doi:10.1016/j.jmb.2005.01.037 J. Mol. Biol.

(2005) 347, 231241

C OMMUNICATION
A Survey of Left-handed Helices in Protein Structures
Marian Novotny1,2 and Gerard J. Kleywegt1*
1
Department of Cell and All naturally occurring amino acids with the exception of glycine contain
Molecular Biology, Uppsala one or more chiral carbon atoms and can therefore occur in two different
University, Box 596, SE-751 24 configurations, L (levo, left-handed) and D (dextro, right-handed). Proteins
Uppsala, Sweden are almost exclusively built from L-amino acids. The stereochemical bias of
2 nature is further reflected at the secondary structure level where right-
Linnaeus Centre for
handed helices are strongly preferred over left-handed helices.
Bioinformatics, Biomedical
The handedness of helices has not received much attention in the past
Centre, Box 598, SE-751 24
and is often overlooked during the analysis, description and deposition of
Uppsala, Sweden
experimentally solved protein structures. Therefore, an extensive survey of
left-handed helices in the Protein Data Bank (PDB) was undertaken to
analyse their frequency of occurrence, length, amino acid composition,
conservation and possible structural or functional role.
All left-handed helices (of four or more residues) in a non-redundant
subset of the PDB, were identified using hydrogen-bonding analysis,
comparison of related structures, and experimental electron density
assessment to filter out likely spurious and artefactual hits. This analysis
yielded 31 verified left-handed helices in a set of 7284 proteins. The f
angles of the residues in the left-handed helices lie between 308and 1308
and the j angles lie between K508 and 1008. Most of the helices are short
(four residues) and for 87% of them, it was possible to determine that they
are important for the stability of the protein, for ligand binding, or as part of
the active site. This suggests that, even though left-handed helices are rare,
when they do occur, they are structurally or functionally significant.
Four secondary structure assignment programs were tested for their
ability to identify the handedness of the helices. Of these programs, only
DSSP correctly assigns the handedness.
q 2005 Elsevier Ltd. All rights reserved.
Keywords: left-handed helix; protein fold; protein structure; secondary
*Corresponding author structure; structurefunction relations

Nature shows a profound right-left asymmetry.1 side-chain atoms and the main-chain carbonyl
DNA appears mainly in the right-handed moiety.2 Conversely, D-amino acids will form more
B-conformation as do a-helices in proteins. This stable a-helices with a left-handed than with a right-
right-handed preference of biological macro- handed conformation. This phenomenon is perhaps
molecules is a consequence of the selective incor- most convincingly illustrated by the fact that the
poration of L-amino acids into proteins and of structures of all-D proteins, such as D-rubredoxin3
D-monosaccharides into DNA. The basis for this and D-monellin,4 are the mirror images of the
selectivity remains unknown, although a number of corresponding all-L proteins.
explanations have been proposed.1 a-Helices are characterised by a typical hydrogen
a-Helices composed of L-amino acids are ener- bonding pattern in which the main-chain carbonyl
getically more favourable in a right-handed confor- oxygen of residue i forms a hydrogen bond with the
mation than in the left-handed mirror image of main-chain amide hydrogen of residue iC4. Other
this arrangement due to steric hindrance between types of helices can also occur in proteins,
namely 310 and p-helices with (i, iC3) and (i, iC5)
Abbreviations used: PDB, Protein Data Bank; EDS, hydrogen bonds, respectively. 310-Helices occur
Electron Density Server. fairly often in proteins, although they are frequently
E-mail address of the corresponding author: shorter than a-helices.5 p-Helices are relatively rare
[email protected] motifs, but when they do occur they can be of

0022-2836/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.
232 Left-handed Helices

functional importance due to their unique successfully) four secondary structure assign-
structural parameters.5 To our knowledge, the ment programs assign the handedness of left-
handedness of helices in proteins has not been handed helices.
studied systematically, and not much is known
about left-handed helices other than that they are Detection and verification of left-handed helices
very rare.
Stretches of amino acids with unusual backbone Initially, ideal left-handed helices of different
conformations (e.g. left-handed helices) often lengths (four to ten residues) were constructed
appear at ligand-binding sites, proteinprotein and used as templates in SPASM searches15 against
interfaces or other functional sites. It has been subsets of the PDB.17 The hits obtained in these
suggested that proteins may sacrifice a part of their searches were used to define putative left-handed
stability to form an effective functional site.6,7 It has helices as being continuous stretches of at least four
further been suggested that searches for regions residues whose f angles are all between 308 and
with unusual backbone conformations could be 1308 and whose j angles are all between K508 and
used for the annotation of novel protein structures, 1008. Subsequently, a non-redundant subset of the
since such regions are candidates for being PDB (version of September, 2003) was generated
functional sites.7 with the PISCES server.18 To produce a large
Little is known about how residues with left- enough subset, relaxed criteria were used in the
handed helical conformations affect the stability of generation of the subset: none of the pairs of protein
proteins. It is often assumed that such residues chains had more than 90% sequence identity, and all
(with the exception of glycine residues) would crystal structures with a resolution better than 3.5 A
suffer steric clashes between main-chain and Cb and an R-value less than 0.4 were included, and so
atoms, thereby reducing the stability of the protein.2 were all proteins whose structure had been solved
However, Takano et al. found that five of six non- by nuclear magnetic resonance (NMR) spec-
glycine left-handed residues in lysozyme do not troscopy. The minimum chain length was set to
significantly impair the stability of lysozyme.8 On ten amino acid residues. The resulting subset
the other hand, these residues are scattered contained 7284 protein chains from 6535 PDB
throughout the lysozyme sequence and do not entries and included 1,687,315 amino acids. A Perl
form a left-handed helix. Scattered amino acids program was written to find all instances of left-
with a left-handed helical conformation also occur handed helices (according to the definition above)
in type I 00 turns, where two amino acids have in this subset. Initially, wider ranges for the
this conformation, and in helix stop signal9 and torsion angles were used (f between K208and
Schellman motifs.10 1458 and j between K708 and 1458), and the hits
To our knowledge, there are currently only three that were obtained were examined to check if
protein structure entries in the PDB that have a left- their hydrogen-bonding pattern was compatible
handed helix assigned. Two of these are four with left-handed a or 310-helices. The final ranges
residues long and they are found in thermolysin yielded no false negatives and relatively few false
(PDB code 8tln)11 and neutral protease (1npc).12 The positives.
third one is a three-residue long helix in granulo- If any part of a structure satisfied our definition of
cyte-colony-stimulating factor (1rhg).13 A PubMed a left-handed helix it was designated a putative hit.
search yielded one more left-handed helix, namely The hits were visualised with Deep View,19 and
in spinach glycolate oxidase (1gox).14 Interestingly, hydrogen-bonding patterns were analysed with
this left-handed helix is a part of the active site of Deep View, DSSP20 and HBPLUS.21 Putative left-
the enzyme. The left-handed helices in glycolate handed helices were subjected to additional checks
oxidase and granulocyte-colony-stimulating factor to validate them or to dismiss them as probable
both have a hydrogen-bonding pattern consistent artefacts. Any hits in NMR structures were only
with a 310-helix. accepted if the left-handed helix occurred in at least
A few years ago we discovered a left-handed 50% of the models of the ensemble. For those crystal
helix that plays an important role in alanine structures for which electron-density maps were
racemase15 and that had previously been classified available from the Uppsala Electron Density Server
incorrectly as right-handed.16 We also noted at (EDS),22 the density for the putative helices was
the time that several secondary structure assign- inspected. In cases where no map was available
ment programs failed to annotate the handedness of (including all hits found in NMR structures), the
this helix. These findings prompted us to undertake PDB was checked for structures of the same or
related proteins and, if any were found, it was
a survey of left-handed helices in the PDB with
checked if the corresponding residues in the related
respect to their frequency of occurrence, their
structures had similar f and j angles as the
length, their sequence characteristics (if any),
residues in the putative left-handed helix. If no
their conservation and, in particular, their pos-
related structures could be found in the PDB, we
sible functional or structural roles. We also
trusted the authors and accepted the hit as a true
investigated in more detail if (and how
left-handed helix.
For each of the accepted left-handed helices,
http://pubmed.gov/. many sources of information (literature, contacts
Left-handed Helices 233

