR, and for the amplification of long DNA
fragments, one should use polymerases with low error rates. Polymerases with
low error rates typically have lower incorporation rates and a proofreading ability
where the polymerase can correct misincorporated nucleotides in the growing
strand of DNA being synthesized.
Another common source of undesired point mutations is the use of impure
oligonucleotides for synthetic cassettes or as primers in the PCR. The error rates
of synthetic oligonucleotides rapidly increase with the length of the oligonucleotides. There are various techniques for cleaning primers by using chromatography
(see, e.g., Ausubel et al., 1992). A simple time-saving method is to use for
hybridization an annealing temperature just low enough to enable binding of
primers with the correct sequence, but too high to permit a significant binding of
primers with additional point mutations.
Obviously, the adjustment of the annealing temperature is very important:
Rapid cooling to a too low temperature usually increases the fraction of unspecific
binding of contaminations; at too high annealing temperatures, the hybridization
of the primers is less efficient, and undesired side-reactions may be more favored.
In PCR and cassette mutagenesis, and many other DNA reactions, it is of great
importance to use oligonucleotides of sufficient length. Oligonucleotides that are
too short will bind poorly, and thus unspecific binding and binding of contaminations will be more favored.
128
Point mutations are usually detected by DNA sequencing (Ausubel et al.,

1992). High-resolution mass-spectrometry may provide a high probability for the
absence of undesired mutations in a protein: Among the 380 possibilities for the
mutation of one of the 20 natural amino acid residues to another, there are only
four cases which introduce a mass change of less than 0.9 dalton under
measurement conditions in which the sidechains are neutralized (see Sect. 2.1):
LeuIle, IleLeu, GlnLys, and LysGln.
Post-translational modifications. Problems connected with post-translational
modifications are relatively common and obviously cannot be detected by
sequencing the gene, but rather by mass-spectrometry, and occasionally also by
chromatography. Examples for post-translational modifications are:
1. Partial digestion of the protein by proteases.
2. Attachments of additional amino acid residues to the polypeptide chain, for
example, of a methionine to the N-terminus.
3. Covalent attachment of prosthetic groups. An example is horse heart cytochrome c in which the heme group is attached to the protein via two thioether
bonds involving two cysteine residues.
4. Glycosylation. Carbohydrates are attached to the protein, usually via hydroxyl
groups of threonine or serine or via the amine group of the asparagine or lysine
sidechain.
5. N-terminal acetylation. Acetylation of the -amino terminus usually blocks
the degradation of the protein by the energy-dependent breakdown pathway in
eucaryotes.
6. Phosphorylation. The attachment of a phosphate group to the hydroxyl group
of threonine, serine or tyrosine, to the -amino group of lysine, or to the ring
nitrogen of histidine usually serves as a message signal.
7. Adenylation of the hydroxyl group of a tyrosine.
8. Methylation of the lysine amino group.
9. Hydroxylation of lysine or proline.
10. Disulfide bond formation.
Table 9.2. Methods for the detection of expression errors.
Problem connected with:
Method of detection
Wild-type contamination
Sequencing, or restriction digest if the desired

mutation introduces or removes an additional
restriction site
Point mutation
Sequencing, and in most cases, but not always,

detectable by mass-spectrometrya
Post-translational modifications
Mass-spectrometry for almost all casesa
The mass assignment accuracy of electrospray ionization mass spectrometers is usually in the
range of 0.001 0.05% for a protein of up to 100 kDa molecular weight. Note, that in exceptional
cases, two errors that occur simultaneously have no detectable effect on the mass.
9.2 Aggregation
129
9.2
Aggregation
9.2.1
Detection
Aggregation of proteins is not always connected with flocculation and
precipitation that is observable with the naked eye. To rule out the occurrence of
aggregation in a protein sample, more sensitive methods must be used. In kinetic
folding experiments transient aggregation might occur, i.e., aggregates might
form only during the folding reaction and vanish thereafter. Transient aggregates
may easily be confused with folding intermediates.
Absorption spectroscopy. Aggregation causes increased light scattering (Fig.
9.1). This scattering gives rise to an extinction at wavelengths where the protein
usually does not absorb. For example, many proteins do not have chromophores
for wavelengths around 400 nm. The extinction caused by light scattering rapidly
increases with decreasing wavelength which causes a characteristic shape of the
spectrum (Fig. 9.1). For proteins which absorb around 400 nm, often a different
wavelength suitable for the observation of scattering can be found.
Dynamic light scattering. The scattering of small aggregates, for example
dimers or trimers, is often too small to be easily detectable by absorption
spectroscopy. Dynamic light scattering (Fig. 9.2) is generally more sensitive to
changes of the association state of macromolecules. In this method the rotation
correlation time of the molecular tumbling is used to determine the molecular
weight of the particles. For proteins of >10 kDa, typically quantities of 0.01 1 mg
protein dissolved in 10 100 L buffer are needed.
Fig. 9.1. Example for the detection of protein aggregation by using absorption spectroscopy.
130
Fig. 9.2. Principle of dynamic light scattering. Light is scattered from a solution of
macromolecules. The light intensity is measured as function of time and the autocorrelation
function of the intensity fluctuations is calculated. Fluctuations of the scattered light intensity are
related to the molecular weight of the scattering molecules. The angular dependence of
scattering contains information about the shape of the macromolecules. Dynamic light scattering
is an exquisitely sensitive method for the detection of aggregation.
Ultracentrifugation. Upon ultracentrifugation, the protein molecules move

towards the bottom of the centrifugation flask (Fig. 9.3). Because the Brownian
motion acts against this movement, the protein concentration close to the bottom
will approach an equilibrium. From the concentration profile in equilibrium,
measured by absorption spectroscopy, the concentrations of monomers, dimers,
and multimers are determined.
Fig. 9.3. Detection of aggregation of

proteins by using ultracentrifugation.
The sedimentation equilibrium is
monitored at different positions of the
centrifugation flask by using absorption
spectroscopy. The light beam (not
shown) is perpendicular to the force of
acceleration. Larger particles tend to
move closer to the bottom of the flask.
Kinetic analysis. In order to prove that kinetic folding experiments are not
affected by aggregation, it is not sufficient to show that the observed rate constant
is concentration-independent over a few orders of magnitude of concentration.
For example, the observed relaxation may be caused by, or affected by, the rapid
dissociation of slowly forming aggregates, which is generally a concentrationindependent process. To provide further indication for the absence of these
complications, it is useful to measure also the relative amplitude of the kinetic
event as function of protein concentration (see Sect. 8.3.3.4).
9.2 Aggregation
131
9.2.2
Avoidance of aggregation
In general, the magnitude of aggregation and structure of aggregates strongly
depends on a number of chemical and physical factors:
Protein concentration. Obviously, the magnitude of protein association events
depends on the protein concentration, and thus, may usually be diminished by
lowering the protein concentration. For a simple monomerdimer equilibrium:
(9.2)
the equilibrium concentrations of monomer, [F]eq , and dimer, [F2]eq , are given by
Eq. 9.3 (see Fig. 9.4 and Sect. 4.5):
[F]eq =
2
k 1
k
[F ]k
+ tot 1 1
16k12
2k1
4k1
(9.3)
[F 2 ]eq = 0.5 ( [F tot ] [F]eq )

where [Ftot] is the total protein concentration, expressed in units of monomers.
Practically, the concentration of dimers becomes significant for about (Fig. 9.4):
[Ftot]k1/k1 > 0.1
(9.4)
Fig. 9.4. Concentration of monomers, [F]eq , and dimers, [F2]eq , as function of the total protein
F2 . k1 and k1 are the forward
concentration, [Ftot] , in a monomerdimer equilibrium, 2 F
(on-) and backward (off-) rate constant, respectively (see Eq. 9.2).
132
Fig. 9.5. Example for the protein aggregation as function of urea concentration. Moderate and
very high concentrations of urea may often dissolve aggregates. At urea concentrations where
the protein starts to unfold, sometimes increased aggregation is observed.
Denaturant concentration. Aggregation may often be reduced by adding small

amounts of denaturants, for example urea. When further increasing the concentration of denaturant, most proteins will partially unfold which occasionally leads
to increased aggregation (Fig. 9.5). At very high concentrations of denaturant
where the protein is mainly unfolded, usually the aggregation decreases again.
Co-solvents. Co-solvents, such as ethanol, propanol, trifluoroethanol, often
increase aggregation levels, mainly because they modify the hydration shell of the
protein molecule and tend to enhance the intermolecular hydrophobic interaction,
which is a main driving force of many aggregation processes.
Fig. 9.6. Solubility of carbon monoxide horse hemoglobin in different salt solutions as function
of the ionic strength (Cohn and Edsall, 1942). Similar to observations for many other proteins,
addition of moderate concentrations of these salts causes increased solubility of hemoglobin.
However, some salts decrease the protein solubility at very high concentrations of salt (salting
out effect).
9.3 Misfolding
133
Surfactants. Addition of a variety of surfactants, for example, simple alkanols

and micelle-forming surfactants, can often reduce aggregation and substantially
increase renaturation yields of unfolded proteins (see Wetlaufer and Xie, 1995).
Salt-concentration. Aggregation is not only affected by hydrophobic interactions but also by electrostatic forces. For example, occasionally aggregation may
be reduced by engineering charge mutations which cause a stronger charge repulsion between the protein molecules at low salt concentration. In those cases for
which the charge repulsion between the protein molecules is important for the
suppression of aggregation, the use of denaturants with high ionic strength (e.g.,
guanidine hydrochloride) in experiments for the study of folding is problematical.
On the other hand, the solubility of many proteins decreases at very low ionic
strength (Fig. 9.6).
pH. At the isoelectric point (pH = pI) the net charge of the protein is zero. In
the vicinity of this point, often a particularly high tendency for aggregation is
found (Fig. 9.7).
Temperature. Below 100oC, temperature elevation leads to a strengthening of
hydrophobic interactions (Sect. 3.4) which often causes increased aggregation.
Chaperonins. Chaperonins, for example GroEL and GroES (Goloubinoff et al.,
1989a, b; Martin et al., 1993, 1994; Hunt et al., 1996), are proteins and protein
complexes which facilitate folding and can often reduce misfolding and aggregation in vivo and in vitro.
Fig. 9.7. Example for the course of aggregation as function of pH for a protein which is
significantly charged at pH > pI. Vastly different curves are observed for different proteins.
9.3
Misfolding
Usually, the energy landscape of a protein under folding conditions displays a
global minimum, which is occupied by the folded conformation, and local minima
which may trap the protein into misfolded conformations if the folding reaction
proceeds under unfavorable conditions (Figs. 1.3 and 9.8). Only in very rare cases
is the native conformation not at the global energy minimum (Sohl et al., 1998).
134
Fig. 9.8. Example for misfolding. Misfolding is the occurrence of an incorrectly folded protein
conformation which has a stability that is usually lower than the stability of the folded (native)
conformation, but whose rate of transition to the correctly folded conformation is very slow. The
folded and misfolded conformations in this example do not necessarily differ significantly in
stability or surface burial.
Fig. 9.9. Reduction of misfolding and aggregation by using unfolding-refolding cycles. Some
preparations of overexpressed recombinant proteins yield low fractions of correctly folded
protein. In many cases the property of a misfolded protein usually to unfold faster than the
correctly folded protein may be used for improved reconstitution. Incubating the protein for a
short period of time under unfolding conditions causes unfolding of a larger percentage of the
misfolded protein than of the correctly folded protein. In the subsequent refolding reaction, a
fraction of the unfolded (previously mainly misfolded) protein folds into the native conformation. Correctly folded protein accumulates in these unfoldingrefolding cycles.
A simple method to decrease the fraction of misfolded protein in preparations of

expensive proteins is shown in Fig. 9.9. This method makes use of the property of
most misfolded proteins to unfold faster than the correctly folded protein:
9.4 Unstable curve fit
135
1. The protein is incubated for a certain period of time under conditions that favor
unfolding. This period is long enough to enable unfolding of a large fraction of
misfolded and aggregated protein, but too short to enable a significant degree
of unfolding of correctly folded protein.
2. To refold the protein, the denaturant concentration is lowered, for example, by
mixing with a buffer.
3. Because again only a fraction refolds correctly, steps 1 and 2 are repeated several times. Prior to return to step 1, aggregates are collected, e.g., by centrifugation, and the soluble protein is concentrated by ultra-filtration.
9.4
Unstable curve fit
Many of the methods presented rely on fitting equations to experimental data.
However, one may easily generate huge errors if equations with too many free
parameters are fitted to curves of simple shape. This is shown for the example of
the determination of the Gibbs free energy change upon folding (Fig. 9.10): Table
9.3 displays the results for curve fits with two different equations: Fit 1 is made
according to Eq. 8.13 (Sect. 8.3.2) which contains slopes for the folded and
unfolded state. Fit 2 is made with the same equation, but using no slope for the
folded state (F = 0). Data set 1 contains all the points shown in Fig. 9.10. In data
set 2 the point at zero concentration of urea, indicated by a square in Fig. 9.10, is
left out. The two fits with slope for the folded state for the complete and
incomplete set of data, respectively, are shown in Fig. 9.10. Even though the two
fits are almost superimposable, the results are quite different. The use of a slope
for the folded state makes the fit quite unstable: The removal of a single data point
changes the result by 0.8 kJ mol1 (0.2 kcal mol1). Better stability is obtained
when using fewer free parameters. In this case the slope for the folded state is
very small and might be neglected or be fixed (Fit 2 in Table 9.3). This changes
the result for the Gibbs free energy, but for the -value analysis (Sect. 8.3) only
differences in Gibbs free energies between mutants and wild-type protein are
needed. When using the same type of equation for all curve fits, usually the
systematic errors for the energies will partially cancel each other out.
Table 9.3. Gibbs free energy change upon folding, GFU , obtained for two different fits and
two different data sets (see text and Fig. 9.10). (1 kJ mol1 = 0.24 kcal mol1)
Fit 1 (with slope for
the folded state,
shown in Fig. 9.10)
Fit 2 (with no slope

for the folded state)
Data set 1 (all data)
14.7 kJ mol1
13.9 kJ mol1
Data set 2 (without point

at zero urea concentration)
15.5 kJ mol1
13.6 kJ mol1
136
Fig. 9.10. Determination of the Gibbs free energy change upon folding from an equilibrium
denaturant titration curve of a protein by using an equation (Eq. 8.13 in Sect. 8.3.2) that contains
a large number of free parameters. The two, almost superimposable, fits for the complete and
incomplete data set, respectively, are shown. In the incomplete data set one point is removed at
zero concentration of denaturant. Quite different Gibbs free energy changes upon folding,
GFU , are obtained for the two different data sets when using too many free parameters for the
curve fit. A more stable fit is obtained in this example when using no slope or a fixed slope for
the signal of the folded state (see Table 9.3). When comparing energy differences between
mutants and wild-type, the introduced systematic errors partially cancel each other out.
10 The folding pathway of a protein (barstar)

at the resolution of individual residues
from microseconds to seconds
10.1
Introduction
This chapter reports the first resolution of a folding pathway of a protein (barstar,
the 10 kDa polypeptide inhibitor of the ribonuclease barnase, see Fig. 10.1) from
a well-characterized unfolded state (Sect. 10.4) at the level of individual amino
acid residues on a microsecond to second time scale. The presentation, which is
largely based on several articles by the author and coworkers (Nlting et al., 1995,
1997a, b; Nlting, 1996, 1998a, b, 1999; Nlting and Andert, 2000), is intended to
illustrate the method of -value analysis (see Sect. 8.3) in detail and in a more
applied way than in the previous chapters.
A number of studies (Nlting et al., 1995, 1997a, b; Nlting, 1998a) has shown
that a folding intermediate with marginal stability, 2.5 kJ mol1, is formed within
a few hundred s after rapidly raising the temperature of the (partially) coldunfolded solution of barstar from 2 to 10oC (Sect. 10.5). -value analysis (Sects.
10.6 10.8) shows that the first and fourth helix become substantially consolidated
as the intermediate is formed, stabilized by long-range interactions. A native-like
structure is then formed within about 100 ms as the whole structure consolidates.
The overall folding pathway fits the nucleationcondensation model in which
structure is (partly) formed in a diffuse nucleus and then becomes consolidated as
further structure condenses around the nucleus or as several modules of structure
dock (Nlting et al., 1997a). The structure of the diffuse nucleus and its growth in
three stages, 500 s, 1 ms, and 100 ms after initiation of the folding reaction, is
mapped out by correlating -values for a set of single mutants with inter-residue
contact charts (Sects. 10.7, 10.8; Nlting, 1998a, 1999). The folding nucleus
detected in barstar initially comprises mainly secondary and tertiary interactions
of strand1 , loop1 , and helix1 (Nlting et al., 1997a; Nlting, 1998a). The
discovered highly anisotropic folding behavior may explain the high speed of the
folding reaction, compared with the speed of a random sampling. In agreement
with the nucleationcondensation model of folding (Sect. 10.11), the non-uniform
structure consolidation is most pronounced in the early stages of folding. The late
folding events of barstar are characterized by a propagation of structure
consolidation from the N- and C-termini towards amino acid residues located
close to the center of the polypeptide chain (Nlting, 1998a).
138
10 The folding pathway of a protein ...
Fig. 10.1. Folded structure of barstar, drawn with the program MOLSCRIPT (Kraulis, 1991) and
Raster3D (Bacon and Anderson, 1988; Merritt and Murphy, 1994).
10.2
Materials and methods
Protein expression and purification. The pseudo-wild-type barstar used in the
studies which are presented in Sects. 10.310.11 is C40A/C82A/P27A barstar
containing no cysteines and only one proline residue that is at position 48 (Nlting
et al., 1995, 1997a, b; Nlting, 1998a). Mutants were engineered by cassette
mutagenesis or by PCR mediated mutagenesis (see Sect. 8.3.1 and Ausubel et al.,
1992). Screening of mutant plasmids (pML2bs), expressed in Escherichia coli
BL21(DE3)(pLysE), was simplified by introducing a silent mutation for a
restriction site. For expression, freshly transformed cell colonies were transferred
directly into a 2-L flask with 800 mL 2-TY medium with 200 M ampicillin and
incubated at 30oC for 30 h with addition of 4 mM isopropyl-1-thio--galactoside
(IPTG) at 10 h and 20 h each. Intense pulsed ultrasonication of the ice-cold cellsuspension was used instead of lysozyme for cell lysis, resulting in >95% purity
of barstar mutants in the extracted inclusion bodies before ion-exchange
chromatography. Electrospray mass spectrometry confirmed the size of the
mutants within 0.8 dalton. Mass spectrometry and N-terminal sequencing
showed that the N-terminal methionine was not cleaved in these highly expressed
barstar mutants, including pseudo-wild-type. The expression level was typically
100 mg L1, and the protein was concentrated to 12 mM in water without any
precipitation occurring (Nlting et al., 1995, 1997a, b; Nlting, 1998a).
10.2 Materials and methods
139
Circular dichroism (CD) spectra. CD was measured using a JASCO (Easton,

MD) model J-720 spectropolarimeter that was interfaced with a computercontrolled Neslab (Newington, NH) RTE-111 waterbath. CD calibration was
performed using (1S)-(+)-10-camphor-sulfonic acid (Aldrich, Milwaukee, WI)
with a molar ellipticity of 2.36 L mol1 cm1 at 290.5 nm and a molar extinction
coefficient of 34.5 L mol1 cm1 at 285 nm. CD spectra were obtained with a
spectral resolution of 2 nm, an optical pathlength of 1 cm and a protein
concentration of about 5 M. For each spectrum, 16 scans of about 1 minute
duration were accumulated and averaged. Barstar concentrations were determined
using an extinction coefficient at 280 nm of 22,690 L mol1 cm1 (Lubienski et al.,
1994; Nlting et al., 1995).
CD temperature scans. CD temperature scans were done with a heating rate of
50oC h1. The absolute error of temperature is 1.5oC and the relative error of
temperature between curves in one diagram is less than 0.3oC. The absolute error
of the CD signal, resulting from the error in the determination of the protein
concentration and errors in the CD calibration, is less than 6% at 25oC. The
reversibility of thermal unfolding in 3 M urea was only about 85%, but the
reversibility of the temperature-induced changes in the unfolded state was better
than 99.5%, as judged by heating a 5-M barstar sample in 7 M urea at 20oC h1
from 1oC to 95oC, then cooling down to 1oC and heating again to 95oC at 60oC h1.
CD measurements shown in each diagram were done with the same batch of
protein and sample cell. The relative error of R,230 for the scans on each side of
Fig. 10.4 is 0.01 L mol1 cm1 0.01R,230 as judged by the reproducibility. No
differences for different temperature scan rates above 20oC were noticed. Below
20oC, the systematic error of R,230 due to kinetic effects is up to 0.02 L mol1
cm1. Temperature scans were started at a temperature 5oC lower than those
shown in Fig. 10.4 (Nlting et al., 1995, 1997a, b).
Stopped-flow circular dichroism studies. Kinetic stopped-flow experiments
used an Applied Photophysics BioSequential DX-17 MV stopped-flow spectrometer complemented with a CD.1 circular dichroism accessory (Leatherhead,
UK) with a dead time of 7 ms. Refolding was initiated by 11-fold dilution of urea
solutions with 260 M protein. Measurements of the changes in CD at 222 nm
were performed at 15oC in 10 mM sodium phosphate buffer at pH 8.0. Only a
single slow rate constant of 310 s1, depending on urea concentration, was
observed. Its amplitude was obtained as a function of the concentration of urea by
fitting the refolding curves to a single-exponential function. A small correction
was made for the amplitude lost in the dead time. The denaturant dependence of
the amplitude (Fig. 10.12) mainly reflects the stability curves of the secondary
structure of the intermediate and native-like folded state with trans-conformation
of the peptidyl-prolyl (48) bond, respectively (Nlting et al., 1997a).
Fluorescence spectra. For equilibrium fluorescence experiments with the
fluorescence excitation at 280 nm, a sample cell of 0.4 cm 1 cm and a protein
concentration of 4 M were used. At 2oC, the sample was equilibrated for 2 h
prior to the measurement (Nlting et al., 1995).
140
Equilibrium unfolding. Free energies of unfolding for pseudo-wild-type and

mutants in 0 M urea, GFU , were determined as described in Sect. 8.3.2 by using
CD. The CD signal at 222 nm for protein samples in 50 mM Tris-HCl buffer at
pH 8 with 100 mM KCl at 10oC was detected as a function of urea concentration
using a protein concentration of 10 20 M and an optical pathlength of 0.1 cm,
and the spectral resolution was adjusted to 2 nm (Nlting et al., 1995, 1997a, b;
Nlting, 1998a).
T-jump measurements. To avoid artifacts, only small-amplitude T-jumps from
2oC to 10oC and low protein concentrations were used with a T-jump fluorimeter
from DIA-LOG (Dsseldorf, Germany), equipped with a 50-nF capacitor, a
200-W mercuryxenon lamp, and supplemented with a NICOLET (Madison, WI)
model Pro 90 storage oscilloscope (see Fig. 5.6). The conditions were, unless
stated otherwise: Fluorescence signals at 90o angles were detected with two
photomultipliers, summed and filtered with a 5-s response time for the fast
transition and a 1-ms response time for the slow transition. 2000 data points at 12bit resolution were recorded for each trace. Fluorescence excitation was at
280 nm, and a cut-off filter at 295 nm was used for emission. Folding proceeds
under isothermal conditions since after the T-jump, the temperature of the interior
of the protein molecule equilibrates with the bulk solvent in the nanosecond time
scale (Nlting, 1995, 1998b). The conditions were typically 210 M protein,
50 mM Tris-HCl buffer at pH 8 with 100 mM KCl, urea concentration as stated.
At 2oC, about 1% of the pseudo-wild-type barstar molecules are in the unfolded
state. Changing the concentration of pseudo-wild-type from 5 to 30 M at 0 M
urea does not change the rate constants and relative amplitudes of the fast folding
transition by more than 5%, but deviations could be seen at protein concentrations
in the mM range (Nlting et al., 1997a, b; Nlting, 1998a).
NMR studies on peptide fragments. Peptides of barstar comprising residues
1129, 28 44, 33 44 and 14 43 were synthesized with a Synergy Personal
Peptide Synthesizer, using F-moc protection. All were purified by reverse HPLC
chromatography. NMR spectra of peptides at 5oC in acetate buffer at pH 5.3 were
acquired using standard pulse sequences (Wthrich, 1986) with an AMX-500
Bruker spectrometer. Peptide concentrations were in the range of 1 to 1.5 mM
(Nlting et al., 1997a).
10.3
Structure of native barstar
Barstar is an 89-amino acid residue protein that has evolved to be the specific
intracellular inhibitor of the ribonuclease barnase. Both are expressed from
Bacillus amyloliquefaciens (Hartley, 1988; Schreiber and Fersht, 1993a; Nlting
et al., 1995, 1997a). Barstar (Fig. 10.1) has four -helices and three strands of
parallel -sheet; helix1 from Ser14 to Ala25, helix2 from Asn33 to Gly43, helix3
from Gln55 to Thr63, helix4 from Glu68 to Gly81, strand1 from Lys1 to Asn6,
strand2 from Leu49 to Arg54, and strand3 from Asp83 to Ser89. Residues 26 44
10.4 Residual structure in the cold-unfolded state
141
comprise a loop as well as helix2 of barstar, forming the binding site for barnase.
The inter-residue contact map (Fig. 10.2) shows that the divisions between
possible subdomains, such as residues 150 and 5189, are very weak and so
barstar is mainly a single domain protein (Nlting, 1998a).
Fig. 10.2. Inter-residue contact map for folded barstar (Nlting, 1998a).
10.4
Residual structure in the cold-unfolded state
To assess the structural changes along the pathway of folding (Nlting et al.,
1997a; Nlting, 1998a), it is important to know the spectroscopic properties,
thermodynamics and structures of the unfolded states. In particular, it is crucial to
know the differences in the residual structure between the different unfolded
states. Important questions regarding unfolded states of protein remain to be
answered: How can the structural distribution of unfolded protein be resolved?
What is the energy landscape of unfolded protein? What is the nature of the
transitions between different unfolded states? Do the transitions involve cooperative melting of structure or gradual changes? How might structures be funneled at
the beginning of the folding reaction before reaching the first detectable transition
state (Nlting et al., 1997b)?
142
The cold-unfolded state of barstar has been extensively analyzed by nuclear

