Vous êtes sur la page 1sur 65

Progress in Biophysics & Molecular Biology 77 (2001) 111–175

Review

The architecture of parallel b-helices and related folds


John Jenkinsa,*, Richard Pickersgillb
a
Institute of Food Research, Norwich Research Park, Colney Lane, Norwich NR4 7UA, UK
b
Biological Sciences, Queen Mary, University of London, Mile End Road, London E1 4NS, UK

Abstract

Three-dimensional structures have been determined of a large number of proteins characterized by a


repetitive fold where each of the repeats (coils) supplies a strand to one or more parallel b-sheets. Some of
these proteins form superfamilies of proteins, which have probably arisen by divergent evolution from a
common ancestor. The classical example is the family including four families of pectinases without
obviously related primary sequences, the phage P22 tailspike endorhamnosidase, chrondroitinase B and
possibly pertactin from Bordetella pertusis. These show extensive stacking of similar residues to give
aliphatic, aromatic and polar stacks such as the asparagine ladder. This suggests that coils can be added or
removed by duplication or deletion of the DNA corresponding to one or more coils and explains how
homologous proteins can have different numbers of coils.
This process can also account for the evolution of other families of proteins such as the b-rolls, the
leucine-rich repeat proteins, the hexapeptide repeat family, two separate families of b-helical antifreeze
proteins and the spiral folds. These families need not be related to each other but will share features such as
relative untwisted b-sheets, stacking of similar residues and turns between b-strands of approximately
901often stabilized by hydrogen bonding along the direction of the parallel b-helix.
Repetitive folds present special problems in the comparison of structures but offer attractive targets for
structure prediction. The stacking of similar residues on a flat parallel b-sheet may account for the
formation of amyloid with b-strands at right-angles to the fibril axis from many unrelated peptides. r 2001
Elsevier Science Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
1.1. Scope of the review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
1.2. Nomenclature, definitions and general features of parallel b-helices . . . . . . . . . . . . 116
1.2.1. Parallel b-helix and its b-sheets . . . . . . . . . . . . . . . . . . . . . . . . . . 116

*Corresponding author.
E-mail address: john.jenkins@bbsrc.ac.uk (J. Jenkins).

0079-6107/01/$ - see front matter r 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 7 9 - 6 1 0 7 ( 0 1 ) 0 0 0 1 3 - X
112 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

1.2.2. Coils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116


1.2.3. Stacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
1.2.4. Turns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
1.2.5. Packing of b-sheets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

2. Description of known structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126


2.1. Pectinases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
2.1.1. The extra-cellular pectate lyase family . . . . . . . . . . . . . . . . . . . . . . . 126
2.1.2. Polygalacturonases and rhamnogalacturonase A . . . . . . . . . . . . . . . . . . 130
2.1.3. Pectin methylesterase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
2.1.4. PelL from Erwinia chrysanthemi . . . . . . . . . . . . . . . . . . . . . . . . . . 134
2.1.5. Pectate lyase Pel-15 from Bacillus sp. strain KSM-P15 . . . . . . . . . . . . . . 134
2.2. The P22 phage tailspike endorhamnosidase . . . . . . . . . . . . . . . . . . . . . . . . . 135
2.3. Chrondroltinase B from flavobacterium hepinarum . . . . . . . . . . . . . . . . . . . . . 138
2.4. P69 pertactin from Bordetella pertussis . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
2.5. Glutamate synthase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
2.6. The antifreeze protein from Tenebrio molitor . . . . . . . . . . . . . . . . . . . . . . . 141
2.7. The leucine-rich repeat family . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
2.7.1. Ribonuclease inhibitors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
2.7.2. The GTPase-activating protein Ma1P from Schizosaccharomyces pombe . . . . . 144
2.7.3. Human insulin-like growth factor receptor . . . . . . . . . . . . . . . . . . . . . 144
2.7.4. Human spliceosomal protein U2A0 , Rab geranylgeranyltransferase and the
mRNA export factor TAP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
2.7.5. Internalin B from Listeria monocytogenes . . . . . . . . . . . . . . . . . . . . . 146
2.8. Left-handed parallel b-helix structures . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
2.8.1. Left-handed parallel b-helix structures containing hexapeptide repeats . . . . . . 147
2.8.2. The left-handed parallel b-helix antifreeze protein from spruce budworm . . . . . 150
2.9. Parallel b-rolls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
2.10. Spiral folds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152

3. The prediction and design of parallel b-helix structures . . . . . . . . . . . . . . . . . . . . . 152

4. Are amyloid fibrils related to parallel b-helices? . . . . . . . . . . . . . . . . . . . . . . . . . 156

5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
5.1. Evolutionary relationships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
5.2. Future directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165

1. Introduction

1.1. Scope of the review

In 1993 Yoder et al. (1993a) reported the structures of the first parallel b-helix and Baumann
et al. (1993) that of the first b-roll, which were at first regarded as revolutionary in using parallel
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 113

b-sheet alone to form complex structures by repeating individual coils. Subsequently the group of
proteins built using similar principles, for which Jurnak et al. (1994) used the term coiled folds and
Kobe and Deisenhofer (1995a, b) used solenoid proteins, has greatly extended. However, new
discoveries do not remain shocking for long and structural biology has rapidly absorbed the
lessons of these structures. Purely parallel b-structures are no longer described as unstable and the
parallel b-helix has simply taken its place amongst the globular proteins with all b-folds. This is
justified because the parallel b-helix architecture is simply one of many ways to form a globular
folded protein, which are stabilised by the same types of interactions. Thus we can easily imagine
a parallel b-helical enzyme and one with a different architecture converging to use the same
catalytic mechanism.
However, structures with the parallel b-helix architecture do possess some unusual common
features and the ambiguity of finding a unique solution when aligning these structures does
pose special problems when comparing them. The unusual simplicity of the architecture also
suggests that understanding their evolution or predicting their occurrence may be unusually
easy. This is not unique and may also apply to folds such as the b-propeller fold (Full . and
. op
Jones, 1999).
Parallel b-helix domains appear to fold as a single co-operative unit as do the domains of other
globular proteins. However, there is a sense in which parallel b-helices and other coiled folds are
intermediate between most globular enzymes and structures made up of domains arranged as
beads on a string, where the folding unit is the individual repeating unit. In the parallel b-helices
the unit of folding is the domain while the unit of evolution may have been the individual ‘‘coil’’
(see Section 1.2.2 below).
Parallel b-helices also form a bridge between globular and fibrous proteins. The rather flat
(untwisted) parallel b-sheets and the ‘‘stacks’’ of similar or alternating residues are special and
probably related features of these folds and we may expect to find similarities between the parallel
b-helix architecture and other systems where untwisted b-sheets pack against each other. This has
led to analogies being made between parallel b-helices and some models of amyloid.
Kobe and Kajava (2000) have recently briefly reviewed the whole family of coiled folds or
solenoid proteins and identified 18 solenoid folds (see http://cmm.info.nih.gov/kajeva/solenoidta-
ble.html). The structures containing only a-helices have been recently reviewed by Kobe et al.
(1999) and we will focus on the structures with parallel b-sheets, which are listed in Table 1, and
especially the parallel b-helical proteins. Even this is a long list and is growing rapidly (for
example two different parallel b-helical antifreeze protein families have been published in 2000,
glutamate synthase shows a parallel b-helical domain and the abstract by Rozwarski et al. (1999)
appears to promise a new family). The main focus will be the architecture of these proteins rather
than the details of their function. All these parallel b-sheet containing proteins have some
common architectural features. These may also share a common mechanism of evolution (as
distinct from a common ancestor). They show the more general properties of solenoid proteins
identified by Kobe and Kajava (2000). However, the focus on the parallel b-sheet containing
proteins makes it hard to discuss Kobe and Kajava’s suggestion that evolution may relate
the leucine-rich repeat (LRR) proteins discussed below with the LRR variant family with a
3/10-helix and an a-helix per coil (Peters et al., 1996). We will also include a brief description of
the evidence for proposed amyloid structures, especially on the suggestions of similarity to parallel
b-helices.
114 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

Table 1
Structures with repetitive folds containing b-sheetsa
Name and origin of the protein Short name PDB code (
Resolution in A, Comments
reference

Right-handed parallel b-helix proteins


Pectate lyase PelC from Erwinia PelC 1AIR 2.20 [1]
chrysanthemi 1PLU 2.20 [2] Lu3+ complex
Pectate lyase from Bacillus subtilis Bspel 1BN8 1.80 [3]
2BSP 1.80 R279K
Pectate lyase PelE from Erwinia PelE 2.20 [2,4]
chrysanthemi
Pectin lyase PnlA from Aspergillus niger PnlA 1IDJ 2.40 [5] pH 6.5
1IDK 1.93 [5] pH 8.5
Pectin lyase PnlB from Aspergillus niger PnlB 1QCX 1.70 [6]
Rhamnogalacturone A from Aspergillus RGase A 1RMG 2.00 [7]
aculeatus
Polygalacturonase PehA from Erwinia PehA 1BHE 1.90 [8]
carotivora
Polygalacturonase II from Aspergillus PG II 1CZF 1.68 [9]
niger
Pectin methylesterase from Erwinia PemA 1QJV 2.40 [10]
chrysanthemi
Pectate lyase Pel-15 from Bacillus Pel-15 1EE6 2.30
sp. strain KSM-P15
Salmonella P22 phage tailspike TSP 1TSP 2.00 [11,12]
endorhamnosidase 1TYU 1.80 [13] Complex
1TYV 1.80 [13] Complex
1TYW 1.80 [13] Complex
1CLW 2.0 [14] V331A
1QA1 2.0 [14] V331G
1QA2 2.0 [14] A334V
1QQ1 1.8 [15] E359G
1QRB 2.0 [15] T326F
1QRC 2.5 [15] W391A
Chrondroitinase B from Flavobacterium 1DBG 1.70 [16]
hepinarum 1DBO 1.70 [16] Complex
P69 pertactin from Bordetella pertussis Pertactin 1DAB 2.60 [17]
Glutamate synthase from Azospirillum 1EAO 3.0 [18]
brasilense
Antifreeze protein from Tenebrio molitor TmAFP 1EZG 1.40 [19]

Leucine-rich repeat proteins


Porcine ribonuclease inhibitor RI 2BNH 2.30 [20]
1DFJ 2.50 [21] Ribonuclease
Complex
Human ribonuclease inhibitor HRI 1A4Y 2.00 [22] Angiogenin
complex
Human insulin-like growth factor receptor IGF 1IGR 2.60 [23]
Human spliceosomal protein U2A0 U2A0 1A9N 2.4 [24] U2b00 -U2A0 U2
RNA complex
Rab geranylgeranyltransferase RabGGT 1DCE 2.0 [25]
Human mRNA export factor TAP TAP 1FO1 2.9 [26]
GTPase-activating protein Ma1P from Ma1P or 1YRG 2.66 [27]
Schizosaccharomyces pombe RNA1P
Internalin B from Listeria monocytogenes InlB 1DOB 1.86 [28]
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 115

Table 1 (continued)

Name and origin of the protein Short name PDB code (


Resolution in A, Comments
reference

Left-handed parallel b-helix hexapeptide


repeat proteins
UDP-N-acetylglucosamine LpxA 1LXA 2.60 [29]
acyltransferase from Escherichia coli
Carbonic anhydrase from Cam 1THJ 2.80 [30]
Methanosarcina thermophila
Tetrahydrodipicolinate DapD 1TDT 2.20 [31]
N-succinyltransf erase
2TDT 2.00 [32] Complex
3TDT 2.00 [32] Complex
Xenobiotic acetyltransferase from PaXAT 1XAT 3.20 [33]
Pseudomonas aeruginosa
2.25 [34]
N-acetylglucosa 2.30 [34] Complex
mine 1-phosphate uridyltransferase
from E. coli (truncated)
N-acetylglucosamine 1-phosphate 1HV9 2.1 [35] Complex
uridyltransferase from E. coli
N-acetylglucosamine 1-phosphate 1G95 2.33 [36]
uridyltransferase from 1G97 1.96 [36] Complex
Streptococcus pneumoniae 1HMO 2.3 [37]
1HM8 2.5 [37] Complex
1HM9 1.75 [37] Complex

Left-handed parallel b-helix pentapeptide repeat proteins


Antifreeze protein from spruce budworm sbwAFP 1EWW NMR [38]
b-roll proteins
Alkaline protease from Pseudomonas 1KAP 1.64 [39]
aeroginosa
50 kDa metalloprotease from Serratia 1SAT 1.75 [40]
marcescens