with authors, and bioinformatics resources such as cases, one or more closely related structures were
SWISS-PROT,23 ProSite,24 PDBsum25 and Omim) available in which no support could be found for
were consulted to find information about the any left-handed helical conformation, and in one
possible functional or structural significance of the case there was no support in the electron density
residues in the helix. (although the structure had been determined at
We also investigated if the left-handed helices are 1.9 A resolution and refined to a crystallographic
recognised as such by the secondary structure R-value of 0.2). Finally, one entry was accepted
assignment tools DSSP,20 STRIDE,26 Promotif27 in good faith. After these validation steps, there
and SecStruct,5 and in the annotation in the original remained 31 hits that were deemed to be genuine
PDB files. The frequencies of occurrence of the 20 left-handed helices (Table 1). Based on hydrogen-
natural amino acids in the left-handed helices were bonding analysis, the left-handed helices were
calculated for comparison with their frequencies in further divided into a-helices (11 cases) and 310-
the entire subset of the PDB. The sequences of the helices (20 cases). Two of the hits occurred in NMR
left-handed helices were used to search a 95% structures.
sequence identity subset of the PDB with the The left-handed helices were short; the longest
PATINPROT server 28 to find out if identical helix was six residues long, but the majority were
sequences occur in any other proteins whose just four residues long (Table 1). The distribution of
structure is known. Any hits were then analysed the f and j torsion angle values of all the left-
to determine if the sequence displayed strong handed helices is shown in Figure 1. For a-helices,
secondary structure preferences. the average values of f and j were 598 (sZ128) and
We also generated a non-redundant subset of the
428 (sZ138), respectively; for 310-helices, these
PDB with PISCES18 using a 25% cut-off for the
values were 678 (sZ218) and 238 (sZ258), respect-
sequence identity (all other criteria were the same as
ively. The 310-helices thus had a wider distribution
in the generation of the 90% subset) and located all
of f and j angles than the a-helices. The observed
left-handed helices in that subset. For comparison
purposes, the set of proteins from the 90% subset torsion angle ranges for the left-handed helices are
that contained a left-handed helix was also pruned similar to those defined by Gunasekaran et al. (208 to
with PISCES to yield a set of left-handed-helix 1258 for f and K458 to 908 for j).9 The average f
containing proteins with no more than 25% and j angles for the left-handed helices are also
sequence identity. similar to those observed for right-handed helices
(ignoring the sign changes).29 The proteins that
contain left-handed helices show no preference for
Analysis of the left-handed helices in the PDB overall secondary structure contents; they belong
to the classes of mainly alpha, mainly-beta and
Ideal left-handed helices of four to ten residues mixed alpha-beta proteins. Most of the left-
were used in SPASM searches for similar motifs in handed helices are located on the protein surface,
the PDB. This yielded 21 left-handed helices that but there are no obvious patterns in the types
were used to define the initial ranges of f and j and spatial orientations of flanking secondary
angle values for residues in left-handed helices, structure elements.
namely 08 to 1258 for f and K508 to 758 for j. Later, All amino acid types except proline were encoun-
these ranges were extended (K208 to 1458 for f and tered in left-handed helices. Proline residues are
K708 to 1458 for j) to make sure that no true left- unlikely to appear in helices in general, because
handed helices would be missed (false nega- their imido nitrogen atom cannot donate a hydro-
tives). Eventually, the ranges were narrowed gen bond and because the bulky Cd methylene
down to the final values (308 to 1308 for f and group attached to the nitrogen introduces severe
K508 to 1008 for j), which ensured that there steric clashes. Moreover, due to the inability of
were no false negatives, and not too many false proline residues to assume positive f values, they
positives. A minimum length of four residues are not expected to occur in any left-handed helices
was imposed on potential left-handed helices to at all, and this expectation is borne out by our
ensure that any such helices would contain at observations. The second position in the helix is
least one full turn. occupied by asparagine in 11 cases and by
The criteria for detecting left-handed helices were glutamate or glutamine in six more cases. The last
implemented in a Perl program that was run on all two positions in the helix are often occupied by two
the members of the non-redundant subset of the identical residues, usually Gly-Gly (14 cases), and
PDB described earlier, an exercise that yielded 56 an even higher prevalence of glycine is apparent at
putative hits. All hits in structures determined by the last position in the helices (19 cases). This is
NMR were required to have a left-handed helical most likely because glycine is achiral and lacks a
conformation in more than half of the models in the side-chain, so that left-handed and right-handed
ensemble (to filter out spurious hits with little conformations are equally favourable. Although the
support in the experimental NMR data), which sample size is very small, tryptophan has a
reduced the number of hits to 38. In six of the 38 surprisingly high propensity for being in left-
handed helices, especially compared to its pro-
http://www.ncbi.nlm.nih.gov/omim/ pensity for assuming a left-handed backbone
234 Left-handed Helices