magnetic resonance (NMR) and circular dichroism (CD) spectroscopy (Wong et
al., 1996; Nlting et al., 1997b). From NMR, three regions have been shown to
contain residual structures which are in the -helical region of (,) space. Two
of these are native-like since they are also helical in the native protein: Ser12 to
Lys21 of helix1 ; Tyr29 to Glu46 (which contains helix2). The third, Leu51 to
Phe56, is part of strand2 plus the loop connecting it to helix3 and so is non-native.
The C-terminal region, Asn65 to Ser89 (which contains helix4 and strand3), is
indistinguishable from a random coil (Nlting et al., 1997a).
Consistent with the NMR data, the far-UV CD of unfolded pseudo-wild-type
(C40A/C82A/P27A) barstar displays a surprising sensitivity to the point mutations
Q18G and A25G, located in helix1 in the native structure (Fig. 10.3; Nlting et al.,
1997b). In contrast, the chemically similar mutants Q72G and A77G, located in
helix4 , and Q58G, located in helix3 , display little, if any, difference of the far-UV
CD signal relative to pseudo-wild-type. Similarly, the CD at 230 nm of S14A in
helix1 is more negative than that of S59A in helix3 (Nlting et al., 1997b).
The dependence of the CD signal of the unfolded state on the concentration of
urea (Nlting et al., 1997b) and on the temperature (Fig. 10.4; Nlting et al.,
1997b) indicates that the residual structure in unfolded barstar is of very low
stability and does not involve significant surface burial. For example, no
cooperative transitions in the unfolded state in high concentrations of urea are
observed in CD temperature scans (Fig. 10.4).
Fig. 10.3. Spectra of the mean residual ellipticity, R , of the urea-unfolded states of
C40A/C82A/P27A barstar (wt b*) and its mutants: I5V, Q18G, A25G, L34V, Q72G and A77G
(Nlting et al., 1997b). Different symbols are used for better visibility and do not reflect the
number of points measured.
10.5 Gross features of the folding pathway of barstar
143
Fig. 10.4. CD temperature scans of C40A/C82A/P27A barstar at different concentrations of urea

and KCl, as indicated (Nlting et al., 1997b).
10.5
Gross features of the folding pathway of barstar
10.5.1
Equilibrium studies
The C40A/C82A/P27A (pseudo-wild-type) barstar displays cold-unfolding at
moderately low temperatures (0oC; see Fig. 10.5). There is a significant loss of
far-UV signal upon cold- and heat-unfolding, indicative of the melting of the
secondary structure (Nlting et al., 1995).
Fig. 10.5. Cold- and heat-unfolding of C40A/C82A/P27A barstar at different urea concentrations, as indicated (Nlting et al., 1995).
144
The equilibrium unfolding curves on addition of urea are identical when

monitored by far-UV CD at 222 nm, which follows the secondary structure, and
fluorescence emission at 330 nm, which follows tertiary structure interactions
(Fig. 10.6). The m-value (m = (GFU [urea]) / [urea], where GFU [urea] is the
Gibbs free energy change upon folding at various concentrations of urea, [urea]) is
as high as that for wild-type barstar, at 5.0 5.4 kJ L mol2 (1.21.3 kcal L mol2)
and the population of any intermediates at equilibrium is small. Both heat- and
cold-induced unfolding are cooperative (Nlting et al., 1995, 1997a).
Fig. 10.6. Urea-induced unfolding of C40A/C82A/P27A barstar, monitored by fluorescence

emission at 330 nm (F330) with excitation at 280 nm and CD at 222 nm (CD222), as indicated
(Nlting et al., 1995).
Fig. 10.7. Cold- and

heat- unfolding of
pseudo-wild-type barstar in 2 M urea,
monitored by fluorescence (Nlting et al.,
1995).
145
The fluorescence spectrum (Fig. 10.7) displays a significant red-shift upon cold
and heat-unfolding, indicative of the solvent exposure of aromatic side-chains
upon unfolding (see also next section; Nlting et al., 1995).
10.5.2
Kinetic studies
There is a cis-peptidyl-prolyl bond to the proline at position 48, which complicates
the folding pathway since the major conformation present in the unfolded state, U,
is the trans. The major pathway for the folding of barstar is the sequence (Nlting
et al., 1995, 1997a; Nlting, 1998a):
(10.1)
where Itrans is the early intermediate and Ftrans is a native-like state which binds
barnase and has similar fluorescence properties to the fully folded state, Fcis. The
sequence Itrans
Ftrans
Fcis has been detected by stopped-flow studies (Schreiber
and Fersht, 1993b; Agashe and Udgaonkar, 1995; Nlting et al., 1997a; Nath and
Udgaonkar, 1997b). It was found from temperature-jump studies that Itrans is a
state that has a compactness in-between that of the unfolded and folded states: On
the one hand it is clearly more collapsed than the unfolded state, on the other hand
the solvent exposure of aromatic sidechains is significantly larger than in the
folded state (Nlting et al., 1995, 1997a).
Fig. 10.8. Kinetic difference spectra of C40A/C82A/P27A barstar (Nlting et al., 1995). The
amplitudes of the relaxations are measured as function of the wavelength of fluorescence
emission for the transitions UtransItrans and ItransFtrans , as indicated.
146
There is evidence from the analysis of spectral changes for the burial of hydrophobic side chains in the slow transition, IF (Nlting et al., 1995). The fluorescence spectrum of pseudo-wild-type barstar shows a characteristic blue shift upon
cold- and heat-refolding (Fig. 10.7), indicative of the decrease in solvent exposure
of the tryptophan side chains on folding. The significant increase in fluorescence
accompanied by a blue shift during the slow kinetic event (Fig. 10.8) indicates a
burial of the side chains of tryptophan. In contrast, the shape of the difference
spectrum of the fast transition, UI, derived from monitoring the amplitudes of
the transition between 300 and 400 nm, resembles that of the equilibrium spectrum
of the cold-unfolded protein (Fig. 10.7), although there is also an increase in
amplitude. At a wavelength of 418 nm, however, the amplitude of the kinetic
difference spectrum (Fig. 10.8) is negative, suggestive of a small blue shift. These
observations indicate an increase in hydrophobic burial of tryptophan side chains
for the fast phase, too (Eftink and Shastry, 1997). The position of the wavelength
maximum in the difference spectrum, however, suggests that the fast transition
occurs between solvent-exposed states (Nlting et al., 1995).
Fig. 10.9. Temperature-jump trace for the folding of 6 M pseudo-wild-type

(P27A/C40A/C82A) barstar in the absence of denaturants at 10oC (Nlting et al., 1995, 1997a).
42 traces were accumulated with time constants of 5 s and 1 ms for the fast (Utrans
Itrans)
Ftrans) transition, respectively. There is no evidence for a lag phase preceding
and slow (Itrans
the formation of Itrans. The constant for the slow transition matches with that obtained using
rapid mixing.
When monitoring folding from the combined changes in fluorescence of

tryptophans 38, 44, and 53, it is found that the folding intermediate Itrans is formed
with a rate constant k1 = 2300 s1 and decays with k1 = 800 s1 at 10oC
(Figs. 10.9 10.11), so that GItrans Utrans = 2.5 kJ mol1 ( 0.6 kcal mol1). The
147
Itrans proceeds to Ftrans with k2 = 11 s1, and k2 is about 2 s1, so that GFtrans Itrans =
4.0 kJ mol1 (1.0 kcal mol1). The overall Gibbs free energy change upon
folding is 12.5 kJ mol1 (3.0 kcal mol1), so that GFcis Ftrans is 6.0 kJ mol1
(1.4 kcal mol1). The conversion of FtransFcis has a half life of several minutes
at 10oC (Nlting et al., 1995, 1997a).
The change in solvent exposure of each state during folding has been estimated
from the sensitivity of the folding and unfolding rate constants of both transitions
to the effect of the urea concentration, [urea], (Fig. 8.26) compared with that of the
overall equilibrium constant for folding, KFU (Figs. 8.21, 8.22). For the overall
folding reaction, the slope is logKFU / [urea] = 0.9 L mol1. The individual
folding and unfolding rate constants have been found to change with the urea
concentration according to: logk1 / [urea] = 0.2 L mol1; logk1 / [urea] = 0.3
L mol1; logk2 / [urea] = 0.3 L mol1; and logk2 / [urea] = 0.1 L mol1. Thus,
the compactness in surface area of the first transition state, #1, the intermediate,
Itrans , and the second transition state, #2, is approximately 20%, 50%, and 90%,
respectively, (Nlting et al., 1995, 1997a).
Fig. 10.10. Temperature-jump traces for the fast folding transition, Utrans
Itrans , of
P27A/C40A/C82A barstar at 10oC under different instrument settings and concentrations
(Nlting et al., 1995, 1997a). 36 traces were accumulated, each with 500 ns step width,
corresponding to 5000 points per trace and a time constant of 1 s for the measurement with
50 M protein. Noise was reduced by digital smoothing using a moving window of 50 data
points. For the measurement with 6 M protein, 42 traces were accumulated with a time constant
of 5 s. A 5 times higher light intensity relative to the measurement with 50 M was used. The
observed rate constant, kobs = k1 + k1 , does not deviate by more than 5% of 3100 s1 over the
concentration range 6 to 50 M protein and a wide range of instrument settings and heating
times.
148
Itrans is thus a fairly compact state with about 50% burial of surface area
(relative to the burial of surface in the folded state, taking the unfolded state as the
reference state), and with about 40% of -helical content of the folded structure
(Fig. 10.12). The formation of Itrans is cooperative since its rate constant is affected
by many mutations all over the molecule (Sects. 10.6 10.8; Nlting et al., 1997a;
1998a).
Fig. 10.11. The

amplitude
of
the fast phase,
Utrans
Itrans ,
is proportional
to the amplitude
of the T-jump.
Fig. 10.12. Amplitude of the ellipticity at 222 nm on the folding of barstar at 15oC in a stoppedflow circular dichroism spectrometer (Nlting et al., 1997a). The burst of signal (open circles) in
the dead time relative to the baseline corresponds to the circular dichroism signal of Itrans. The
total amplitude of the signal on folding (closed circles) corresponds to that of Ftrans (combined
changes of the signals for the two reactions).
10.6 -value analysis
149
10.6
-value analysis
The degree of formation of structure at individual positions in the intermediate and
transition states, X, on the folding pathway of barstar has been estimated from the
changes in their Gibbs free energy on mutation, GXU , relative to the change in
the overall Gibbs free energy of folding, GFU , (Fig. 10.13; Fersht et al., 1992;
Fersht, 1995a; Nlting et al., 1995, 1997a; Nlting, 1998a, 1999). For example,
when the -value for the state X, X = GXU/GFU , is = 1, then X is
destabilized by mutation by the same amount of energy as is the fully folded state,
F. When X = 0 , X is as unaffected by mutation as is the unfolded state U.
Intermediate values of indicate either a mixture of states of different degrees of
formation of structure or partial formation of structure (for more details see Sect.
8.3, especially Sect. 8.3.4; Fersht et al., 1992, 1994; Nlting et al., 1997a).
Figs. 10.14 and 10.15 present the -values for the transition state for the
formation of Itrans (#1) and for the equilibrium constant for its formation (I), as
well as those for the transition state #2 for the formation of Ftrans (Nlting et al.,
1997a; Nlting, 1998a). Helix1 has significant values of for #1 which slightly
increase in I and show that it is nearly completely formed in #2. Mutation of
amino acid residues in helix2 indicates that it is less well formed in #1, but these
mutations are more radical than those in helix1 because of the large hydrophobic
xxxxxxxxxxxx
Fig. 10.13. Mutants used for the -value analysis of barstar are distributed throughout the
molecule; mutated sidechains are indicated as a ball-and-stick structure. Each single mutant
probes the role of a part of the protein molecule in the folding reaction (Lubienski et al., 1994;
Nlting et al., 1997a; Nlting, 1998a). [Figure was drawn with MOLSCRIPT (Kraulis, 1991).]
150
groups involved and so there are also contributions from changes in their
interactions with the hydrophobic core. In particular, Leu34 which shows a high
in I, has strong interactions with Ser69 and Val70 in helix4. Helix3 is only very
weakly formed in #1, I and #2. Helix4 is not formed in #1, is partially consolidated
in I and is significantly formed in #2 (Nlting et al., 1997a).
Fig. 10.14. -value analysis of barstar (Nlting et al., 1995, 1997a; Nlting, 1998a). The
-values for the transition state for the formation of Itrans (#1) are calculated from k1 and those
for the equilibrium constant for the formation Itrans are calculated from the ratio k1/k1. -values
for the transition state for the formation of Ftrans (#2) are calculated from the values of k2 (from
a foldingunfolding double-jump experiment). The error bars indicate the estimated maximum
errors.
10.6 -value analysis
151
Fig. 10.15. -value analysis of a set of mutants of barstar for the early transition state (#1), the
intermediate (I), and late transition state (#2), as indicated (Nlting et al., 1995, 1997a; Nlting,
1998a). The error bars indicate the estimated maximum errors.
Hydrophobic amino acid residues in the core show varying degrees of structural
consolidation in #1 and I (Fig. 10.14). L16V, which has a high in #1 and I,
mainly probes interactions in the helix1loop1strand1 motif and between helix1
and helix4. The five probes in the strands tend to show that the -sheet is formed
primarily in I and #2 (Nlting et al., 1997a; Nlting, 1998a).
These results are further refined in the following sections. But, it is clear at this
stage of refinement that the first detectable intermediate, I, is structured around
helix1 (being already considerably formed in #1) and helix4 (being in the process
of consolidation in I).
152
10.7
Inter-residue contact maps
In order to simplify the structural interpretation of the -values it is important to
use mutants which do not cause a significant disruption of the native structure
outside the position of the mutation but only local changes (Matouschek et al.,
1989; Fersht et al., 1992). Unfortunately it is often unavoidable that mutations
probe several interactions simultaneously. In this case it is useful, for an improved
interpretation of the -values, to correlate the -values with inter-residue contact
maps (Figs. 10.16 10.18; Nlting, 1998a, 1999, 2003; Nlting and Andert, 2000).
When a mutant probes n contacts, for a particular state is
=
(i GFU,i) /GFU
GFU =
GFU,i ,
(10.2)
(10.3)
where i and GFU,i are the -value and Gibbs free energy contribution upon
mutation, respectively, for the contact number i , and GFU is the total observed
change in Gibbs free energy on mutation. For correlation with contacts, the values are equally assigned to all contacts that are predicted to be probed by
mutation.
Fig. 10.16. Inter-residue contact map for the microsecond folding transition state, #1, of barstar
(Nlting, 1998a).
10.7 Inter-residue contact maps
153
Fig. 10.17. Inter-residue contact map for the early folding intermediate, I, of barstar (Nlting,
1998a).
Fig. 10.18. Inter-residue contact map for the late folding transition state, #2, of barstar (Nlting,
1998a, 2003). See also Figs. 11.2 and 11.3.
154
Table 10.1. Predictions for the contacts which are probed upon mutation (Nlting, 1998a).
Mutant
Location
Typea
I5V
Q9G
I13A
S14A
L16V
Q18G
A25G
L26V
L34V
A36G
L37A
W38F
L41A
V45A
L49A
V50G
L51V
F56A
Q58G
Q61G
T63A
A67G
E68A
V70A
Q72G
R75E
A77G
A79G
T85A
strand 1
loop 1
loop 1
helix 1
helix 1
helix 1
helix 1
loop 2
helix 2
helix 2
helix 2
helix 2
helix 2
loop 3
strand 2
strand 2
strand 2
helix 3
helix 3
helix 3
helix 3
loop 4
helix 4
helix 4
helix 4
helix 4
helix 4
helix 4
strand 3
T
S
T
S
T
S
S
T
T
S
T
T
T
T
T
T
T
T
S
S
S
T
C
T
S
C
T
S
T
Interactions that are predicted to be modified by

mutation (S = strand; H = helix; L = loop) b
S2: W53; H1: L16, L20
L1: N6, I10
H1: S14, L16, H17; H3: T63; L4: G66
H1: D15; L2: E32
S1: I5; L1: I10, I13; L4: A67; H4: V70
H1: S14, D15, H17, T19
H1: K22, E23; L2: L26
H1: K21, L24; L2: Y30; H2: A40
H4: S69, V70, V73; H2: N33
L2: G31, E32; H2: N33, L37
H4: V70, V73, F74; L2: Y30
H4: V73, E76, A77
H2: L37, W38, A40; H4: V73, F74, A77
H1: L24; H2: A40, L41; L3: E46, Y47
S1: K1, K2, A3; L3: Y47; S2: V50; H4: F74; S3: I84
S1: K2, A3; S2: L49, L51, E52; S3: T85, I87
S1: I5; S2: W53; H4: F 74
L4: A67; H4: E68, L71; S2: W53; S3: L88
H3: Q55, E57, S59
H3: E57, K60, L62
L1: I13; H3: S59, L62; L4: G66
S2: W53; H3: F56, S59, K60
H3: K60
H1: L16, H17, L20
H4: E68, S69, L71, V73
H4: E76
H2: W38, L41; H4: V73, F74; S3: I84
H4: R75, E76, K78, E80
S2: V50; S3: I86, I87
S = Mainly secondary structure probing mutants (solvent-exposed sidechain); T = Mainly

tertiary structure probing mutants (buried sidechain); C = Charge mutants (solvent-exposed
sidechain).
b Predicted by using the NMR structure of wild-type barstar (Lubienski et al., 1994). Because
the precise structure of the mutants is not known, this column reflects only probabilities.
However, all mutants show a high activity.
a
In case of a -value of 1 the assignment gives exact -values for all individual
contacts since = 1 can only be found if all i = 1 and similarly for = 0 all
i = 0 unless there are non-native interactions with a significant Gibbs free energy
contribution (Nlting, 1998a, 1999, 2003; Nlting and Andert, 2000). The
occurrence of such non-native interactions is analyzed in Sect. 10.9.
When the -value is in-between 0 and 1, the contact map reflects a probability
of structure formation (Nlting, 1998a). When a cluster of a large number of
contacts from different mutations shows consistently high -values, it is
reasonable to assume that the average of for this cluster is non-negligible. If,
10.8 The highly resolved folding pathway of barstar
155
for example, two contacts of about similar energy contribution are probed by a
mutation, then one of the contacts can have a negligible only if the average is
found to be 0.5 or less (Nlting, 1998a, 1999).
Further, if some mutants probe only one specific structural element E and other
mutants probe the two structural elements E and F, then one can obtain a
probability of structure formation in F by comparing the two sets of -values. For
example, the mutants in helix1 L16V, Q18G, S14A, A25G with #1 = 0.59, 0.47,
0.32, and 0.17, are predicted to probe 100%, 0%, 50%, and 33% tertiary structure
contacts, respectively. No decreased for a large fraction of tertiary structure
contacts is obvious, suggesting a high probability for a significant consolidation of
tertiary structure interactions at this stage, too (Nlting, 1998a).
On average, 3 4 intramolecular contacts are predicted to be significantly modified in the folded state upon mutation (Table 10.1). Often, several of the contacts
for one mutant cluster in a relatively limited region of the inter-residue contact
map (Fig. 10.16 10.18), so that often only one or two clusters of interactions are
predicted to be probed by mutation. Further, to simplify the interpretation, preferentially mutants have been used which probe either only secondary or mainly
tertiary structure interactions (Nlting et al., 1995, 1997a; Nlting, 1998a).
10.8
The highly resolved folding pathway of barstar
Fig. 10.19. Consolidation of structure along the folding pathway of barstar as measured by
solvent exclusion (circles), Gibbs free energy (squares), CD at 222 nm (diamonds), CD at
270 nm (upward triangles) and fluorescence detection at 330 nm with excitation at 280 nm
(downward triangles), respectively (Nlting et al., 1995, 1997a; Nlting, 1998a). At 10oC, the
folding rate constants for #1 and #2 are (435 s)1 and (91 ms)1, respectively, and at 6oC they
are (562 s)1 and (181 ms)1, respectively.
156
Unfolded state
strand1
#1 (500 s)
20-30%
solvent
exclusion
Nucleation
in and around
helix1
helix1
helix4
strand1
I (1 ms)
50% solvent
exclusion
helix1
Further
consolidation
#2 (100 ms)
90% solvent
exclusion
and
hierarchical
Folded
state
assembly
Fig. 10.20. Structural consolidation along the folding pathway of barstar (Nlting et al., 1995,
1997a; Nlting, 1998a, 2003). #1, I, and #2 are the microsecond transition state, intermediate
state, and late transition state, respectively. At 10oC, the intermediate and folded states are
populated to 70 80% after 1 ms and 200 ms, respectively. The figure was drawn using the
program MOLSCRIPT (Kraulis, 1991). See also Fig. 11.3.
157
10.8.1
Microsecond transition state
The evidence collected from the kinetic, spectroscopic, equilibrium thermodynamic characterization and -value analysis (Sects. 10.6 and 10.7) suggests that
in the absence of denaturants, about 500 s after initiation of the folding reaction,
the barstar molecule collapses under roughly 25% solvent exclusion and the
polypeptide chain kinks to enable the formation of attractive interactions in the
helix1loop1strand1 motif (Figs. 10.16, 10.19, 10.20; Nlting et al., 1995, 1997a,
b; Nlting, 1998a).
10.8.2
Intermediate
The structure of the early transition state allows the molecule to condense further
within a few 100 s to form an intermediate with strengthened tertiary structure
interactions of helix1 , helix2 , and helix4. Secondary structure interactions in helix1
are further strengthened, and at this stage, helix4 has formed significant amounts
of secondary structure too. As judged by the Gibbs free energy, the average
structure consolidation of the early intermediate is only 20%. Few, if any, residues
form very strong non-native interactions, as suggested by the absence of
significantly negative changes in -values (Fig. 10.15). The solvent exclusion has
already reached 50%, as judged by the effect of urea on the rate constants of
folding and unfolding (Figs. 10.17, 10.19, 10.20; Nlting et al., 1995, 1997a, b).
10.8.3
Late transition state
100 ms later, the still largely fluid-like molecule of small stability has significantly
further consolidated and is passing through the slow folding transition state with
an average of 50% structure consolidation, as judged by Gibbs free energy, and
90% solvent exclusion, as judged by m-values. This state enables then the rapid
formation of the fully folded conformation. Tertiary structure interactions in #2
are detected at many positions in helix1 , helix2 , helix4 , strand1 , strand2 , and strand3.
Strong secondary structure interactions have built up in helix1 and helix4. Nearly
complete structure formation is found for helix1 at the positions Ser14 and Leu16
and in helix4 at the position Arg75. The inter-residue contact map suggests that
there is little, if any, consolidation of helix3 (Figs. 10.1810.20; Nlting, 1998a).
10.8.4
Directional propagation of folding
Figs. 10.16 and 10.17 show that initially in the folding reaction, i.e., in #1
(500 s) and I (1 ms), amino acid residues that are located close to the middle
of the polypeptide chain display little, if any, structure consolidation. In I , mainly
helix1 and helix4 have significant -values. In the course of the folding reaction,
158
the consolidation of structure propagates from the termini towards helix3 , which
not yet formed in #2 (100 ms). In #2 for most of the residues located close to the
middle of the polypeptide chain, the formation of secondary structure is weak, but
N- and C-termini are largely consolidated as suggested by -values close to 1.
The consolidation of structure in the middle of the sequence is the latest folding
event that proceeds mainly between #2 and the fully folded state, F (Figs. 10.18
and 10.20; Nlting, 1998a).
Further, -values close to 1 found for #2 at several positions in helix1 and
helix4 suggest a high degree of correct sidechain interlocking at these positions
(Figs. 10.14, 10.15, 10.18). Probably, there are still weak non-native interactions
in most other parts of the molecule as suggested by -values of less than 1 and by
a lower average in Gibbs free energy change of 50% compared with the overall
degree of solvent exclusion of 90% (Nlting et al., 1995, 1997a; Nlting, 1998a).
A considerable likelihood of terminal proximity has been predicted for randomflight chains (Flory, 1969; Ptitsyn, 1981), and has been observed in folded
structures of proteins (Christopher and Baldwin, 1996). This study (Nlting,
1998a) presents evidence for terminal consolidation in the early stages of folding
of barstar (Fig. 10.20). One might speculate that an early consolidation of the Nterminal part of the molecule could represent an advantage in the in-vivo protein
synthesis. However, the -value analysis of barnase did not reveal this feature
(Fersht, 1993) and for CI2 the effect of early terminal consolidation is at the
borderline of statistical significance (Itzhaki et al., 1995a; Nlting, 1999),
suggesting that this feature of barstar folding is not found in all proteins. A more
important aspect of this high resolution of a protein folding pathway is that the
discovered nucleationcondensation events (see Sect. 10.11) involve the
formation of tertiary structure interactions of structural elements that are distant in
the primary structure. A similar behavior has been observed also in CI2 on a
lower time scale (Fersht, 1993; Itzhaki et al., 1995a; Nlting, 1999). The obvious
advantage of a folding nucleus which is stabilized by interactions of distant
structural elements is that it can prevent misfolding more efficiently than a nucleus
that would comprise only short-range interactions (Sect. 10.11; Nlting, 1998a).
10.8.5
Cis trans isomerization
Folded barstar with the cis-conformation of the prolyl(48)-peptidyl bond has a
significantly higher stability than that with the trans-conformation. However, the
small, if any, effect of mutation on the ratio of the unfolding rate constant of
barstar with cis-conformation of the prolyl(48)-peptidyl bond, Fcis , versus that for
the trans-conformation, Ftrans , (Fig. 10.21) suggests that the cistrans isomerization of this bond is mainly a local effect and has no significant effect on the
degree of consolidation of structure in most parts of the molecule (Nlting et al.,
1997a; Nlting, 1998a).
Taking the cistrans isomerizations of the prolyl(48)-peptidyl bond into
account, the scheme for barstar folding has to be extended according to Fig. 10.22.
159
Fig 10.21. Unfolding rate constants for pseudo-wild-type barstar (wt) and several of its mutants
with cis-conformation of the prolyl(48)-peptidyl bond versus those for the trans-conformation in
3 M urea at 11.6oC. The line indicates the ratio of the rate constants which is obtained for
pseudo-wild-type. A similar ratio of rate constants is obtained for most of the mutants, which
suggests that the cistrans isomerization is mainly a local and not a global folding event.
Fig. 10.22. Extended scheme for the folding pathway of barstar. The pathway which leads from
the unfolded state with trans-conformation of the prolyl(48)-peptidyl bond, Utrans , to the folded
state with trans-conformation, Ftrans , to the fully folded state with cis-conformation, Fcis , is the
major pathway.
10.8.6
Are there further folding events?
There are no indications of further global folding events with a significant Gibbs
free energy contribution: 1. The total of the -value increments calculated using
Eq. 8.36 comes to 1, within the experimental error for all mutants, with exception
of two mutations located close to Pro48. 2. The high kinetic resolution down to a
few s (Fig. 10.23) suggests that there is no further phase with a fluorescence
change of significant amplitude.
160
Fig. 10.23. Deviation of the temperature-jump fluorescence trace for the transition
Utrans
Itrans from the single-exponential curve fit. Only part of the trace is shown; the
complete time range for trace and curve fit was 3 ms. There is no evidence for a lag phase
preceding the formation of the early intermediate, Itrans (Nlting et al., 1997a).
10.9
Structural disorder and misfolding
For a comprehensive -value analysis it is important to distinguish between
correctly and wrongly formed interactions. Non-native interactions must exist in
the individual stages of sampling of conformations along the folding pathway, in
particular in the intermediate and transition states, because otherwise the folding
reaction is predicted to take significantly less than 1 ms (Nlting, 1998a, 1999).
Intriguingly, few if any significantly negative -value changes are observed (Figs.
10.14 and 10.15), which suggests little, if any, formation of strongly attractive
non-native interactions. This finding indicates the absence of significant
misfolding on the folding pathway of barstar, i.e., the absence of deep energy
traps with slow exchange rates to the native conformation, but rather the presence
of structural disorder, i.e., small energy contributions from a large number of
interactions. A reason for this behavior may be that the intermediate of barstar
corresponds to a relatively shallow valley in the energy-landscape of folding of
only 2.5 kJ mol1 (0.6 kcal mol1) (Nlting et al., 1997a). However, the redshifted kinetic difference fluorescence spectrum for the fast transition relative to
the slow transition (Fig. 10.8), the large CD signal at 222 nm of the intermediate
(40% of the folded state), and its large solvent exclusion (50% of the folded state),
compared with its stability of only 20% of the stability of the folded state, suggest
a significant sidechain disorder in the intermediate (Pfeil, 1993; Pfeil et al., 1993a;
Nlting et al., 1995, 1997a; Nlting, 1998a, 1999).
Because for #1 there is no significant cross-correlation between tertiary
structure formation in helix1 and helix4 (Fig. 10.16) and since helix4 is still in a
disordered state with no stable secondary structure formed yet, no statement about
a correct or incorrect tertiary structure alignment can be made for this stage
(Nlting, 1998a).
10.9 Structural disorder and misfolding
161
There are, however, indications for a significant degree of correct tertiary