Spiral folds
4-Chlorobenzoyl coenzyme A dehalogenase 1NZY 1.80 [41]
Enoyl-coenzyme A hydratase 1DUB 2.50 [42]
2DUB 2.40 [43] Complex
ATP-dependent Clp protease from ClpP 1TYF 2.20 [44]
Escherichia coli
a
Publications describing new or refined structures are given as [1] above. These are 1. Yoder et al. (1993a); 2. Lietzke
et al. (1994); 3. Pickersgill et al. (1994); 4. Lietzke et al. (1996); 5. Mayans et al. (1997); 6. Vitali et al. (1998); 7. Petersen
et al. (1997); 8. Pickersgill et al. (1998); 9. Van Santen et al. (1999); 10. Jenkins et al. (2001); 11. Steinbacher et al. (1994);
12. Steinbacher et al. (1997); 13. Steinbacher et al. (1996); 14. Baxa et al. (1999); 15. Schuler et al. (2000); 16. Huang et al.
(1999); 17. Emsley et al. (1996); 18. Binda et al. (2000); 19. Liou et al. (2000); 20. Kobe and Deisenhofer (1993); 21.
Kobe and Deisenhofer (1995b); 22. Papageorgiou et al. (1997); 23. Garrett et al. (1998); 24. Price et al. (1998); 25. Zhang
et al., 2000; 26. Liker et al. 2000; 27. Hillig et al. (1999); 28. Marino et al. (1999); 29. Raetz and Roderick (1995); 30.
Kisker et al. (1996); 31. Beaman et al. (1997); 32. Beaman et al. (1998a); 33. Beaman et al. (1998b); 34. Brown et al.
(1999); 35. Olsen and Roderick (2001); 36. Kostrewa et al. (2001); 37. Sulzenbacher et al. (2001); 38. Graether et al.
(2000); 39. Baumann et al. (1993); 40. Baumann (1994); 41. Benning et al. (1996); 42. Engel et al. (1996); 43. Engel et al.
(1998); 44. Wang et al. (1997).
116 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

1.2. Nomenclature, definitions and general features of parallel b-helices

1.2.1. Parallel b-helix and its b-sheets


The term ‘‘parallel b-helix’’ was introduced by Yoder et al. (1993a) when reporting the structure
of the pectate Lyase PelC, the first protein structure displaying this fold. It is important to note
that the ‘‘parallel b-helix’’ is not related to the term ‘‘b-helix’’, used earlier to describe the
structure of gramicidin (Wallace and Ravikumar, 1988; Langs, 1988). The publication of the first
left-handed structure of UDP-N-acetylglucosamine acyltransferase by Raetz and Roderick (1995)
required that ‘‘parallel b-helix’’ be further extended to either ‘‘right-handed parallel b-helix’’ or
‘‘left-handed parallel b-helix’’. The nomenclature generally used to describe the basic architecture
of the parallel b-helical proteins was essentially fixed by Yoder et al. (1993b), who introduced the
nomenclature PB1, PB2 and PB3 for the three parallel b-sheets of pectate lyase and T1, T2 and T3
(turn) for the regions following the b-sheets. This was revised to include PB1a and implicitly T1a
by Petersen et al. (1997) when the structure of rhamnogalacturonase A from Aspergillus aculeatus
clearly showed a fourth b-sheet. Figs. 1 and 2 illustrate this convention. An alternative
nomenclature was proposed for a fragment of the P22 phage tailspike protein by Steinbacher et al.
(1994) in which the PB1 is equivalent to b-sheet C, PB2 equivalent to A and PB3 equivalent to B.
Van Santen et al. (1999) in describing polygalacturonase II from Aspergillus niger use PB2a and
PB2b as an alternative nomenclature corresponding to PB1a and PB2 used for RGase A and
PehA.

1.2.2. Coils
As ‘‘turn’’ was used by Yoder et al. (1993b) to describe the regions between the b-stands by
analogy with b-turns in hairpin loops, there is a need for an unambiguous term for a complete
turn of the parallel b-helix and the term ‘‘coil’’ is used in this review. However, there remains the
difficult problem of whether to count the coils using topological arguments or to accept only coils
with the regular three-stranded parallel b-helical structure. This ambiguity results in great
confusion on how many coils occur in known structures and in how to number them. A good
example is provided by the family of hydrolases containing RGase A, PehA and PG II because the
reasoning for the different choices between 10 and 13 coils can be explicitly described. RGase A
might be said to have 12 coils starting with the first PB2 (residues 20–22) and finishing with the

—————————————————————————————————————————————————c
Fig. 1. Ribbon diagrams and views of the cross-section of single coils running clockwise for various types of repetitive
fold. The atoms of the cross-section are coloured by type: black for carbon, blue for nitrogen, red for oxygen and yellow
for sulphur. (a) Part of the b-roll of the alkaline protease from Pseudomonas aeroginosa (1 kap) with the tandem
repeated sequence GGXGXDXLX giving rise to the b-roll architecture. The two b-sheets are coloured red and green.
Calcium ions are shown as yellow balls. The single coil runs clockwise starting with the sequence Ile–Leu–Tyr. (b)
Bacillus subtilis pectate lyase (1bn8) with the N-terminal end of the parallel b-helix to the left and the T3 region above.
The three b-sheets PB1 (Thr–Asp–Ala), PB2 (Ile–Thr–Met) and PB3 (Tyr–Tyr–His) are coloured yellow, green and red.
An asparagine of the asparagine ladder can be seen at T2 (just before PB3) and a residue of the aromatic stack can be
seen on PB3. (c) Porcine ribonuclease inhibitor with the b-sheets in red and the helices in green. In the ribbon diagram
the chain direction or direction of the superhelix is anti-clockwise. However, the single coil runs clockwise and shows
the leucines from the a-helix and b-sheet packing in the hydrophobic core. (d) The left-handed UDP-N-
acetylglucosamine acyltransferase from Escherichia coli with the N-terminal end of the parallel b-helix to the right.
The coil shows the isoleucines that form its hydrophobic core.
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 117
118 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

Fig. 2. Ribbon diagrams of some of the longer right-handed parallel b-helices. The three common b-sheets PB1, PB2
and PB3 are coloured yellow, green and red, while the extra sheet PB1a found in the polygalacturonase family is
coloured blue. The N-terminal end of the right-handed parallel b-helices are to the right. (a) The Salmonella P22 phage
tailspike endorhamnosidase. (b) P69 pertactin from Bordetella pertussis showing the longest known parallel b-helix. (c)
The polygalacturonase PehA from Erwinia carotavora showing the PB1a sheet within the T1 region.

last PB1a (residues 353–356) as shown in Table 1 of Petersen et al. (1997), which lists 12 PB2
stands. Petersen et al. suggest that there are 13 coils because the main chain forms a very tight coil
with residues 354 and 356 forming hydrogen bonds to the previous PB2 at residues 331–333 (i.e.
the last PB1a acts as both a PB1a and a PB2), forms a disulphide to Cys 350 in the previous PB1
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 119

from Cys 359 and then reverses direction to give a last PB1 strand (residues 369–375), so that the
topological winding number is 13. However, the strand of residues 369–375 is anti-parallel to the
previous PB1 making the nomenclature PB1 rather confusing, so that we might reject the residues
after 356 as not being part of the parallel b-helix. We might also reject the first coil because it
contains a single b-strand (PB2) and an a-helix by arguing that this is not a parallel b-helix,
leaving us with 11 coils. Finally, using the original definition of Yoder et al. (1993b) and requiring
that each coil starts with PB1 and ends with PB3 would leave 10 coils, which is how the very
similar structure of PehA is described by Pickersgill et al. (1998). PehA forms the same final PB1a
to PB2 hydrogen bonds but does not form the disulphide or the anti-parallel structure. PGII does
not form the PB1a to PB2 hydrogen bonds as it forms a rather different structure with proline 357
in approximately a–R conformation at the end of PB1a (or PB2a), terminates by forming the
disulphide without any anti-parallel extension and is also described as having 10 complete turns
(van Santen et al., 1999). What is more significant than the number of coils used to describe the
structure is that these hydolases have two complete coils and a PB1 and PB1a more than the
pectate lyases or the pectin methylesterases and three fewer coils than the P22 phage tailspike
endorhamnosidase. It is important to check the number of coils visually rather than to accept
statements in the literature since an unexpected definition may have been used.

1.2.3. Stacks
Yoder et al. (1993b) first discussed the ‘‘stacking’’ of similar residues at the equivalent positions
in neighbouring coils and this was extended by Raetz and Roderick (1995), who revived the term
‘‘cupped stacks’’ for the most common stacking of valines and isoleucines. Petersen et al. (1997)
introduced the useful distinction between ‘‘stacked’’ and ‘‘aligned’’ residues. ‘‘Stacked’’ residues
had to be similar residues with similar w angles while ‘‘aligned’’ described the same or chemically
similar residues packing with different w angles (especially w1 ). The distinction is between regular
and irregular structures but ‘‘stacks’’ that mix valines and isoleucines with leucines or methionines
with similar w angles do occur. The stacks of valines and isoleucines packing on each other tend to
be more regular. There are at least three different types of stacks found in the parallel b-helix
proteins as illustrated in Figs. 3 and 4. Aliphatic stacks such as the ‘‘cupped stacks’’ are the most
common and are found in every protein, aromatic stacks are formed by the offset face to face
packing of phenylalanine and tyrosine side chains (Hunter et al., 1991) and polar stacks include
the well-known asparagine ladder (Yoder et al., 1993b), several variations of which are shown in
Figs. 4 and 5.
‘‘Aligned’’ residues are ubiquitous in b-sheet containing proteins and occur in both parallel and
anti-parallel sheets. ‘‘Cupped stacks’’ of valines and isoleucines were discussed by Richardson
(1981) long before the parallel b-helix was observed, who noted that the type of packing seen in
the ‘‘stacks’’ will not be stable in anti-parallel b-sheet because it would demand that the w1 angles
differ by 1801. However, such stacking is rather rare except in the repetitive folds. In a highly
twisted parallel b-sheet, stacking will be restricted to a single ridge if it occurs, which may explain
its rarity in TIM barrels. An example of a single ridge is seen in glyceraldehyde-3-phosphate
dehydrogenase from Sulfolobus solfataricus, PDB code 1B7G (Isupov et al., 1999) where residues
Ile 50, Val 31, Val 6, Val 82, Ile 105, Ile 134 and Ile 117 form a ridge and Val 6 to Ile 134 clearly
form a cupped stack. By contrast, the relationship between stacking and the repetitive folds is one
of the most basic features of their architecture so that no parallel b-helix has been observed
120 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

Fig. 3. Stacking of non-polar residues in parallel b-helices. Carbons are drawn in light grey with oxygens and nitrogens
in dark grey. (a) Aliphatic ‘‘cupped’’ stacks of isoleucines 212, 214, 240 and 242 and valines 265 and 267 on PB2 of the
polygalacturonase PehA. (b) The internal aromatic stack of residues phenylalanine 159 and 201 and the tyrosines 242,
273 and 295 on PB3 of Bacillus subtilis pectate lyase. The tyrosine oxygens form hydrogen bonds with threonine 226,
tryptophan 310 and two buried waters inside the parallel b-helix.

without stacking. However, the source of the relationship is less obvious. The average spacing
( which means that the distances between the equivalent atoms are
between coils is about 4.8 A
slightly longer than would give ideal packing of valines or isoleucines. However, in general each
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 121

Fig. 4. Stereo views of T2 turns and asparagine ladders viewed from within parallel b-helices. Carbons are drawn in
light grey with oxygens and nitrogens in dark grey with some residue numbers indicted to the right of the a-carbons.
Hydrogen bonds of the central coils main chain and asparagine side chain are indicated by thin lines. The hydrogen
bonding of PB2 (above) and PB3 (below) continues regularly. (a) Three coils of BsPel near T2 showing the first three
asparagines of the internal asparagine ladder. (b) Three coils of the T2 turn from RGase A showing continuous (if not
regular) hydrogen bonding from PB2 to PB3.

sheet packs against at least one other so that small residues on, for example, PB1 and PB2 can
partially intercalate (see Section 1.2.5).
Although in general anti-parallel b-sheets do not allow stacking, an alternating stacking of
glutamines across an anti-parallel b-sheet, called polar zippers, has been proposed by Perutz et al.
(1994) as a model for the interactions leading to aggregation between glutamine rich proteins.
Such aggregates are involved in several neurodegenerative diseases such as Huntingdon’s disease.
Only the w3 angles need to be moved from optimal values to give good hydrogen bonding and
there is good packing of the Cb and Cg atoms and thus some hydrophobic stabilisation of the b-
sheet. Both asparagines and glutamines can form hydrogen bonds across an anti-parallel b-sheet
but asparagines in a parallel sheet can also form hydrogen bonds from Od1 to the NH of the other
asparagine, as well as the extra hydrogen bond of the T2 type turn (see Figs. 4 and 5). By contrast,
no glutamine stacks have been reported in the experimental structures of parallel b-helical
proteins. Thus, a high ratio of asparagine to glutamine in a b-sheet protein may be suggestive of
untwisted parallel b-structure and of the turns described below.
122 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

Fig. 5. Asparagine ladders and T2 turns in leucine-rich repeat proteins. Carbons are drawn in light grey with oxygens
and nitrogens in dark grey with some residue numbers indicted to the right of the a-carbons or carbonyl carbons.
Hydrogen bonds from the main chain of the central coil and the central asparagine are indicated by thin lines. The
hydrogen bonding of the main sheet (PB2) above is regular but the residues below form less regular hydrogen bonds. (a)
Stereo view of the asparagine ladder of insulin-like growth factor receptor domain 3. (b) Stereo view of the asparagine
ladder of U2A0 . Only three of the four asparagines are shown.