Table 1. Left-handed helices encountered in a subset of the PDB

PDB code and CATH Helix

chain IDa Protein name classificationb Residues Sequence typec Reference

1bd0A Alanine racemase 2.40.37.10 4044 ANAYG a 16

1autL Activated protein C 2.10.25.10 101104 DNGG 310 34
1b9wA Merozoite surface protein 1 5255 KNGG 310 38
(P. cynomolgi)
1g2lB Coagulation factor X 2.10.25.10 258261 DNGD 310 59
1kliL Coagulation factor VII 2.10.25.10 9497 ENGG 310 60
1n1iA Merozoite surface protein 1 5760 NNGG 310 61
(P. knowlesi)
1ob1C Merozoite surface protein 1 5255 NNGG 310 62
(P. falciparum)
1rfn_ Coagulation factor IX 2.10.25.10 9194 KNGR 310 63
1pb5A Lnr module from Notch 2832 GWDGG 310 44
1kdgA Cellobiose dehydrogenase 532535 YENW 310 45
1hzmA Protein phosphatase 6 6164 IMLR a 46
1gtxA 4-Aminobutyrate 3.30.70.160 7073 SQIS a 49
aminotransferase
1qj5A 7, 8-Diaminopelargonic acid 3.30.70.160 5053 SSWW 310 48
synthase
2gsaA Glutamate semialdehyde 3.30.70.160 6568 GTWG 310 47
aminotransferase
2oatA Ornithine aminotransferase 3.30.70.160 8386 SSYS 310 50
1bnlA Endostatin (Homo sapiens) 3.10.100.10 135138 CETW a 52
1dy2A Endostatin (M. musculus) 3.10.100.10 207210 CEAW a 54
1koeS Endostatin (M. musculus) 3.10.100.10 266269 CETW a 53
1bqbA Aureolysin 1.10.390.10 223226 DNGG a 64
1npcE Neutral protease 1.10.390.10 227230 DNGG a 12
8tlnE Thermolysin 1.10.390.10 226229 DNGG a 11
1j9qA Nitrate reductase (A. feacalis) 2.60.40.240 105108 ALGG 310 65
1nifA Nitrate reductase (A. cycloclates) 2.60.40.240 105108 ALGG 310 55
1oe1A Nitrate reductase (A. xylosoxidans) 2.60.40.240 99102 ALGG 310 66
1ptmA 4-Hydroxythreonine-4-phosphate 211216 HAGEGG 310 56
dehydrogenase
1ak0E P1 nuclease 1.10.575.10 131134 AVGG a 57
1mzr_ 2, 5-Diketo-D-gluconate reductase 191194 AQGG 310 d
1h21A Split-soret cytochrome C 7781 GGISD 310 67
1hxxA Ompf porin 2.40.160.10 143146 NFFG 310 68
1jv1A Glcnac1p uridyltransferase 182185 KYFG 310 69
1kwsA Beta-1,3-glucuronyltransferase 3 3.90.550.10 298301 AANC a 70
a
The four-character PDB code, followed by the chain identifier (where an underscore signifies a blank chain identifier). The entries
are sorted in the order in which they are discussed in the text (i.e. grouped by function). Entries 1hzm and 1pb5 are NMR structures; all
other structures were determined by X-ray crystallographic methods.
b
The CATH33 classification; a dash indicates that the PDB entry had not been classified by CATH yet.
c
The type of the left-handed helices, as assigned manually based on hydrogen-bonding patterns.
d
C. Abergel, S. Jeudy, V. Monchois and J.M. Claverie, Structure of DkgA from Escherichia coli at 2.13 A resolution solved by molecular
replacement (unpublished results).

conformation. Some additional data can be found helix only contains an RL nest, whereas 14 helices
on our web site. contain only an LR nest.
Sometimes right-handed helices have a Schell- To assess if the short amino acid sequences of the
man turn as a C-cap.10,30 In such a motif, the helix is left-handed helices always occur in such secondary
terminated by a residue with a left-handed confor- structure elements, all occurrences of each of the 31
mation, and 52 or 61 hydrogen bonds are usually sequences were located in a 95% sequence-identity
formed. Of the present set of left-handed helices, 13 subset of the PDB with the PATINPROT server.28 As
contain a left-handed version of the Schellman turn. expected,32 none of the sequences found in the left-
Whenever two consecutive residues have a helical handed helices showed any strong secondary
conformation of opposite handedness (i.e. RL or structure preferences. Most were found to occur in
LR), they form a motif called a nest.31 Analysis of all three secondary structure types (helix, strand
the 31 left-handed helices and their flanking and loop), e.g. the sequence AQGG originally found
residues revealed that six of them contain both an in P1 nuclease (PDB code 1ak0) as a left-handed
RL nest at their N terminus and an LR nest at their C helix was found in ten other proteins in the PDB
terminus, whereas ten of them contain neither. One subset where it appeared as a helix, a strand, a loop
and combinations thereof. The sequence of the
longest left-handed helix (HAGEGG) was not
found in any other protein in the PDB.
http://xray.bmc.uu.se/wmarian/left The extent to which left-handed helices are
Left-handed Helices 235

Figure 1. Distribution of f and j torsion angles of residues found in left-handed helices. Pink squares represent
residues in 310-helices and blue diamonds represent residues in a-helices. (For interpretation of the references to colour
in this Figure legend, the reader is referred to the web version of this article.)

conserved between different sequence families at addition, the assignments of these helices in the
the same homologous superfamily level of the parent PDB files were retrieved. Significant differ-
CATH classification33 was also investigated. It was ences in the results of the programs were observed.
found that none of the helices that occurred in In fact, in none of the cases did the programs
proteins that had been classified in CATH are produce an identical result. Only two of the
conserved across all corresponding sequence programs, DSSP and SecStruct, provided any
families. This suggests that the left-handed helix information at all about the handedness of the
motifs have evolved relatively recently and that secondary structure elements (both of them in a
they serve a specific purpose. column labelled chirality in the output). The
To assess the effect of using different cut-offs for results obtained with the oldest program, DSSP,
the allowed level of sequence identity during the agree best with our own manual assignments. Good
generation of reduced subsets of the PDB, the agreement was also obtained with the results of
analysis was repeated with a 25% subset. Whereas SecStruct, but the chirality assignments of this
the 90% subset yielded 31 left-handed helices, only program were often misleading. Interestingly, the
13 such helices were found in the 25% subset. handedness of the helices appears to have escaped
However, if the set of 31 proteins that contain a left- the attention of most of the depositors and
handed helix is in turn reduced to a subset of annotators of the structures. As far as we could
proteins that have no more than 25% sequence determine, only four of the 31 left-handed helices
identity, the number of hits is as high as 18. This were described as such in the corresponding PDB
discrepancy can be explained by realising that the file (1npc, 8tln) or original papers (1b9w, 1n1i).
two processes of selecting a subset of the PDB using
a cut-off on the allowable sequence identity level Function of left-handed helices
and that of testing a set of proteins for a particular
property (in this case, the presence of left-handed For each of the 31 left-handed helices it was
helices) are not commutative. In this case, 13 is the investigated if they are important for the structure
number of left-handed helices found in a 25% or function of the parent protein. The results are as
subset of the PDB, whereas 18 is the size of a 25% follows (the hits have been grouped according to
PDB subset of the proteins that are known to sequence and structure similarity):
contain a left-handed helix.
Finally, four secondary structure assignment
Alanine racemase
programs (DSSP, STRIDE, PROMOTIF and Sec-
Struct) were tested for their ability to recognise the This enzyme mediates the interconversion of L
handedness of left-handed helices (Table 2). In and D-alanine, which is indispensable for cell wall
236 Left-handed Helices