structure alignment in the early intermediate: 1. Peptides comprising the helix1
loop1helix2 motif and helix4 , respectively, do not have significant secondary
structure under folding conditions (see Sect. 10.10; Nlting et al., 1997a). 2. The
inter-residue contact map (Fig. 10.17) shows a strong cross-correlation for
secondary and tertiary structure contacts of helix1 and helix4. 3. The average of
-values for mainly secondary structure probing and mainly tertiary structure
probing mutants is found to be the same (Nlting, 1998a).
The absence of misfolding in the barstar intermediate is in agreement with
computer simulations of folding (Wolynes et al., 1995; Nymeyer et al., 1998),
which suggest that the so-called glass-transition, i.e., the freezing of conformations
into deep energy traps, may occur after formation of most of the native contacts,
but is less likely to occur in fluid-like early states (Fig. 10.24; Nlting, 1998a).
Fig. 10.24. Folding funnel model (Wolynes et al., 1995, 1996; Shoemaker et al., 1997).
Computer simulations suggest that early in the folding reaction there is a significant number of
weak non-native contacts. The main transition state barrier arises from an entropic bottleneck
(Onuchic et al., 1996). After passing through the glass transition, misfolding may occur, i.e.,
protein molecules may become trapped into conformations that correspond to deep energy
valleys from which escape is possible only at a very low rate (Hagen et al., 1995; Shortle et al.,
1998). Good folding sequences have energy landscapes where the energy bias towards the fully
folded state is larger than the ruggedness of the folding funnel (Nymeyer et al., 1998).
162
10.10
Structure of peptides of barstar
Four peptides that correspond to sequences in barstar have been synthesized and
investigated (Nlting et al., 1997a): 1129, 33 44, 28 44, and 14 43, that
contain the sequences of helix1 , helix2 , loop1helix2 , and the entire helix1
loop1helix2 motif, respectively. 40% trifluoroethanol induces helical structure in
the 4 peptides, as judged by far-UV circular dichroism (CD). However, in H2O at
pH 5.3 and 5oC the two-dimensional 1H-NMR spectra of the peptides 1129,
33 44, and 28 44 resemble those expected for a random coil state, apart from
some specific interactions involving the tryptophan residues 38 and 44 (the 14 43
peptide could not be studied in aqueous solutions because of signal broadness).
CD spectra in H2O are also close to the random coil spectrum (Fig. 7.11) with less
than 510% of the molar ellipticity at 222 nm expected for a helical structure.
Less than 5% helical structure is calculated from a titration procedure (Jasanoff
and Fersht, 1994). Thus, helix1 and helix2 do not form significant amounts of
stable helical structure under folding conditions in the absence of the rest of the
protein. The formation of the relatively stable structure in helix1 in the folding
intermediate I must, accordingly, be coupled with the formation of long-range
interactions that stabilize the helix (Nlting et al., 1997a).
10.11
Nucleation condensation mechanism of folding
The data presented in Sects. 10.310.10 are consistent with a nucleationcondensation model (Abkevich et al., 1994b; Itzhaki et al., 1995a; Fersht, 1995c;
Shakhnovich et al., 1996; Shoemaker et al., 1997; Guo and Thirumalai, 1997;
Klimov and Thirumalai, 1998; Michnick and Shakhnovich, 1998; Ptitsyn, 1998;
Nlting et al., 1997a; Nlting, 1998a, 1999) for folding of barstar in the submillisecond time scale. Although the early stages of barstar folding represent collapsed
states, they are clearly not uniformly consolidated. Analogous to the growth of a
crystal, a part of the molecule, the so-called nucleation site, forms significantly
earlier than other parts of the molecule. In the course of the reaction, the initially
diffuse folding nucleus becomes increasingly stabilized as further structure
condenses around it. The nucleus of barstar (Fig. 10.20) comprises mainly helix1
and some surrounding structural elements (Nlting et al., 1997a; Nlting, 1998a).
The investigation on barstar was inspired by earlier studies (Matouschek et al.,
1989; Fersht et al., 1992; Fersht, 1993; Serrano et al., 1992c) on the ribonuclease
barnase that has a clear modular structure (Yanagawa et al., 1993). Indirect
methods indicated the occurrence of a distinct early forming folding intermediate,
and -value analysis of the later stages of folding suggested a framework
mechanism in which a preformed secondary structure of -helix docked on that of
-sheet (Serrano et al., 1992c). A general scheme has been proposed in which
modules of structures in larger globular proteins were formed initially by
10.11 Nucleationcondensation mechanism of folding
163
nucleationcondensation (Abkevich et al., 1994b; Itzhaki et al., 1995a). These

modules would then dock, either in a purely stepwise manner or by the processes
of docking and nucleationcondensation being coupled, depending on the
stabilities of the modules (Itzhaki et al., 1995a). It was predicted, that the more the
stabilities of the individual modules, the greater the hierarchical tendency of the
folding reaction. However, it was not possible to measure the rate of formation of
the folding intermediate of barnase and so there was no direct evidence for an
initial nucleationcondensation process (Nlting et al., 1997a).
Application of new T-jump methods (Nlting et al., 1995, 1997a; Nlting,
1996, 1998b), in combination with protein engineering, to the study of early
folding events of barstar enabled to test the nucleationcondensation mechanism
directly. The results of -value analysis and spectroscopic studies on barstar are
directly consistent with a nucleationcondensation model (Fig. 10.25): Peptides
that contain helix1 and correspond to parts of a folding nucleus are mainly random
under folding conditions in the absence of the rest of the protein, but a nucleus
centered around helix1 is substantially formed in the first transition state on the
microsecond time scale, and further consolidated in the early-formed intermediate.
Many of the rest of the amino acid residues in the protein make weak interactions
in the intermediate, which are then more highly consolidated in the later transition
state (Figs. 10.1610.18; Nlting et al., 1997a; Nlting, 1998a).
Considering the astronomically high number of possible conformations of a
random polypeptide chain (see Fig. 1.1), proteins may fold with amazingly high
rate constants, and folding is an astonishingly efficient process. One of the fastestfolding proteins, a thermostable variant of monomeric lambda repressor, can fold
in approximately 20 s (Burton et al., 1997) ! As pointed out in the Introduction,
folding cannot proceed via random sampling of all possible conformations.
The nucleationcondensation model of protein folding may explain the high
speed of folding reactions (Itzhaki et al., 1995a). Nucleationgrowth theory
predicts that the nuclei of fast-folding protein sequences contain a certain number
xxxxxxxxxxxxx
Fig. 10.25. Nucleationcondensation model for protein folding (Abkevich et al., 1994b; Itzhaki
et al., 1995a; Fersht, 1995c; Shakhnovich et al., 1996; Nlting et al., 1997a; Nlting, 1998a,
1999). The formation of a few correct secondary and tertiary structure interactions in the folding
nucleus catalyzes further folding. The nucleus becomes increasingly stabilized as further
structure condenses around it.
164
of native contacts (Michnick and Shakhnovich, 1998; Mirny et al., 1998).

Theoretical studies suggest that the formation of a specific nucleus is a necessary
and sufficient condition for subsequent rapid folding to the folded state, that
strengthening of interactions in the nucleus is accompanied by acceleration of
folding, and that the amino acid residues involved in the nucleus are the most
conserved ones within families of evolved sequences (Abkevich et al., 1994b;
Shakhnovich et al., 1996; Mirny et al., 1998).
A nucleus which has a (marginal) stability in the presence of some correctly
formed tertiary structure interactions is predicted to catalyze folding by preventing
misfolding: If secondary structure would form significantly without correct
tertiary structure interactions, misalignment of these secondary structure elements
could easily occur in proteins which have a complicated chain topology (Nlting,
1998a). This conclusion is consistent (a) with observations on a slower time scale
of a nucleationcondensation folding mechanism for chymotrypsin inhibitor 2
(CI2) which suggest that the folding nucleus is not stable in the absence of longrange interactions (Fersht, 1995c; Nlting, 1999), and (b) with observations of the
effect of a large number of long-range interactions in the folded state on the speed
of folding: Proteins of similar size fold several orders of magnitude faster if they
exhibit a relatively simple chain topology with mainly strong local contacts,
especially those conductive to helix formation (Viguera et al., 1997; Chan, 1998;
Plaxco et al., 1998): Obviously, the more the folded structure is dominated by
complex chain topologies with many long-range interactions, the longer it takes to
find the conformation of the transition state and the slower the protein will fold.
The nucleationcondensation model is consistent with the folding funnel model
(see Fig. 10.24; Wolynes et al., 1995, 1996; Onuchic et al., 1996; Shoemaker et
al., 1997). Predictions about the residues involved in the folding nucleus agree for
both models (Shoemaker et al., 1997). Fast folding sequences have a low
ruggedness of the folding funnel (Nymeyer et al., 1998). Energy landscape theory
predicts, in agreement with the experimental results on barstar and CI2 (which
contain - and -structure), that the folding funnel for small fast-folding -helical
proteins has a transition state roughly half-way to the fully folded state (Onuchic
et al., 1996).
For barstar, the speed of the formation of the early intermediate, (500 s)1, is
well below the diffusion limit of folding which is estimated to be around (1 s)1
for a protein of this size (Hagen et al., 1996), suggesting that a large number of
conformational sampling steps takes place, roughly 500 500,000 steps if one
assumes 1 ns 1 s per step, to funnel the unfolded conformations into the
intermediate state. This large number of sampling steps indicates the existence of
a significant number of important interactions in the folding nucleus and represents
evidence against a diffusion-limited hydrophobic collapse model for this folding
step (Nlting et al., 1995, 1997a; Nlting, 1998a; 1999).
11
Highly resolved folding pathways and

mechanisms of six proteins
Similar investigations as described for barstar in the previous chapter were done
with five further small proteins: barnase, chymotrypsin inhibitor 2 (CI2), the src
SH3 domain, Arc repressor, and a tetrameric p53 domain. As detailed in the
following sections, it was found that for the main transition states for formation of
the native structure of barstar, barnase, CI2, src SH3 domain, and Arc repressor:
1. on average over the molecule, secondary and tertiary structure interactions have
built up to the same degree, or at least a high degree, but the built-up of
interactions is non-uniformly distributed over the molecule;
2. the most consolidated parts of the molecules form clusters, and these clusters
contain a particularly high fraction of amino acid residues that belong to
secondary structure elements of the native state;
3. elements of secondary structure have on average a larger relative consolidation
than loops, as judged by -values for the main transition states (Nlting and
Andert, 2000).
As discussed in the following sections, these observations further reconcile the
framework model with the nucleationcondensation mechanism for folding: the
amazing speed and efficiency of folding of many proteins can be understood as
caused by the catalytic effect of the formation of folding nuclei which comprise
significant amounts of tertiary structure interactions, but have a preference for the
early formation of regular secondary structure (Nlting and Andert, 2000).
The -values in this chapter are from Matouschek et al., 1992; Serrano et al.,
1992a, 1992b; Matouschek and Fersht, 1993; Milla et al., 1995; Itzhaki et al.,
1995a; Lpez-Hernndez and Serrano, 1996; Nlting et al., 1997a; Nlting,
1998a; Riddle et al., 1999; Mateu et al., 1999; Chiti et al., 1999; Fulton et al.,
1999. For reasons of a higher precision, only data from mutants which cause a
change of stability |GFU| > 0.5 kcal mol1 were used. For more details
regarding the study presented in this chapter the reader may also refer to Nlting
and Andert (2000).
11.1
General features of the main transition states for the
formation of the native structures
As discussed in Chap. 10, different models have been established to explain the
surprisingly high speed and efficiency of folding, e.g., the framework model
166
11 Highly resolved folding pathways and mechanisms of six proteins
Fig. 11.1. Models for protein folding (Nlting and Andert, 2000). a: Framework model (Ptitsyn
and Rashin, 1975; Kim and Baldwin 1982, 1990; Udgaonkar and Baldwin, 1988). Protein folding
is thought to start with the formation of elements of secondary structure. These elements form
independently of tertiary structure, or at least before tertiary structure is locked in place. The
elements then assemble into the tightly packed native tertiary structure either by diffusion and
collision (Karplus and Weaver, 1994) or by propagation of structure in a stepwise manner
(Wetlaufer, 1973). b: Hydrophobic collapse model for folding (Rackovsky and Scheraga, 1977;
Dill, 1985, 1990a, 1990b). The initial event of the folding reaction is thought to be a relatively
uniform collapse of the protein molecule, mainly driven by the hydrophobic effect, i.e., the
tendency of non-polar groups dissolved in water to cluster together (see Sect 3.4). Stable
secondary structure elements can only form in the resulting collapsed state. c: Nucleationcondensation mechanism (Fersht 1995c; Itzhaki et al., 1995a; Nlting et al., 1995, 1997a;
Shakhnovich et al., 1996; Fersht, 1997, 1999; Kiefhaber et al., 1997; Nlting, 1998a, 1999;
Michnick and Shakhnovich, 1998; Otzen and Fersht, 1998): Early formation of a folding nucleus
catalyzes further folding. The nucleus is diffuse, but comprises secondary structure interactions
and approximately correct tertiary structure interactions (see also Sect. 10.11). This model is
consistent with the funnel model (Wolynes et al., 1995; Shoemaker et al., 1999) which focuses
on the rapid decrease of the conformational dispersity in the course of the reaction. Some
proteins, in particular the ones with larger numbers of amino acid residues, may have several
nuclei. The three models may analogously be applied on proteins with multiple transition states
on their pathways (not shown). See also Sect. 10.11.
11.1 General features of the main transition states ...
167
Fig. 11.2. Inter-residue contact map for the main transition state of barstar. The x (abscissa) and
y (ordinate) axes indicate the sequence of the protein. Every circle corresponds to a contact of the
amino acid residue number x with the residue number y. In this plot, secondary structure contacts
are displayed on the diagonal, and tertiary structure contacts are contained in the bulk (see also
Fig. 10.18 which shows earlier results on a somewhat different scale of -values). For reasons of
simplification, only the bottom right triangle of contacts is shown and the top left triangle with
the same information is left out. The sizes and colors of the circles indicate the magnitudes of the
-values of the contacts between the residues in the native state that are predicted to be altered
by mutation. High -values in the map (large symbols) suggest a high degree of consolidation of
structure (about native interaction energies) at the individual positions in the inter-residue contact
space. ~ 0 (small symbols) indicates little, if any, formation of stable structure. -values in the
range of ~ 0.20.8 indicate different probabilities of the consolidation of structure (see Sect.
10.7): for around 0.5 usually only clusters with at least 5 contacts may be used to draw
statistically significant conclusions about the presence or absence of a certain degree of structural
consolidation. Because of the possibility of non-native interactions, the same region of contacts
should be probed by several mutants. Thick bars and the axis labels H1, H2, .. show the positions
of helices and thin bars and the axis labels S1, S2, .. show the positions of strands of -sheets in
the native state. For further details see the Chap. 10 and (Nlting, 1998a, 1999, 2003; Nlting
and Andert, 2000).
168
Fig. 11.3. Consolidation of structure in the main transition state of barstar as measured by
-value analysis (Nlting and Andert, 2000; Nlting, 2003). The most consolidated secondary
structures elements of the molecule are highlighted in yellow, and the parts which are drawn in
blue have no fixed structure. Amino acid residues with high -values ( 0.8) are highlighted
as red spheres.
(Fig. 11.1a), the nucleationgrowth mechanism, the diffusioncollision mechanism, the hydrophobic collapse model (Fig. 11.1b), the funnel model (Fig. 10.24),
and the nucleationcondensation mechanism (Fig. 11.1c; Sect. 10.11).
As mentioned in Sect. 10.11, the folding nucleus of the nucleationcondensation mechanism is diffuse and mainly consists of several neighboring amino
acid residues whose conformations are stabilized by long-range interactions with
residues that are remote in sequence. An essential component of this mechanism is
the formation of secondary and tertiary structures at the same time (Nlting,
1998a, 1999, 2003; Nlting and Andert, 2000).
The significance of the nucleationcondensation mechanism is that it can make
plausible the extreme efficiency of protein folding and can resolve the folding
paradox: for example, in a 100-residue protein, the formation of a folding nucleus
169
Fig. 11.4. Inter-residue contact map for the main transition state of barnase (Nlting and Andert,
2000; Nlting, 2003). For explanation of the symbols see the legend to Fig. 11.2.
which needs, e.g., 10 20 inter-residue contacts for a sufficient degree of stability

probably requires only roughly 1010 1020 random sampling steps compared to
roughly 10100 sampling steps for a completely random folding process of the
whole molecule (see Chap. 1). Taking into account that in the course of the
nucleation the sampling is not completely randomly because steric hindrance and
already small energy differences affect the probabilities of different molecular
motions, and considering that a sampling step takes roughly 109 s, one may
obtain quite realistic folding times.
A thorough analysis of the structures of the main transition states for the
formation of the native state of six proteins by -value analysis presented in the
following paragraphs (see also Nlting and Andert, 2000; Nlting, 2003) shows
170
Fig. 11.5. Consolidation of structure in the main transition state of barnase (Nlting and Andert,
2000; Nlting, 2003). Amino acid residues with high -values ( 0.8) are highlighted as red
spheres. For further explanation see the legend to Fig. 11.3.
that the nucleationcondensation mechanism is not only valid for a few

exceptional proteins, but for at least five of the investigated six proteins, but that
their transition states have also some framework-model-like properties. In the
following paragraphs, the main transition state structures of the folding pathways
of the six proteins are visualized by inter-residue contact maps (Figs. 11.211.12,
even numbers) and ribbon representations (Figs. 11.311.13, odd numbers). This
visualization is shown to contribute significantly to a mechanistic understanding
of the surprising speed and efficiency of protein folding and to the resolution of
the folding paradox (Nlting and Andert, 2000; Nlting, 2003).
The X-ray and NMR measurements show that all six proteins contain
considerable amounts of fixed secondary and tertiary structures in their native
conformations. Consistent with predictions from the funnel model, the structural
consolidation of the main transition states of the four monomeric proteins (barstar,
barnase, CI2, src SH3 domain) is about quarter to half way to the native state, as
judged by the average of the -values (Tables 1 and 2 in Nlting and Andert,
2000). Accordingly, the average free energy of interactions in the main transition
state is about 2550% of the free energy of interactions in the native state (Nlting
and Andert, 2000).
171
Fig. 11.6. Inter-residue contact map for the main transition state of chymotrypsin inhibitor 2
(CI2) (Nlting and Andert, 2000; Nlting, 2003). For explanation of the symbols see the legend
to Fig. 11.2.
Barstar (Figs. 11.2, 11.3) folds via a three-state transition (U