1.2.4. Turns
The b-sheets of the right-handed parallel b-helical proteins often change direction sharply with
a single residue in aL -conformation interrupting the residues in b-conformation. To avoid
confusion it is important to distinguish between the angle through which the sheet is turned from
its extended direction and the interior angle (taking a single coil as a polygon). The chain turns 801
at the T2 turn between PB2 and PB3 of the pectate lyases, giving an interior angle of 1001. By
contrast, in a typical b-hairpin turn the chain changes direction by 1801. Many aL turns occur in
the right-handed parallel b-helix proteins but can be subdivided into at least two forms. The most
common form is shown in Fig. 4a. This was the form initially identified in the structure of PelC by
Yoder et al. (1993b). This turn involves two residues not forming all the possible main chain
hydrogen bonds to the neighbouring coils. However, the simplest arrangement is that illustrated
in Fig. 4b which occurs at the start of PB2 in BsPel and extensively in RGase A and the
polygalacturonases. The T2 turn of TSP also shows this regular hydrogen bonding towards the
carboxy-terminus of the parallel b-helix. This type of hydrogen bonding was identified as possible
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 123

by Chothia and Murzin (1993). The turn shown in Fig. 4a can be derived from the simpler
Chothia and Murzin turn by rotating the amides of the residue after the a–L residue (denoted
therefore as residue i þ 1) inwards. At T2 in TSP at the amino terminal end of the parallel b-helix
these amides form hydrogen bonds with buried water molecules although the NH of 352 can
hydrogen bond to Og of Ser 384. Similar structures also occur at T1 in TSP. In pertactin Ser 121
and Ser 142, and in pectin methylesterase Ser 249 and Thr 294, can hydrogen bond to the inward
pointing NH from the previous coil. In the lyases a much more elaborate structure has evolved,
the asparagine ladder (Fig. 4a) in which the main chain NH of residue i þ 1 hydrogen bonds to the
carbonyl of the asparagine side chain and the NH2 of the asparagine side chain hydrogen bonds to
the main chain carbonyl of the residue i  2: Ribonuclease inhibitor shows a Cys–Asn ladder
related to the asparagine ladder (Kobe and Deisenhofer, 1995b) but with the difference that the
asparagines (and thus also the cysteines) only hydrogen bond to the main chain atoms. These have
a peptide rotated away from b-conformation so that the i  2 carbonyls can bond to the
asparagine NH2 group rather than with the main chain from neighbouring coils. The human
spliceosomal protein U2A has ladders with only asparagines but follows ribonuclease inhibitor in
that the asparagine NH2 groups now bridge between each pair of i  2 carbonyls, which are again
directed inwards (Fig. 5b). However, the insulin-like growth factor receptor (Garrett et al., 1998)
shows asparagine ladders resembling those of the pectate lyases with main chain hydrogen bonds
along the ‘‘helix’’ axis.
The left-handed parallel b-helices also have aL -containing turns but their coils are much closer
in shape to an equilateral triangle than are those of the righthanded parallel b-helices. Thus they
have turns that give approximately 1201 changes of direction and 601 interior angles. These turns
often form the 1–4 hydrogen bond of the classical b-turn but are clearly different from the latter as
the chains subsequently diverge. Interestingly, a similar turn also occurs in pertactin and Figs. 6a
and b show that these turns are locally similar in parallel b-helices with the opposite hand.
The turns seen in pectate lyase were termed a distorted gbE turn by Yoder et al. (1993b) after the
classification of Wilmot and Thornton (1990). Chothia and Murzin (1993) viewed the turn as a
kink in a single continuous b-sheet, similar to the b-bulge (Richardson et al., 1978; Chan et al.,
1993), whilst Pickersgill et al. (1994) identify the occurrence of aL -bounded b-strands as a new
motif. The turns in the left-handed parallel b-helices have generated less controversy, possibly
because they were assumed to resemble those of the right-handed ‘‘family’’. A Wilmot and
Thornton definition for them might be bP gL ; which is the classical type II turn of Richardson
(1981) rather than bE gL : However, this turn also involves four residues rather than two outside the
normal b-sheet conformation and only one of the three peptides between them can form hydrogen
bonds along the axis of the parallel b-helix.
So far turns involving an external aR residue have only been found in pertactin (Fig. 7a) but
turns in which the aR residue is inside the parallel b-helix (and the surface is therefore concave)
occur in all the right-handed parallel b-helix enzymes at the start of PB1 (Fig. 7b). These can also
form hydrogen bonds along the axis of the parallel b-helix. The groove formed by this turn is
normally part of the active site of these enzymes.

1.2.5. Packing of b-sheets


The packing of two b-sheets in a sandwich is a common feature of many protein folds and has
been extensively studied Chothia et al. (1997). There is a correlation between the optimal relative
124 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

Fig. 6. Stereo views of LpxA (UDP-N-acetylglucosamine acyltransferase from E. coli) together with P69 pertactin after
superposing all the main chain atoms of residues 116–123 of LpxA on residues 352–359 of pertactin with an RMSD of
( The side chains pointing out of the parallel b-helix towards the viewer are reduced to alanines for clarity. (a)
0.78 A.
Stereo view of the bonds of LpxA drawn as cylinders. Carbons are drawn in light grey with oxygens and nitrogens in
darker grey. Some LpxA residues are numbered. (b) Stereo view of the bonds of LpxA around residues 121 drawn as
thick grey lines together with the bonds of pertactin drawn as thinner black lines. Some pertactin residues are
numbered. (c) Stereo view of the bonds of pertactin drawn as cylinders. Carbons are drawn in light grey with oxygens
and nitrogens in darker grey. Some pertactin residues are numbered.

orientation of the b-sheets and their twist, with low twist allowing the sheets to pack with their
strands (anti)parallel. Most b-sandwich proteins have more twisted sheets and the strands are
rotated by an angle of about 301, or pack approximately orthogonally. In the right-handed
parallel b-helices, PB1 and PB2 form such a sandwich in RGase A and the polygalacturonases.
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 125

Fig. 7. Stereo views of right-handed a-helical residues forming turns in the right-handed parallel b-helix family.
Carbons are drawn in light grey with oxygens, nitrogens and a sulphur in darker grey. (a) The turn from T3 into PB1 at
the active site of the polygalacturonase PehA viewed from along PB1. The active site residues Asp 202 and Asp 224 are
shown but Asp 205 is omitted. (b) Part of the stack of a-helical residues at T3 near the C-terminal end of the pertactin
parallel b-helix. Hydrogen bonds of the main chain are indicated by thin lines. The hydrogen bonding is not completely
regular with two protons being donated to the carbonyl of Asn 442 (note that the residue numbers are to the left of the
residues in this image only).

The helix becomes more triangular in shape in the other proteins and especially in pertactin and
pectin methylesterase, while the left-handed parallel b-helices are close to an equilateral triangle in
cross-section. In RGase A and the polygalacturonases, PB1 and PB2 are rather flat and pack
almost anti-parallel. When viewed down the helix axis, the g-carbons from PB1 and PB2 are
nearly aligned so that they approach each other rather than packing as knobs into holes.
However, this sandwich is unusual in being formed from a purely parallel sheet. The g-carbons are
slightly off the perpendicular to the helix axis with those of PB1 directed slightly towards the
C-terminus while those of PB2 are directed towards the N-terminus. This allows opposing
residues to avoid steric clashes between the sheets by simply having w1 near 1801 which also avoids
clashes within the sheet. The g-carbons from both sheets are generally directed towards the N-
terminus. However, there is no steric clash because the b-carbons from PB1 of one coil slot
between the g-carbons from PB2 of the same coil and the equivalent residue in the next coil. Thus
in this case the local preference for w1 ; ‘‘cupped stacking’’, the rather flat b-sheets, and the optimal
126 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

anti-parallel packing combine to stabilise the right-handed parallel b-helix architecture. However,
the other, more triangular structures are also stabilised by the packing of stacked residues from
two or three b-sheets. The b-roll protease structures (Section 2.9) all show almost exact anti-
parallel packing but the sheets pack unusually close together so that the residues must
interdigitate unless one is a glycine. Thus the pleating in these b-sheets is anti-correlated between
the sheets. In the thicker b-roll like section of pertactin (Section 2.4) the packing involves a
tryptophan from each coil packing against two leucines from the other sheet. The right-handed
antifreeze protein from Tenebrid molitor has similar spacing to the b-roll structures with cysteines
on opposite sides forming disulphide bridges but only two adjacent sides of the helix are b-sheet
and thus only one of the opposed cysteines comes from a b-sheet.

2. Description of known structures

2.1. Pectinases

2.1.1. The extra-cellular pectate lyase family


The family contains most known pectate lyases, including all those with published three-
dimensional structures starting with the archetype PelC (Yoder et al., 1993a), and all known
pectin lyases (Henrissat et al., 1995). Henrissat and Coutinho have developed a web site, http://
afmb.cnrsmrs.fr/Bpedro/CAZY/lya.html, displaying lists of polysaccharide lyases classified into
homologous families where this family is listed as family 1.
The X-ray structures of five members of this family have been published (see Table 1) and
crystals of one more have been reported (Doan et al., 2000). The known structures comprise three
rather distantly related pectate lyases, PelC and PelE from Erwinia chrysanthemi and the Bacillus
Subtilis pectate lyase, and the two pectin lyases A and B from Aspergillus niger, PnlA and PnlB,
which are more closely related with more than 60% sequence identity. The overall fold of these
structures is shown in Fig. 8. There is a structurally conserved core comprising the parallel b-helix,
the N-terminal helix and the N- and C-terminal extensions. The shape of the parallel b-helix is
formed by three b-sheets and is very similar in all five enzymes. It is often described as L shaped,
which arises because strands PB2 and PB3 are relatively long and make a slightly obtuse interior
angle at T2 (about 1001) and PB1 is often preceded by a residue in (a–R conformation and then
runs roughly antiparallel to PB2 (Figs. 1 and 7b).
All five structures are similar in the general pattern of long and short T1 and T3 loops
especially in having long T3 loops in the coils to the N-terminal end of the parallel b-helix.
These T3 loops are sometimes long enough to bury some hydrophobic residues as well as forming
hydrogen bonds. Thus this region sometimes appears to form a non-contiguous ‘‘domain’’.
However, there is no information to suggest that any of these structures can fold independently
of the parallel b-helix. It is also worth noting that there is little similarity in the detailed structures
of the T3 and T1 loops between these enzymes except between the two pectin lyases. For example,
although each of the pectate lyases has a helix in its first T3 loop, this coil in PelC does not
have a PB3 but after a few residues in irregular conformation starts an a-helix which packs
against PB3 of the next two coils, while PelE and BsPel have a PB3 and thus leave the coil
at a different angle. The T3 loop is much shorter in PelE (residues 58–74) than in BsPel
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 127

Fig. 8. Comparison of the pectinase folds, showing that even the obviously homologous lyases have very different loop
regions. Bacillus subtilis pectate lyases with a single bound calcium at (b), Aspergillus niger pectin lyase A at (d), Erwinia
chrysanthemi pectin methylesterase, PemA, at (a) and the Erwinia carotivora polygalacturonase PehA at (c) are shown.
The three common b-sheets PB1, PB2 and PB3 are coloured yellow, green and red, while the extra sheet PB1a of the
polygalacturonase family is dark blue. The b-hairpins in the T1 region of pectin methylesterase are cyan. All the active
sites are towards the viewer and the N-terminal ends of the parallel b-helices are at the top.