Table 2. Comparison of secondary structure assignments of left-handed helices by different methods

PDB code and

chain IDa Type b PDB c,d PROMOTIF d,e DSSP d,e STRIDE d,e SECSTR d,e

1bd0A H HHHHH HHHHH HHHTT (L) TGGGC HHHH- (R)

1autL G GGGG GGGG GGGG (L) GGGG GGGG (L)
1b9wA G GGGG GGGG GGGG (L) GGGG GGG- (L)
1g2l_ G GGGG GGGG GGGG (L) GGGG GGG- (L)
1kliL G GGGG GGGG GGGG (L) GGGG GGG- (L)
1n1iA G GGGG GGGG GGGG (L) GGGG GGG- (L)
1ob1C G GGGG GGGG GGGG (L) GGGG GGGG (L)
1rfn_ G GGGG GGGG GGGG (L) GGGG GGGG (L)
1pb5A G GGGG GGGG GGGG (L) GGGC
1kdgA G GGGG GGGG GGGG (L) GGGG GGGG (L)
1hzmA H HHHH HHHH HHHH (L) TTTT HHHH (L)
1gtxA H HHHH HHHH HHHH (L) HHHH HHHH (L)
1qj5A G TTTT (L) HCCC
2gsaA G GGGG HHHT GGGT (L) GGGT
2oatA G HHHH GGGG GGGG (L) GGGG GGGG(R)
1bnlA H HHHH HHHH HHHH (L) TTTT HHHH (L)
1dy2A H HHHH HHHH HHHH (L) GGGT HHHH (L)
1koeS H HHHH HHHH HHHH (L) TTTT HHHH (L)
1bqbA H HHHH HHHH HHHH (L) GGGC HHHH (R)
1npcE H HHHH/left HHHH HHHH (L) GGGC HHHH (R)
8tlnE H HHHH/left HHHH HHHH (L) GGGC HHHH (R)
1j9qA G GGGG GGGG GGGG (L) GGGC GGGG(R)
1nifA G GGGG GGGG GGGG (L) GGGG GGGG(R)
1oe1A G GGGG GGGG GGGG (L) GGGG GGGG(R)
1ptmA G GGGG GGGGGG GGGGGG (L) GGGGGG GGGGGG (L)
1ak0E H HHH- HHHT HHHT (L) TGGG HHHH (L)
1mzr_ G TTTT TTTT (L) TTTT
1h21A G GGG GGTTT GGTTT (L) GGTTT G (R)
1hxxA G HHHH HHHH HHHH (L) TGGG
1jv1A G GGGG GGGG GGGG (L) GGGG GGGG (L)
1kwsA H HHHH HHHH HHHH (L) TTTT HHHH (L)
a
The four-character PDB code is listed, followed by the chain identifier (where an underscore signifies a blank chain identifier). The
entries are listed in the same order as in Table 1.
b
The type of the left-handed helices is listed as assigned manually based on hydrogen-bonding patterns (HZa-helix; GZ310-helix).
c
Secondary structure assignment for the residues in each left-handed helix in their parent PDB file.
d
, no secondary structure type was assigned, H is a-helix, G 310 -helix, T turn, and S bend. The assigned handedness (if reported) is
listed in parentheses as R for right-handed and L for left-handed.
e
Secondary structure assignment for the residues in each left-handed helix according to the programs listed.

formation. Alanine racemase (PDB code 1bd0) was The seven proteins have a conserved Asn residue
already known to contain a left-handed helix.15 The in their left-handed helices that forms an important
catalytic residue is Lys39 and the left-handed a- hydrogen bond34 either with residue iC5 (protein C
helix covers residues 40 to 44, where Tyr43 is the and factor VII) or iC7 (merozoite surface proteins)
covalent attachment site for the cofactor pyridoxal in the EGF-like domain, or with residues in the
phosphate (Figure 2(a)). The carbonyl oxygen atom protease domain (factor IX and factor X). There is
of Gly44 forms an interdomain hydrogen bond with also some evidence that the left-handed helix in
Arg366.16 The entire helix is part of the PROSITE factor IX is involved in the binding of factor VIII,
pattern (PS00395) for the alanine racemase family. which is its natural interaction partner in the blood-
clotting cascade. Double mutation of Asn89 and
Asn92 to alanine completely abolishes the inter-
EGF-like family action between factor IX and factor VIII.35 However,
A left-handed helix was identified in seven the single mutation of Asn89 to alanine has only a
proteins with EGF-like domains (PDB codes 1aut, marginal effect on binding.36,37
1b9w, 1g2l, 1kli, 1n1i, 1ob1 and 1rfn). Four of these The two consecutive glycine residues in the left-
are blood-clotting proteins (factor VII, factor IX, handed helix have been implicated as protein-
factor X and protein C) and three are merozoite stabilising factors in the merozoite surface pro-
surface proteins from various Plasmodium species. teins.38 The first amino acid in the helix in protein C
These two functionally distinct groups show weak (Asp101) participates in an unusual hydrogen
but detectable sequence similarity, mainly due to bonding interaction with another acidic residue
conserved cysteine residues that form disulphide (Glu85).39 Mutations of three residues in the left-
bridges, but also in a short stretch of sequence handed helix, namely Asp101, Gly103 and Gly104,
containing the left-handed helix. The hydrogen- cause protein C deficiency which may lead to
bonding pattern of these helices classifies them as recurrent venous thrombosis, which in turn may
310-helices rather than a-helices. cause neonatal death.4042
Left-handed Helices 237

domains, one of them being the Lin 12-Notch repeat

module (LNR), which has been suggested to
keep the Notch protein in the idle state.43 A left-
handed 310-helix (residues 2832) was identified in
this module (PDB code 1pb5). It appears to be
important for folding, because Asp30 and Asp33
(immediately following the helix) participate in the
coordination of a Ca2C, whose binding is highly
specific and necessary for folding of LNR.44 The
hydrophobic residue Trp29 lies on the surface of the
module and may be involved in inter-domain
contacts with a second LNR.44

Flavin domain of cellobiose dehydrogenase

Cellobiose dehydrogenase (CDH) is a lignin
and cellulose-decomposing enzyme. Residues 532
to 535 in the flavin domain of this protein (PDB
code 1kdg) form a left-handed 310-helix and line
the entrance to the active site of the protein,
although they are not in contact with the
substrate.45 It has been suggested that the left-
handed helical segment could interact with the
cytochrome domain of CDH (C. Divne, personal
communication).