I
F,
with U, unfolded state; I, intermediate state; F, folded state). Its main transition
state (from I to F) has significant amounts of secondary and tertiary structure
interactions involving helix1 , helix2 , helix4 , and most parts of the -sheet, and
little, if any, consolidation in helix3 (Nlting and Andert, 2000; Nlting, 2003).
For more details see Chap. 10.
Barnase (Figs. 11.4, 11.5) which also folds via a three-state transition (U
I
F) clearly shows a cluster of consolidated residues comprising helix1
(Thr16, and slightly lower at His18) and the -sheet including some of the turns
(Asn58 in turn3 , Ile88 and Leu89 in strand3 , Ser91 and Ser92 in turn5) in its main
transition state (from I to F) (Nlting and Andert, 2000; Nlting, 2003).
172
Fig. 11.7. Consolidation of structure in the main transition state of chymotrypsin inhibitor 2
(CI2) (Nlting and Andert, 2000; Nlting, 2003). Amino acid residues with high -values
( 0.8) are highlighted as red spheres. For further explanation see the legend to Fig. 11.3.
Also CI2 (Figs. 11.6, 11.7) which has been shown to fold according to a twostate transition (U
F) exhibits a clear cluster of consolidated residues which
is located in and around the helix. In particular, Glu14 and Ala16 in the helix have
-values near 1 (Itzhaki et al., 1995a; Nlting and Andert, 2000; Nlting, 2003).
The helix has diffuse contacts with the -sheet.
A similar clear cluster is observed in the transition state structure of the src SH3
domain (Figs. 11.8, 11.9): the highest -values are found in some residues of
strand3 (Ala45, Ser47) and the hairpin (Thr50, Gly51) with connects strand3 with
173
Fig. 11.8. Inter-residue contact map for the main transition state of the src SH3 domain (Nlting
and Andert, 2000; Nlting, 2003). For explanation of the symbols see the legend to Fig. 11.2.
strand4 . The src SH3 domain also displays two-state folding behavior (U
F)
(Nlting and Andert, 2000; Nlting, 2003).
Summarizing, all 4 small monomeric proteins display one or two clusters of
structural consolidation in some residues which are located in the polypeptide
chain about 10 30% apart from the N- and C-termini (Figs. 11.2, 11.4, 11.6,
11.8). This position in sequence of the folding nucleus appears not to be a general
rule, however, as indicated by observations on other proteins, e.g., acylphosphatase for which the data coverage is not sufficient for a more thorough -value
analysis (Nlting and Andert, 2000).
The transition state of the dimeric Arc repressor (Figs. 11.10, 11.11):
1. is in average relatively weakly consolidated,
174
Fig. 11.9. Consolidation of structure in the main transition state of src SH3 domain (Nlting and
Andert, 2000; Nlting, 2003). Amino acid residues with high -values ( 0.8) are highlighted
as red spheres. For further explanation see the legend to Fig. 11.3.
2. has -values larger than 0.4 for only two (Leu19 and Gly30) of the 27 residues
probed by mutation with |GFU| > 0.5 kcal mol1 (see Table 1 in Nlting and
Andert, 2000),
3. has the strongest consolidation near the middle of the sequence (Fig. 11.10),
4. involves a significant number of inter-molecular interactions (Fig. 11.10).
The data show that its transition state structure is affected by both the process of
folding and as well the assembly of the monomers (Nlting and Andert, 2000).
In contrast, the main transition state structure of the p53 domain (Figs. 11.12,
11.13) is highly consolidated almost everywhere. The folding model is a four-state
transition (Nlting and Andert, 2000):
4U
2 I'2
2 I2
F4 ,
(11.1)
where U, I'2, I2, and F4 are monomeric unfolded state, first dimeric intermediate
state, second dimeric intermediate state, and native tetrameric state, respectively.
xxxxxx
175
xxxx
Fig. 11.10. Inter-residue contact map for the main transition state of the dimeric Arc repressor
(Nlting and Andert, 2000). For explanation of the symbols see the legend to Fig. 11.2. For Arc
repressor there are some quaternary structure contacts because its reaction involves folding and
association of the monomers into dimers.
The main transition of this protein is the formation of the tetramer, F4 , from two
dimers, 2 I2 . Only -values of some mutants which probe interactions at the
interface between the two dimers were found to be somewhat lower than 1 which
suggests that the interactions between these two dimers are not completely formed
(Fig. 11.13). On the other hand, the average of of the transition state for the
formation of the early dimers, 2 I'2 , from monomers, 4 U, is only 0.01 0.03, so
the formation of almost all secondary, tertiary, and monomermonomer
quaternary interactions of the molecule occurs in the step 2 I'2
2 I2 (Nlting
and Andert, 2000).
176
Fig. 11.11. Consolidation of structure in the main transition state of Arc repressor (Nlting and
Andert, 2000). For this dimeric protein, folding and association occur nearly concurrently. The
atoms of the residue with the largest (0.46) are highlighted as red spheres. For further
explanation see the legend to Fig. 11.3.
11.2
Nucleation condensation
The analysis of the folding behavior of the four monomeric proteins and of the
dimeric Arc repressor shows that:
1. their early formed structures are non-uniformly distributed over the molecule
(Figs. 11.211.11; see also Chap. 10);
2. the most consolidated parts of these structures form one or very few clusters
(Figs. 11.211.11; see also Chap. 10);
3. these early formed clusters involve secondary and as well tertiary structure
interactions (Tables 1 4 in Nlting and Andert, 2000);
4. peptides of barstar (Nlting et al., 1997a; Sect. 10.10), CI2 (Ladurner et al.,
1997) and barnase (Neira and Fersht, 1999) have a very low stability;
11.2 Nucleationcondensation
177
Fig. 11.12. continued from Sect. 11.1. Inter-residue contact map for the main transition state of
p53 tetramerization domain (Nlting and Andert, 2000). For explanation of the symbols see the
legend to Fig. 11.2. The p53 domain is structurally a dimer of dimers, which involves
monomer monomer (squares) and dimer dimer (diamonds) contacts.
5. folding involves a significant degree of solvent exclusion in the transition state

(Serrano et al., 1992a; Itzhaki et al, 1995a; Milla et al., 1995; Nlting et al.,
1997a; Chap. 10).
The combined information suggests that the folding of these five proteins is
initiated by nucleation condensation or a nucleation condensation-like process
(see Sect. 10.11; Nlting and Andert, 2000; Nlting, 2003).
This mechanism has first experimentally been proven for CI2 (Itzhaki et al.,
1995a). For the src SH3 domain we see a comparably simple situation with a very
clear cluster in and around strand3 and strand4 (Fig. 11.8). In barnase and barstar,
which both have an early intermediate on the folding pathway, the cluster of
178
Fig. 11.13. continued from Sect. 11.1. Consolidation of structure in the main transition state of
the p53 tetramerization domain (Nlting and Andert, 2000). Residues with 0.6 are
highlighted as small red spheres. For this tetrameric protein, the rate-limiting step is the
transition from the dimeric to the tetrameric state. The average -value for the early transition
state for the formation of the dimers is 0.0.
consolidated residues in the main transition state (Figs. 11.3, 11.5) has already a
considerable size. In these two proteins, several nuclei may have docked (Nlting
and Andert, 2000). The transition state of Arc repressor is characterized by a
relatively low consolidation, but the average degree of formation of intermolecular interactions ( = 0.17 0.02) is almost the same like that of intramolecular interactions ( = 0.21 0.01). This indicates that folding of the
monomers proceeds almost at the same time or possibly slightly earlier than the
docking of the monomers (Milla et al., 1995; Nlting and Andert, 2000;
Figs. 11.10, 11.11).
For p53 the mechanism of folding and association appears to be more
complicated (Nlting and Andert, 2000), but the -value analysis cannot rule out
the possibility of formation of one or several nuclei in the early steps of the
reaction, possibly during 2 I'2
2 I2 (see Sect. 11.1) when most of the contact
formation in the molecule proceeds.
One of the interesting observations regarding the folding of proteins is that
many folding-related physical parameters have only little dependence on the chain
length of the protein. Consistent with this general observation, the nine proteins
contained in Table 2 of Nlting and Andert (2000) show only a weak, statistically
11.3 Framework-model-like properties
179
insignificant, tendency of a larger degree of overall consolidation in the transition

state with shorter chain length. This observation is further indication that the
folding mechanism involves physical phenomena which largely reduce the
exponential growth of possibilities with the number of residues. In particular, also
this observation is consistent with a nucleationcondensation mechanism in which
the formation of the folding nucleus represents the rate-limiting step of the
reaction.
11.3
Framework-model-like properties
As shown in Table 2 in Nlting and Andert, 2000 (for barnase see also Matouschek and Fersht, 1993; Serrano et al., 1992), in five of the six transition states, the
average degree of structural consolidation is relatively higher at positions of
secondary structure than at positions which form loops in the folded structure.
Further, there is an increased percentage of secondary structure forming residues
in the most consolidated parts of these five transition states (Table 3 in Nlting
and Andert, 2000). These observations support an interpretation towards a
partial validity of the framework model (Sect. 11.2; Nlting and Andert, 2000).
On the other hand, with the exception of src SH3 domain, the for tertiary
structure contacts is on average similar to for secondary structure contacts
(Table 4 in Nlting and Andert, 2000). In the transition state of the src SH3
domain, tertiary structure interactions are weaker than secondary structure
interactions, but far from absent. Thus, in all six transition state structures, the
degree of tertiary structure consolidation is on average equivalent or almost
equivalent to that of secondary structure consolidation. Furthermore, as pointed
out in Sect. 11.2, many large peptide fragments of barstar, barnase and CI2 are
unstable in the absence of the rest of the molecule. Consequently, the secondary
structure elements can gain some degree of stability only in the presence of
significantly stabilizing tertiary structure interactions (Nlting and Andert, 2000).
So, the picture which emerges is as follows: folding of at least five of the six
investigated proteins proceeds via a nucleationcondensation mechanism, but
secondary structure interactions appear to be a major driving force in this process.
11.4
Conclusions
According to the presented high resolution of the folding processes (Figs.
11.211.13; Tables 1 4 in Nlting and Andert, 2000), nucleationcondensation
mechanism and framework model may further be reconciled in a generalized
nucleationcondensation mechanism for folding (Nlting and Andert, 2000;
Nlting, 2003):
180
1. Protein folding starts with a collapse of the molecule and structure growth in
one or several folding nuclei. These folding nuclei catalyze further folding by
reducing the necessary number of sampling steps.
2. The structure of the main transition state for folding consists of one or several
folding nuclei which may have docked or may already have attracted further
structure around them. This transition state structure contains important tertiary
as well as secondary structure interactions, but on average a relatively higher
fraction of residues which belong to secondary structure elements than the rest
of the molecule.
3. The formation of the structure of the main transition state represents the ratelimiting step of the folding reaction: it enables rapid folding into the native state
or a native-like state in which further small structural reorganizations may take
place, e.g., prolyl-peptidyl cistrans isomerizations (see Sect. 10.8.5).
It appears that this unified model is valid for the folding of a considerable number
of small monomeric and also some dimeric proteins in which additional assembly
processes of the monomers take place.
12
Structural determinants of the rate

of protein folding
It has been a long-standing question: what makes some proteins fold within less
than a millisecond, while others need minutes? This question is directly related to
the question what at all makes protein folding so astonishingly fast. It was
speculated, that the size or helix content of the protein plays an important role.
However, investigations of about 20 small proteins with two-state protein folding
kinetics showed only a moderate correlation between helix content and folding
speed and even less correlation between molecular weight and folding rate. A
breakthrough regarding this question was the discovery that the rate of two-state
folding mainly depends on the complexity of the tertiary structure of the protein
(Doyle et al., 1997; Chan, 1998; Jackson, 1998; Plaxco et al., 1998; Alm & Baker,
1999; Baker & DeGrado, 1999; Muoz & Eaton, 1999; Riddle et al., 1999; Baker,
2000; Grantcharova et al., 2000): Proteins with a very complicated tertiary structure fold slower than proteins with a simple tertiary structure. More precisely, the
so-called chain topology of the molecule plays a major role for the folding kinetics
of small proteins with two-state transitions (Nlting et al., 2003; Nlting, 2003).
12.1
Chain topology as a major structural determinant of
two-state folding
The chain topology is a measure of the average distance in sequence of contacts in
the molecule. In Fig. 12.1a we see a hypothetical protein molecule with a quite
xxxxxxx
Fig. 12.1. Illustration of the effect of the chain topology for a four-helix bundle protein (Nlting
et al., 2003). Three contacts in the molecules are shown as dashed lines.
182
12 Structural determinants of the rate of protein folding
Fig. 12.2. Inter-residue contact map for the folded structure of -repressor (Nlting et al., 2003).
Fig. 12.3. Inter-residue contact map for the folded structure of acylphosphatase (Nlting et al.,
2003).
12.1 Chain topology as a major structural determinant
183
simple chain topology: its tertiary structure contacts are only between amino acid
residues relatively close to each other in sequence. However, if the native structure
of the molecule would be like shown in Fig. 12.1b, it would have contacts of
residues far distant in sequence. This would be the case of a complicated chain
topology (Nlting et al., 2003; Nlting, 2003).
This geometrical meaning of chain topology can probably best be visualized in
inter-residue contact maps. See, for example, Fig. 12.2 which shows the interresidue contact map of -repressor an extremely fast folding protein. One can
see that this protein has many secondary structure contacts, but only few tertiary
structure contacts. Accordingly, its folding rate, kf , is high: roughly 50,000 s1 for
its thermostable variant ! In contrast, acylphosphatase (Fig. 12.3), a protein with
many contacts of residues remote in sequence, folds only with kf = 0.23 s1. So the
effect of the chain topology on kf can be quite dramatic (Nlting et al., 2003).
Originally it was discovered that for proteins with two-state folding kinetics,
log (kf) correlates with the so-called contact order, CO. Later a significantly
better correlation of log (kf) with the so-called chain topology parameter, CTP,
was found (Nlting et al., 2003). CTP is defined as:
CTP =
1
Si,2j ,
L N
(12.1)
where Si,j is the separation in sequence between the contacting residues number i
and j, L the chain length of the protein molecule, i.e., its number of residues, and N
is the total number of inter-residue contacts in the molecule (Nlting et al., 2003).
Fig. 12.4 shows this correlation for 20 small proteins and two peptides. In the
range of 101 s1 < kf < 108 s1, the correlation coefficient is 0.86. The magnitude
of CTP slightly varies with the cut-off used in the calculation of the contacts in the
xxxxxx
Fig. 12.4. Correlation of CTP with log(kf) for 20 small proteins and two peptides with two-state
folding kinetics (Nlting et al., 2003; Nlting, 2003).
184
12 Structural determinants of the rate of protein folding
protein molecule, but a similarly good correlation between CTP and log (kf ) was
found for all cut-off distances between 4 and 8.5 (Nlting et al., 2003).
12.2
Chain topology of the transition state and implications for
the mechanism of folding
The question was raised why the structure of the native state should affect the
speed of folding at all? Wouldn't the structure of the transition state be more
important? The formation of the transition state structure represents the ratelimiting event in the folding reaction, so its structure should matter and not that of
the folded state. This question was answered in Nlting et al. (2003):
1. There is already a good correlation between log (kf) and CTP of the transition
state structure.
2. The chain topology of many protein transition states has already some
similarities to the chain topology of the native state. In particular, it is likely
that the tertiary structure alignment in the transition state is already roughly
correct: otherwise the transition would likely lead to a misfolded and not the
native state. This is because wrong tertiary structure alignment could not easily
be dissolved in the already relatively compact conformations the molecule
attains in and after passing through the main transition state.
The latter point is consistent with the nucleationcondensation model in which an
essential component of the catalytic action of the folding nucleus is its roughly
correct tertiary structure (see Sects. 10.11 and 11.2).
12.3
Further factors
Investigations by Galzitskaya, Finkelstein, and coworkers show that for proteins
with three-state kinetics the chain length is an important determinant of folding
kinetics (Galzitskaya et al., 2003).
The existence of mutants with largely changed kinetics shows that individual
charge interactions can affect the folding rate, kf , by a factor of 5 and more. In
particular, mutations of salt bridges involving lysines or arginines can have a large
effect.
Log (kf) correlates also with the number of the residues which belong to
-sheets (R = 0.73 for the proteins and peptides in Fig. 12.4; Nlting et al., 2003).
This is partially because -sheets usually involve more long-range contacts than
other elements of secondary structure.
Some residues involved in the interlocking of strands of -sheets participate in
the folding nucleus and give a folding-kinetical advantage to members of the
sandwich-like protein family (Wilson and Wittung-Stafshede, 2005).
12.4 Ultrafast folding
185
12.4
Ultrafast folding
Using similar methods as described in Chap. 5, in particular temperature jumping
(see, e.g., Gruebele et al., 1998; Gruebele, 1999; Hagen and Eaton WA, 2000;
Leeson et al., 2000; Mayor et al., 2000; Yamamoto et al., 2000; Hofrichter 2001;
Urbanke and Wray, 2001; Callender and Dyer, 2002; Gillespie et al., 2003;
Gulotta et al., 2003; Kubelka et al., 2003; Maness et al., 2003; Xu et al., 2003;
Arora et al., 2004; Du et al., 2004; Vu et al., 2004; Chung et al., 2005) and optical
triggers (see, e.g., Bredenbeck et al., 2005; Buscaglia et al., 2005), several
ultrafast-folding proteins were identified. Astonishingly, there are folding rates, kf ,
above 100,000 s1. Examples for ultrafast protein folding events are:
1. the 60-residue three-helix bundle B-domain of protein A from Staphylococcus
aureus (BdpA): 1/kf ~ 3 s (Arora et al., 2004; Dimitriadis et al., 2004);
2. 3D, a designed, 73-residue three-helix bundle protein: 1/kf ~ 3 s at ~50oC
(Zhu et al., 2003);
3. the 20-residue Trp-cage miniprotein: 1/kf ~ 4 s (Qiu et al., 2002);
4. a 35-residue subdomain of the villin headpiece: 1/kf ~ 4 s at ~27oC (Kubelka
et al., 2003);
5. cytochrome b562 , a 106-residue four-helix bundle protein: 1/kf ~ 5 s (WittungStafshede et al., 1999);
6. the 61-residue Engrailed homeodomain (En-HD), a three-helix bundle protein:
1/kf ~ 1 s for an intermediate with much native -helical secondary structure
and 1/kf ~ 25 s for folding into the native state (Mayor et al., 2003), probably
through a compact native-like transition state (DeMarco et al., 2004).
According to simulations, the least stable helix of En-HD, helix2 , unfolds in
< 450 ps at high temperature (DeMarco et al., 2004).
The formation of monomeric helices takes approximately 100 500 ns (Zhu et al.,
2004). Ultrafast folding of proteins is mainly limited by diffusion-collision (Myers
and Oas, 2002; Vu et al., 2004), internal friction (Pabit et al., 2004), and the
magnitude of the hydrophobic effect (Zhu et al., 2004). Fast formation of helices
can drive the ultrafast folding of helical proteins (Vu et al., 2004).
This subclass of proteins behaves differently than the average proteins treated
in the previous sections: ultrafast-folding proteins have
1. usually a simple chain topology with an accordingly very low chain topology
parameter, CTP (see Eq. 12.1);
2. very often a small size with a number of amino acid residues of < 80;
3. often a high helix content.
13
Evolutionary computer programming of

protein structure and folding
As demonstrated in the previous chapters, protein folding is an extremely

complicated, but also highly efficient process. Since a long time it is desired to
simulate protein folding on computers. However, one of the major problems to do
so is the lack of an efficient mathematical description of multi-body problems like
the movements of the structural elements of a biological macromolecule in the
course of the folding reaction. It is well known that we do not have an analytical
solution for the general three-body problem, i.e., the non-periodical movement of
three gravitationally or electromagnetically interacting bodies in space. So,
currently it seems hopeless to obtain an analytical solution for the mechanics of a
complicated object like a protein. What makes the problem even worse is that
humans cannot really foresee in their imagination the non-periodical movement of
many bodies in space. This may actually be one of the major reasons why we still
have not found a mathematical description of the multi-body problem that can be
solved in a simple manner. Currently most physicists believe that the reason for
the impossibility of finding an analytical solution for the multi-body problem is
the lack of a sufficient number of constants of motion: the equations one can write
down contain too many free parameters. But nature can solve with high precision
the multi-body problem for macroscopic, non-quantum mechanical bodies. So
possibly one reason for our inability to find a simple solution might be that there is
something wrong in our mathematics or physics. But so far, nobody could find out
the possible shortcoming in our theories, and again this may be because humans
currently are not smart enough to map the complicated motions into their brains
and then, based on thought experiments, find a comprehensible theoretical
description which leads to simple solutions. So what appears to be needed is a
machine which is, at least in certain aspects, smarter than humans. But, how can
humans develop systems partially more intelligent than themselves? Is it possible,
that the ape develops higher mathematics? Probably not.
Fortunately nature shows us the way out of this dilemma: If we cannot directly
develop an ultra-capable system, we have to make it evolve itself to a level
beyond our direct intellectual abilities. Is this possible? The answer is yes! For
example, recently a self-evolving computer program method was created and
applied on the calculation of protein structures (Nlting et al., 2004). After a small
number of evolution steps, the program is already much more efficient than a
program based on rational design (see Fig. 13.1), and intriguingly, it was
impossible to say precisely why the evolved structural features of the program
188
13 Evolutionary computer programming ...
code make this program so efficient. The code of the evolved program differs
significantly from the original wild-type program, but it appears to be impossible
to fully rationalize its higher success.
13.1
Evolution method
The computer evolution method proceeds as follows (Fig. 13.1):
1. An initial, so-called wild-type program is developed. The program contains
parts which can be changed (mutated) without causing a complete failure of the
program.
2. The following steps are passed through for a certain number of cycles
(evolution steps): a) A number of so-called program mutants is created: in the
mutants, the changeable parts are altered. b) The performance of the mutants is
tested. The best-performing variant within the set of mutants and wild-type
program is used as a template for further mutagenesis (step a).
3. The highly evolved program may now be used for the application it was
evolved for, but usually also for other, similar tasks.
Fig. 13.1. Evolution of a computer program: A wild-type program evolves itself according to an
applied evolutionary pressure (see the text and Nlting et al., 2004).
13.2 Protein folding and structure predictions
189
The computer evolution proceeds similar to the evolution of species in nature

according to the principles of mutation and selection. Higher evolved programs
may also take advantage of the technique of gene shuffling.
In contrast to simple self-learning programs, e.g., programs with neuronal
networks, in the course of the evolution, a self-evolving program can significantly
change its code and structure: the program evolution is not limited to a certain
program structure as it would be the case if, e.g., only parameters in given tables
were changed.
As shown in the next section, this evolution method was performed specifically
for a program for protein folding and structure predictions (Nlting et al., 2004).
13.2
Protein folding and structure predictions
For the application of self-evolution of a computer program, protein folding and
structure prediction is a suitable task because:
1. We do not know enough about the mechanism and mechanics of folding to be
able to create, based on rational design, a fast and precise program for folding
simulations.
2. The interaction energies are not known precisely enough to determine the
conformation with the lowest energy.
3. The number of conformations of unfolded proteins is so astronomically large
(see Chap. 1) that it is currently impossible to find the conformation with the
lowest energy by calculation of the energies of all conformations even if the
interaction energies were precisely known.
For details of the program see Nlting et al. (2004). Briefly, the protein structure
is approximated by a hexagonal lattice model (Fig. 13.2). The program was
designed to calculate 64-residue proteins. For larger proteins, only the first 64
residues were selected. The potential of interaction between amino acids is
V-shaped with a strong repulsion at low distances, an attractive force at moderate
distances, and a slowly rising repulsion at larger distances (see Nlting et al.,
2004).
Fig. 13.3 shows the structure of the program: The program generates a number
of start conformations and tries to fold them. All start conformations consist of
random combinations of secondary structure elements. This has been shown to
speed up the calculations compared to purely random start conformations. The
mutatable parts (genes) of the program encode for the structural representation of
the molecule and, most importantly, for the folding reaction (Nlting et al., 2004).
Most genes (gene 1 gene n in Fig. 13.3) encode for molecular movements in the
folding reaction, e.g., rotations around single bonds, simultaneous rotations around
two bonds. Genes of the program can be mutated, e.g., by changing the region of
the molecule on which a certain movement is exerted, changing the direction of
molecular movement, or by deleting, adding, or exchanging genes.
190
Fig. 13.2. Hexagonal lattice model of the protein structure used for the folding and structure
calculations (Nlting et al., 2004). Each amino acid residue is modeled as a sphere. The open
circles indicate the possible positions of a residue relative to another one (closed circle). The
bond angles and in the horizontal and vertical direction, respectively, can attain only
multiples of 30o.
Fig. 13.3. Principle of operation of the self-evolving computer program for protein structure and
folding calculations (see the text and Nlting et al., 2004).
191
Fig. 13.4. Evolution of the program. In this case, the program evolution was performed with the
protein CI2. However, the evolved program has been shown to be applicable also on other small
proteins (see, e.g., Nlting et al., 2004). At the evolution step 20, the lattice was changed from
the geometry described in (Nlting et al., 2004) to the hexagonal geometry shown in Fig. 13.2.
As shown in Fig. 13.4, the evolution proceeds quite rapidly with some jumps at
different stages. The evolutionary pressure applied in this specific case was
towards finding deeper minima in the energy landscape within a given time period
(Nlting et al., 2004). Other additional pressures are feasible, e.g., towards higher
compactness of the protein molecule.
At the cut-off for the contact distance of 8 used for all inter-residue contact
maps in this section, the native structure of phosphatidylinositol 3-kinase displays
extensive clusters of tertiary structure interactions around 265 and 5628 and no
interactions around 4016 in the inter-residue contact space (Fig. 13.5). These
important features of the structure were approximately correctly predicted by the
simulations (Fig. 13.6). The five structures with the lowest energies consistently
display these features in the inter-residue contact maps (not shown).
Analogously, the simulations confirmed important features of the native
structure of acyl-coenzyme A binding protein, in particular its completely
different structure of inter-residue contact map compared to phosphatidylinositol
3-kinase: in large parts of the inter-residue contact space this protein has only few
tertiary structure interactions (Fig. 13.7). In particular, the absence of interactions
in the region around 5010 is correctly predicted (Fig. 13.8).
These are only first results far from perfection, but considering the extreme
number of conformations, 1064, it is quite surprising that important features of the
protein structures could be calculated by the evolved program which does not use
192
Fig. 13.5. Inter-residue contact map of the NMR structure of phosphatidylinositol 3-kinase
(Nlting et al., 2004). The protein has a large number of contacts between the residues around
number 26 and around number 5, and between the residues around 56 and around number 28.
Fig. 13.6. Inter-residue contact map of the calculated structure of phosphatidylinositol 3-kinase.
The map reflects the superposition of the three structures with the lowest energy.
193
Fig. 13.7. Inter-residue contact map of the NMR structure of the acyl-coenzyme A binding
protein.
Fig. 13.8. Inter-residue contact map of the calculated structure of the acyl-coenzyme A binding
protein. The map reflects the superposition of the three structures with the lowest energy.
194
any kind of sequence alignment. It should be pointed out that most other
successful lattice simulations without sequence alignment were restricted to
27-residue miniproteins which have roughly 1037 times fewer conformations.
Within the only 20 minutes of simulations on a PC for Figs. 13.6 and 13.8, the
evolved program achieved a quality of folding that could not be achieved by the
initial program within a 100 times longer period of time, as judged be the energy,
compactness and match with known structures of the calculated conformations.
This shows that, in principle, it is possible to let a self-evolving program learn
how to fold a protein without putting much information about the folding pathway
into the initial program. From time to time there are jumps in the evolution
(Fig. 13.4), thus, significant further improvement should easily be obtainable by
continuing the evolution process and using longer simulations.
13.3
Further potential applications of the evolution method
Obviously, the applicability of the evolution method is by far no means limited to
protein folding and structure predictions. Self-evolving programs are probably
potentially useful wherever scientific or economical systems are too complex to be
rationalized by humans. In fact, it is expected that similar self-evolving systems as
presented here will eventually become one of the major sources of further success
in science and technology. It appears important that in contrast to many previous
approaches to learn from nature (for a survey on evolutionary algorithms see, e.g.,
Back, 1996), here not only parameters in given tables of a program are changed,
but the program is given the ability to change its code. The unprecedented
capability of self-evolving systems is that potentially systems and machines can
emerge which are not only faster, but truly smarter than humans, at least regarding
certain tasks. Potential applications range from the solution of astronomical
movements, descriptions of ecosystems, the advanced design of optical devices,
robots, and nanomachines, to the prediction of stock indices and sociological and
biological phenomena including the evolution of mankind itself.
14 Conclusions
The mystery of submillisecond folding events is in the process of being resolved,