(residues 64–121) where the loop continues out through irregular structure and two turns of a-
helix (residues 85–91) before returning as a long helix (residues 104–121) which overlaps with a
shorter helix in PelE for the last 3 turns (residues 67–74). The pectin lyases have long loops which
roughly but not exactly overlap the loop of BsPel but consist of irregular structure and a long b-
hairpin.
The T1 regions include both extended irregular structure and shorter chains in mostly extended
conformation resembling short b-strands as part of the parallel b-helix, which, however, do not
form all the hydrogen bonds needed for the region to be classified as a b-sheet. In fact the lyases
128 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

are intermediate between the esterases where this region is definitely not a b-sheet and the
hydrolyases where it forms PB1a as shown in Fig. 2.
Jenkins et al. (1998) compared the lyases except for PnlB and determined that PelE and BsPel
have 31–34% identical residues after structural alignment (depending on whether alignment was
automatic or assisted by human intervention), while the other pairs had between 13.6% and 17%
identical residues after structural alignment (PnlA was most different from the pectate lyases).
This low level of sequence identity allows us to have complete confidence that these are
homologues but makes alignment difficult without knowledge of the structure. For example,
Henrissat et al. (1995) used the three pectate lyase structures to align all the then known sequences
but chose not to align several regions of the fungal pectin lyase sequences, including one active site
T3 region, because there was insufficient similarity. Examination of the aligned sequences suggests
that there may also be a bacterial pectin lyase family, which has diverged as far from the pectate
lyase and the fungal pectin lyases as these have from each other. Pissavin et al. (1998) recently
convincingly suggested that PelZ from Erwinia chrysanthemi is another even more remote
homologue.
The structurally aligned primary structures show rather few identical residues overall but there
are two clusters of conserved residues. One represents the active site as expected but the second,
including the sequence W(I/V)DH, is on the opposite side of the parallel b-helix from the active
site near T2 of the coils bearing the active site residues. This second cluster does not seem to be
associated with enzymic activity but may be critical for folding or stability. This region is at least
partially buried by the N-terminal extension and the C-terminus is also nearby. The X-ray
structures tend to show low temperature factors for that region and Jurnak et al. (1996) found
that mutation in that region impaired folding. The conservation of some other residues, such as
the asparagine ladder, shown in Fig. 4, is more simply explained. Some other conserved residues
are involved in the folding of the N-terminal extension, while one asparagine forms a hydrogen
bond to the main chain of the C-terminal region. Kamen et al. (2000) have studied the
denaturation and renaturation of PelC and find that the protein may unfold in two structural
blocks, but these could not be clearly identified in the structure. The unfolding transition is highly
co-operative and the protein was described as unusually stable.
The active sites of the pectate and pectin lyases are very different with mostly small hydrophilic
side chains forming the pectate lyase active site and many large aromatic side chains dominating
that of pectin lyase. As expected pectate lyase has a positive potential while pectin lyase is
negatively charged. Calcium is essential for pectate lyase activity but its precise role has been
poorly understood until very recently. The structure of the calcium complex of Bspel (Pickersgill
et al., 1994) showed that there was a single high affinity site and its affinity could also be measured
calorimetrically as 0.2 mM. However, the kinetic constant Km for calcium was much higher (near
1 mM) suggesting that calcium had a more complex role (Smith, D., unpublished). Scavetta et al.
(1999) overcame many difficulties to determine the structure of a complex of the R218 K mutant
of PelC with calcium and the substrate pentagalacturonic acid. The surprising result was that four
calcium sites were found including one equivalent to that seen in BsPel. All the calcium atoms had
at least one atom from both the protein and from the substrate as ligands and thus formed
bridges. Only four galacturonates were observed implying either that the substrate had been
partially degraded or the extra site was either blocked or energetically unfavourable for binding.
The penta-(or tetra-)galacturonate bound almost parallel to the parallel b-helix along the grove in
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 129

PB1 interacting with ligands from T3, PB1 and T1. Not all the calcium ligands were conserved
even between the three known structures of the pectate lyases but it is possible, for example, that
PelC’s Asp 160 and Asp 162 might be replaced by PelE’s Glu 124 or BsPel’s Asp 170 which come
from different T3 loops to a similar position. Thus, it is likely that Scavetta et al. have solved the
basic problem of how the extra-cellular pectate lyases bind their substrates although it is probable
that a longer substrate would bind at extra sites and possibly use extra bridging calciums. Some
differences in the substrate binding between the different pectate lyases are also suggested by the
alterations in the side-chains binding the calciums. The structure of the complex immediately
accounted for most of the available data accumulated by site directed mutagenesis. Kita et al.
(1996) had shown that mutation of Asp 131, Glu 166, Asp 170, Lys 190, Arg 218 and Arg 223 led
to reduced activity. Except for the already mutated Arg 218, all of these are ligands of either
calcium or the substrate and are generally conserved except that Glu 166 is an aspartate in most
pectate lyases.
As there was no potential base near any of the C5 atoms of the substrate, Scavetta et al.
proposed a mechanism assuming that the sites occupied in the complex were 3 to +1 and that
Arg 218 would have the same position in the productive complex as it had in the free enzyme. This
would place the arginine’s NH2 2.6 A ( from the C5 in an ideal position to act as the base removing
a proton from C5 with the b-elimination at O4 occurring in either a stepwise or a concerted
reaction. This explains the complete conservation of the arginine in the family and the dramatic
loss of activity on its mutation but has not won universal acceptance (Huang et al., 1999). The use
of an uncharged arginine as a base might be plausible if the pH optimum for activity were high
and the arginine was close to one or more calciums. However, only a very small fraction of Arg
236 of pectin lyase A would be uncharged at its pH optimum near 5.5 and no calcium ions are
required for pectin lyase activity. The review by Herron et al. (2000) gives a detailed view of the
mechanism and its relation to pathogenesis by Erwinia.
The pectin lyases have a ‘‘simpler’’ active site than the pectate lyases, consistent with having a
simpler task, as pectin will easily undergo non-enzymatic b-elimination. They have no known
calcium binding, no lysine homologous with Lys 190 of PelC and only one of the three
carboxylates of the high-affinity calcium site of pectate lyase site is conserved. This carboxylate is
the one furthest from the substrate and appears to function to orient the arginine that has
replaced Glu 166 of PelC. The pectin lyases A and B of Aspergillus do present a new problem in
that activity is lost above pH 7. This may be associated with the two different structures
reported by Mayans et al. (1997) at pH 6.5 and 8.5 of PnlA from two different strains of
Aspergillus niger. At pH 6.5, Asp 186 and Asp 221 are buried within the parallel b-helix and
form a hydrogen bond while at pH 8.5, Asp 186 has turned outwards and been replaced
by Thr 183 with a significant rearrangement of the T1 region. This conformational
change is probably associated with pH rather than with the small number of amino acid
substitutions between the strains, differences in glycosylation or altered crystal packing.
PnlB at pH 5.5 and high salt concentration (Vitali et al., 1998) has a very similar conforma-
tion to the pH 6.5 structure of PnlA, crystallised from polyethylene glycol. However, it is
not clear how the conformational change is related to the loss of activity or its biological
function. Jaap Visser (pers. comm.) has suggested that a pectin lyase is unnecessary above pH 7
when pectin is labile and that PnlA may change conformation to accelerate its recycling by
proteases.
130 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

2.1.2. Polygalacturonases and rhamnogalacturonase A


The polygalacturonases and two rhamnogalacturonases were identified as family 28 of the
glycosyl hydrolyases (Henrissat and Bairoch, 1996) and have been shown to act with inversion of
anomeric configuration (Biely et al., 1996; Pitson et al., 1998). Three crystal structures of
rhamnogalacturonase A from Aspergillus aculeatus (Petersen et al., 1997), the endo-polygalactur-
onase PehA from Erwinia carotivora (Pickersgill et al., 1998) and endopolygalacturonase II or PG
II from Aspergillus niger (van Santen et al., 1999) have been determined and the crystallisation of
several further enzymes has been reported (Yoder and Schell, 1995; Federici et al., 1999; Lu et al.,
2000). Although these three enzymes are clearly homologous, they resemble the lyases in that they
have diverged beyond the level of sequence identity where accurate models can be constructed. All
were solved by using heavy atom derivatives rather than molecular replacement (for example, van
Santen et al. (1999) report that superposition of PG II and PehA gives an rms deviation (RMSD)
of 1.8 A( for 265 Ca atoms and 19% sequence identity). Despite the change of function, RGase A
is slightly more closely related to PG II than the two polygalacturonases are to each other and it
may thus have diverged from PG II after the fungi diverged from bacteria. In fact PehA is also
very slightly closer to RGase A than to PG II and superposition by O (Jones et al., 1991) gives
1.6 A( for 280 Ca atoms between RGase A and PG II, 1.67 A ( for 253 Ca atoms between PehA and
PG II, and 1.67 A ( for 280 Ca atoms between PehA and RGase A (using chain B for PG II).
RGase A and PehA have N-terminal extensions which both start with the amino terminus
forming hydrogen bonds to a T1a turn residue in a–L conformation. In PehA this residue is Asn
264 and the OD1 also bonds to the N-terminus while in RGase A the N-terminal residue is an
asparagine and forms an internal hydrogen bond so that only the carbonyl of Asp 235 bonds to
the amino terminus. Ser 3 also makes a conserved hydrogen bond to an a–L residues at T1a (Asn
185 of RGase A or Asn 211 of PehA). The nearby residues lie in the same region with Leu 2 and
Val 6 of RGase A packing in approximately the same position as Asp 2 and Arg 4 of PehA but
forming very different interactions. The chains then diverge with Glu 8 of PehA forming a salt
bridge to another a–L residue, Lys 189, which is at T2. At the equivalent position at T2 RGase A
has Asn 165 which forms a hydrogen bond to the C-terminal extension of RGase A. RGase A
forms a short a-helix and Cys 15 of PehA and Thr 20 of RGase A come together again near the
second residue of PG II, Ser 29 (in fact Asp 28 of PG II is close to Lys 19 of RGase A) and enter
the parallel b-helix.
The parallel b-helix of PehA ends abruptly with the final strand of PB1a, while that of PG II
loops back to form a C-terminal disulphide bridge which may cap the parallel b-helix. RGase A
has a long C-terminal extension which forms a strand of anti-parallel b-sheet with the last strand
of PB1 (residues 369–375) and then packs against PB3 burying several aromatic residues such as
the externally stacked pair of Tyr 215 and 242. It also partially buries the longer stack of Phe 104,
His 141, Tyr 164 and His 189 from the end of PB2 by forming a S bend (residues 386–398–412–
421) which lies against the PB3-T3 region. Thus the C-terminal extension like the N-terminal
extension interacts with a stack of a–L turns (at T2 in this case). The main chain interactions with
the Asp 142, Asn 165 and Asp 190 may be especially significant. A final factor that may stabilise
RGase A is its glycosylation which is unusually well ordered in the crystal structure.
The first obvious distinguishing feature of the central parallel b-helix of the family 28 enzymes is
the occurrence of the fourth PB1a b-stand (PB2a) as shown in Fig. 2. In fact the conformational
differences in this region between a pectate lyase such as BsPel and the hydrolases are not as
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 131

striking as might be imagined. RGase typically has two residues in b-conformation flanked by two
residues in a–L conformation forming hydrogen bonds along the helix axis. The b-arch of BsPel at
residues 148–151, 190–193 and 231–234 has glycines at 190, 191 and 231, which together with the
small Ser 149 and Ala 232 allows these residues to adopt an irregular conformation while the Ca
positions are close to the equivalent residues of RGase A. However, residues 150–151, 192–193
and 233–234 of BsPel have essentially the same conformations and hydrogen bonds as the
equivalent residues of RGase A.
The second feature is that the structure of RGase A is generally much more regular than that of
the lyases and forms many of the hydrogen bonds needed to generate the ‘‘perfect’’ parallel b-helix
imagined by Chothia and Murzin (1993) but not actually seen in the pectate lyases. For example
at T1 Gly 70, Asp 97, Asp 134 and Thr 157 are all in the a–L conformation in successive coils
forming hydrogen bonds along the parallel b-helix. This is followed by a coil with a continuous
PB1-PB1a b-sheet, and finally by three more distorted turns with one residue in a–L conformation
but without regular hydrogen bonding. At T1a these coils have Ala 73, Asp 100, His 137, Asp 160,
Asn 185, Asn 208, Asp 235, Asn 263, Asn 299, and Asp 330 in a–L conformation forming regular
hydrogen bonds along the parallel b-helix before ending with the more continuous sheet as
described in the introduction. Finally at T2 after two irregular turns Asp 142, Asn 165, Asp 190,
Ser 213, Asn 240, Asn 268, Asn 304 and Asp 335 are in a–L conformation with the regular
hydrogen bonding pattern. The differences between the type of a–L turn generally seen in the
lyases and the hydrolases are illustrated in Figs. 4a and b. Some, but not all, of the asparagine
residues in these turns form regular external stacks. There is also often a residue in a–R
conformation at the start of PB1 and again RGase A shows regularity with Met 92, lle 129, His
150, Gly 178 and Cys 199 in a–R conformation, followed by an irregular coil, then Met 249, Ser
277 and Pro 317 in a–R conformation without making all the possible hydrogen bonds and finally
a longer PB1 making anti-parallel sheet with the C-terminal extension.
PehA and PG II have very similar parallel b-helices and interior residues to RGase A. As well as
slowing the evolution of the interior, the conservation of the overall shape of the coils may make
compensating mutations rather more probable than usual. For example, Met 249 in RGase
becomes Gly in PehA and Ala in PG II but Gly 271 of RGase A is replaced by Met both PehA
and PG II to conserve the volume. The four disulphide bridges between Cys 21 and Cys 47, Cys
199 and Cys 216, Cys 322 and 328, and Cys 350 and Cys 359 are conserved between RGase A and
PG II (although the last does not have the same structure) but are not found in PehA, which
however has a single disulphide between Cys 89 and Cys 99 in a T3 loop. An unusual feature of
RGase A is the cluster of three cysteines 199, 216 and 222 where Cys 222 is in the position
equivalent to Cys 199 in the next coil but does not have the same a–R conformation. It is not clear
if Cys 222 assists in the folding. The large cavity reported in RGase A seems to be mostly occupied
by Phe 74 in PG II and at least partially by Leu 65 of PehA.
The family 28 enzymes have very many aliphatic residues both ‘‘aligned’’ and forming ‘‘stacks’’
and these dominate the interior of the parallel b-helices. Although there are several aromatic
residues, there are very few internal aromatic stacks: one of Phe 129, Phe 152 and Phe 182 in PG II
at the start of PB1, Phe 263 and Trp 305 at the asparagine ladder position of RGase A at the start
of PB3 as shown in Fig. 4b (Phe 101 and Phe 138 are aligned) and Tyr 296 and Phe 330 on PB2 in
PehA (Phe 162 and Phe 185 have similar conformations but do not interact strongly). There are
no internal polar stacks in RGase A or PG II but there is a single Asn 245 at the asparagine ladder
132 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