Erk-2 binding domain of MAPK phosphatase MPK-3

MPK-3 negatively regulates the activity of mito-
gen-activated protein kinases (MAPK) by depho-
sphorylating their threonine and tyrosine residues.
MPK-3 consists of a C-terminal catalytic domain
and an N-terminal domain that is responsible for
binding of MAPK. Residues 6164 (PDB code 1hzm)
have a left-handed a-helical conformation, where
residues Leu63, Arg64 and Arg65 are critical for the
binding of MAPK.46 Residues Ile61 and Met62
stabilise the conformation of the helical segment
by interacting with the hydrophobic core of the
protein.46

Type I PLP-dependent aspartate aminotransferase-

Figure 2. (a) The left-handed helix (in blue) in alanine like fold
racemase (PDB code 1bd0) directly follows the active site Four proteins with this fold were found to contain
residue Lys39. Moreover, Tyr43 in the left-handed helix
binds the cofactor pyridoxal phosphate (PLP). Both
a left-handed helix: ornithine amino transferase
panels of this Figure were created with PyMol 58 (PDB code 2oat), aminobutyrate aminotransferase
(http://www.pymol.org/). (b) His211 in the left-handed (1gtx), glutamate-1-semialdehyde aminomutase
helix (in blue) of PdxA (PDB code 1ptm) participates in (2gsa) and 7,8-diaminopelargonic acid synthase
the coordination of a Zn2C, together with two histidine (1qj5). The third residue in the four-residue long
residues from the other subunit of this protein. (For a-helix (1gtx) or 310-helix (the other three struc-
interpretation of the references to colour in this Figure tures) contributes to the substrate specificity of the
legend, the reader is referred to the web version of this enzymes.4750 A structural segment, equivalent to
article.) the left-handed helical region of these transferases,
in another protein with a type I PLP-dependent
aspartate aminotransferase-like fold, namely di-
alkylglycine decarboxylase (PDB code 1d7r),
Lin12-Notch repeat module of human Notch protein shows some similarity with them. Three of the
four residues (2932) have positive f and j values,
Notch is a transmembrane receptor for many but the third residue lies within the right-
signals that initiate conserved signalling pathways handed helical region of the Ramachandran plot.51
in a variety of cellular responses (e.g. cellular Hence, this segment of the protein does not form a
growth or differentiation). Notch consists of several left-handed helix, but a turn. Interestingly, the third
238 Left-handed Helices

residue is the substrate-binding residue in this RNA and DNA. The enzyme contains a short left-
family of proteins as well. handed a-helix (Ala131 to Gly134) that forms a part
of the active site cleft of the enzyme. Asn135,
immediately following the helix, forms a hydrogen
Endostatins bond with the substrate.57 This enzyme contains
Left-handed a-helical segments (with identical zinc ions, and one of them is coordinated by His126,
sequences) were identified in three different endo- which lies five residues upstream of the left-handed
statins (PDB codes 1bnl, 1dy2, 1koe). A conserved helix.
cysteine residue in these helices participates in a
disulphide bridge that is crucial for the stability of
these proteins.5254 2,5-Diketo-D-gluconic acid reductase (Dkg A)
DkgA (PDB code 1mzr) is an enzyme in the
metabolic pathway to vitamin C. A short 310-helix
Proteins from peptidase family M14
was identified involving amino acids 191 to 194.
A left-handed a-helix was found in three related Leu190 of this enzyme coordinates the phosphate
proteases, namely, in thermolysin (PDB code 8tln), group of the cofactor NADPH (S. Jeudy, personal
aurolysin (1bqb) and neutral protease (1npc), and communication).
the sequence of the helix is identical in all three. The
side-chain of the first amino acid in the helix,
aspartate, forms a hydrogen bond with a histidine Remaining cases
that coordinates the zinc ion in the active site.12 For the remaining proteins that contain a left-
There is also a hydrogen bond between the back- handed helix (PDB codes 1hxx, 1kws, 1jv1 and
bone of this aspartate and the zinc-coordinating 1h21) no indication could be found as to their
histidine, which further underscores the import- possible structural or functional roles.
ance of the helical segment for the catalytic
mechanism of these proteases. Concluding remarks

Nitrate reductase An extensive survey of left-handed helices in

known protein structures has been carried out. The
Nitrate reductases mediate the second reaction in results show that left-handed helices are rare motifs
the N2-fixation pathway, the reduction of nitrate to in protein structures (31 cases in 7284 proteins).
NO. Three bacterial nitrate reductases were found Amino acid residues in these 31 left-handed helices
to contain a left-handed 310-helix (PDB codes 1j9q, have f angles between 308 and 1308 and j angles
1nif and 1oe1). The second amino acid in the helix is between K508 and 1008. These ranges agree well
a leucine residue that is part of the active site pocket with previous definitions.9 Most of the left-handed
where it also hydrogen bonds to an ordered water helices were only four residues in length, and they
molecule.55 occur as both a-helices and 310-helices. Most of
them could be shown to play an important role
4-Hydroxythreonine-4-phosphate dehydrogenase either for the stability or for the function of
(PdxA) the protein (e.g. determining substrate specificity,
mediating proteinprotein interactions, or binding
PdxA (PDB code 1ptm) is a homodimeric enzyme a cofactor of an enzymatic reaction). Consideration
in the pathway leading to pyridoxal 5 0 -phosphate, of the handedness of helices could therefore help
an essential cofactor for many enzymes in amino pinpoint functionally interesting parts of the pro-
acid metabolism. PdxA contains the longest left- tein. This can become especially important as we
handed helix encountered in this survey, a six- enter the structural genomics era and the need for
residue long 310-helix that is located in the dimer automatic annotation of structures grows, in par-
interface. A histidine residue in the helix (His211), ticular for structures of proteins of unknown
together with two histidine residues from the other function.
monomer, coordinates a Zn2C that is required for Our results also show that secondary structure
catalytic activity of the enzyme (Figure 2(b)).56 After assignment is not a trivial problem and that
the substrate 4-hydroxy-l-threonine-phosphate different answers can be obtained for a given
(HTP) has entered the protein, a hydrogen bond is piece of structure when different programs are
formed between the Og oxygen atom of HTP and used. The same is true for the designation of
the Nd2 atom of His211. Thus, this histidine residue hydrogen bonds, where discrepancies between the
participates in both metal-binding and substrate- results of DSSP, Deep View and HBPLUS were
binding. It is conserved in all 74 known sequences observed (data not shown).
of the PdxA family56. The Uppsala Electron Density Server22 was a
helpful tool to assess which hits were genuine and
which were artefactual left-handed helices. Of the
Nuclease P1
56 putative left-handed helices that were identified,
Nuclease P1 (PDB code 1ak0) is a phospo- 33 occurred in crystal structures and for 19 of these
diesterase that degrades both single-stranded electron density maps were available from EDS.
Left-handed Helices 239