along with the development of a variety of new, faster and structurally higherresolving, kinetic methods. Novel techniques enable the examination of fast
folding events on a microsecond to picosecond time scale to be made.
Atomic resolution on a submillisecond time scale is achieved by NMR line
broadening experiments if a sufficient population of the species involved is
available, and by H/D exchange NMR experiments.
A combination of rapid kinetic-, equilibrium thermodynamic- and protein
engineering methods, the so-called -value analysis, enables the structures of
folding intermediates at the level of individual amino acid residues, and with
microsecond to nanosecond time resolution, to be mapped out. Beyond this, the
-value analysis can provide detailed maps for the very important structures of
transition states, i.e., for the structures of which formation represents the ratelimiting events in the folding reaction.
A variety of microsecond folding events has been discovered and structurally
characterized, ranging from the folding of the B-domain of staphylococcal protein
A in ~ 3 s to the formation of a subdomain in apo-myoglobin in 7 s at 22oC to
the initial collapse of cytochrome c in 40 s at 40oC to the nucleationcondensation of barstar in 500 s at 8oC (300 s at 15oC).
New techniques now enable the predictions of various theoretical models to be
tested directly. The folding of the 10-kDa protein barstar and four further small
proteins is shown to be consistent with a nucleationcondensation model (Fig.
10.25; Chap. 11). The classical nucleation growth mechanism invokes the initial
formation of a well-defined nucleus followed by growth of structure from it
(Wetlaufer, 1973, 1990). The nucleation condensation mechanism, on the other
hand, has a diffuse nucleus that contains a few neighboring amino acid residues
whose conformations are stabilized by long-range interactions with amino acid
residues that are remote in the primary structure. An essential component of the
nucleationcondensation mechanism is that the nucleus and its stabilizing interactions elsewhere in the protein develop concurrently: For example, the -helices 1,
2, and 4 of barstar are unstable in the absence of long-range interactions, and the
rest of the structure is unstable without the interactions with the helices: There is
cooperative formation of the folding nucleus and the surrounding structure (Chap.
10; Nlting et al., 1997a; Nlting and Andert, 2000; Nlting, 1998a, 2003).
Many important biological processes and disease states which involve protein
folding-, misfolding-, and aggregation reactions have been poorly understood
196
14 Conclusions
because of lack of extensive information at a molecular level. Examples are

spongiform encephalopathies (CreutzfeldtJacob disease in humans; BSE in
cattle; scrapie in sheep; see Masters and Beyreuter, 1997), Huntingtons-, and
Alzheimers diseases.
This book introduces important mathematical, biophysical, and molecular
biological methods which can support the high kinetic and structural resolution of
such biologically and chemically relevant processes. With these truly interdisciplinary methods, different reaction mechanisms can be far better discriminated than was previously possible. Beyond this, protein structures along the
folding pathway can now be visualized at the level of individual amino acid
residues in nearly any biologically relevant time scale. This detailed mechanistic
knowledge will further aid the understanding of biological processes and disease
states, and will eventually help us to find rational ways for re-designing biological
processes, and to find cures for diseases.
References
Abkevich VI, Gutin AM, Shakhnovich EI (1994a) Free energy landscape for protein folding
kinetics. Intermediates, traps and multiple pathways in theory and lattice model simulations.
J. Chem. Phys. 101:60526062
Abkevich VI, Gutin AM, Shakhnovich EI (1994b) Specific nucleus as the transition state for
protein folding: Evidence from the lattice model. Biochemistry 33:1002610036
Abkevich VI, Gutin AM, Shakhnovich EI (1996) Improved design of stable and fast-folding
model proteins. Folding & Design 1:221230
Abola EE, Bernstein FC, Bryant SH, Koetzle TF, Weng J (1987) Protein data bank. In:
Allen FH, Bergerhoff G, Sievers R (eds) Crystallographic databases information content,
software systems, scientific applications. Data Commission of the International Union of
Crystallography, Bonn Cambridge Chester, 107132.
Abola EE, Sussman JL, Prilusky J, Manning NO (1997) Protein data bank archives of
three-dimensional macromolecular structures. Methods Enzymol. 277:556571.
Agashe VR, Udgaonkar JB (1995) Thermodynamics of denaturation of barstar: Evidence for
cold denaturation and evaluation of the interaction with guanidine hydrochloride.
Biochemistry 34:32863299
Alm E, Baker D (1999) Matching theory and experiment in protein folding. Curr. Opin. Struct.
Biol. 9:189196
Arai M, Kuwajima K (1996) Rapid formation of a molten globule intermediate in refolding of
-lactalbumin. Folding & Design 1:275287
Arora P, Oas TG, Myers JK (2004) Fast and faster: A designed variant of the B-domain of
protein A folds in 3 sec. Protein Science 13:847853
Atta-ur-Rahman (1986) Nuclear magnetic resonance. Springer-Verlag, New York Berlin
Heidelberg
Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman JG, Smith JA, Struhl K (eds) (1992)
Short protocols in molecular biology. Wiley & Sons, New York
Baker D (2000) A surprising simplicity to protein folding. Nature 405:3942
Baker D, DeGrado WF (1999) Engineering and design. Curr. Opin. Struct. Biol. 9:485486
Back T (1996) Evolutionary Algorithms in Theory and Practice: Evolution Strategies,
Evolutionary Programming, Genetic Algorithms. Oxford University Press
Bacon DJ, Anderson WF (1988) A fast algorithm for rendering space-filling molecule
pictures. J. Mol. Graphics 6:219220
Balbach J, Forge V, Vannuland NAJ, Winder SL, Hore PJ, Dobson CM (1995) Following
protein folding in real-time using NMR spectroscopy. Nature Struct. Biol. 2:865870
Baldwin RL (1996) On-pathway versus off-pathway folding intermediates.
Folding & Design 1:R1R8
Ballew RM, Sabelko J, Gruebele M (1996a) Observation of distinct nanosecond and
microsecond protein folding events. Nature Struct. Biol. 3:923926
Ballew RM, Sabelko J, Reiner C, Gruebele M (1996b) A single-sweep, nanosecond time
resolution laser temperature-jump apparatus. Rev. Sci. Instrum. 67:36943699
Ballew RM, Sabelko J, Gruebele M (1996c) Direct observation of fast protein folding: The
initial collapse of apomyoglobin. Proc. Natl. Acad. Sci. USA 93:57595764
Becker OM, Karplus M (1997) The topology of multidimensional potential-energy surfaces:
Theory and application to peptide structure and kinetics. J. Chem. Phys. 106:14951517
198
References
Beyer WH (1991) CRC standard mathematical tables and formulae. CRC Press, Boca Raton
Ann Arbor Boston London, 29th Ed.
Billeter M, Schaumann T, Braun W, Wthrich K (1990) Restrained energy refinement with two
different algorithms and force fields of the structure of the -amylase inhibitor tendamistat
determined by NMR in solution. Biopolymers 29:695706
Bismuto E, Irace G, Sirangelo I, Gratton E (1996) Pressure-induced perturbation of ANS
apomyoglobin complex: Frequency domain fluorescence studies on native and acidic
compact states. Prot. Sci. 5:121126
Bjrling SC, Goldbeck RA, Paquette SJ, Milder SJ, Kliger DS (1996) Allosteric intermediates
in hemoglobin: 1. Nanosecond time-resolved circular dichroism spectroscopy.
Bredenbeck J, Helbing J, Kumita JR, Woolley GA, Hamm P (2005) -helix formation in a
photoswitchable peptide tracked from picoseconds to microseconds by time-resolved IR
spectroscopy. Proc. Natl. Acad. Sci. USA 102:23792384
Booth DR, Sunde M, Bellotti V, Robinson CV, Hutchinson WL, Fraser PE, Hawkins PN,
Dobson CM, Radford SE, Pepys MB, Blake CCF (1997) Instability, unfolding and
aggregation of human lysozyme variants underlying amyloid fibrillogenesis.
Nature 385:787793
Boxer SG, Anfinrud PA (1994) Casting a cold eye over myoglobin. Nat. Struct. Biol. 1:749751
Brandl G, Kastner F, Mannschreck A, Nlting B, Andert K, Wetzel R (1991) Chiroptical
detection during liquid chromatography. J. Chromatogr. 586:249254
Bryngelson JD, Onuchic JN, Socci ND, Wolynes PG (1995) Funnels, pathways, and the energy
landscape of protein folding: A synthesis. Proteins: Struct. Funct. Genetics 21:167195
Bundi A, Wthrich K (1979) 1H NMR parameters of the common amino acids residues
measured in aqueous solution of the linear tetrapeptides HGlyGlyXAlaOH.
Biopolymers 18:285297
Burova TV, Bernhardt R, Pfeil W (1995) Conformational stability of bovine holo- and apoadrenodoxin: A scanning calorimetric study. Prot. Sci. 4:909916
Burton RE, Huang GS, Daugherty MA, Calderone TL, Oas TG (1997) The energy landscape of
a fast-folding protein mapped by AlaGly substitutions. Nat. Struct. Biol. 4:305310
Buscaglia M, Kubelka J, Eaton WA, Hofrichter J (2005) Determination of ultrafast protein
folding rates from loop formation dynamics. J. Mol. Biol. 2005 347:657664
Bychkova VE, Ptitsyn OB (1995) Folding intermediates are involved in genetic diseases?
FEBS Lett. 359, 68
Bycroft M, Matouschek A, Kellis JT, Serrano L, Fersht AR (1990) Detection and
characterization of a folding intermediate in barnase by NMR. Nature 346:488490
Cabani S, Gianni P, Mollica V, Lepori L (1981) Group contributions to the thermodynamic
properties of non-ionic organic solutes in dilute aqueous solution. J. Sol. Chem. 10:563595
Callender R, Dyer RB (2002) Probing protein dynamics using temperature jump relaxation
spectroscopy. Curr. Opin. Struct. Biol. 12:628633
Cantor CR, Schimmel PR (1980) Biophysical chemistry. W. H. Freeman and Co., New York
Cathou RE, Hammes GG (1964) Relaxation spectra of ribonuclease. I. The interaction of
ribonuclease with cytidine 3'phosphate. J. Am. Chem. Soc. 86:32403245
Cathou RE, Hammes GG (1965) Relaxation spectra of ribonuclease. III. Further investigation of
the interaction of ribonuclease and cytidine 3'phosphate. J. Am. Chem. Soc. 87:46744680
Causgrove TP, Dyer RB (1993) Protein response to photodissociation of CO from
carbonmonoxy myoglobin probed by time-resolved infrared spectroscopy of the amide I
band. Biochemistry 32:1198511991
Chalikian TV, Sarvazyan AP, Breslauer KJ (1994) Hydration and partial compressibility of
biological compounds. Biophys. Chem. 51:89109
Chalikian TV, Gindikin VS, Breslauer KJ (1995) Volumetric characterizations of the native,
molten globule and unfolded states of cytochrome c at acidic pH. J. Mol. Biol. 250:291306
References
199
Chalikian TV, Totrov M, Abagyan R, Breslauer KJ (1996) The hydration of globular proteins as
derived from volume and compressibility measurements: Cross correlating thermodynamic
and structural data. J. Mol. Biol. 260:588603
Chan HS (1995) Kinetics of protein folding. Nature 373:664665
Chan HS (1998) Protein folding: Matching speed and locality. Nature 392:761763
Chan CK, Hofrichter J, Eaton WA (1996) Optical triggers of protein folding.
Science 274:628629
Chan CK, Hu Y, Takahashi S, Rousseau DL, Eaton WA, Hofrichter J (1997) Submillisecond
protein folding kinetics studied by ultrarapid mixing.
Proc. Natl. Acad. Sci. USA 94:17791784
Chen EF, Lapko VN, Song PS, Kliger DS (1997) Dynamics of the N-terminal -helix unfolding
in the photoreversion reaction of phytochrome A. Biochemistry 36:49034908
Chen E, Wood MJ, Fink AL, Kliger DS (1998) Time-resolved circular dichroism studies of
protein folding intermediates of cytochrome c. Biochemistry 37:55895598
Cherepanov AV, De Vries S (2004) Microsecond freeze-hyperquenching: development of a new
ultrafast micro-mixing and sampling technology and application to enzyme catalysis.
Biochim. Biophys. Acta 1656:131
Chiti F, Taddei N, White PM, Bucciantini M, Magherini F, Stefani M, Dobson CM (1999)
Mutational analysis of acylphosphatase suggests the importance of topology and contact order
in protein folding. Nature Struct. Biol. 6:10051009
Chothia C, Hubbard T, Brenner S, Barns H, Murzin A (1997) Protein folds in the all-beta and
all-alpha classes. Annu. Rev. Biophys. Biomol. Struct. 26:597627
Chou PY (1989) Prediction of protein structural class from amino acid compositions. In:
Fasman GD (ed) Prediction of protein structure and the principles of protein conformation.
Plenum, New York, 549586
Chou PY, Fasman GD (1977) -turns in proteins. J. Mol. Biol. 115:135175
Chou PY, Fasman GD (1978a) Prediction of the secondary structure of proteins from their
amino acid sequence. Adv. Enzymol. 47:45148
Chou PY, Fasman GD (1978b) Empirical predictions of protein conformation.
Annu. Rev. Biochem. 47:251276
Christopher JA, Baldwin TO (1996) Implications of N-terminal and C-terminal proximity for
protein folding. J. Mol. Biol. 257:175187
Chung HS, Khalil M, Smith AW, Ganim Z, Tokmakoff A (2005) Conformational changes during
the nanosecond-to-millisecond unfolding of ubiquitin. Proc. Natl. Acad. Sci. USA
102:612617
Clarke J, Fersht AR (1993) Engineered disulfide bonds as probes of the folding pathway of
barnase: Increasing the stability of proteins against the rate of denaturation.
Cohn EJ, Edsall JT (1942) Proteins, amino acids, and peptides as ions and dipolar ions.
Reinhold, New York
Coligan JE, Dunn BM, Ploegh HL, Speicher DW, Wingfield PT (eds) (1996) Current protocols
in protein science. Wiley & Sons, New York
Cowan SW, Schirmer T, Rummel G, Steiert M, Ghosh R, Pauptit RA, Jansonius JN,
Rosenbusch JP (1992) Crystal structures explain functional properties of two E. coli porins.
Nature 358:727733
Creighton TE (1993) Proteins structures and molecular properties. W.H.Freeman and
Company, New York, 2nd Ed.
Croasmun WR, Carlson MK (eds) (1994) Two-dimensional NMR spectroscopy. Methods in
stereochemical analysis. VCH, New York
Cvijovic D, Klinowski J (1995) Taboo search: An approach to the multiple minima problem.
Science 267:664666
200
References
Damaschun G, Damaschun H, Gast K, Misselwitz R, Mller JJ, Pfeil W, Zirwer D (1993) Cold
denaturation-induced conformational change in phosphoglycerate kinase from yeast.
Dauber P, Hagler AT (1980) Crystal packing, hydrogen bonding, and the effect of crystal forces
on molecular conformation. Accts. Chem. Res. 13:105112
Dawson RMC, Elliott DC, Elliott WH, Jones KM (1969) Data for biochemical research. Oxford
University Press, 2nd Ed.,
DeMaeyer LCM (1969) Electric field methods. Methods Enzymol. 16:80118
DeMarco ML, Alonso DOV, Daggett V (2004) Diffusing and colliding: The atomic level
folding/unfolding pathway of a small helical protein. J. Mol. Biol. 341:11091124
Dill KA (1985) Theory for the folding and stability of globular proteins. Biochemistry
24:15011509
Dill KA (1990a) Dominant forces in protein folding. Biochemistry 29:71337155
Dill KA (1990b) The meaning of hydrophobicity. Science 250:297298
Dill KA, Chan HS (1997) From Levinthal to pathways to funnels. Nature Struct. Biol. 4:1019
Dimitriadis G, Drysdale A, Myers JK, Arora P, Radford SE, Oas TG, Smith DA (2004)
Microsecond folding dynamics of the F13W G29A mutant of the B domain of staphylococcal
protein A by laser-induced temperature jump. Proc. Natl. Acad. Sci. USA 101:38093814
Dolgikh DA, Gilmanshin RI, Brazhnikov EV, Bychkova VE, Semisotnov GV,
Venjaminov SY, Ptitsyn OB (1981) -Lactalbumin compact state with fluctuating
secondary structure. FEBS Lett. 136:311315
Doyle R, Simons K, Qian H, Baker D (1997) Local interactions and the optimization of protein
folding. Proteins 29:282291
Du D, Zhu Y, Huang CY, Gai F (2004) Understanding the key factors that control the rate of
-hairpin folding. Proc. Natl. Acad. Sci. USA 101:1591515920
Duddeck H (1995) Gnther Snatzke. Liebigs Annalen 6:IXIII
Dyer RB, Einarsdttir O, Killough PM, Lpez-Garriga JJ, Woodruff WH (1989) Transient
binding of photodissociated CO to CuB+ of eukaryotic cytochrome oxydase at ambient
temperature. Direct evidence from time-resolved infrared spectroscopy.
J. Am. Chem. Soc. 111:76577659
Dyer RB, Peterson KA, Stoutland PO, Woodruff WH (1994) Picosecond infrared study of the
photodynamics of carbonmonoxy cytochrome c oxydase. Biochemistry 33:500507
Dyson HJ, Wright PE (1996) Insights into protein folding from NMR.
Annu. Rev. Phys. Chem. 47:369395
Eaton WA, Thompson PA, Chan CK, Hagen SJ, Hofrichter J (1996a) Fast events in protein
folding. Structure 4:11331139
Eaton WA, Henry ER, Hofrichter J (1996b) Nanosecond crystallographic snapshots of protein
structural changes. Science 274:16311632
Eaton WA, Muoz V, Thompson PA, Chan CK, Hofrichter J (1997) Submillisecond kinetics of
protein folding. Curr. Opin. Struct. Biol. 7:1014
Eftink MR, Shastry MCR (1997) Fluorescence methods for studying kinetics of protein folding
reactions. Methods Enzymol. 278:258286
Eggers F, Kustin K (1969) Ultrasonic methods. Methods Enzymol. 16:5580
Eigen M (1996) Prionics or the kinetic basis of prion diseases. Biophys. Chem. 63:A1A18
Eigen M, deMaeyer L (1963) In: Friess SL, Lewis ES and Weissberger A (eds)
Techniques of Organic Chemistry. Wiley Interscience, New York, pp. 8951054
Eigen M, Hammes GG, Kustin K (1960) Fast reactions of imidazole studies with relaxation
spectroscopy. J. Am. Chem. Soc. 82:34823483
Einterz CM, Lewis JW, Milder SJ, Kliger DS (1985) Birefringence effects in transient
circular dichroism measurements with application to the photolysis of
carbonmonoxy hemoglobin and carbonmonoxy myoglobin. J. Phys. Chem. 89:38453853
Eliezer D, Yao J, Dyson HJ, Wright PE (1998) Structural and dynamic characterization of
partially folded states of apomyoglobin and implications for protein folding.
References
201
Nat. Struct. Biol. 5:148155

Elve GA, Chaffotte AF, Roder H, Goldberg ME (1992) Early steps in cytochrome c folding
probed by time-resolved circular dichroism and fluorescence spectroscopy.
Elve GA, Bhuyan AK, Roder H (1994) Kinetic mechanism of cytochrome c folding:
Involvement of the heme and its ligands. Biochemistry 33:69256935
Englander SW, Mayne L (1992) Protein folding studied using hydrogen exchange labeling and
two-dimensional NMR. Annu. Rev. Biophys. Biomol. Struct. 21:243265
Ernst RR, Anderson WA (1966) Application of Fourier transform spectroscopy to magnetic
resonance. Rev. Sci. Instrum. 37:93103
Esquerra RM, Lewis JW, Kliger DS (1997) An improved linear retarder for time-resolved
circular dichroism studies. Rev. Sci. Instrum. 68:13721376
Feltch SM, Stuehr JE (1979) Relaxation studies of enzymes: Rapid isomerization in
desoxyribonuclease I. Biochemistry 18:20002004
Fersht AR (1985) Enzyme structure and mechanism. Freeman and company, New York, 2nd Ed.
Fersht AR (1992) Pathway of protein folding. Faraday Discussions 93:183193
Fersht AR (1993) Protein folding and stability: The pathway of folding of barnase.
FEBS Lett. 325:516
Fersht AR (1995a) Characterizing transition states in protein folding: An essential step in the
puzzle. Curr. Opin. Struct. Biol. 5:7984
Fersht AR (1995b) Mapping the structures of transition states and intermediates in folding:
Delineation of pathways at high resolution.
Phil. Transactions Royal Soc. London B: Biol. Sci. 348:1115
Fersht AR (1995c) Optimization of rates of protein folding: The nucleationcondensation
mechanism and its implications. Proc. Natl. Acad. Sci. USA 92:1086910873
Fersht AR (1997) Nucleation mechanisms in protein folding. Cur. Opin. Struct. Biol. 7:39
Fersht AR (1999) Structure and mechanism in protein science. New York. WH Freeman and Co.
Fersht AR, Bycroft M, Horovitz A, Kellis JT, Matouschek A, Serrano L (1991) Pathway and
stability of protein folding. Phil. Transactions Royal Soc. London B: Biol. Sci. 332:171176
Fersht AR, Matouschek A, Serrano L (1992) The folding of an enzyme. 1. Theory of protein
engineering analysis of stability and pathway of protein folding. J. Mol. Biol. 224:771782
Fersht AR, Itzhaki LS, ElMasry NF, Matthews JM, Otzen DE (1994) Single versus parallel
pathways of protein folding and fractional formation of structure in the transition state.
Fink AL (1995) Compact intermediate states in protein folding.
Annu. Rev. Biophys. Biomol. Struct. 24:495522
Fink AL, Oberg KA, Seshadri S (1998) Discrete intermediates versus molten globule models for
protein folding: Characterization of partially folded intermediates of apomyoglobin.
Folding & Design 3:1925
Finkelstein AV (1997) Protein structure: What is it possible to predict now?
Curr. Opin. Struct. Biol. 7:6071
Finkelstein AV, Badretdinov AY (1997) Rate of protein folding near the point of thermodynamic
equilibrium between the coil and the most stable chain fold. Folding & Design 2:115121
Fisher MT, Sligar SG (1987) Temperature-jump relaxation kinetics of the P-450cam spin
equilibrium. Biochemistry 26:47974803
Flory PJ (1969) Statistical mechanics of chain molecules. Wiley, New York
Foguel D, Weber G (1995) Pressure-induced dissociation and denaturation of allophycocyanin
at subzero temperatures. J. Biol. Chem. 270:2875928766
Franzen S, Bohn B, Poyart C, Martin JL (1995) Evidence for sub-picosecond heme doming in
hemoglobin and myoglobin: A time-resolved resonance Raman comparison of
carbonmonoxy and deoxy species. Biochemistry 34:12241237
French TC, Hammes GG (1969) The temperature-jump method. Methods Enzymol. 16:330
202
References
Freund SMV, Wong KB, Fersht AR (1996) Initiation sites of protein folding by NMR analysis.
Frieden C, Hoeltzli SD, Ropson IJ (1993) NMR and protein folding: Equilibrium and stoppedflow studies. Protein Science 2:20072014
Fulton KF, Main ERG, Daggett V, Jackson SE (1999) Mapping the interactions present in the
transition state for unfolding/folding of FKBP12. J. Mol. Biol. 291:445461
Galzitskaya OV, Garbuzynskiy SO, Ivankov DN, Finkelstein AV (2003) Chain length is the
main determinant of the folding rate for proteins with three-state folding kinetics. Proteins:
Struct. Funct. Genetics 51:162166
Gast K, Damaschun G, Damaschun H, Misselwitz R, Zirwer D (1993) Cold denaturation of
yeast phosphoglycerate kinase: Kinetics of changes in secondary structure and compactness
on unfolding and refolding. Biochemistry 32:77477752
Gast K, Damaschun G, Desmadril M, Minard P, Mller-Frohne M, Pfeil W, Zirwer D (1995)
Cold denaturation of yeast phosphoglycerate kinase: Which domain is more stable.
FEBS Lett. 358:247250
Gast K, Noppert A, Mller-Frohne M, Zirwer D, Damaschun G (1997) Stopped-flow dynamic
light scattering as a method to monitor compaction during protein folding.
Eur. Biophys. J. Biophys. Lett. 25:211219
Gavish B, Gratton E, Hardy CJ (1983a) Adiabatic compressibility of globular proteins.
Gavish B, Gratton E, Hardy CJ, Stdenis A (1983b) Differential sound velocity apparatus for the
measurement of protein solutions. Rev. Sci. Instrum. 54:17561760
Gillespie B, Vu DM, Shah PS, Marshall SA, Dyer RB, Mayo SL, Plaxco KW (2003) NMR and
temperature-jump measurements of de novo designed proteins demonstrate rapid folding in
the absence of explicit selection for kinetics. J. Mol. Biol. 330:813819
Gilmanshin R, Williams S, Callender RH, Woodruff WH, Dyer RB (1997a) Fast events in
protein folding: Relaxation dynamics and structure of the I-form of apomyoglobin.
Gilmanshin R, Williams S, Callender RH, Woodruff WH, Dyer RB (1997b) Fast events in
protein folding: Relaxation dynamics of secondary and tertiary structure in native
apomyoglobin. Proc. Natl. Acad. Sci. USA 94:37093713
Gilmanshin R, Callender RH, Dyer RB (1998) The core of apomyoglobin E-form folds at the
diffusion limit. Nat. Struct. Biol. 5:363365
Goldbeck RA (1988) Sign variation in the magnetic circular dichroism spectra of -substituted
porphyrins. Accts. Chem. Res. 21:95101
Goldbeck RA, Kliger DS (1993) Nanosecond time-resolved absorption and polarization
dichroism spectroscopies. Methods Enzymol. 226:147177
Goldbeck RA, Dawes TD, Einarsdttir O, Woodruff WH, Kliger DS (1991) Time-resolved
magnetic circular dichroism spectroscopy of photolyzed carbonmonoxy cytochrome c
oxidase (cytochrome aa3). Biophys. J. 60:125134
Goldbeck RA, Einarsdttir O, Dawes TD, O'Connor DB, Surerus KK, Fee JA, Kliger DS
(1992) Magnetic circular dichroism study of cytochrome ba3 from Thermus thermophilus:
Spectral contributions from cytochrome b and a3 and nanosecond spectroscopy of CO
photodissociation intermediates. Biochemistry 31:93769387
Goldenberg DP, Frieden RW, Haack JA, Morrison TB (1989) Mutational analysis of a proteinfolding pathway. Nature 338:127133
Goloubinoff P, Gatenby AA, Lorimer GH (1989a) GroE heat shock proteins assist assembly of
foreign prokaryotic ribulose bisphosphate carboxylase oligomers in Escherichia coli.
Nature 337:4447
Goloubinoff P, Christeller JT, Gatenby AA, Lorimer GH (1989b) Reconstruction of active
dimeric ribulose bisphosphate carboxylase from an unfolded state depends on 2 chaperonin
proteins and MgATP. Nature 342:884888
References
203
Grantcharova VP, Riddle DS, Baker D (2000) Long-range order in the src SH3 folding transition
state. Proc. Natl. Acad. Sci. USA 97:70847089
Griebenow K, Klibanov AM (1997) Can conformational changes be responsible for solvent and
excipient effects on the catalytic behavior of subtilisin Carlsberg in organic solvents?
Biotechnol. Bioeng. 53:351362
Griko YV, Privalov PL (1992) Calorimetric study of the heat and cold denaturation of
-lactoglobulin. Biochemistry 31:88108815
Gross M, Jaenicke R (1994) Proteins under pressure: The influence of high hydrostatic pressure
on structure, function and assembly of proteins and protein complexes.
Eur. J. Biochem. 221:617630
Gruebele M (1999) The fast protein folding problem. Annu. Rev. Phys. Chem. 50:485516
Gruebele M, Sabelko J, Ballew RM, Ervin J (1998) Laser temperature jump induced protein
refolding, Acc. Chem. Res. 31:699707
Gulotta M, Rogatsky E, Callender RH, Dyer RB (2003) Primary folding dynamics of sperm
whale apomyoglobin: core formation. Biophys J. 84:19091918
Guo Z, Thirumalai D (1997) The nucleationcollapse mechanism in protein folding: Evidence
for the non-uniqueness of the folding nucleus. Folding & Design 2:377391
Gussakovsky EE, Haas E (1995) 2 Steps in the transition between the native and acid states of
bovine -lactalbumin detected by circular polarization of luminescence: Evidence for a
premolten globule state. Prot. Science 4:23192326
Hagen SJ, Eaton WA (2000) Two-state expansion and collapse of a polypeptide. J. Mol. Biol.
301:10191027
Hagen SJ, Hofrichter J, Eaton WA (1995) Protein reaction kinetics in a room-temperature glass.
Science 269:959962
Hagen SJ, Hofrichter J, Szabo A, Eaton WA (1996) Diffusion-limited contact formation in
unfolded cytochrome c: Estimating the maximum rate of protein folding.
Hagen SJ, Hofrichter J, Eaton WA (1997) The rate of intrachain diffusion of unfolded
cytochrome c. J. Phys. Chem. B 101:23522365
Hagler AT, Dauber P, Lifson S (1979) Consistent force field studies of intermolecular forces in
hydrogen bonded crystals. III. The C=O...HO hydrogen bond and the analysis of the
energetics and packing of carboxylic acids. J. Am. Chem. Soc. 101:51315141
Hammes GG, Roberts PB (1969) Dynamics of helix-coil transition in poly-L-ornithine.
J. Am. Chem. Soc. 91:18121816
Harper ET, Rose GD (1993) Helix stop signals in proteins and peptides: The capping box.
Hartley RW (1988) Barnase and barstar: Expression of its cloned inhibitor permits expression
of a cloned ribonuclease. J. Mol. Biol. 202:913915
Hoeltzli SD, Frieden C (1995) Stopped-flow NMR spectroscopy: Real-time unfolding studies of
6-19F tryptophan-labeled Escherichia coli dihydrofolate reductase.
Hoeltzli SD, Frieden C (1996) Real-time refolding studies of 6-19F-tryptophan-labeled E. coli
dihydrofolate reductase using stopped-flow NMR spectroscopy.
Hoeltzli SD, Frieden C (1998) Refolding of 6-19F-tryptophan-labeled E. coli dihydrofolate
reductase in the presence of ligand: A stopped-flow NMR spectroscopy study.
Hofrichter J (2001) Laser temperature-jump methods for studying folding dynamics. Methods
Mol. Biol. 168:159191
Hofrichter J, Henry ER, Szabo A, Murray LP, Ansari A, Jones CM, Coletta M, Falconi G,
Brunori M, Eaton WA (1991) Dynamics of the quaternary conformational change in trout
hemoglobin. Biochemistry 30:65836598
204
References
Hu XH, Frei H, Spiro TG (1996) Nanosecond step-scan FTIR spectroscopy of hemoglobin:

Ligand recombination and protein conformational changes. Biochemistry 35:1300113005
Huang GS, Oas TG (1995) Submillisecond folding of monomeric -repressor.
Hubbard T, Tramontano A (1996) Update on protein structure prediction: Results of
the 1995 IRBM workshop. Folding & Design 1:R55R63
Hunt JF, Weaver AJ, Landry SJ, Gierasch L, Deisenhofer J (1996) The crystal structure of the
GroES co-chaperonin at 2.8 resolution. Nature 379:3745
Itzhaki LS, Evans PA, Dobson CM, Radford SE (1994) Tertiary interactions in the folding
pathway of hen lysozyme: Kinetic studies using fluorescent probes.
Itzhaki LS, Otzen DE, Fersht AR (1995a) The structure of the transition state for folding of
chymotrypsin inhibitor 2 analyzed by protein engineering methods: Evidence for a
nucleationcondensation mechanism for protein folding. J. Mol. Biol. 254:260288
Itzhaki LS, Neira JL, Ruiz-Sanz J, PratGay Gde, Fersht AR (1995b) Search for nucleation sites
in smaller fragments of chymotrypsin inhibitor 2. J. Mol. Biol. 254:289304
Jackson SE (1998) How do small single-domain proteins fold? Fold. Des. 3:R81R91
Jasanoff A, Fersht AR (1994) Quantitative determination of helical propensities from
trifluoroethanol titration curves. Biochemistry 33:21292135
Jentoft JE, Neet E, Stuehr JE (1977) Relaxation spectra of yeast hexokinases. Isomerization of
the enzyme. Biochemistry 16:117121
Johnson WC Jr (1990) Protein secondary structure and circular dichroism: A practical guide.
Proteins: Struct. Funct. Genetics 7:205214
Jones CM, Henry ER, Hu Y, Chan CK, Luck SD, Bhuyan A, Roder H, Hofrichter J, Eaton WA
(1993) Fast events in protein folding initiated by nanosecond laser photolysis.
Jung C, Hoa GHB, Davydov D, Gill E, Heremans K (1995) Compressibility of the heme pocket
of substrate-analog complexes of cytochrome P450cam-CO: The effect of hydrostatic
pressure on the Soret band. Eur. J. Biochem. 233:600606
Jung C, Ristau O, Schulze H, Sligar SG (1996) The CO stretching mode infrared spectrum of
substrate-free cytochrome P450cam-CO: The effect of solvent conditions, temperature, and
pressure. Eur. J. Biochem. 235:660669
Kalnin NN, Kuwajima K (1995) Kinetic folding and unfolding of staphylococcal nuclease and
its six mutants studied by stopped-flow circular dichroism.
Proteins: Struct. Funct. Genetics 23:163176
Karplus M, Sali A (1995) Theoretical studies of protein folding and unfolding.
Karplus M, Weaver DL (1994) Protein folding dynamics: the diffusioncollision model and
experimental data. Protein Science 3:650668.
Katayanagi K, Miyagawa M, Matsushima M, Ishikawa M, Kanaya S, Nakamura H, Ikehara M,
Matsuzaki T, Morikawa K (1992) Structural details of ribonuclease H from Escherichia coli
as refined to an atomic resolution. J. Mol. Biol. 223:10291052
Kharakoz DP, Bychkova VE (1997) Molten globule of human -lactalbumin: Hydration,
density, and compressibility of the interior. Biochemistry 36:18821890
Kharakoz DP, Sarvazyan AP (1993) Hydrational and intrinsic compressibilities of globular
proteins. Biopolymers 33:1126
Khorasanizadeh S, Peters ID, Butt TR, Roder H (1993) Folding and stability of a tryptophancontaining mutant of ubiquitin. Biochemistry 32:70547063
Kiefhaber T, Baldwin RL (1995) Intrinsic stability of individual -helices modulates
structure and stability of the apomyoglobin molten globule form. J. Mol. Biol. 252:122132
Kiefhaber T, Labhardt AM, Baldwin RL (1995) Direct NMR evidence for an intermediate
preceding the rate-limiting step in the unfolding of ribonuclease A. Nature 375:513515
References
205
Kiefhaber T, Bachmann A, Wildegger G, Wagner C (1997) Direct measurement of nucleation

and growth rates in lysozyme folding. Biochemistry 36:51085112
Kim PS, Baldwin RL (1982) Specific intermediates in the folding reactions of small proteins and
the mechanism of protein folding. Annu. Rev. Biochem. 51:459489
Kim PS, Baldwin RL (1990) Intermediates in the folding reactions of small proteins. Annu. Rev.
Biochem. 59:631660
Kim Y, Grable JC, Love R, Greene P, Rosenberg JM (1990) Refinement of EcoRI endonuclease
crystal structure: A revised protein chain tracing. Science 249:13071309
Klibanov AM (1989) Enzymatic catalysis in anhydrous organic solvents.
Trends Biochem. Sci. 14:141144
Klibanov AM (1997) Why are enzymes less active in organic solvents than in water?
Trends Biotechnol. 15:97101
Klimov DK, Thirumalai D (1998) Lattice models for proteins reveal multiple folding nuclei for
nucleationcollapse model. J. Mol. Biol. 282:471492
Korolev S, Nayal M, Barnes WM, DiCera E, Waksman G (1995) Crystal structure of the large
fragment of Thermus aquaticus DNA polymerase I at 2.5 resolution: Structural basis for
thermostability. Proc. Natl. Acad. Sci. USA 92:92649268
Kraulis PJ (1991) MOLSCRIPT: A program to produce both detailed and schematic plots of
protein structures. J. Appl. Cryst. 24:946950
Kubelka J, Eaton WA, Hofrichter J (2003) Experimental tests of villin subdomain folding
simulations. J. Mol. Biol. 329:625630
Kumaraswamy VS, Lindley PF, Slingsby C, Glover ID (1996) An eye lens proteinwater
structure: 1.2 resolution structure of B crystallin at 150K. Acta Cryst. D 52:611622
Kunugi S, Suzuki N, Nishimoto S, Morisawa T, Yoshida M (1997) Kinetic study of
carboxypeptidase Y catalyzed peptide condensation reactions in aqueous-organic solvent.
Biocat. Biotrans. 14:205217
Kuwajima K (1996) The molten globule state of -lactalbumin. FASEB J. 10:102109
Kuwajima K, Yamaya H, Miwa S, Sugai S, Nagamura T (1987) Rapid formation of secondary
structure framework in protein folding studied by stopped-flow circular dichroism.
FEBS Lett. 221:115118
Kuwajima K, Semisotnov GV, Finkelstein AV, Sugai S, Ptitsyn OB (1993) Secondary structure
of globular proteins at the early and the final stages in protein folding.
FEBS Lett. 334:265268
Kuwajima K, Yamaya H, Sugai S (1996) The burst-phase intermediate in the refolding of
-lactoglobulin studied by stopped-flow circular dichroism and absorption spectroscopy.
J. Mol. Biol. 264:806822
Ladurner AG, Itzhaki LS, dePratGay G, Fersht AR (1997) Complementation of peptide
fragments of the single domain protein chymotrypsin inhibitor 2. J. Mol. Biol. 273:317329
Leeson DT, Gai F, Rodriguez HM, Gregoret LM, Dyer RB (2000) Protein folding and unfolding
on a complex energy landscape. Proc. Natl. Acad. Sci. USA 97:25272532
Levinthal C (1968) Are there pathways for protein folding? J. Chim. Phys. 85:4445
Lewis JW, Tilton RF, Einterz CM, Milder SJ, Kuntz ID, Kliger DS (1985) New Technique for
measuring circular dichroism changes on a nanosecond time scale. Application to
carbonmonoxy myoglobin and carbonmonoxy hemoglobin. J. Phys. Chem. 89:289294
Lewis JW, Yee GG, Kliger DS (1987) Implementation of an optical multichannel analyzer for
nanosecond flash photolysis measurements. Rev. Sci. Instrum. 58:939944
Lewis JW, Goldbeck RA, Kliger DS, Xie XL, Dunn RC, Simon JD (1992) Time-resolved
circular dichroism spectroscopy: Experiment, theory, and applications to biological systems.
J. Phys. Chem. 96:52435254
Lide DR (ed) (1993) CRC handbook of chemistry and physics. CRC Press, Boca Raton
Ann Arbor London Tokyo, 74th Ed.
Lin SH, Cheung HC (1992) The kinetics of a two-state transition of myosin subfragment 1: A
temperature-jump relaxation study. FEBS Lett. 304:184186
206
References
Lin Y, Gerfen GJ, Rousseau DL, Yeh SR (2003) Ultrafast microfluidic mixer and
freeze-quenching device. Anal Chem. 75:53815386
Lpez-Hernndez E, Serrano L (1996) Structure of the transition state for folding of the 129-aa
protein CheY resembles that of a smaller protein, CI2. Folding & Design 1:4355
Lubienski MJ, Bycroft M, Freund SMV, Fersht AR (1994) Three-dimensional solution structure
and 13C-assignments of barstar using nuclear magnetic resonance spectroscopy.
Luchins J, Beychok S (1978) Far-ultraviolet stopped-flow circular dichroism.
Science 199:425426
Luo YZ, Baldwin RL (1998) Trifluororethanol stabilizes the pH 4 folding intermediate of sperm
whale apomyoglobin. J. Mol. Biol. 279:4957
Luthardt G, Frmmel C (1994) Local polarity analysis: A sensitive method that discriminates
between native proteins and incorrectly folded models. Protein Engineering 7:627631
Makhatadze GI, Privalov PL (1993) Contribution of hydration to protein folding
thermodynamics. 1. The enthalpy of hydration. J. Mol. Biol. 232:639659
Maness SJ, Franzen S, Gibbs AC, Causgrove TP, Dyer RB (2003) Nanosecond temperature jump
relaxation dynamics of cyclic -hairpin peptides. Biophys. J. 84:38743882
Martin J, Mayhew M, Langer T, Hartl FU (1993) The reaction cycle of GroEL and GroES in
chaperonin-assisted protein folding. Nature 366:228233
Martin J, Goldie KN, Engel A, Hartl FU (1994) Topology of the morphological domains of the
chaperonin GroEL visualized by immunoelectron microscopy.
Biol. Chem. HoppeSeyler 375:635639
Masters CL, Beyreuther K (1997) Tracking turncoat prion proteins. Nature 388:228229
Mateu MG, DelPino MMS, Fersht AR (1999) Mechanism of folding and assembly of a small
tetrameric protein domain from tumor suppressor p53. Nature Struct. Biol. 6:191198
Matouschek A, Fersht AR (1991) Protein engineering in analysis of protein folding pathways
and stability. Methods Enzymol. 202:82112
Matouschek A, Fersht AR (1993) Application of physical organic chemistry to engineered
mutants of proteins: Hammond postulate behavior in the transition state of protein folding.
Matouschek A, Kellis JT, Serrano L, Fersht AR (1989) Mapping the transition state and
pathway of protein folding by protein engineering. Nature 340:122126
Matouschek A, Kellis JT, Serrano L, Bycroft M, Fersht AR (1990) Transient folding
intermediates characterized by protein engineering. Nature 346:440445
Matouschek A, Serrano L, Fersht AR (1992) The folding of an enzyme. IV. Structure of an
intermediate in the refolding of barnase analysed by a protein engineering procedure. J. Mol.
Biol. 224:819835
Matthew JB (1985) Electrostatic effects in proteins.
Annu. Rev. Biophys. Biophys. Chem. 14:387417
Mayor U, Johnson CM, Daggett V, Fersht AR (2000) Protein folding and unfolding in
microseconds to nanoseconds by experiment and simulation. Proc. Natl. Acad. Sci. USA
97:1351813522
Mayor U, Guydosh NR, Johnson CM, Grossmann JG, Sato S, Jas GS, Freund SM, Alonso DO,
Daggett V, Fersht AR (2003) The complete folding pathway of a protein from nanoseconds
to microseconds. Nature 421:863867
McCaldon P, Argos P (1988) Oligopeptide biases in protein sequences and their use in
predicting protein coding regions in nucleotide sequences.
Proteins: Struct., Funct., Genetics 4:99122
McMaster TJ, Miles MJ, Walsby AE (1996) Direct observation of protein secondary structure in
gas vesicles by atomic force microscopy. Biophys. J. 70:24322436
Merritt EA, Murphy MEP (1994) Raster3D version 2.0: A program for photorealistic molecular
graphics. Acta Cryst. D50:869873
References
207
Michels PC, Hei D, Clark DS (1996) Pressure effects on enzymatic activity and stability at high
temperatures. Adv. Prot. Chem. 48:341376
Michnick SW, Shakhnovich E (1998) A strategy for detecting the conservation of foldingnucleus residues in protein superfamilies. Folding & Design 3:239251
Milder SJ, Bjrling SC, Kuntz ID, Kliger DS (1988) Time-resolved circular dichroism and
absorption studies on the photolysis reaction of carbonmonoxy myoglobin.
Biophys. J. 53:659664
Milla ME, Brown BM, Waldburger CD, Sauer RT (1995) P22 Arc repressor: transition state
properties inferred from mutational effects on the rates of protein unfolding and refolding.
Mines GA, Pascher T, Lee SC, Winkler JR, Gray HB (1996) Cytochrome c folding triggered by
electron transfer. Chemistry & Biology 3:491497
Miranker A, Kruppa GH, Robinson CV, Aplin RT, Dobson CM (1996a) Isotope-labeling
strategy for the assignment of protein fragments generated for mass spectrometry.
J. Am. Chem. Soc. 118:74027403
Miranker A, Robinson CV, Radford SE, Dobson CM (1996b) Investigation of protein folding
by mass spectrometry. FASEB J. 10:93101
Mirny LA, Abkevich VI, Shakhnovich EI (1998) How evolution makes proteins fold quickly.
Mozhaev VV, Heremans K, Frank J, Masson P, Balny C (1996) High-pressure effects on
protein structure and function. Proteins: Struct., Funct., Genetics 24:8191
Muoz V, Thompson PA, Hofrichter J, Eaton WA (1997) Folding dynamics and mechanism of
-hairpin formation. Nature 390:196199
Muoz V, Henry ER, Hofrichter J, Eaton WA (1998) A statistical-mechanical model for
-hairpin kinetics. Proc. Natl. Acad. Sci. USA 95:58725879
Murzin AG, Brenner SE, Hubbard T, Chothia C (1995) SCOP: A structural classification of
proteins database for the investigation of sequences and structures.
J. Mol. Biol. 247:536540
Myers JK, Oas TG (2002) Mechanism of fast protein folding. Annu. Rev. Biochem. 71:783815
Nakanishi K, Berova N, Woody RW (eds) (1994) Circular Dichroism. VCH, New York
Weinheim Cambridge
Narasimhulu S (1993) Substrate induced spin-state transition in cytochrome P-450LM2: A
temperature-jump relaxation study. Biochemistry 32:1034410350
Nath U, Udgaonkar JB (1997a) How do proteins fold? Current Science 72:180191
Nath U, Udgaonkar JB (1997b) Folding of tryptophan mutants of barstar: Evidence for an initial
hydrophobic collapse on the folding pathway. Biochemistry 36:86028610
Neira JL, Fersht AR (1999) Exploring the folding funnel of a polypeptide chain by biophysical
studies on protein fragments. J. Mol. Biol. 285:13091333
Nishi I, Kataoka N, Tokunaga F, Goto Y (1994) Cold denaturation of the molten globule states
of apomyoglobin and a profile for protein folding. Biochemistry 33:49034909
Nlting B (1991) A new spectrometer for the simultaneous measurement of absorption and
circular dichroism (in German). Ph.D. thesis, University of Bochum.
Nlting B (1995) Relation between adiabatic and pseudoadiabatic compressibility in ultrasonic
velocimetry. J. theor. Biol. 175:191196
Nlting B (1996) Temperature-jump induced fast refolding of cold-unfolded protein.
Biochem. Biophys. Res. Comm. 227:903908
Nlting B (1998a) Structural resolution of the folding pathway of a protein by correlation of
-values with inter-residue contacts. J. theor. Biol. 194:419428
Nlting B (1998b) The distribution of temperature in globular molecules, cells, or droplets in
temperature-jump-, sound velocity-, and pulsed LASER-experiments.
J. Phys. Chem. B. 102:75067509
Nlting B (1999) Analysis of the folding pathway of chymotrypsin inhibitor by correlation
of -values with inter-residue contacts. J. theor. Biol. 197:113121
208
References
Nlting B (2003) Methods in Modern Biophysics. Springer-Verlag, Berlin Heidelberg New York
Nlting B, Sligar SG (1993) Adiabatic compressibility of molten globules.
Nlting B, Andert K (2000) Mechanism of protein folding. Proteins: Struct. Funct. Genetics
41:288298
Nlting B, Jung C, Snatzke G (1992) Multichannel circular dichroism investigations of the
structural stability of bacterial cytochrome P-450. Biochim. Biophys. Acta 1100:171176
Nlting B, Jiang M, Sligar SG (1993) The acidic molten globule state of -lactalbumin probed
by sound velocity. J. Am. Chem. Soc. 115:98799882
Nlting B, Golbik R, Fersht AR (1995) Submillisecond events in protein folding.
Nlting B, Golbik R, Neira JL, Soler-Gonzlez AS, Schreiber G, Fersht AR (1997a) The
folding pathway of a protein at high resolution from microseconds to seconds.
Nlting B, Golbik R, Soler-Gonzlez AS, Fersht AR (1997b) Circular dichroism of denatured
barstar suggests residual structure. Biochemistry 36:98999905
Nlting B, Schlike W, Hampel P, Grundig F, Gantert S, Sips N, Bandlow W, Qi PX (2003)
Structural determinants of the rate of protein folding. J. theor. Biol. 223:299307
Nlting B, Jlich D, Vonau W, Andert K (2004) Evolutionary computer programming of protein
folding and structure predictions. J. theor. Biol. 229:1318
Nymeyer H, Garcia AE, Onuchic JN (1998) Folding funnels and frustration in off-lattice
minimalist protein landscapes. Proc. Natl. Acad. Sci. USA 95:59215928
O'Connor DB, Goldbeck RA, Hazzard JH, Kliger DS, Cusanovich MA (1993) Time-resolved
absorption and magnetic circular dichroism spectroscopy of cytochrome c3 from
Desulfovibrio. Biophys. J. 65:17181726
O'Neil KT, DeGrado WF (1990) A thermodynamic scale for the helix-forming tendencies of the
commonly occurring amino acids. Science 250:646651
Oliveberg M, Fersht AR (1996a) Formation of electrostatic interactions on the protein folding
pathway. Biochemistry 35:27262737
Oliveberg M, Fersht AR (1996b) Thermodynamics of transient conformations in the folding
pathway of barnase: Reorganization of the folding intermediate at low pH.
Onuchic JN, Wolynes PG, Luthey-Schulten Z, Socci ND (1995) Toward an outline of the
topography of a realistic protein folding funnel. Proc. Natl. Acad. Sci. USA 92:36263630
Onuchic JN, Socci ND, Luthey-Schulten Z, Wolynes PG (1996) Protein folding funnels: The
nature of the transition state ensemble. Folding & Design 1:441450
Otzen DE, Fersht AR (1998) Folding of circular and permuted chymotrypsin inhibitor 2:
retention of the folding nucleus. Biochemistry 37:81398146
Otzen DE, Itzhaki LS, ElMasry NF, Jackson SE, Fersht AR (1994) Structure of the transition
state of the folding/unfolding of the barley chymotrypsin inhibitor 2 and its implications for
mechanisms of protein folding. Proc. Natl. Acad. Sci. USA 91:1042210425
Pabit SA, Roder H, Hagen SJ (2004) Internal friction controls the speed of protein folding from a
compact configuration. Biochemistry 43:1253212538
Pappu RV, Weaver DL (1998) The early folding kinetics of apomyoglobin.
Protein Sci. 7:480490
Park SH, O'Neil KT, Roder H (1997) An early intermediate in the folding reaction of the B1
domain of protein G contains a native-like core. Biochemistry 36:1427714283
Pascher T, Chesick JP, Winkler JR, Gray HB (1996) Protein folding triggered by electron
transfer. Science 271:15581560
Petrich JW, Martin JL, Houde D, Poyart C, Orszag A (1987) Time-resolved Raman
spectroscopy with subpicosecond resolution: Vibrational cooling and delocalization of strain
energy in photodissociated carbonmonoxy hemoglobin. Biochemistry 26:79147923
Pfeil W (1981) Thermodynamics of -lactalbumin unfolding. Biophys. Chem. 13:181186
References
209
Pfeil W (1988) Protein unfolding. In: Jones MN (ed) Biochemical thermodynamics.