position of PehA. This hydrogen bonds to the main chain NH of Ser 217 in the previous coil. OG
of Ser 217 in turn hydrogen bonds to the main chain NH of Thr 190 but OG1 of Thr 190 only
hydrogen bonds weakly to the carbonyl oxygen of Leu 164. To the C-terminal side, ND2 of Asn
245 hydrogen bonds to a buried water and to the carbonyl oxygen of Val 214. The buried water is
capable of bonding to three carbonyl oxygens of residues 242, 268 and 269 (as Asp 269 forms an
a–L turn). The existence of a stable ‘‘asparagine step’’ suggests that the evolution of new
asparagine ladders is possible.
As discussed in the introduction, the parallel b-helices of RGase A and PG II finish with a
disulphide and may have a topological winding number of 13, while PehA finishes with the final
PB1 a and has a winding number of 12. All of the family 28 hydrolases have 10 coils as defined by
Yoder et al. (1993b). The pattern of long T3 loops especially at the N-terminal end of the parallel
b-helix and long T1 loops towards the C-terminal end is also seen in these hydrolases. Both PehA
and PG II have longer T3 and T1 loops than RGase A, so that the active sites become deeper and
narrower clefts. However, there is no detailed similarity in the loop conformations suggesting that
this similarity reflects the requirements of polygalacturonase as opposed to rhamnogalacturonase
activity. RGase A has several unusually well-ordered N-inked sugar residues at the C-terminal
end of the parallel b-helix.
The active sites of the family 28 hydrolyases have not yet been fully identified by determining
the structure of a complex with a substrate, product or inhibitor. However, sequence conservation
and analogy with other enzymes suggests that substrates bind to the cleft formed by the T3–PB1–
T1 region and that catalysis involves three carboxyl residues: Asp 177 (Asp 202, Asp 180), Asp
197 (Asp 223, Asp 201) and Glu 198 (Asp 224, Asp 202), where the RGase A residue is given
followed by the PehA and PG II equivalent residue in brackets. Van Santen et al. (1999) reported
that site directed mutation of PG II gave activities of 0.01% and 0.08% for D180E and D180N,
0.01% and 0.01% for D201E and D201N, and 0.6% and 0.01% for D202E and D202N
while only slightly changing the Km values. The H223A mutant also had only 0.5% of the WT
specific activity and no change in Km but R256N and K258N showed 14% and 0.8% of the
WT specific activity with an order of magnitude increase in Km. Histidine 223 of PG II
(His 251 of PehA) is not conserved in RGase A but may thus have a critical role in binding the
substrate in polygalacturonases in an altered conformation to assist the catalysis. Armand et al.
(2000) report the properties of these mutants in more detail as well as mutations at Arg 256 and
Lys 258.
As this family of enzymes was known to catalyse hydrolysis with inversion of anomeric
configuration, both Pickersgill et al. (1998) and van Santen et al. (1999) noted conserved water
sites in contact with two of the conserved carboxylates. Similarities to the active site of the P22
phage tailspike endorhamnosidase were noted as well as the fact that the distances between the
carboxylates did not fit the rules used previously for identifying inverting and retaining
glycosidases. However, these rules were developed for b-linked sugars such as the substrates of
xylanases and cellulases where the lone pair on the glycosidic oxygen which is protonated by the
( distance
acid catalyst is on the opposite side of the sugar to the attacking nucleophile. The 7–8 A
for these inverting family 28 hydrolases suggests that the lone pair is not pointed directly away
from the nucleophilic water in these enzyme substrate complexes. Van Santen et al. are more
explicit in proposing the water in contact with and activated by Asp 180 and Asp 202 (water 2 of
PehA and water 37 of RGase A) as the nucleophile attaching at C1. Asp 201 of PG II is assumed
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 133

to act as an acid to protonate O1 (i.e. the leaving group). This is plausible as Asp 202 is in contact
with the positively charged Arg 256 and is likely to be charged.
Page" s et al. (2000) reported the properties of the mutants N186E and D282 K, which implied
that the substrate binds with its reducing end towards the C-terminus of the parallel b-helix as is
seen in PelC by Scavetta et al. (1999) and that N186 was part of subsite 4 while D282 was part of
subsite +2. Mutations were also reported of residues forming the more central subsites and these
often showed greater effects on the overall activity. Examples include D183N at subsite 2, and
Y291L or Y291F at subsite +1. Finally the mutation E252A allowed PG II to accept a
methylated substrate more easily.
As might be expected given the same substrate, the general nature of the residues at the active
sites of these enzymes resembles those at the active sites of the pectate lyases. However, calcium is
not required for substrate binding or activity. Some aromatic residues are found such as Phe 151,
Trp 182, Tyr 276 and Trp 284 of RGase A, Phe 175 and Tyr 231 of PehA and Tyr 130, Tyr 254,
Tyr 283 and Tyr 326 of PG II (as the exact binding site is unknown we cannot assert that all these
residues are in contact with the substrate in a productive complex). None of these aromatic side
chains are conserved in all three structures.

2.1.3. Pectin methylesterase


The structure of the pectin methylesterase PemA from Erwinia chrysanthemi has been reported
by Jenkins et al. (2001) at 2.4 A ( resolution. It has the same number of coils as the pectate and
pectin lyases and the coils of this right-handed parallel b-helix have the same general shape. PemA
has an a-helix at the N-terminal end of the parallel b-helix, long T3 loops and long T1 loops
towards the C-terminal end of the parallel b-helix, and a C-terminal extension. The C-terminal
extension interacts with PB2 rather than PB3 as seen in the lyases and is longer than that of the
lyases. It also makes interactions with the parallel b-helix in the region which interacts with the N-
terminal extensions of the lyases and the hydrolyases.
In terms of the shape of the coils, the structure represents a divergence from the lyases in the
opposite direction from the hydrolyases with little trace of the PB1a sheet and a short T1 arch
region. Stacked a–R residues are again found at the start of PB1 but the angle between PB2 and
PB3 is less obtuse than in the other pectinases so that the distance between PB3 and the start of
PB1 is shorter. This arises because there is no internal stack of aromatic residues on PB3 as seen in
the lyases. The overall shape of the coil is similar to that at the N-terminal end of the parallel b-
helix of pertactin (see Section 2.4).
PemA shows many internal aliphatic stacks and an internal aromatic stack, which occur on
PB2. An external asparagine stack is found but the asparagine ladder position of the lyases is
frequently a cysteine in the pectin methylesterases. In PemA the disulphide between Cys 192 and
Cys 212 appears to be partially formed but the crystals had been treated with the reducing agent
DTT as 100 mM DTT did not inhibit the activity (K. Worboys, unpublished observations). It is
not clear if the cysteines in PemA and in its homologues generally form disulphides or stacks of
cysteines although some sequences contain cysteines which cannot form disulphides.
There is a deep cleft along the parallel b-helix formed the T3–PB1–T1 region which contains the
most conserved sequences and corresponds to the substrate binding site of the lyases. The T3 and
T1 loops of PemA have no detailed similarity to those of the lyases and the cleft is rather deeper
(as is the active site cleft of the polygalacturonases). Two of the T1 loops form b-hairpins and with
134 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

one hydrogen bond between the hairpins almost form a four stranded anti-parallel b-sheet. The
cleft resembles the pectin lyases in having many aromatic residues including an external aromatic
stack of Tyr 158, Tyr 181 and Phe 202 on PB1. The similarity between pectin methylesterases and
pectin lyases results from convergent evolution to bind the same substrate because only one
aromatic residue, Phe 202 of PemA and Tyr 215 of PnlA, occurs at an equivalent position.
Only the sequence conservation (Markovic and Jornvall, 1992; Laurent et al., 1993) and the
observed pH activity profile gives us any information on the catalytic mechanism of PemA. The
enzyme has a broad roughly bell-shaped pH optimum from about pH 5–9 (Pitk.anen et al., 1992).
However, if the active site has been correctly identified, the only conserved potentially catalytic
residues are Asp 178, Asp 199 and Arg 267, and the last two form an ion pair. As Asp 199 is likely
to be deprotonated, only Asp 178 is likely to act as an acid and either Asp 178 after it has donated
a proton or Asp 199 or both may activate a water molecule to act as a nucleophile. PemA is thus
the first member of a new class of aspartyl esterases.

2.1.4. PelL from Erwinia chrysanthemi


The structure of PelL from Erwinia chrysanthemi is currently being determined in our
laboratories. This is a member of family 9 in Henrissat and Coutinho’s classification of the
polysaccharide lyases. Although the sequences are unrelated (Lojkowska et al., 1995), PelL
resembles the ‘‘extra-cellular’’ pectate lyases in requiring calcium for activity. PelL is an endo-
pectate lyase and makes a significant contribution to tissue maceration in vivo. However, its
expression in Erwinia is not regulated by the same mechanism that controls expression of the
‘‘extra-cellular family’’ enzymes. Crystallographic refinement has not yet been completed but the
current model with an overall R factor below 17% and a free R factor below 19% to 1.6 A (
resolution allows almost all the residues to be clearly defined.
The architecture of PelL is a right-handed parallel b-helix and the overall shape of the coils is
similar to that of the extra-cellular lyases with three b-sheets, PB1, PB2 and PB3. The T1 region
resembles BsPel, generally with an a–L turn at the start of PB2 but without a regular PB1 a sheet
as seen in the polygalacturonases and rhamnogalacturonase A. However, there are 10 coils as in
polygalacturonase rather than eight, as in the other lyases. There is an N-terminal a-helix as in all
the right-handed parallel b-helix proteins except pertactin (Emsley et al., 1996). There is a short
N-terminal extension of residues 26–39, numbering residues from the gene so that Ala 26 is the
first residue of the mature enzyme, and a long C-terminal extension of residues 357–425 which
packs against PB3. PelL has 12% of asparagine in its sequence and as initially suggested by
Lojkowska et al. (1995), these form both internal and external stacks. However, the internal
asparagine ladders occur at the start of PB2 (T1a) and at T3 rather than at T2. There is a cluster
of conserved residues including Asp 209, Asp 233, Asp 234, Asp 237 and Lys 273 in the T3–PB1–
T3 region where all the known pectinase active sites are located and this region can be tentatively
identified as the active site of PelL.

2.1.5. Pectate lyase Pel-15 from Bacillus sp. strain KSM-P15


Akita et al. (2000) have reported the crystallization of a pectate lyase Pel-15 from Bacillus sp.
strain KSM-P15 with a very alkaline pH optimum of 10.5 and although no article has yet been
published describing the structure, the coordinates have been deposited to the Protein Data Bank
as accession 1EE6. This structure is the first for a member of the family 3 lyases according to
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 135

Henrissat and Coutinho’s classification. The structure is a right-handed parallel b-helix with a
similar shape for each coil to that of the extra-cellular lyases and also has 8 coils together with an
extra PB3 strand. However, the sequence is significantly shorter than that of the other pectinases
and there is no N-terminal helix, there is only one notable T3 loop (residues 27–42) and there is no
C-terminal extension. The small size is not associated with an unusually repetitive structure. There
are left-handed a-helical turns and some stacking of aliphatic groups. However, the only
asparagines stack is external and there are no obvious aromatic stacks. Alignment with Bacillus
subtilis pectate lyase gives an RMSD of 1.94 A ( for 158 Cas (of 197), showing that this enzyme is
another member of the pectinase superfamily. The active site contains several striking similarities
to the other lyases and some surprises, including the site of calcium binding. The calcium position
is different from that of the single calcium seen in BsPel and PelL. This site also does not
correspond to any of the sites reported by Scavetta et al. (1999) for PelC but is near aspartates in
both PelE and BsPel. A full description of this interesting structure is eagerly awaited and
hopefully it will be possible to model or observe substrate binding and thus understand in detail
the role of calcium, which is also necessary for the pectate lyase activity of this family.