Thus, even low-resolution structures (that are References

otherwise often excluded in the generation of
subsets through strict criteria for resolution or 1. Cintas, P. (2002). Chirality of living systems: a helping
R-value) could be included in the PDB subset. hand from crystals and oligopeptides. Angew. Chem.
Conversely, one of the putative left-handed helices Int. Ed. Engl. 41, 11391145.
was rejected on the basis of its poor electron density, 2. Branden, C.-I. & Tooze, J. (1999). Introduction to Protein
even though the parent structure had been Structure, Garland Publishing, New York.
determined at 1.9 A resolution. 3. Zawadzke, L. E. & Berg, J. M. (1993). The structure of a
centrosymmetric protein crystal. Proteins: Struct.
This study was concerned exclusively with left-
Funct. Genet. 16, 301305.
handed helices (i.e. motifs of at least four consecu- 4. Hung, L. W., Kohmura, M., Ariyoshi, Y. & Kim, S. H.
tive residues that form at least one turn). However, (1998). Structure of an enantiomeric protein, D-
a cursory analysis of motifs that consist of only monellin at 1.8 A resolution. Acta Crystallog. sect. D,
three consecutive residues with left-handed helical 54, 494500.
conformations was also carried out (data not 5. Fodje, M. N. & Al-Karadaghi, S. (2002). Occurrence,
shown). About 600 such motifs were identified in conformational features and amino acid propensities
the 90% sequence-identity subset used throughout for the pi-helix. Protein Eng. 15, 353358.
this study. The large number of hits precluded 6. Herzberg, O. & Moult, J. (1991). Analysis of steric
analysis and validation at the same level of detail strain in the polypeptide backbone of protein models.
and accuracy that was used in the rest of this study. Proteins: Struct. Funct. Genet. 11, 223229.
However, the distribution of torsion angles and 7. Petock, J. M., Torshin, I. Y., Weber, I. T. & Harrison,
amino acid frequencies were examined. The distri- R. W. (2003). Analysis of protein structures reveals
regions of rare backbone conformation at functional
bution of f and j torsion angles turned out to be
sites. Proteins: Struct. Funct. Genet. 53, 872879.
similar to the one observed for the residues in left- 8. Takano, K., Yamagata, Y. & Yutani, K. (2001). Role of
handed helices. The amino acid frequencies are non-glycine residues in left-handed helical confor-
more similar to the frequencies observed for the mation for the conformational stability of human
amino acids in left-handed conformations than lysozyme. Proteins: Struct. Funct. Genet. 44, 233243.
those in the verified left-handed helices. For a 9. Gunasekaran, K., Nagarajaram, H. A., Ramakrishnan,
random subset of 52 of the three-residue motifs, C. & Balaram, P. (1998). Stereochemical punctuation
taken exclusively from X-ray structures, the sec- marks in protein structures: glycine and proline
ondary structure assignment of the motifs and their containing helix stop signals. J. Mol. Biol. 275, 917932.
neighbouring residues according to Promotif25,27 10. Schellman, C. (1980). The aL-conformation at the ends
was inspected. Most of the motifs had been of helice. In Protein Folding (Jaenicke, R., ed.),
classified as 310-helices (33 instances), but turns pp. 5361, Elsevier/New Holland, New York.
were also frequent (14 instances), and five of the 11. Holmes, M. A. & Matthews, B. W. (1982). Structure of
motifs were considered a-helical. A few of these thermolysin refined at 1.6 A resolution. J. Mol. Biol.
160, 623639.
motifs had been annotated in PDBsum as part of an 12. Stark, W., Pauptit, R. A., Wilson, K. S. & Jansonius,
active, metal-binding or ligand-binding site or were J. N. (1992). The structure of neutral protease from
found in close proximity to such a site. Therefore, Bacillus cereus at 0.2-nm resolution. Eur. J. Biochem.
while annotating structures, it would appear to be 207, 781791.
useful to carefully examine any motif with a left- 13. Hill, C. P., Osslund, T. D. & Eisenberg, D. (1993). The
handed conformation or, indeed, any motif with structure of granulocyte-colony-stimulating factor
unusual main-chain conformation.6,7 and its relationship to other growth factors. Proc.
Natl Acad. Sci. USA, 90, 51675171.
14. Lindqvist, Y. (1989). Refined structure of spinach
glycolate oxidase at 2 A resolution. J. Mol. Biol. 209,
151166.
15. Kleywegt, G. J. (1999). Recognition of spatial motifs in
Acknowledgements protein structures. J. Mol. Biol. 285, 18871897.
16. Stamper, G. F., Morollo, A. A., Ringe, D. & Stamper,
The authors thank Professor T. Alwyn Jones C. G. (1998). Reaction of alanine racemase with 1-
(Uppsala University) for a critical reading of aminoethylphosphonic acid forms a stable external
the manuscript, and Terese Bergfors (Uppsala aldimine. Biochemistry, 37, 1043810445.
University) for indispensable linguistic comments 17. Madsen, D. & Kleywegt, G. J. (2002). Interactive motif
and corrections. We also thank an anonymous and fold recognition in protein structures. J. Appl.
referee for several useful suggestions. M.N. is Crystallog. 35, 137139.
supported through grants from the Linnaeus 18. Wang, G. & Dunbrack, R. L., Jr (2003). PISCES: a
protein sequence culling server. Bioinformatics, 19,
Centre for Bioinformatics (Uppsala) and Uppsala
15891591.
University. G.J.K. is a Royal Swedish Academy of 19. Guex, N. & Peitsch, M. C. (1997). SWISS-MODEL and
Sciences (KVA) Research Fellow, supported the Swiss-PdbViewer: an environment for compara-
through a grant from the Knut and Alice Wallen- tive protein modeling. Electrophoresis, 18, 27142723.
berg Foundation. He is further supported by 20. Kabsch, W. & Sander, C. (1983). Dictionary of protein
Uppsala University, the Linnaeus Centre for Bioin- secondary structure: pattern recognition of hydrogen-
formatics, and the Swedish Structural Biology bonded and geometrical features. Biopolymers, 22,
Network (SBNet). 25772637.
240 Left-handed Helices