Elsevier, Amsterdam, 5399
Pfeil W (1993) Thermodynamics of apocytochrome b5 unfolding. Protein Science 2:14971501
Pfeil W, Welfle K, Bychkova VE (1991) Guanidine-hydrochloride titration of the unfolded
apocytochrome c studied by calorimetry. Studia Biophysica 140:512
Pfeil W, Nlting B, Jung C (1993a) Apocytochrome P-450cam is a native protein with some
intermediate-like properties. Biochemistry 32:88568862
Pfeil W, Nlting B, Jung C (1993b) Thermodynamic properties of apocytochrome P-450 cam.
In: Tweel WJJvd, Harder A, Buitelaar RM (eds) Stability and Stabilization of Enzymes.
Elsevier, Amsterdam, 407414
Pflumm M, Luchins J, Beychok S (1986) Stopped-flow circular dichroism.
Methods Enzymol. 130:519534
Phillips CM, Mizutani Y, Hochstrasser RM (1995) Ultrafast thermally-induced unfolding of
ribonuclease A. Proc. Natl. Acad. Sci. USA 92:72927296
Plaxco KW, Dobson CM (1996) Time-resolved biophysical methods in the study of protein
folding. Curr. Opin. Struct. Biol. 6:630636
Plaxco KW, Simons KT, Baker DJ (1998) Contact order, transition state placement and the
refolding rates of single-domain proteins. J. Mol. Biol. 277:985994
Porschke D (1996) Analysis of chemical and physical relaxation processes of polyelectrolytes
by electric field pulse methods. A comparison of critical comments with facts. Ber. Bunsen
Gesellschaft - Phys. Chem. Chem. Phys. 100:715720
Porschke D, Obst A (1991) An electric-field-jump apparatus with ns time resolution for electrooptical measurements at physiological salt concentrations. Rev. Sci. Instrum. 62:818820
Poulos TL, Finzel BC, Howard, AJ (1986) Crystal structure of substrate-free Pseudomonas
putida cytochrome P-450. Biochemistry 25:53145322
Presta LG, Rose GD (1988) Helix signal in proteins. Science 240:16321641
Prince SM, Papiz MZ, Freer AA, McDermott G, Hawthornthwaite-Lawless AM, Cogdell RJ,
Isaacs NW (1997) Apoprotein structure in the LH2 complex from Rhodopseudomonas
acidophila strain 10050: Modular assembly and proteinpigment interactions.
J. Mol. Biol. 268:412423
Privalov PL (1979) Stability of protein: Small globular proteins. Adv. Prot. Chem. 33:167241
Privalov PL (1990) Cold denaturation of proteins. Crit. Rev. Biochem. Mol. Biol. 25:281305
Privalov PL (1996) Intermediate states in protein folding. J. Mol. Biol. 258:707725
Privalov PL, Makhatadze GI (1993) Contribution of hydration to protein folding
thermodynamics. 2. The entropy and Gibbs energy of hydration. J. Mol. Biol. 232:660679
Pryse KM, Bruckman TG, Maxfield BW, Elson EL (1992) Kinetics and mechanism of the
folding of cytochrome c. Biochemistry 31:51275136
Ptitsyn OB (1981) Protein folding: General physical model. FEBS Lett. 131:197202
Ptitsyn OB (1994) Kinetic and equilibrium intermediates in protein folding.
Protein Eng. 7:593596
Ptitsyn OB (1995) Molten globule and protein folding. Adv. Prot. Chem. 47:83229
Ptitsyn OB (1998) Protein folding: Nucleation and compact intermediates.
Biochemistry (Moscow) 63:367373
Ptitsyn OB, Rashin AA (1975) A model of myoglobin self-organization. Biophys. Chem. 3:120
Qi PX, Beckman RA, Wand AJ (1996) Solution structure of horse heart ferricytochrome c and
detection of redox-related structural changes by high-resolution 1H NMR.
Qiu L, Pabit SA, Roitberg AE, Hagen SJ (2002) Smaller and faster: the 20-residue Trp-cage
protein folds in 4 s. J. Am. Chem. Soc. 124:1295212953
Rackovsky S, Scheraga HA (1977) Hydrophobicity, hydrophilicity, and the radial and
orientational distributions of residues in native proteins. Proc. Natl. Acad. Sci. USA
74:52485251
210
References
Radzicka A, Wolfenden R (1988) Comparing the polarities of the amino acids: Side chain
distribution coefficients between the vapor-phase, cyclohexane, 1-octanol, and neutral
aqueous solution. Biochemistry 27:16641670
Regenfuss P, Clegg RM (1987) Diffusion-controlled association of a dye, 1-anilinonaphthalene8-sulfonic acid, to a protein, bovine serum albumin, using a fast-flow microsecond mixer and
stopped-flow. Biophys. Chem. 26:8389
Regenfuss P, Clegg RM, Fulwyler MJ, Barrantes FJ, Jovin TM (1985) Mixing liquids in
microseconds. Rev. Sci. Instrum. 56:283290
Richards FM (1974) The interpretation of protein structures: Total volume, group volume
distributions and packing density. J. Mol. Biol. 82:114
Richardson JS, Richardson DC (1988) Amino acid preferences for specific locations at the ends
of -helices. Science 240:16481652
Riddle DS, Grantcharova VP, Santiago JV, Alm E, Ruczinski I, Baker D (1999) Experiment and
theory highlight role of native state topology in SH3 folding. Nature Struct. Biol.
6:10161024
Riek R, Hornemann S, Wider G, Billeter M, Glockshuber R, Wthrich K (1996) NMR structure
of the mouse prion protein domain PrP(121231) and inherited human prion diseases.
Nature 382:180182
Robinson CR, Sauer RT (1996) Equilibrium stability and submillisecond refolding of a
designed single-chain Arc repressor. Biochemistry 35:1387813884
Roder H, Colon W (1997) Kinetic role of early intermediates in protein folding.
Roder H, Elve GA, Englander SW (1988) Structural characterization of folding intermediates
in cytochrome c by H-exchange labeling and proton NMR. Nature 335:700704
Rose GD (1987) Protein hydrophobicity: Is it the sum of its parts?
Proteins: Struct., Funct., Genetics 2:7980
Sabelko J, Erwin J, Gruebele M (1998) Cold-denatured ensemble of apo-myoglobin:
Implications for the early steps of folding. J. Phys. Chem. B 102:18061819
Sarvazyan AP (1991) Ultrasonic velocimetry of biological compounds.
Annu. Rev. Biophys. Biophys. Chem. 20:321342
Sauder JM, Mackenzie NE, Roder H (1996) Kinetic mechanism of folding and unfolding of
Rhodobacter capsulatus cytochrome c2. Biochemistry 35:1685216862
Scheraga HA (1996) Recent developments in the theory of protein folding: Searching for the
global energy minimum. Biophys. Chem. 59:329339
Schindler T, Herrler M, Marahiel MA, Schmid F-X (1995) Extremely rapid protein folding in
the absence of intermediates. Nat. Struct. Biol. 2:663673
Schreiber G, Fersht AR (1993a) Interaction of barnase with its polypeptide inhibitor barstar
studied by protein engineering. Biochemistry 32:51455150
Schreiber G, Fersht AR (1993b) The refolding of cis-peptidylprolyl and trans-peptidylprolyl
isomers of barstar. Biochemistry 32:1119511203
Schreiber G, Fersht AR (1995) Energetics of the proteinprotein interactions: Analysis of the
barnasebarstar interface by single mutations and double mutant cycles.
J. Mol. Biol. 248:478486
Schreiber G, Buckle AM, Fersht AR (1994) Stability and function: Two constraints in the
evolution of barstar and other proteins. Structure 2:945951
Schwarz G, Seelig J (1968) Kinetic properties and the electric field effect of the helixcoil
transition of poly(benzyl-L-glutamate) determined from dielectric relaxation measurements.
Biopolymers 6:12631277
Semisotnov GV, Kihara H, Kotova NV, Kimura K, Amemiya Y, Wakabayashi K, Serdyuk IN,
Timchenko AA, Chiba K, Nikaido K, Ikura T, Kuwajima K (1996) Protein globularization
during folding: A study by synchrotron small-angle X-ray scattering.
J. Mol. Biol. 262:559574
References
211
Serrano L, Kellis JT Jr, Cann P, Matouschek A, Fersht AR (1992a) The folding of an enzyme.
II. Substructure of barnase and the contribution of different interactions to protein stability.
J. Mol. Biol. 224:783804
Serrano L, Matouschek A, Fersht AR (1992b) The folding of an enzyme. III. Structure of the
transition state for unfolding of barnase analyzed by a protein engineering procedure.
J. Mol. Biol. 224:805818
Serrano L, Matouschek A, Fersht AR (1992c) The folding of an enzyme. VI. The folding
pathway of barnase: Comparison with theoretical models. J. Mol. Biol. 224:847859
Service RF (1996) Folding proteins caught in the act. Science 273:2930
Shakhnovich EI (1997) Theoretical studies of protein folding thermodynamics and kinetics.
Cur. Opin. Struct. Biol. 7:2940
Shakhnovich EI, Abkevich V, Ptitsyn O (1996) Conserved residues and the mechanism of
protein folding. Nature 379:9698
Shapiro DB, Goldbeck RA, Che DP, Esquerra RM, Paquette SJ, Kliger DS (1995) Nanosecond
optical rotatory dispersion spectroscopy: Application to photolyzed hemoglobinCO
kinetics. Biophys. J. 68:326334
Shastry MCR, Udgaonkar JB (1995) The folding mechanism of barstar: Evidence for multiple
pathways and multiple intermediates. J. Mol. Biol. 247:10131027
Shastry MCR, Roder H (1998) Evidence for barrier-limited protein folding kinetics on the
microsecond time scale. Nat. Struct. Biol. 5:385392
Shastry MCR, Luck SD, Roder H (1998) A continuous-flow capillary mixing method to monitor
reactions on the microsecond time scale. Biophys. J. 74:27142721
Shoemaker BA, Wang J, Wolynes PG (1997) Structural correlations in protein folding funnels.
Shoemaker BA, Wang J, Wolynes PG (1999) Exploring structures in protein folding funnels with
free energy functionals: the transition state ensemble. J. Mol. Biol. 287:675694
Shortle D, Simons KT, Baker D (1998) Clustering of low-energy conformations near the native
structures of small proteins. Proc. Natl. Acad. Sci. USA 95:1115811162
Sligar SG (1976) Coupling of spin, substrate, and redox equilibria in cytochrome P-450.
Sohl JL, Jaswal SS, Agard DA (1998) Unfolded conformations of -lytic protease are more
stable than its native state. Nature 395:817819
Sosnick TR, Mayne L, Hiller R, Englander SW (1994) The barriers in protein folding.
Sosnick TR, Mayne L, Englander SW (1996) Molecular collapse: The rate-limiting step in twostate cytochrome c folding. Proteins: Struct. Funct. Genetics 24:413426
Stark B, Nlting B, Jahn H, Andert K (1992) Method for determining the electron number in
charge-coupled measurement devices. Optical Engineering 31:852856
Steinhoff HJ, Lieutenant K, Redhardt A (1989) Conformational transition of
aquomethemoglobin: Intramolecular histidine E7 binding reaction to the heme iron in the
temperature range between 220 K and 295 K as seen by EPR and temperature-jump
measurements. Biochim. Biophys. Acta 996:4956
Stouten PFW, Frmmel C, Nakamura H, Sander C (1993) An effective solvation term based on
atomic occupancies for use in protein simulations. Molecular Simulation 10:26
Takahashi MT, Alberty RA (1969) The pressure-jump method. Methods Enzymol. 16:3155
Takahashi S, Yeh SR, Das TK, Chan CK, Gottfried DS, Rousseau DL (1997) Folding of
cytochrome c initiated by submillisecond mixing. Nat. Struct. Biol. 4:4450
Tamura Y, Gekko K (1995) Compactness of thermally and chemically denatured ribonuclease A
as revealed by volume and compressibility. Biochemistry 34:18781884
Tanaka N, Kunugi S (1996) Effect of pressure on the deuterium-exchange of -lactalbumin and
-lactoglobulin. Int. J. Biol. Macromol. 18:3339
Teilum K, Maki K, Kragelund BB, Poulsen FM, Roder H (2002) Early kinetic intermediate in
the folding of acyl-CoA binding protein detected by fluorescence labeling and ultrarapid
212
References
mixing. Proc. Natl. Acad. Sci. USA 99:98079812

Thompson PA (1997) Laser temperature-jump for the study of early events in protein folding.
In: Marshak DR (ed) Techniques in Protein Chemistry VIII. San Diego: Academic Press.
Thompson PA, Eaton WA, Hofrichter J (1997) Laser temperature-jump study of the helixcoil
kinetics of an alanine peptide interpreted with a 'kinetic zipper' model.
Thornton JM, Jones DT, MacArthur MW, Orengo CM, Swindells MB (1995) Protein folds:
Towards understanding folding from inspection of native structures.
Phil. Transactions Royal Soc. London B 348:7179
Tilton RF Jr, Dewan JC, Petsko GA (1992) Effects of temperature on protein structure and
dynamics: X-ray crystallographic studies of the protein ribonuclease A at nine different
temperatures from 98 to 320 K. Biochemistry 31:24692481
Topchieva IN, Sorokina EM, Kurganov BI, Zhulin VM, Makarova ZG (1996) High pressure
induced complexes of -chymotrypsin with block-copolymers based on ethylene and
propylene oxides. Biochemistry (Moscow) 61:746749
Tsong TY (1982) Viscosity-dependent conformational relaxation of ribonuclease A in the
thermal unfolding zone. Biochemistry 21:14931498
Tsong TY, Baldwin RL, Elson EL (1971) The sequential unfolding of ribonuclease A: Detection
of a fast initial phase in the kinetics of unfolding. Proc. Natl. Acad. Sci. USA 68:27122715
Udgaonkar JB, Baldwin RL (1988) NMR evidence for an early framework intermediate on the
folding pathway of ribonuclease A. Nature 335:694699
Urbanke C, Wray J (2001) A fluorescence temperature-jump study of conformational transitions
in myosin subfragment 1. Biochem. J. 358:165173
Varotsis C, Babcock GT (1993) Nanosecond time-resolved resonance Raman spectroscopy.
Methods Enzymol. 226:409431
Velluz L, Legrand M, Grosjean M (1965) Optical circular dichroism. Verlag Chemie,
Weinheim
Viguera AR, Villegas V, Aviles FX, Serrano L (1997) Favorable native-like helical local
interactions can accelerate protein folding. Folding & Design 2:2333
Vu DM, Myers JK, Oas TG, Dyer RB (2004) Probing the folding and unfolding dynamics of
secondary and tertiary structures in a three-helix bundle protein. Biochemistry 43: 35823589
Wallenhorst WF, Green SM, Roder H (1997) Kinetic evidence for folding and unfolding
intermediates in staphylococcal nuclease. Biochemistry 36:57955805
Walz FG (1992) Relaxation kinetics of ribonuclease T1 binding with guanosine and 3'-GMP.
Biochim. Biophys. Acta 1159:327334
Wang MS, Gandour RD, Rodgers J, Haslam JL, Schowen RL (1975) Transition-state structure
for a conformation change of ribonuclease. Bioorg. Chem. 4:392406
Wangikar PP, Michels PC, Clark DS, Dordick JS (1997) Structure and function of subtilisin
BPN' solubilized in organic solvents. J. Am. Chem. Soc. 119:7076
Warshel A, Levitt M (1976) Theoretical studies of enzymatic reactions: Dielectric, electrostatic
and steric stabilization of the carbonium ion in the reaction of lysozyme.
J. Mol. Biol. 103:227249
Weber G (1993) Thermodynamics of the association and the pressure dissociation of oligomeric
proteins. J. Phys. Chem. 97:71087115
Weber G (1996) Persistent confusion of total entropy and chemical-system entropy in chemical
thermodynamics. Proc. Natl. Acad. Sci. USA 93:74527453
Wen YX, Chen EF, Lewis JW, Kliger DS (1996) Nanosecond time-resolved circular dichroism
measurements using an up-converted Ti-sapphire LASER. Rev. Sci. Instrum. 67:30103016
Wetlaufer DB (1962) Ultraviolet spectra of proteins and amino acids.
Adv. Protein Chem. 17:303390
Wetlaufer DB (1973) Nucleation, rapid folding, and globular intrachain regions in proteins.
References
213
Wetlaufer DB (1990) Nucleation in protein folding: Confusion of structure and process.

Trends in Biochemical Sciences 15:414415
Wetlaufer DB, Xie Y (1995) Control of aggregation in protein refolding: A variety of
surfactants promote renaturation of carbonic anhydrase II. Protein Sci. 4:15351543
Williams DH, Fleming I (1995) Spectroscopic methods in organic chemistry. McGraw-Hill,
London
Williams S, Causgrove TP, Gilmanshin R, Fang KS, Callender RH, Woodruff WH, Dyer RB
(1996) Fast events in protein folding: Helix melting and formation in a small peptide.
Wilson CJ, Wittung-Stafshede P (2005) Role of structural determinants in folding of the
sandwich-like protein Pseudomonas aeruginosa azurin. Proc. Natl. Acad. Sci. USA
102:39843987
Wittung-Stafshede P, Lee JC, Winkler JR, Gray HB (1999) Cytochrome b562 folding triggered
by electron transfer: approaching the speed limit for formation of a four-helix-bundle protein.
Wolynes PG (1996) Symmetry and the energy landscapes of biomolecules.
Wolynes PG, Onuchic JN, Thirumalai D (1995) Navigating the folding routes.
Science 267:16191620
Wolynes PG, Luthey-Schulten Z, Onuchic JN (1996) Fast folding experiments and the
topography of protein folding energy landscapes. Chem. & Biol. 3:425432
Wong KB, Freund SMV, Fersht AR (1996) Cold denaturation of barstar: 1H, 15N and 13C
NMR assignment and characterization of residual structure. J. Mol. Biol. 259:805818
Woodruff WH, Einarsdttir O, Dyer RB, Bagley KA, Palmer G, Atherton SJ, Goldbeck RA,
Dawes TD, Kliger DS (1991) Nature and functional implications of the cytochrome a3
transients after photodissociation of CO-cytochrome oxydase.
Wthrich K (1986) NMR of proteins and nucleic acids. Wiley, New York
Xie X, Simon JD (1989) Picosecond time-resolved circular dichroism spectroscopy:
Experimental details and applications. Rev. Sci. Instrum. 60:26142627
Xie X, Simon JD (1991) Protein conformational relaxation following photodissociation of CO
from carbonmonoxy myoglobin: Picosecond circular dichroism and absorption studies.
Xu Y, Oyola R, Gai F (2003) Infrared study of the stability and folding kinetics of a 15-residue
-hairpin. J. Am. Chem. Soc. 125:1538815394
Yamamoto K, Mizutani Y, Kitagawa T (2000) Nanosecond temperature jump and time-resolved
Raman study of thermal unfolding of ribonuclease A. Biophys. J. 79:485495
Yanagawa H, Yoshida K, Torigoe C, Park JS, Sato K, Shirai T, Go M (1993) Protein anatomy:
Functional roles of barnase module. J. Biol. Chem. 268:58615865
Yang JT, Wu CSC, Martinez HM (1986) Calculation of protein conformation from circular
dichroism. Methods Enzymol. 130:208269
Yeh SR, Rousseau DL (1998) Folding intermediates in cytochrome c.
Yeh SR, Takahashi S, Fan BC, Rousseau DL (1997) Ligand exchange during cytochrome c
folding. Nat. Struct. Biol. 4:5156
Yuzawa T, Kato C, George MW, Hamaguchi HO (1994) Nanosecond time-resolved infrared
spectroscopy with a dispersive scanning spectrometer. Appl. Spectroscopy 48:684690
Zhang CF, Lewis JW, Cerpa R, Kuntz ID, Kliger DS (1993) Nanosecond circular dichroism
spectral measurements: Extension to the far-ultraviolet region.
J. Phys. Chem. 97:54995505
Zhu X, Zhao X, Burkholder WF, Gragerov A, Ogata CM, Gottesman ME, Hendrickson, WA
(1996) Structural analysis of substrate binding by the molecular chaperone DnaK.
Science 272:16061614
214
References
Zhu Y, Alonso DO, Maki K, Huang CY, Lahr SJ, Daggett V, Roder H, DeGrado WF, Gai F
(2003) Ultrafast folding of 3D: a de novo designed three-helix bundle protein.
Zhu YJ, Fu XR, Wang T, Tamura A, Takada S, Savan JG, Gai F (2004) Guiding the search for a
protein's maximum rate of folding. Chem. Phys. 307:99109
Zubay G (1993) Biochemistry. Wm. C. Brown Publishers, Dubuque Melbourne Oxford, 3rd Ed.
Index
Acid-base catalysis of H/D exchange, 9596

Acoustic
relaxation, 6971, 7677
resonator, 70
Activation energy barrier, 28
Acyl-coenzyme A binding protein, 193
Aggregation, 2, 28, 76, 80, 96, 115, 117,
119120, 126, 129135, 165
avoidance, 131135
concentration-dependence, 131
co-solvent-induced, 132
denaturant-dependence, 132
detection, 129130
discrimination from folding events,
119120
effect on NMR studies, 80
effect on observed rate constant, 117120
lysozyme-induced, 126
pH-dependence, 133
role in diseases, 2
salt-dependence, 133
temperature-dependence, 133
transient occurrence, 80, 115, 117, 129
Alzheimers disease, 2, 166
Amino acids, 2, 511, 13, 2223, 80, 107,
128
abundance in proteins, 7
cistrans isomerization, 4, 8, 145147,
158159
conformational preference, 11
cysteine, 58, 11, 2223, 128, 138
hydrophilicity / hydrophobicity, 2225
molecular weight, 7
NMR, 86, 8889
phenylalanine, 59, 11, 2223
pKa , 8
proline, 1, 48, 11, 2223, 61, 128,
138139, 145, 158159
structure, 58
surface area, 22
tryptophan, 58, 11, 2223, 79, 93,
146, 162
tyrosine, 58, 11, 2223, 128
volume, 7
Amino group, 5, 8, 128

Ampicillin, 138
Analog-to-digital converter (ADC), 55, 57,
90, 98
Annealing temperature in PCR, 109, 127
Arc repressor, 165, 170, 173176, 178, 185
Aromatic amino acid residues, 48, 11,
2223, 89, 93, 128129
absorption, 5, 129
circular dichroism, 4, 69, 89, 93, 155
fluorescence, 5557, 61, 6364,
144148, 155, 160
Association / dissociation, 4147, 8082,
119120, 129133
Autocorrelation, 130
Barnase, 18, 60, 73, 82, 137, 140141, 145,
158, 162163, 165, 169171, 173,
176177, 179
Barstar, 4, 18, 5960, 82, 9293, 111, 116,
137165, 167168, 170171, 173,
176177, 179, 184, 195
Bimolecular reactions, 28, 4147, 49, 69,
119120, 129133
Boltzmann constant (kB), XV, 28, 83
Brookhaven Protein Data Bank, 14
BSE, 166
Burial of sidechains on folding, 23, 95, 146
Burst-phase detection, 34, 60, 148
13C, 75, 83, 87
Calorimetry, 113114
Camphor-sulfonic acid (CSA), 139
Cassette mutagenesis, 108, 127, 138
Carbon monoxide (CO), 6569
Carboxyl group, 89, 2021
CD, see Circular dichroism
Chain topology of proteins, 164, 181184
Chain topology parameter, 183, 184
Chaperonins, 13, 133
Charge-coupled device (CCD), 5354,
98101
Charged amino acid residues, 58, 18,
2223, 133
216
Index
Charge-interactions, 1720, 133