2.2. The P22 phage tailspike endorhamnosidase

The P22 phage tailspike endorhamnosidase (TSP) has been for many years one of the principal
model systems used to study protein folding and misfolding both in vivo and in vitro (King et al.,
1996; Betts et al., 1997; Seckler, 1998; Betts and King, 1999). The final rate-limiting trimer
maturation reaction, which has the same rate in vivo (Goldenberg and King, 1982) and in vitro
(Danner et al., 1993), produces a folded trimeric enzyme which is unusually stable both to high
temperatures and even more usefully in presence of denaturants including SDS. Thus this system
has allowed a clear separation of issues relating to folding from those relating to the stability of
the folded protein. Many temperature sensitive folding (tsf ) mutants have been found which are
stable once folded but which will not fold above some non-permissive temperature. Several
supressor mutations (su) that restore folding to tpf mutants at the non-permissive temperature
have also been identified. The determination of the TSP structure depended on the fragmentation
of the enzyme by recombinant expression and crystallisation of the two fragments, residues 1–124
and 109–666 (Miller, 1995; Miller et al., 1998a). The structures of these fragments of TSP
(Steinbacher et al., 1994, 1997) revealed that the major part of the protein forms a parallel b-helix
while the N-terminal (residues 1–108) and both C-terminal (residues 541–666) domains have anti-
parallel b-folds. This information opened the way for the genetic and in vitro folding studies to be
interpreted in terms of atomic interactions.
The nomenclature describing the structure of TSP was developed independently from that of
the pectinases. The articles describing the structure of TSP describe the three stranded parallel
b-helix in terms of sheets A, B and C corresponding to PB2, PB3 and PB1 of the pectinases. The
overall shape of the monomer resembles a fish with the parallel b-helix domain taken as the body
and the extreme C-terminal domain taken as the ‘‘Caudal fin’’. The long loops off the parallel
b-helix are then named ‘‘Dorsal and Ventral fins’’, corresponding to T3 and T1 loops,
respectively. There are 13 complete coils in the parallel b-helix domain of TSP and these coils have
the characteristic L or kidney shape also seen in the pectinases and the N-terminal region of
pertactin with a stack of a–R residues at the start of sheet C (PB1). Like the pectinases but unlike
136 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

pertactin, the parallel b-helix has an N-terminal a-helix instead of a PB1 in its first coil followed by
the first strand of sheet A (PB2). The detailed shape of the parallel b-helix is more similar to the
pectic lyases than pectin methylesterase or pertactin, mostly because the a–R residue at the start of
sheet C (PB1) and the second inward pointing residue of sheet B (PB3) tends to be larger, forcing
sheets B and C (PB3 and PB1) apart. One possible alignment of the parallel b-helix with BsPel
gives an RMSD of 2.02 A ( for 174 Ca atoms. There is one long T3 loop from residues 197 to 260
forming the Dorsal fin. There are some long T1 loops, forming the Ventral fin, one of which is
poorly ordered in the X-ray structure. Although TSP is a hydrolyase, the T1 region is more similar
to the pectic lyases than the hydrolyases and there is only a very small region where the hydrogen
bonds of a sheet PB1a are present between residues 267–268 and 287–288. However, the T1 region
has an interesting example of internal stacked cysteines. The turn at T2 is generally similar to that
seen in the pectic lyases and has similar geometry. The interior of the parallel b-helix is almost
exclusively hydrophobic, without polar stacks such as the asparagine ladders of the pectic lyases,
and with few buried waters, mostly associated with the amides of the T2 turn. There are several
aliphatic stacks especially at the C-terminal end of PB3 and an aromatic stack on PB1 (residues
Phe 284, Phe 308 and Phe 336). However, there are also aligned rather than stacked interactions
and edge to face aromatic interactions.
The fragment 109–666 had the same oligosaccharide binding and endoglycosidase activity as
the full length protein (Miller et al., 1998a) which acts both as an adhesion factor in phage binding
and as an enzyme. Steinbacher et al. (1996) determined the structure of three complexes of TSP
with receptor lipopolysaccharide fragments comprising two O-antigenic repeating units from
three Salmonella species, giving the first direct evidence for the substrate binding site of a parallel
b-helix enzyme. The substrate bound to the same cleft created by sheet C and the loops on either
side (i.e. T3–PB1–T1) and was bound almost parallel to the helix axis as was later seen in pectate
lyase C (Scavetta et al., 1999) and chrondroitinase B (Huang et al., 1999). The binding site shows
‘‘wobble’’ with alternative binding sites for fragments from different strains but the terminal
rhamnose is in almost the same position in the three complexes. The binding of the O-antigen
involves mostly small hydrophilic side-chains and only the side chain of Lys 302 shows significant
displacement on forming the complex. However, the binding does bury a considerable surface
area, showing the excellent complimentarity of the protein and the polysaccharide. The reducing
end of the polysaccharide was bound towards the C-terminus of the parallel b-helix. Aspartates
392 and 395 and glutamate 359 were identified as likely to be involved in catalysis. Mutation of
any of these residues to the amide, either asparagine or glutamine, caused a dramatic loss of
activity (Baxa et al., 1996) but all three mutants continued to bind substrate and product with
wild-type affinity. An inverting endorhamnosidase mechanism was suggested (Steinbacher et al.,
1996) in which a water molecule observed to bind in contact with Glu 359 and Asp 395 acts as the
nucleophile, while Asp 392 protonates the glycosidic oxygen.
The structural data revealed that all the 62 tsf mutants were in the parallel b-helix region of
TSP (Haase-Pettingell and King, 1997). Miller et al. (1998a) had reported that the removal of the
N-terminal domain to give the fragment 109–666 unmasked the effects of two tsf mutants, G244R
and D238S, and four suppressor mutants, V331G, V331 A, A334 V and A3341, on the kinetics of
folding and unfolding. Subsequently Miller et al. (1998b) reported the properties of a tailspike
fragment corresponding to the isolated parallel b-helix domain consisting of residues 109–544
(bhx). This fragment has low but measurable endorhamnosidase activity (0.2% of the wild-type)
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 137

and is a mixture of monomers and trimers. Unlike the larger species, it folds and unfolds
reversibly. Further truncation at the C-terminus of the parallel b-helix produced increasingly
unstable proteins which could not be purified showing the importance of capping the parallel b-
helix. Schuler and Seckler (1998) conclude that the P22 tailspike folding mutants act by globally
stabilising or destabilising thermolabile monomeric folding intermediates in which the central
parallel b-helix is topologically similar to the native structure but less tightly packed. Betts and
King (1999) point out the alternative that the rate of the self aggregation process is reduced. An
obvious question might be why are the su mutant sequences not selected by evolution? Val 331 is
close to the substrate and the V331G or V331A mutants are less active than wildtype. The case of
A334 V and A3341 is more complex in that the intermediate and bhx are stabilised by these
mutations, which produce a classical aliphatic stack, but the final trimer is destabilised (Beissinger
et al., 1995) although it has a similar structure to the wild type enzyme (Baxa et al., 1999).
Recently Schuler et al. (2000) have described the rational design of mutants aimed at producing
the tsf and su phenotypes. A tsf mutant was produced by the mutation T326F, which introduced a
much larger residue into the parallel b-helix. The most surprising feature was how little the
structure was distorted with very small shifts of the main chain and only a slight but significant
increase in the dissociation constant for an octasaccharide (no change was seen with T326S and
T326V). The steric strain was mostly absorbed in an unusual conformation of Phe 352 although
strain was transmitted as far as Val 362 on the opposite side of the helix. The rigidity of the
parallel b-helix (or the plasticity of its internal residues) may suggest the mechanism which has
preserved the shape of parallel b-helices while erasing all trace of sequence similarity. Potential su
mutants were designed by mutation of residues Glu 359 and Trp 391 at the active site following
the hypothesis that these mutants had not been observed because of the loss of activity. These
residues’ conformation lies on the edge of the allowed region of the Ramachandran plot.
However, E359A was much less stable than wild-type and E359G was still less stable than wild-
type, suggesting that the side chain of Glu 359 is important for stability as well as for activity.
W391A and W391G were also less stable than wild type. The binding of octasaccharide was also
seriously affected but the crystal structures of E359G and W391A were similar to wild type except
for some small changes near the side chain of Trp 391.
After describing the role of the parallel b-helix in folding, it is necessary for balance to note that
Robinson and King (1997) showed that the formation and breaking of disulphides involving Cys
613 and Cys 635 in the C-terminal ‘‘Caudal fin’’ domain was critical for the folding to the native
conformation. This domain is formed by chains from each monomer of the trimer and has been
called a triple b-helix by Seckler (1998). This domain contains anti-parallel b-sheet but Kreisburg
et al. (2000) note some similarity in the packing compared to the parallel b-helix and suggest that
this packing might be a model for amyloid.
Misfolding of TSP has been studied by King and colleagues (King et al., 1996; Speed et al.,
1996, 1997) with the surprising result that aggregation is specific rather than random and that
folding intermediates do not coaggregate with each other but only with themselves. It was possible
to directly identify by native gel electrophoresis sequential multimers as the earliest intermediates
along the in vitro aggregation pathway. Schuler et al. (1999) have similarly shown that bhx
aggregates via a linear polymerisation mechanism observing monomers, dimers, trimers and
tetramers. The secondary structure was relatively unchanged but tryptophan fluorescence was
quenched. Fibrils were observed which bound Congo Red and gave the green birefringence
138 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

characteristic of amyloid. Foguel et al. (1999) used hydrostatic pressure to rescue native TSP from
aggregated material and suggest that the dissociation of the misfolded aggregates may be similar
to the pressure induced dissociation of oligomeric proteins rather than the behaviour expected
from a random aggregate.

2.3. Chrondroltinase B from flavobacterium hepinarum

The structure of chondroitinase B from Flavobacterium hepinarum, which degrades dermatan


sulfate, has recently been determined at 1.7 A ( resolution (Huang et al., 1999). It is a lyase forming
an unsaturated 4–5 carbon–carbon bond at the non-reducing end of the product but unlike the
pectin and pectate lyases it catalyses the cleavage of a b-(1–4) glycosyl bond. The mature enzyme
is produced by removing a 25 amino acid leader peptide from the 506 amino acid precursor (the
residues numbering used below is that of the precursor). This structure is a right-handed parallel
b-helix with 13 coils, making it the longest of the parallel b-helix enzymes. There is no obvious
relationship between the sequence of chondroitinase B and any other protein whose structure is
known. However, there is clear homology to two poly-b-d-mannuronate lyases from
Pseudomonas sp. (Maki et al., 1993) and Photobacterium sp. (Malissard et al., 1993) and these
lyases constitute family 2 of the alginate lyases as defined by Chavagnat et al. (1996) and family 6
of Henrissat and Coutinho’s polysaccharide lyases.
The overall fold is similar to the other right-handed parallel b-helix enzymes both in the general
L-shape of the three sheets and in several specific features. Thus the parallel b-helix starts with an
a-helix, finishes with a long C-terminal extension from residue 421 to 506 and including two a-
helices (residues 452–458 and 491–503) which interacts with the PB2–T2–PB3 region. The parallel
b-helix also has only the two residue gbE turn (Yoder et al., 1993b) at T2 but has long T3 loops,
especially to the N-terminal side of the active site and long T1 loops to the C-terminal side of the
active site. However, chondroitinase B differs from the other right-handed parallel b-helix
enzymes in lacking a large N-terminal extension and having an a-helix (residues 360–371) inserted
in a very extended T1 turn in coil 11 which forms part of the active site. The final a-helix (residues
491–503) is also unusual in interacting with the T3 loops forming the other side of the active site
cleft.
There are many examples of both stacking and alignment of similar residues in the
chondroitinase B structure. The interior of the parallel b-helix contains aliphatic, aromatic and
polar stacks, including a very long stack of Phe 83, Phe 103, Phe 135, Phe 167, Phe 205, Phe 237,
Tyr 261, Tyr 283, Phe 309, Phe 347, and Phe 388 on PB3. There is a classical asparagine ladder
(together with cysteines) two residues before at the T2 turns from Cys 133, Cys 165, Cys 203, Asn
235, Asn 259, Asn 281, Asn 307, Asn 345 and Asn 386. There are also external aromatic stacks.
However, there are at least two aromatic clusters (Tyr 256–Phe 270–His 302 and Phe 357–Phe
388–Phe 399–Phe 404–Trp 411) showing that the hydrophobic core of the parallel b-helix is not
exclusively built from stacking interactions although stacked residues participate in the clusters.
The structure forms a L shaped cleft from the T3–PB1–T1 region to which Huang et al. (1999)
bound a dermatan disaccharide product and determined the 1.7 A ( resolution structure of the
complex. There are some aromatic side chains lining the cleft (His 116, Tyr 222, Phe 296, Trp 298
and Tyr 324). However, the most obvious interactions of the disaccharide are with polar and
charged residues (Asn 213, Glu 243, Glu 245, Lys 250, Asn 269, Arg 271, Arg 318, His 334, Arg
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 139

363 and Arg 384), while His 116, Arg 184 and Glu 333 might interact with a substrate rather than
a disaccharide. The only side chains moving significantly on forming the complex belong to Asn
213 and Arg 363 (4 and 1 A, ( respectively). Of the residues near the active site, conservation
between the three sequences suggests that Lys 250, Arg 271 and Glu 333 may have important roles
in catalysis while Arg 318 and Arg 363 may be significant in defining the specificity.
The most surprising feature of the complex is that the disaccharide has bound with the reducing
end towards the C-terminus of the parallel b-helix, in the reverse direction to that seen in the
complexes of pectate lyase (Scavetta et al., 1999) and the phage P22 tailspike endorhamnosidase
(Steinbacher et al., 1996). However, we should expect the mechanism of a lyase acting on a b-(1–4)
bond to be very different from one acting on an a-(1–4) bond. Huang et al. (1999) argue that the
mechanism should be a two-step process of proton abstraction followed by b-elimination as
suggested by Gacesa (1987), Gerlt and Gassman (1992) and Guthrie and Kluger (1993) rather
than the concerted mechanism likely in the pectate lyases.