21. McDonald, I. K. & Thornton, J. M. (1994). Satisfying Residues 88109 of factor IXa are important for
hydrogen bonding potential in proteins. J. Mol. Biol. assembly of the factor X activating complex. J. Biol.
238, 777793. Chem. 277, 57255733.
22. Kleywegt, G. J., Harris, M. R., Zou, J. Y., Taylor, T. C., 38. Chitarra, V., Holm, I., Bentley, G. A., Petres, S. &
Wahlby, A. & Jones, T. A. (2004). The Uppsala Longacre, S. (1999). The crystal structure of C-
electron-density server. Acta Crystallog. sect. D, 60, terminal merozoite surface protein 1 at 1.8 A resolu-
22402249. tion, a highly protective malaria vaccine candidate.
23. Boeckmann, B., Bairoch, A., Apweiler, R., Blatter, Mol. Cell. 3, 457464.
M. C., Estreicher, A., Gasteiger, E. et al. (2003). The 39. Mather, T., Oganessyan, V., Hof, P., Huber, R.,
SWISS-PROT protein knowledgebase and its sup- Foundling, S., Esmon, C. & Bode, W. (1996). The
plement TrEMBL in 2003. Nucl. Acids Res. 31, 365370. 2.8 A crystal structure of Gla-domainless activated
24. Falquet, L., Pagni, M., Bucher, P., Hulo, N., Sigrist, protein C. EMBO J. 15, 68226831.
C. J., Hofmann, K. & Bairoch, A. (2002). The PROSITE 40. Gandrille, S. & Aiach, M. (1995). Identification of
database, its status in 2002. Nucl. Acids Res. 30, mutations in 90 of 121 consecutive symptomatic
235238. French patients with a type I protein C deficiency.
The French INSERM Network on Molecular Abnor-
25. Laskowski, R. A., Hutchinson, E. G., Michie, A. D.,
malities Responsible for Protein C and Protein S
Wallace, A. C., Jones, M. L. & Thornton, J. M. (1997).
deficiencies. Blood, 86, 25982605.
PDBsum: a web-based database of summaries and
41. Zheng, Y. Z., Sakata, T., Matsusue, T., Umeyama, H.,
analyses of all PDB structures. Trends Biochem. Sci. 22,
Kato, H. & Miyata, T. (1994). Six missense mutations
488490. associated with type I and type II protein C deficiency
26. Frishman, D. & Argos, P. (1995). Knowledge-based and implications obtained from molecular modelling.
protein secondary structure assignment. Proteins: Blood Coagul. Fibrinolysis, 5, 687696.
Struct. Funct. Genet. 23, 566579. 42. Ireland, H., Thompson, E. & Lane, D. A. (1996). Gene
27. Hutchinson, E. G. & Thornton, J. M. (1996). PROMO- mutations in 21 unrelated cases of phenotypic
TIF-a program to identify and analyze structural heterozygous protein C deficiency and thrombosis.
motifs in proteins. Protein Sci. 5, 212220. Protein C Study Group. Thromb. Haemost. 76, 867873.
28. Combet, C., Blanchet, C., Geourjon, C. & Deleage, G. 43. Greenwald, I. & Seydoux, G. (1990). Analysis of
(2000). NPS@: network protein sequence analysis. gain-of-function mutations of the lin-12 gene of
Trends Biochem. Sci. 25, 147150. Caenorhabditis elegans. Nature, 346, 197199.
29. Creighton, T. (1993). Structures and Molecular Properties 44. Vardar, D., North, C. L., Sanchez-Irizarry, C., Aster,
(2nd edit.), W.H. Freeman, New York. J. C. & Blacklow, S. C. (2003). Nuclear magnetic
30. Penel, S., Morrison, R. G., Mortishire-Smith, R. J. & resonance structure of a prototype Lin12-Notch
Doig, A. J. (1999). Periodicity in a-helix lengths and repeat module from human Notch1. Biochemistry, 42,
C-capping preferences. J. Mol. Biol. 293, 12111219. 70617067.
31. Watson, J. D. & Milner-White, E. J. (2002). A novel 45. Hallberg, B. M., Henriksson, G., Pettersson, G. &
main-chain anion-binding site in proteins: the nest. A Divne, C. (2002). Crystal structure of the flavoprotein
particular combination of f,j values in successive domain of the extracellular flavocytochrome cello-
residues gives rise to anion-binding sites that occur biose dehydrogenase. J. Mol. Biol. 315, 421434.
commonly and are found often at functionally 46. Farooq, A., Chaturvedi, G., Mujtaba, S., Plotnikova,
important regions. J. Mol. Biol. 315, 171182. O., Zeng, L., Dhalluin, C. et al. (2001). Solution
32. Kabsch, W. & Sander, C. (1984). On the use of structure of ERK2 binding domain of MAPK phos-
sequence homologies to predict protein structure: phatase MKP-3: structural insights into MKP-3
identical pentapeptides can have completely activation by ERK2. Mol. Cell. 7, 387399.
different conformations. Proc. Natl Acad. Sci. USA, 47. Hennig, M., Grimm, B., Contestabile, R., John, R. A. &
81, 10751078. Jansonius, J. N. (1997). Crystal structure of glutamate-
33. Orengo, C. A., Michie, A. D., Jones, S., Jones, D. T., 1-semialdehyde aminomutase: an alpha2-dimeric
Swindells, M. B. & Thornton, J. M. (1997). CATHa vitamin B6-dependent enzyme with asymmetry in
hierarchic classification of protein domain structures. structure and active site reactivity. Proc. Natl Acad. Sci.
USA, 94, 48664871.
Structure, 5, 10931108.
48. Kack, H., Sandmark, J., Gibson, K., Schneider, G. &
34. Mather, T., Oganessyan, V., Hof, P., Huber, R.,
Lindqvist, Y. (1999). Crystal structure of diaminope-
Foundling, S., Esmon, C. & Bode, W. (1996). The
largonic acid synthase: evolutionary relationships
2.8 A crystal structure of Gla-domainless activated
between pyridoxal-5 0 -phosphate-dependent enzymes.
protein C. EMBO J. 15, 68226831. J. Mol. Biol. 291, 857876.
35. Chang, Y. J., Wu, H. L., Hamaguchi, N., Hsu, Y. C. & 49. Storici, P., Capitani, G., De Biase, D., Moser, M., John,
Lin, S. W. (2002). Identification of functionally R. A., Jansonius, J. N. & Schirmer, T. (1999). Crystal
important residues of the epidermal growth factor-2 structure of GABA-aminotransferase, a target
domain of factor IX by alanine-scanning mutagenesis. for antiepileptic drug therapy. Biochemistry, 38,
Residues Asn(89)-Gly(93) are critical for binding 86288634.
factor VIIIa. J. Biol. Chem. 277, 2539325399. 50. Storici, P., Capitani, G., Muller, R., Schirmer, T. &
36. Celie, P. H., Van Stempvoort, G., Fribourg, C., Jansonius, J. N. (1999). Crystal structure of human
Schurgers, L. J., Lenting, P. J. & Mertens, K. (2002). ornithine aminotransferase complexed with the
The connecting segment between both epidermal highly specific and potent inhibitor 5-fluoromethyl-
growth factor-like domains in blood coagulation ornithine. J. Mol. Biol. 285, 297309.
factor IX contributes to stimulation by factor 51. Malashkevich, V. N., Strop, P., Keller, J. W., Jansonius,
VIIIa and its isolated A2 domain. J. Biol. Chem. 277, J. N. & Toney, M. D. (1999). Crystal structures of
2021420220. dialkylglycine decarboxylase inhibitor complexes.
37. Wilkinson, F. H., London, F. S. & Walsh, P. N. (2002). J. Mol. Biol. 294, 193200.
Left-handed Helices 241