Chemical shift, 76, 8485, 89
Chirality, 89, 92
Chlorophyll, 14
Chymotrypsin inhibitor (CI2), 158, 164165,
170173, 176177, 179, 191
Clusters of amino acid residues, 154, 155,
166, 167, 169, 172, 173, 174, 177, 191
Chain topology parameter (CTP), 183185
Circular dichroism (CD) spectroscopy, 4, 51,
60, 6869, 8993, 98104, 111, 139,
142144, 148, 155, 162
aromatic sidechains, 4, 69, 89, 93, 155
burst-phase detection, 4, 148
calibration, 139
decomposition of spectra for the determination of secondary structure, 9293
definitions of , R, and [], 90
detection of residual structure in the
unfolded (denatured) state, 142143
far-UV, 6869, 89, 92, 142144, 148,
155
measurement of structure, 8993,
98104, 142144, 148, 155
multichannel spectrometers, 98101, 104
near-UV, 4, 69, 89, 93, 155
scanning spectrometers, 9091, 100104,
139
signal-to-noise ratio, 99100
stopped-flow, 4, 51, 139, 148
temperature scans, 139, 143
time resolved, 6869, 98104
158159
Cloning, 107108, 111, 126127
Co-aggregation, 126
Cold-denaturation, cold-unfolding,
55, 5861, 63, 93, 137, 141146
Collapse
diffusion limited, 4, 164
hydrophobic, 164, 166, 168, 179, 195
of intermediate, 145, 162, 165
of transition state, 157, 162
Compactness, size, 51, 145, 147, 148, 157,
162, 189
Compressibility of proteins, 11, 71
Computer evolution, 187193
Concentration-dependence of aggregation,
131
Conformation, conformational,
changes, gating, movement, relaxations,
transitions, 3, 11, 27, 51, 58, 6669,
7172, 75, 77, 85, 89, 95
cis / trans, 139, 145, 158159

dispersity, 166
folded, 1, 17, 20, 39, 85, 133134, 157
folding nucleus, 60, 137, 158, 162180
misfolded, 39, 54, 133134, 160161
number in the unfolded state, 12, 163
preference of amino acids, 9, 11
sampling during the folding reaction, 160,
163164
secondary structure elements, 10
Conserved residues, 164
Contacts of residues, 137, 141, 152155, 161,
164, 167172, 179
Continuous-flow method, 5254, 96
Continuous wave NMR spectrometers,
8788
Cooperativity, 20, 80, 141142, 144, 148,
164165
Coulomb force, 1720, 7374
Coupling constant, 8485
Co-solvents, 2, 22, 132
CreutzfeldtJacob disease, 166
CTP (chain topology parameter), 183185
Cysteine, 58, 11, 2224, 128, 138
Cytochrome c, 13, 15, 24, 54, 6566, 69,
128, 165
D2O, 62, 9596
Data bank, Brookhaven, 14
deMoivres identity, 40
Denaturation (unfolding) of protein, by
cold, 55, 5861, 63, 93, 137, 141146
denaturants, 31, 60, 66, 76, 9296,
111113, 116119, 121, 126, 132,
135136, 139, 142143
heat, 59, 93, 114115, 143
Denatured (unfolded) state, 13, 27, 5861,
63, 72, 7576, 89, 97, 105, 112113,
115, 118, 120123, 135, 139149, 159
Density meters, 71
Deoxyribonucleic acid, see DNA
Deuterium, 54, 81, 9597
Dielectric relaxation, 27, 44, 7374, 7677
Differential scanning calorimetry (DSC),
113114
Diffusion limit of folding, 164
Diffusion of heat upon T-jump, 6465
Dimermonomer equilibrium, -transition,
4147, 115, 119120, 131
Dipoledipole interactions, 19
Diseases, folding related,
Alzheimer, 2, 166
Huntington, 2, 166
Index
prion (CreutzfeldtJacob, BSE, scrapie),
2, 166
Discrimination between folding and
association events, 42, 115, 119120,
131
Dispersion forces, van der Waals, 1921
Dissociation, 47, 6667, 8082, 109, 115,
119120, 130
Distance constraints between nuclei, 85, 89
Disulfide formation, 7, 11, 15, 128
DNA, plasmid, 107111, 126128, 138
DNA polymerase, 13, 107111, 127
Domain, 11, 13, 15, 76, 141, 165
Double-jump techniques, 54, 61, 9597
Dynamic light scattering, 51, 129130
Elasto-optical (photoelastic) modulator,
9091, 100, 102
Electrical-discharge-induced T-jumping,
5561, 7677, 140, 146148, 163
Electric-field-jump apparatus, 7374, 7677
Electron-transfer-induced refolding, 69, 77
Electro-optical modulator, 98, 100101
Electrostatic forces, 1720, 7374, 133
Energy, Gibbs energy, free energy
change upon folding, 59, 111115,
135136, 144, 147
change upon mutation, 105122, 149, 152
landscape, 23, 30, 33, 3637, 5859,
105106, 133, 141, 161, 164
of activation, 28
of intermediate, 121123, 146147, 155
of transfer, 18, 2225
of transition state, 2829, 120123,
150151
Enthalpy, 24, 29, 5860, 114115
Entropy, 17, 2425, 29, 60, 161
Enzymatic activity in organic co-solvent
mixtures, 2
Enzymeinhibitor complexes, 28, 8082,
140141
Enzymesubstrate complexes, 28, 49, 8082
Equilibrium unfolding, 111115
Error rates of DNA polymerases and PCR,
127
Euler's formula, 40
Evolution of computer programs, 187193
Evolution of nanomachines, 194
Evolution of optical devices, 194
Evolutionary computer programming,
187193
Expression of proteins 125128
errors, 126128
217
inclusion bodies, 126

inducible systems, 125126, 138
level, 125126, 138
19F, 79, 83
Far-infrared, 64, 67
detectors, 67
diode LASER, 67
generation by using difference
frequencies, 67
Field-jump-induced relaxation, 7374
Flash-photolysis-induced relaxation, 6569,
7677
Flips of sidechains, 11
Fluorescence
detection, 51, 5458, 61, 6364, 7274,
139140, 144148, 155
kinetic difference spectra, 145
labeling, 8082
red-shift upon unfolding, 144146
T-jump traces, 146147
Foldability, 3
Folding
bottleneck, 161
burial of hydrophobic sidechains, 23, 95,
142, 146, 148
diffusion limit, 164
directional propagation, 157158
energy landscape, 23, 30, 33, 3637,
5859, 105106, 133, 141, 161, 164
fast events, 14, 5177, 146147,
163165
funnel, 141, 161, 164, 166, 168, 171
highly resolved, 152160, 165180
intermediates, 23, 27, 3140, 4349,
5861, 7981, 85, 89, 9596, 105106,
113, 117118, 123, 129, 137180, 195
kinetic resolution, 5177, 7982
nucleus, nucleationcondensation, 60,
137, 158, 162180, 184, 195
paradox, 1, 168, 170
pathway, 14, 137180
predictions, 187194
solvent exclusion, 146147, 155158,
160, 177
structural resolution, 95123, 137180
transition states, 2829, 7677, 105123,
149180
under isothermal conditions, 63, 140
Folds, classes of, 1215
Fourier transformation, 8688
Fourier transform
NMR spectrometer, 84
218
Index
principle for NMR, 8688

Four-state transitions, 40, 174
Free energy, see Energy
Funnel model of folding, 161, 164
Gene, 107108, 111, 126127
Gibbs free energy, see Energy
Glass transition, 161
1H
(H, protonium, hydrogen), 54, 7576, 81,

8387, 9597, 162
2H (D, deuterium, heavy hydrogen), 54, 81,
9597
3H (tritium, super heavy hydrogen), 82
h (Planck constant), XV, 28, 83
H/D exchange, 51, 54, 76, 8182, 85, 9597,
195
Heat capacity,
molar, 59, 114115
specific, 57, 6465
Heat-denaturation, heat-unfolding
DNA, 109
protein 59, 93, 114115, 143
Heat diffusion in T-jump experiments, 64,
140
-Helix, 911, 60, 6364, 73, 92, 137138,
140142, 149158, 160164
Helixcoil transition, 4, 64, 73
Heme, 1115, 54, 58, 66, 89, 128
Highly resolved folding pathway, 152180
Huntingtons disease, 2, 166
Hydrogen bonding, 910, 2123, 95
Hydrogen exchange, 51, 54, 76, 8182, 85,
9597, 195
Hydrophilicity, 18, 2225
Hydrophobicity, 9, 17, 2225, 60, 107,
132133, 146, 149151, 157, 164
Inclusion bodies, 126, 138
Infrared (IR) and far infrared
absorption of water, 6263
detectors, 67
diode LASER, 67
generation by using difference
frequencies, 67
generation by using Nd:YAG LASER, 63,
6768, 102104
spectroscopy, 64, 67
Interactions in proteins and between proteins,
165180
Coulomb, 1719, 7374
electrostatic, 1719, 7374, 133
hydrogen bonding, 910, 2122, 95
hydrophobic, 2225
LennardJones potential, 21
London dispersion force, 19
non-native, 154, 160161, see also
Misfolding and Aggregation
Pauli-exclusion, 19
quantum-mechanical, 1921
van der Waals, 17, 1921
Intermediates
compactness, 51, 148
cooperative formation, 80, 148, 164
detection of the occurrence by using
-value analysis, 123, 159160, 171, 174,
175, 177, 195
detection under conditions that favor
folding, 23, 27, 3140, 4349, 5860,
8081, 85, 89, 9596, 117118, 129,
137165
detection under conditions that favor
unfolding, 27, 61, 117118
early, 54, 5860, 6364, 6669, 77, 137,
146148, 155157, 159165
kinetic implications of the occurrence,
117118
molten globule, 3, 60, 71, 161
stability, 146, 155, 160
structural resolution by using CD, 4, 51,
69, 89, 144, 148, 155
structural resolution by using -value
analysis, 60, 7677, 105123, 149164
structural resolution by using NMR, 51,
54, 7981, 85, 89, 9597
Inter-residue contact maps, 141, 152155,
165180
Isoelectric point, 133
Isotopes
abundance, 83
exchange kinetics, 54, 76, 8182, 9597
labeling, 8082, 9597
k, see Rate constants
kB (Boltzmann constant) XV, 28, 83
Kinetic implications of the occurrence of
folding and unfolding intermediates,
117118
Kinetic rate constants, see Rate constants
Kinetic resolution, 34, 5182
Kinetics
four-state, 40, 174
multi-state, 121122
three-state, 3139, 117118, 122123,
156, 171, 172, 184
Index
two-state, 2931, 4344, 105, 112113,
116118, 120121, 173, 181, 183
Labeling
fluorescence, 8082
isotope, 8082, 9596
radioactive, 8082
-Lactalbumin, 71
Lactoglobulin, 61
Laplace equation, 71
LASER-induced
flash photolysis, 6569
temperature-jumping, 6264, 77
LASER spectroscopy, 6268, 102104, 130
LeChateliers principle, 72
LennardJones potential, 21
Light harvesting complex, 1415
Light scattering
dynamic, 51, 129130
static, 129
Lock-in amplifier, 90
London dispersion forces, 19
Lysozyme, 24, 71, 126, 138
Magnetic circular dichroism (MCD), 69
Magnetic field of NMR spectrometers, 84
Magnetogyric ratio, 83
Mass spectrometry, 51, 128, 138
Maximum
protein stability as function of T, 59
rate of folding, 3
time resolution in T-jump experiments,
6465
MCT detector, 67
Mean residual ellipticity (R), 90, 9293,
111, 139, 142143
Mechanisms of folding,
framework, 162, 165, 166, 169, 179
funnel, 141, 161, 164, 166, 168, 171
hydrophobic collapse, 164, 166, 168, 179
nucleationcondensation, 60, 137, 158,
162180, 184, 195
Membrane proteins, 1415
Misfolding, 2, 54, 126, 133135, 158,
160161, 164165
Mixing
double-jump, 54, 9697
stopped-flow, 5154, 145, 148
ultrafast, 5154, 7677
Models for folding,
framework, 162, 165, 166, 169, 179
funnel, 141, 161, 164, 166, 168, 171
hydrophobic collapse, 164, 166, 168, 179
219
nucleationcondensation, 60, 137, 158,

162180
Modulators of polarization, 9091, 98,
100102, 104
electro-optical, 98, 100101
elasto-optical (photoelastic), 9091, 100,
102
strain plate, 103104
Molar (universal) gas constant (R), XV, 22,
28, 58, 113, 115
Molecular dynamics, 11, 70
Molecular weight
calculation using amino acid
composition, 7
measurement of proteins, 128130
Molten globule intermediates, 3, 60, 71, 161
Molscript, 1214, 138, 149, 156
Monomerdimer equilibrium, -transition,
27, 4147, 115, 119120, 131, 171, 172,
174178, 180
Motif, 9, 151, 157, 161162
Multichannel detector, 5354, 98101, 104
Multi-state transitions, 27, 113, 115,
121122
Mutagenesis
cassette, 107108, 127, 138
for the assignment of NMR peaks, 79
oligonucleotide-directed, 107111,
126128
PCR-mediated, 107108, 110111, 127,
138
site-directed, 79, 107111, 127, 138,
149155
use for the -value analysis, 105106,
149, 154
Mutants for the -value analysis, 105107,
116, 120123, 135138, 149155, 161
m-value, 112, 116, 144, 157
Myoglobin, 24, 6364, 68, 165
15N, 75, 83
Nd:YAG LASER, 63, 6768, 102, 104
NMR, see Nuclear magnetic resonance
Non-native interactions in and between
proteins, 66, 123, 142, 154, 157158,
160161, see also Misfolding and
Aggregation
Non-polar groups, 5, 17, 2325, 166
N-terminal
consolidation, 158
methionine, 128, 138
sequencing, 138
Nucleationcondensation mechanism
220
Index
60, 137, 158, 162180

Nuclear magnetic resonance (NMR)
spectroscopy, 51, 54, 57, 60, 7589,
9597, 140, 142, 154, 162, 165
chemical shift, 76, 8485, 89
coupling constant, 8485
detection of H/D exchange kinetics, 51,
54, 76, 8182, 9597, 165
Fourier-transform principle, 8688
line broadening, 7577, 165
multiplet, 85
nuclear Overhauser effect (NOE),
through-space interaction, 8485, 8889
real-time, 51, 7980
solution of protein structures, 8389, 142
spectrometer, 84, 140
stopped-flow, 51, 7980
tetramethylsilane (TMS), 85
two-dimensional, 88
Nuclear Overhauser effect (NOE), throughspace interaction, 8485, 8889
Nucleus of folding, 60, 137, 158, 162165
Oligonucleotide-directed (oligodeoxynucleotide-directed) mutagenesis, 107111,
127, 138
Optical parametric oscillator (OPO), 62
Organic solvents, co-solvents, 2, 18, 2225,
132
p53 domain, 165, 170, 174, 177178
31P, 83
PCR-mediated mutagenesis, 107111, 127,
138
PDB, Brookhaven National Laboratory, 14
Peptide bond, 5, 78
Peptides, 2, 78, 28, 64, 73, 140, 161163,
177, 179, 183
Permittivity, 1718
Perturbation methods, see Relaxation
methods
Phenylalanine, 59, 11, 2223
pH-dependence
H/D exchange, 95
protein aggregation, 133
sound velocity, 71
pH-jump, 9597
Phosphatidylinositol 3-kinase, 192
Photoelastic (elasto-optical) modulator,
9091, 100, 102
Photolysis
aromatic residues, 55, 80
COheme bond, 6569, 77
Photomultiplier, 56, 90, 98, 130, 140

Photon shot noise, 56, 99101, 104
pI, 133
pKa of amino acids, 8
Planck constant (h), XV, 28, 83
Plasmid, DNA, 107111, 126128, 138
/4-Plate, 91, 102
Point charges, 1720
Polar amino acid residues, polar groups, 57,
9, 17, 22, 2425
Polarizer, 90, 98, 102104
Polarization of light
circular, 8991, 98104
circular dichroism, 4, 51, 60, 6869,
8993, 98104, 113, 139144, 148, 155,
160, 162
linear, 9091, 98, 100104
modulation, 9091, 98, 100104
Polylinker, 107
Polymerase chain reaction (PCR), 107111,
127, 138
Post-translational modifications, 127128
Prediction of protein structure, 187193
Prediction of secondary structure,
conformational preference, 9, 11
Pressure-jump method, 7273, 77
Primary structure of proteins, 58, 158, 165
Primer for PCR, 109110, 127
Prion diseases, 2, 166
Prion protein domain, 13
Proline
158159
hydroxylation, 128
properties, 58, 11, 2223
structure, 57
Promoter, 107
Protection of protons against exchange, 81,
95
Protein
engineering, 107111, 165180
expression problems, 125128
folding, see Folding
high expression, 138
interactions between proteins, see Interactions
post-translational modifications, 127128
primary structure, 58
secondary structure, 911, 9293, 144
149164
stability, see Energy
structure, 515, 9293, 144, 149164
structure prediction, 187193
Index
tertiary structure, 1115, 93, 144,
149164
Proton (H/D) exchange, 51, 54, 76, 8182,
8485, 9597, 165
Quantum efficiency, 98
Quantum-mechanical repulsion, 1921
Quenched-flow method, 54, 81, 85, 9597
R (molar gas constant), XV, 22, 28, 58, 113,
115
Raman
scattering, 54, 6263, 68
shifter, 6263
spectroscopy, 54, 68
Radioactive labeling, 8082
Random coil, 12, 92, 123, 142, 162
Rapid mixing techniques, 5154, 148
Rate constants, 4, 15, 2749, 54, 6061, 64,
69, 7582, 95, 105, 115123, 130131,
139140, 146148, 155159, 163
calculation, 2749
dissociation, 4147, 115, 119120, 131
extrapolation, 116117
four-state transition, 40
monomerdimer transition, 4147,
119120, 131
multi-state transitions, 116, 121123
perturbation methods, 4449
relaxation methods, 4449
three-state transitions, 3139, 4849, 117,
121123
two-state transitions, 2931, 4849,
116, 120121
Red-shift of fluorescence spectrum upon
unfolding, 144145, 160
Refolding methods, 5177, 9697, 126,
134135, 137, 139, 146
Relaxation methods
acoustic, 6971, 7677
dielectric, 27, 44, 73, 7677
discrimination between folding and
association events, 42, 115, 119120
electric field-jump, 7374
electron transfer, 69, 77
flash photolysis, 6569, 77
pressure-jump, 7273, 77
rate constants, 4449
repetitive pressure application, 7273
temperature-jump, 27, 44, 5565, 7374,
7677, 119, 140, 145147, 160, 163
time resolution, 64, 7677
ultrasonic, 6971, 7677
221
Repetitive pressure-perturbation method

(RPPM), 27, 7273
-Repressor, 76, 163
Residual structure in the unfolded (denatured)
state, 60, 105, 123, 141143
Resonance Raman spectroscopy, 54, 68
Resonator sound velocity meter, 70
Restriction sites, 107108, 111, 128, 138
Roll-over effect, 117118
Salt bridges, 184
Salt-dependence of
electrostatic interactions, 18
protein solubility, 132133
Scrapie, 166
Secondary structure in proteins, 911,
9293, 144, 149164
Second-harmonic generation (SHG), 64,
6768
Self-evolving computer programs, 187193
Self-evolution of complex systems, 193
SH3 domain, 165, 170, 172174, 177, 179
-Sheet, strands, 915, 92, 138, 140,
149157, 162
Shot noise, 56, 99101, 104
Sidechain mobility, 11, 145
Site-directed mutagenesis, 79, 107111, 127,
138, 149155
Slow reactions, 4, 61, 7982
Solvent exclusion upon folding, 146147,
155158, 160, 177
Soret region, 69
Sound
absorption, 6971
velocity, 6971
Spin of nucleus, 58, 83, 85, 87
Stability of protein, see Energy
Stokes lines, 63
Stopped-flow method, 5152, 7980, 98,
139, 145, 148
Strain plate, 103104
Structural preferences of amino acids, 9, 11
Structural resolution
by -value analysis, 123, 159-180
for kinetic methods, 77
of proteins in solution, 8393, 165180
of transient protein conformations,
95123, 137165
Structure,
amino acids, 58
domain, subdomain, 11, 13, 15, 76, 141,
165
222
Index
globular proteins, 515, 9293, 138, 144,

156
intermediates, 80, 9596, 105106, 123,
137, 147158, 160162, 171, 174, 175,
177, 195
motif, 9, 151, 157, 160162
prediction, 187193
primary, 58, 158, 165
resolution by using CD spectroscopy, 4,
51, 60, 69, 8993, 98104, 142144, 148,
155, 160, 162
resolution by using -value analysis, 60,
76, 105123, 149180
resolution by using NMR spectroscopy,
51, 54, 60, 7589, 9597, 142, 154, 162,
165
secondary, 911, 60, 64, 69, 89, 9293,
137, 139, 143144, 149180
tertiary, 1115, 60, 71, 89, 93, 144,
149180
transition states of folding, 105106,
120123, 149158, 162180
Surfactants, 133
Thermus aquaticus polymerase, 13, 109
Temperature-jumping, 27, 44, 5565, 7677,
119, 140, 145148, 160, 163
discrimination between folding and
association events, 115, 119120
electrical-discharge-induced, 5561,
77, 140, 145148, 160, 163
LASER-induced, 6265, 77
maximum time resolution, 6465, 77
Tertiary structure of proteins, 1115, 60, 71,
89, 93, 144, 149180
Tetramethylsilane (TMS), 85
Three-state transitions, 27, 3139, 4849,
117118, 122123, 156,
Time resolution, 34, 5182, 96, 101103,
137, 165
Time scale of folding events, 34
Transducers, 6970, 72
Transformation of DNA, 108, 125126, 138
Transitions
four-state, 40, 48, 174
multi-state, 116, 121123
three-state, 27, 3139, 4849, 117118,
122123, 156, 166, 171, 172
two-state, 2931, 4849, 112113,
116121, 173175, 181, 183
Transition state
structural resolution by using -value
analysis, 76, 105123, 149180
theory, 2829, 120123
Trifluoroethanol, 22, 132, 162
Tritium (3H), 82
Tryptophan, 58, 11, 2223, 79, 93, 146,
162
Turns, 911, 92
Two-dimensional NMR, 8889
Two-state transitions, 2931, 4849,
112113, 116121, 173175, 181, 183
Tyrosine, 58, 11, 2223, 128
Ultracentrifugation, 80, 130
Ultrafast folding, 185
Ultrafast mixing, 5154, 77
Ultrafiltration, 96
Ultrasonication for cell lysis, 126, 138
Ultrasonic relaxation, ultrasonic velocimetry,
27, 44, 6971
Unfolded state, see Denatured state
Unfolding intermediate, 61, 117118
Unfolding of protein, induced by cold,
denaturants or heat, see Denaturation
Unimolecular reactions, 2740, 48, 116118
Unstable curve fit, 115, 135136
-Value analysis, 38, 60, 7677, 105123,
135, 137, 149180
Van der Waals potential, 7, 17, 1921
Vector, cloning-, 107108, 110111, 127
Volume
change upon folding, 7072
fluctuations, 70
of amino acids, 7
of proteins, 11
Water,
compressibility, 11
infrared absorption, 6263
permittivity, 18
sound velocity, 70
thermal expansion, 62
ultra-violet absorption, 92, 99
X-ray scattering, 51
YAG LASER, 63, 6768, 102, 104

Bengt Nolting Protein Folding Kinetics Biophysic PDF

Uploaded by

Copyright:

Available Formats

Bengt Nolting Protein Folding Kinetics Biophysic 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

Bengt Nolting Protein Folding Kinetics Biophysic PDF

Uploaded by

Copyright:

Available Formats

Protein Folding Kinetics

Dr. Bengt Nlting

Library of Congress Control Number: 2005929411

ISBN-10 3-540-27277-1 2nd Edition Springer Berlin Heidelberg New York

Preface to the first edition

Preface to the first edition

along the folding pathway will lead to an understanding of function and

Physical interactions that determine

Calculation of the kinetic rate constants

Transition state theory . . . . . . . . . . . . . . . .

High kinetic resolution of protein folding events

Stopped-flow nuclear magnetic resonance (NMR)

Nuclear magnetic resonance . . . . . . . . . . . . . . . . . . .

High structural resolution of transient . . . . . . . . . . . . . .

Resolution of protein structures in solution

Kinetic methods for slow reactions

Reversible two-state folding transition linked . . . . . .

NMR detection of H/D exchange kinetics . . . . . . . . . . . .

10 The folding pathway of a protein (barstar) at the . . . . . . . .

Experimental problems of the kinetic and structural

Protein expression problems . . .

11 Highly resolved folding pathways and mechanisms

11.1 General features of the main transition states for the . .

12 Structural determinants of the rate of protein folding . . . . .

12.1 Chain topology as a major structural determinant of two-state

13 Evolutionary computer programming of protein structure . .

13.1 Evolution method . . . . . . . . . . . . . . . . . . . . . . . . .

observed rate constant, relaxation

ment of new methods which enable us to access the submillisecond, microsecond,

2.1 Primary structure

Table 2.2. Properties of amino acid residues commonly found in proteins.

Van der Waals

Fig. 2.1. The peptide

pKa of amino acidsa

pKa of amino acid

(Dawson et al., 1969; Fersht, 1985; Zubay, 1993; Lide, 1993).

2.2 Secondary structure

Fig. 2.2. Right-handed helix consisting of 8 amino

Fig. 2.4. Type I turn. Turns

2.3 Tertiary structure

2.3 Tertiary structure

Nature has evolved a gigantic variety of protein three-dimensional structures,

2.3 Tertiary structure

Alpha and beta with mainly parallel -sheets (/)

Alpha and beta with mainly antiparallel

Multi domain (alpha and beta)

3 Physical interactions that determine the

where d1,2 is the distance between Z1 and Z2 , o = 8.8541012 C V1 m1 is the

3 Physical interactions that determine the properties of proteins

3.2 Quantum-mechanical short-range repulsion

3 Physical interactions that determine the properties of proteins

3.3 Hydrogen bonding

C12 = C12, i C12, j

3 Physical interactions that determine the properties of proteins

Hydrogen bonding of proteins in aqueous solution is profoundly altered by

Estimated with the rolling ball method.

3.4 Hydrophobic interaction

3 Physical interactions that determine the properties of proteins

The Gibbs free energy of transfer, Ghyd , of a non-polar molecule from a