2.4. P69 pertactin from Bordetella pertussis

The structure of P69 pertactin (referred to as pertactin below) was determined at 2.5 A (
resolution by Emsley et al. (1996) and has the longest parallel b-helix observed so far with 16 coils.
Pertactin is the amino-terminal external domain of a 93 kDa precursor, P93 pertactin, encoded by
the Bordetella pertussis gene prn. It is a surface-exposed domain of an outer membrane protein of
B. pertussis and is a component of some acellular whooping-cough vaccines. It has been identified
as a virulence factor and a role in adhesion to mammalian target cells proposed (Leininger et al.,
1991). However, the mechanism and importance of pertactins role in adhesion remains open
(Everest et al., 1996; van den Berg et al., 1999).
Whilst the N-terminal region of pertactin does not start with an a-helix, it is otherwise rather
similar in shape to the parallel b-helix enzymes. The shape and size of each coil of the parallel
b-helix is similar and there are detailed features which enable the sheets PB1, PB2 and PB3
to be aligned unambiguously. The T2 turns resemble those in the lyases, pectin methylesterase
or TSP rather than the family 28 hydrolyases. However, pertactin is possibly the most
regular of the parallel b-helical proteins with spectacular internal aliphatic stacks on rather
flat b-sheets. Examples include I17–I45–V79–L107–V120–L149–V186–V209–V264–I287–I310–
L341–I363, I35–V61–L93–L123–I144–V174–I201–I224–V278–L302, L54–L86–A116–V137–
V167–L194–V217–V271 V295–I327–L351–V381–L399–L417–L444, V81–V109–I131–I151–Vl88–
A211–V266–V289–T312 and I118–V139–L169–L196–L219–L273–V297–L329–L353–L383–I401–
L419. However, not all of these are real ‘‘stacks’’ as opposed to ‘‘aligned’’ residues, so that L196
and L219 are only aligned and some alanines are included. Leucines can partially join the cupped
stacking of valines and isoleucines but do introduce some irregularity.
The shape of the substrate binding sites of the parallel b-helical enzymes is maintained with a
features such the stack residues in a–R conformation at the start of PB1. No known activity is
associated with this region in pertactin and the RDG (226–228) sequence which may be
biologically important (Everest et al., 1996; van den Berg et al., 1999) is in the PB3-T3 region at
the start of a very long loop which folds over the region equivalent to the active sites of the
enzymes. Most of the pertactin residues in this region are small to accommodate this but Asp 265
and Arg 288 are reminiscent of the enzyme’s active sites and may suggest that pertactin evolved
140 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

from an active enzyme. However, the C-terminal part of the long loop around Leu 258 at least
partially occludes these residues, in agreement with the absense of any obvious enzymic activity.
Pertactin has a narrow shape for its individual coils, with a short distance between PB1 and PB2
and no trace of a PB1a sheet. This suggests it should be structurally aligned with either pectin
methylesterase or the P22 tailspike endorhamnosidase. However, alignment with the endorham-
nosidase is difficult because the angle between PB2 and PB3 of the endorhamnosidase at T2 is
more open. Thus PB3 does not fit well when PB1 and PB2 are aligned. The parallel b-helix in the
endorhamnosidase is also more curved than in pertactin. By contrast PemA aligns rather well with
pertactin’s N-terminal region. Five different alignments were tried giving RMSDs in A ( (with
number of Cas aligned in brackets) of 1.93 (155), 2.04 (157), 1.79 (141), 1.99 (156) and 1.91 (154)
as PemA is moved one coil at a time towards the N-terminus of pertactin. The PemA sheets are
slightly more twisted than those of pertactin but it is now possible to use a single rotation and
translation matrix to correctly align the whole parallel b-helix. The turns of this N-terminal region
of pertactin also resemble those of PemA surprisingly well. However, there is no evidence of
conserved sequences. Even when both have unusual features such as a buried polar residues which
can be aligned at T3 (T78 and N95 of pertactin and T114 and N131 of PemA) the residues form
different detailed interactions.
As we move towards the C-terminus of the pertactin parallel b-helix, the sheets rotate slowly, as
expected from their twist. The shape then changes, with first the L shape disappearing as the T3
region becomes smaller, the T2 turn becoming sharper and the T1 region expanding. This is
related to the change to triangular shape in polygalacturonase but differs because of the expansion
of T1. The requirement for a small residue at the a–R position at the start of PB1 is lost and
residue 340 is a leucine. In the next coil the a–R turn vanishes and the turn goes to the end of PB3
near Gly 361. The coil is now almost a sandwich or a thick b-roll but PB3 persists and T1 is long
enough to form almost a parallelogram. Then the T1 region is cut short but the thick sandwich is
maintained. Two tryptophans are stacked from the end of PB3 (W389 and W406) and Gly 391
occurs at the turn. The coils then return to a triangular form of three sheets as PB3 and PB1
lengthen, the turn at T1 changing into a very simple a–R turn. There is no sign of an indentation
to give an L-shape. The stacked a–R turn of residues Asp 397, Gly 415, Asn 442 and GIn 468 may
be the most interesting feature of this region. However, another interesting feature is the turn at
T3. A very similar turn can also be found in the left-handed parallel b-helix proteins and the
comparison is discussed in Section 2.8 because of its relevance to the question of how the hand of
the b-helix is determined. Finally there is a return to the sandwich form and then the appearance
of anti-parallel b-sheet as a C-terminal b-hairpin folds back to cap the parallel b-helix.

2.5. Glutamate synthase

Binda et al. (2000) recently determined the structure of glutamate synthase from Azospirillum
( resolution. This is a very large enzyme of 1472 amino acids with a four domain
brasilense at 3.0 A
architecture. The C-terminal domain from residues 1203 to 1472 of the mature enzyme has a right-
handed parallel b-helical fold with 7 coils. This domain does not seem to have a direct
involvement in catalysis or electron transport but plays a critical structural role and stabilises the
domain interface where ammonia is tunnelled between the active sites. The b-helix is reported to
be regular but not to resemble the others in the Protein Data Bank and seems to have a
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 141

hydrophobic interior. The nature of the turns between the sheets has not yet been reported but the
figures in the article seem to suggest a slight resemblance to the insulin-like growth factor receptor
described below.

2.6. The antifreeze protein from Tenebrio molitor

Liou et al. (2000) report that the antifreeze protein from Tenebrio molitor, TmAFP, forms an
exceptionally regular right-handed parallel b-helix. This is the highest resolution parallel b-helix
structure so far solved and with two molecules in the asymmetric unit, excellent refinement
statistics and very similar structure in several coils, the unusual structure is very precisely defined.
The helix is much the smallest known with 12 residues per coil and is not obviously related to any
other structure. Each coil forms a rectangle with two adjacent sides formed by a short b-sheet of
three residues and two residues in b-sheet conformation but forming only a single conventional b-
sheet hydrogen bond as shown in Fig. 9. These ‘‘sheets’’ have an aL residue between them making
an almost exactly right angle turn which is a typical T2 turn (Fig. 4a). The two remaining sides are
formed of irregular structures related to b-turns, which is repetitive in the sense that it repeats in
each coil. Arbitrarily taking residues 39–50 as a representitive coil (and thus aligning with the
numbering of Liou et al. used below as 10 –120 ), the Ramachandran (phi, psi) angles show the
repeat as b–b–b–aL–b–b–aL–[82.2, 56.9]–[66.6, 20.4 (near 3/10 helix)]–[125.1, 14.0]–
[66.1, 141.6 (polyproline II)]–[84.2, 23.2 (near 3/10 helix)], where angles not obviously a or b
are given in square brackets with some nearby regions indicated. Automatic assignment of this
structure by PROMOTIF (Hutchinson and Thornton, 1996) defines a coil as:
b-sheet–type IV b-turn–g-turn–type I b-turn–type VIII b-turn
PROMOTIF does not accept the shorter b-sheet because of its unconventional hydrogen bonds,
with the first residue’s NH bonding to the Og of a serine from the next coil while the carbonyl of
the second residue bonds to the main chain NH of the residue after the next aL turn (i.e. one offset

Fig. 9. Stereo image of a single coil of the antifreeze protein from Tenebrio molitor with carbons coloured khaki,
nitrogens blue, oxygens red and sulphurs yellow. Residues 35–46 and a disulphide bridge are shown. The internal cavity
is shown with the contact surface coloured by the nearest atom of the protein. The ice-binding threonines are at the
bottom of the figure.
142 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

from the expected pattern). The turn between these ‘‘sheets’’ is very similar to the T2 turn of the
lyases but perhaps the best analogy is to pectin methylesterase, where Ser 249 and Thr 294 show
the same hydrogen bonding (cysteines 192 and 212 in the preceeding coils have the same
conformation but are dubious hydrogen bond acceptors while Ser 330 participates via a water
molecule in the next coil). Thus there is an analogy between these regions and PB2 and PB3 of the
pectinases. The definition in terms of the b-turns is almost equally dubious because there are no b-
hairpins. The unusual nature of the antifreeze protein’s main chain is shown by the startling G-
value for a good 1.4 A ( resolution structure of 0.46 for the Phi–Psi distribution given by
PROCHECK (Laskowski et al., 1993)!
The interior of the parallel b-helix has no space for large hydrophobic residues. Stability is
ensured by the formation of the disulphide bridge between Cys 20 and Cys 80 and although several
oxygens and nitrogens are buried, most form hydrogen bonds. The coordinates deposited do not
include any water molecules but Liou et al. (2000) describe the presence of an internal bound
water with low B factors between the six most regular coils occupying the cavity shown in Fig. 9.
This enables the carbonyls of Cys 80 and possibly Ala 110 from one coil and the NH of 100 and
perhaps 110 from the next coil to form hydrogen bonds. Apart from bridges via water, the
hydrogen bonds between the coils are from NH of 10 to the carbonyl of 120 (i  13), those of the b-
sheet, NH of 40 to the carbonyl of 30 (i þ 11), NH of 50 to the Ser Og of 50 (i þ 12), the carbonyl of
50 to the NH of 60 (i þ 13), the carbonyl of 60 to the NH of 80 (i  10) and NH of 120 to the
carbonyl of 120 (i  12). In some coils there is also a hydrogen bond from a Thr Og at 100 to the 90
carbonyl (i  13). There is only one main chain hydrogen bond within the coil from the NH of 110
to the carbonyl of 80 . Thus several peptides are close packed without hydrogen bonding to each
other. Stacking within the b-helix is restricted to the cysteines and the serines. However, the
threonines of the sheet form an obvious external stack. There are two stacked aromatics, Phe 59
and Tyr 71, at position 120 and there are also hydrogen bonding residues at 70 forming an irregular
ladder.
The function of the antifreeze protein almost certainly involves the single well defined b-sheet
with Thr 10 and Thr 30 . The b-sheet is flat, that is both unbent and untwisted. This causes the two
Thr Ogs and a line of bound water molecules to lie on an array that approximately matches both
the primary prism plane and (less perfectly) the basal plane of ice. Crystal packing causes ice-like
water structure to be unexpectedly clear in the crystal structure because two two-fold related
molecules pack with the threonines facing each other (although TmAFP is monomeric in
solution). Liou et al. (2000) argue that TmAFP functions (like other antifreeze proteins) by
binding to the surface of ice crystals so as to inhibit further growth. From the occurrence of two
differently folded antifreeze proteins amongst the parallel b-helices, Liou et al. suggest that
parallel b-helices suited to this function.

2.7. The leucine-rich repeat family

Leucine-rich repeat (LRR) structures are characterised by a rather variable motif such as
LxxLxxLxLxxNxLxxLpxxoFxx (Buchanan and Gay, 1996; Kajava, 1998) but varying from
20 to 30 residues in length. Porcine ribonuclease was the first of these structures to be
determined by Kobe and Deisenhofer (1993), revealing a spectacular horseshoe of 16
right-handed coils. These structures have been described in reviews by Kobe and Deisenhofer
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 143

(1995a, b), Kobe (1996) and Kobe et al. (1999) and thus our description will focus on the
relationships and differences between LRRs and the classical parallel b-helices. We will
not discuss the LRR variant structure solved by Peters et al. (1996), which is a member of the
coiled helical folds and is not closely related to the parallel b-helices. We also use
the nomenclature of PB1, PB2 and PB3 below (sometimes with quotes when the correspon-
dence is very poor) to illustrate the relationships between LRR and parallel b-helical
structures. However, this nomenclature may not be the best description if the LRR structures
are treated separately.