52. Ding, Y. H., Javaherian, K., Lo, K. M., Chopra, R., S., Dartevelle, S. et al. (2003). Crystal structure of a Fab
Boehm, T., Lanciotti, J. et al. (1998). Zinc-dependent complex formed with PfMSP1-19, the C-terminal
dimers observed in crystals of human endostatin. fragment of merozoite surface protein 1 from Plasmo-
Proc. Natl Acad. Sci. USA, 95, 1044310448. dium falciparum: a malaria vaccine candidate. J. Mol.
53. Hohenester, E., Sasaki, T., Olsen, B. R. & Timpl, R. Biol. 328, 10911103.
(1998). Crystal structure of the angiogenesis inhibitor 63. Hopfner, K. P., Lang, A., Karcher, A., Sichler, K.,
endostatin at 1.5 A resolution. EMBO J. 17, 16561664. Kopetzki, E., Brandstetter, H. et al. (1999). Coagulation
54. Sasaki, T., Larsson, H., Tisi, D., Claesson-Welsh, L., factor IXa: the relaxed conformation of Tyr99 blocks
Hohenester, E. & Timpl, R. (2000). Endostatins substrate binding. Struct. Fold. Des. 7, 989996.
derived from collagens XV and XVIII differ in 64. Banbula, A., Potempa, J., Travis, J., Fernandez-
structural and binding properties, tissue distribution Catalan, C., Mann, K., Huber, R. et al. (1998).
and anti-angiogenic activity. J. Mol. Biol. 301, Amino-acid sequence and three-dimensional struc-
11791190. ture of the Staphylococcus aureus metalloproteinase at
55. Adman, E. T., Godden, J. W. & Turley, S. (1995). The 1.72 A resolution. Structure, 6, 11851193.
structure of copper-nitrite reductase from Achromo- 65. Boulanger, M. J. & Murphy, M. E. (2001). Alternate
bacter cycloclastes at five pH values, with NO2- bound substrate binding modes to two mutant (D98N and
and with type II copper depleted. J. Biol. Chem. 270, H255N) forms of nitrite reductase from Alcaligenes
2745827474. faecalis S-6: structural model of a transient catalytic
56. Sivaraman, J., Li, Y., Banks, J., Cane, D. E., Matte, A. & intermediate. Biochemistry, 40, 91329141.
Cygler, M. (2003). Crystal structure of Escherichia coli 66. Ellis, M. J., Dodd, F. E., Sawers, G., Eady, R. R. &
PdxA, an enzyme involved in the pyridoxal Hasnain, S. S. (2003). Atomic resolution structures of
phosphate biosynthesis pathway. J. Biol. Chem. 278,
native copper nitrite reductase from Alcaligenes
4368243690.
xylosoxidans and the active site mutant Asp92Glu.
57. Romier, C., Dominguez, R., Lahm, A., Dahl, O. &
J. Mol. Biol. 328, 429438.
Suck, D. (1998). Recognition of single-stranded DNA
67. Abreu, I. A., Lourenco, A. I., Xavier, A. V., LeGall, J.,
by nuclease P1: high resolution crystal structures of
Coelho, A. V., Matias, P. M. et al. (2003). A novel iron
complexes with substrate analogs. Proteins: Struct.
Funct. Genet. 32, 414424. centre in the split-Soret cytochrome c from Desulfovi-
58. DeLano, W. L. (2002). The PyMOL Molecular Graphics brio desulfuricans ATCC 27774. J. Biol. Inorg. Chem. 8,
System, DeLano Scientific, San Carlos, CA, USA. 360370.
59. Nar, H., Bauer, M., Schmid, A., Stassen, J. M., Wienen, 68. Phale, P. S., Philippsen, A., Widmer, C., Phale, V. P.,
W., Priepke, H. W. et al. (2001). Structural basis for Rosenbusch, J. P. & Schirmer, T. (2001). Role of
inhibition promiscuity of dual specific thrombin and charged residues at the OmpF porin channel con-
factor Xa blood coagulation inhibitors. Structure, 9, striction probed by mutagenesis and simulation.
2937. Biochemistry, 40, 63196325.
60. Sichler, K., Banner, D. W., DArcy, A., Hopfner, K. P., 69. Peneff, C., Ferrari, P., Charrier, V., Taburet, Y.,
Huber, R., Bode, W. et al. (2002). Crystal structures of Monnier, C. & Zamboni, V. (2001). Crystal structures
uninhibited factor VIIa link its cofactor and substrate- of two human pyrophosphorylase isoforms in com-
assisted activation to specific interactions. J. Mol. Biol. plexes with UDPGlc(Gal)NAc: role of the alterna-
322, 591603. tively spliced insert in the enzyme oligomeric
61. Garman, S. C., Simcoke, W. N., Stowers, A. W. & assembly and active site architecture. EMBO J. 20,
Garboczi, D. N. (2003). Structure of the C-terminal 61916202.
domains of merozoite surface protein-1 from Plasmo- 70. Pedersen, L. C., Darden, T. A. & Negishi, M. (2002).
dium knowlesi reveals a novel histidine binding site. Crystal structure of beta 1,3-glucuronyltransferase I in
J. Biol. Chem. 278, 72647269. complex with active donor substrate UDP-GlcUA.
62. Pizarro, J. C., Chitarra, V., Verger, D., Holm, I., Petres, J. Biol. Chem. 277, 2186921873.

Edited by J. Thornton

(Received 8 September 2004; received in revised form 17 January 2005; accepted 17 January 2005)