2.7.1. Ribonuclease inhibitors


The structures of porcine ribonuclease, both free (Kobe and Deisenhofer, 1993) and in a
complex with ribonuclease (Kobe and Deisenhofer, 1995b), are known and the structure of the
complex of human placental ribonuclease with angiogenin at 2.0 A ( resolution has been
determined more recently (Papageorgiou et al., 1997). As expected, these structures are broadly
similar and the two complexes form with the ribonuclease or the homologous angiogenin inside
the horseshoe in contact with the same gross regions of the inhibitor. The two porcine
ribonuclease structures show small local differences in structure which lead to quite large
displacements when they are superposed (RMSD of 1.46 A), ( corresponding to a widening of the
horseshoe opening. Surprisingly, the human ribonuclease complex with angiogenin is more similar
to the free porcine structure with RMSDs of 1.24 A ( against free and 1.88 A
( against the complex.
Also the details of the intermolecular interactions are very different with only the interaction
with the highly conserved catalytic Lys 40 of ribonuclease conserved as a general point of
attachment throughout the family. However, the 77% sequence identity means that the
architecture of these structures is very similar with the caveat that the overall shape clearly shows
some flexibility.
Each coil of RI has an almost rectangular shape with a strand of the 18 strand parallel b-sheet
on one edge, an a-helix on the opposite edge and two shorter connecting regions mostly in an
extended conformation without forming regular b-sheets. It is possible to align coils of the
classical parallel b-helices with a coil from RI so that PB2 matches the b-sheet and is
approximately the same length. The connecting loops then match PBla and PB3 and the a-helix
matches PB1. PB3 is longer and PB1 shorter than the matching structure so that the L-shape of
the parallel b-helix is not seen in RI. There is some detailed similarity at the turns at the ends of
PB2 where the a–L conformation occurs as well as the occurrence of both stacked and aligned
residues on the b-sheet and the residues immediately before and after the turns. The aliphatic
stacks are naturally dominated by leucines which pack against two leucines from the a-helix.
There are no aromatic stacks. The polar alternating stack of asparagines and cysteines forms
similar hydrogen bonds to those seen in the asparagine ladder at the equivalent position in the
pectate lyases (note that cysteine is also frequently found at this position in parallel b-helix
proteins).
The b-sheet of RI shows even less twist than the slightly twisted sheets seen in the right-handed
parallel b-helix proteins. However, the a-helices in each coil of RI cannot pack as close as the PB1
strands of a parallel b-helix and this is accommodated by bending the sheet into the horseshoe (it
might also be possible to design a structure in which the lengths of the loops alternate as in the
spiral folds described in Section 2.10 and retain a flat b-sheet).
144 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

2.7.2. The GTPase-activating protein Ma1P from Schizosaccharomyces pombe


The crystal structure of the GTPase-activating protein Ma1P from Schizosaccharomyces pombe
(Hillig et al., 1999) resembles the ribonuclease inhibitor family in its overall fold, but is shorter
with ‘‘only’’ 11 coils. As in the ribonuclease inhibitors, each coil has an a-helix as well as a b-sheet
which forces the structure into a crescent. A superposition of map1p on human ribonuclease
inhibitor gave an RMSD of 2.4 A ( for 313 Ca atoms. In general the 11 coils are very similar with
only three repeats, LRR1, LRR3 and LRR5, identified as deviating from the consensus structure.
The coil LRR3 and especially Arg 74 at the centre of the most divergent region, is identified as
critical for activating the GTPase by mutagenesis. Sequence comparisons also suggests that a
patch of invariant or conservatively mutated exposed residues in this region after ‘‘T2’’ are likely
to be involved in the biological function. It is not yet clear if Arg 74 is solely involved in protein–
protein interactions or if it is a catalytic residue in the active GTPase complex.

2.7.3. Human insulin-like growth factor receptor


Garrett et al. (1998) determined the structure of a fragment (residues 1–462) of the insulin-like
growth factor receptor at 2.6 A ( resolution, which showed two closely related examples of a fold
much more closely related than the other LRR folds to the parallel b-helices. The insulin-like
growth factor receptor is one of a large class of growth factor receptors coupling binding of a
messenger at the exterior of a cell with signal transduction to the cytoplasm, generally via
activation of a tyrosine kinase. The fragment consists of three domains: an N-terminal parallel b-
helix domain (L1), a central cysteine-rich region and a C-terminal parallel b-helix (L2) and
corresponds to approximately half of the extra-cellular region of the receptor but does not bind
insulin-like growth factor. However, mutation within the equivalent L1 domain of the insulin
receptor does reduce insulin binding (Rouard et al., 1999). The two domains L1 and L2 are clearly
related and can be superimposed with RMSD of 1.6 A ( for 109 Ca.
These domains contain five complete right-handed coils plus an extra extended strand and form
a bridge between the LRR structures such as the ribonuclease inhibitors and the parallel b-helix
proteins. It is possible to align the central five strand b-sheet of L1 or L2 with PB2 and this time
the connecting regions also form hydrogen bonds to suggest a real b-sheet which aligns with PB1a
and a few hydrogen bonds corresponding to PB3. The central sheet is flat because only some of
the strands matching PB1 form a-helices. However, even when these strands are mostly extended,
they do not form the hydrogen bonds of a b-sheet. The shape of each coil is closer to RI but the
bending of the RI sheets causes superposition of RI to be as poor as a superposition on a parallel
b-helix (polygalacturonase) (with RMSD of 2.5 A ( for 60 and 62 Cas; respectively). Compared with
other LRR proteins, the stacking of residues is more like that seen in the parallel b-helices as
isoleucine and valine often replace leucine (for example residues Ile 9, Ile 31, Leu 57, Ile 89, Ile 113
and Ile 139 in the interior and Ile 2, Val 24, Val 50 and Val 75 outside but buried by interaction
with the cysteine-rich domain). The Cys–Asn alternating stack has been replaced by an asparagine
ladder at T2, resembling that of the pectate Lyases (Fig. 5a). The turn between PB1a and PB2
often uses glycines such as Gly 27 and Gly 109 in L1 and Gly 328, Gly 356 and Gly 414 in L2. The
turn from ‘‘PB1’’ to ‘‘PB1a’’ is a rare example of an a–R turn formed by residues 23, 49, 74, 104 in
L1 and residues 324, 352 and 376 in L2. However, only ‘‘PB1a’’ is a b-sheet. This turn may be
stabilised by one of the two disulphides in each domain between Cys 3 and Cys 22 (L1) and Cys
302 and Cys 323 (L2) at the N-terminal end of the domains.
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 145

Possibly because the L1 domain interacts closely with the cysteine-rich domain, the L2 domain
is in general more regular than the L1 with a longer asparagine ladder and a more regular C-
terminal a-helix. However, L1 has the only (external) pairs of stacked aromatics (Tyr 28 and Tyr
54 and in a more edge to face conformation Phe 58 and Phe 90). The face loosely equivalent to
PB1 is a mixture of a-helical and irregular in both L1 and L2.
However, L1 has less helix but includes some repeating units of main chain such as the three b-
turn-like conformations centred on residues 46–47, 71–72 and 101–102 which seem to extend our
idea of what can be efficiently stacked. In particular Pro 46 and Pro 71 show one way that proline
can be included in a repeating structure.

2.7.4. Human spliceosomal protein U2A0 , Rab geranylgeranyltransferase and the mRNA export
factor TAP
The structure of the human spliceosomal protein U2A0 was determined by Price et al. (1998) as
part of a complex with U2B00 and RNA. At first sight, this structure is generally similar to the
LRR region of the insulin-like growth factor receptor in that the ‘‘PB2’’ sheet is not strongly
twisted and only slightly bent. One alignment of U2A0 with the more regular C-terminal region
(domain 3) gave an RMSD of 2.1 A ( for 78 Cas; which is clearly closer than the alignment of either
with a parallel b-helix structure or with ribonuclease inhibitor. However, this is essentially an
alignment of the b-sheet alone because the irregular or helical region of one molecule does not
superpose on the equivalent region of the other molecule. Instead, this pair of structures shows an
entirely new evolutionary possibility seen nowhere else amongst the coiled folds, in which the two
regions of the structure have slipped by one coil relative to one another. Thus the b-sheet residues
A68–A73 of U2A0 aligns on A357–A362 of IGR but residues A365–A370 of IGR superpose onto
residues A53–A58 of U2A0 . Similarly, residues A76–A80 of U2A0 superpose onto A398–A402 of
IGR as shown in Fig. 10. Naturally superpositions of a complete coil can be forced and these do
seem to roughly align several leucines. However, fewer residues are aligned with a higher RMSD
(i.e. 44 Cas with an RMSD of 2.2 A)( and the planes of the b-sheets do not coincide. In general, as

Fig. 10. Comparison of the folding of the insulin-like growth factor receptor domain 3 and human spliceosomal protein
U2A0 . Only backbone atoms are shown, excluding carbonyl oxygens. The complete U2A0 chain is shown as thin bonds
with residues 1 and 163 marked N and C, respectively. The residues 298 and 458 of the insulin-like growth factor
receptor defining domain 3 are marked and the domain is shown with thick bonds. The molecules were aligned by fitting
the backbone atoms of residues 22–27, 46–50, 68–73, 92–98 and 117–122 of U2A0 against residues 307–312, 329–333,
357–362, 389–395 and 415–420 of IGR. These residues are all part of the main b-sheet.
146 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

might be expected from the structural differences there are very few identical residues aligned by
any of the possible alignments. One apparently common feature is that both have an asparagine
ladder at the standard ‘‘T2’’ position. However, the detailed pattern of hydrogen bonds in this
region are very different as shown in Figs. 5a and b. This shows how a residue in each coil of U2A0
forms hydrogen bonds to the asparagine –NH2 group, rather than participating in PB2, and the
asparagine side chains hydrogen bond only to main chain atoms rather than to each other.
Rab geranylgeranyltransferase, RabGGT, comprises three domains: a helical domain, an
immunoglobulin domain and a C-terminal LRR domain from residue 418 to 567 (Zhang et al.,
2000). The 2.0 A( resolution structure showed that the LRR domain is clearly similar to that of
0
U2A with an RMSD of 1.2 A ( over 116 Ca atoms from Zhang et al. (2000) (1.38 A ( over 122 Ca
atoms using O). The similarity is especially clear at the C-terminal coil despite that having a
different conformation to the other coils. No relationship has yet been suggested between the
function of RabGGT and U2A0 .
The structure of a fragment (residues 102–372) of the mRNA Export factor TAP was
( resolution. There are two
determined by Liker et al. (2000) in two crystal forms at 2.9 and 3.15 A
globular domains in this fragment. Residues 119–198 have a ribonucleoprotein (RNP) fold and
residues 203–362 have an LRR fold and show homology to U2A0 (the RNP domain is
homologous to U2B00 ). The homology extends beyond the LRR region and includes the C-
terminal extension and the Tyr 131–Asp 146 interaction of U2A0 . Thus the structure is very
similar to that of U2A0 with an RMSD of 1.56 A ( for 103 Ca atoms.
0
Like U2A , which does not bind RNA in vitro, the isolated LRR domain of TAP did not bind
RNA. Thus in both systems the role of the LRR domain is to bind the RNP domain and, while
the RNP domain of TAP does bind RNA, in both systems association is essential for the specific
co-operative binding. Mutation of the surface equivalent to that involved in the U2A0 to U2B00
interaction, suggested that the interaction survived the introduction of several alanines but was
disrupted by reverse charge mutants. However, surprisingly some mutations at the concave b-face
of the LRR domain did not prevent the RNA interaction. The C-terminal region of the LRR
domain seems to be the critical region for the RNP-LRR binding.

2.7.5. Internalin B from Listeria monocytogenes


The recent 1.86 A ( structure of internalin B (InlB) from Listeria monocytogenes (Marino et al.,
1999) is the highest resolution structure of an LRR protein currently available (from crystals
diffracting significantly beyond the nominal resolution). Due to its resolution, this structure
perhaps best reveals the structural principles of the LRR family. The central feature is the packing
of the Leu–X–Leu sequence at the centre of the main b-sheet (‘‘PB2’’) with w1 angles of
approximately 1801 and 601, respectively. These leucine pairs then stack onto the equivalent
residues in the neighbouring coils. The other leucines of the motif then pack rather less regularly
onto the leucine pair. Only the insulin-like growth factor receptor L1 and L3 domains have
diverged away from this very regular packing so that this pair of leucines are no longer conserved.
Structural alignment reveals that InlB is very similar in its architecture to U2A0 (especially at its
N-terminal end) but is rather more bent. Thus the first five coils and the extra loop and b-sheet of
U2A0 superpose on the N-terminal coils of InlB with an RMSD of 1.92 A ( for 102 residues and
many identical residues (especially leucines and asparagines) are superposed. The asparagine
ladders of InlB and U2A0 are similar as is the general form of each coil. However, InlB is longer
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 147

and has 7 complete coils and an additional b-strand and loop region. Thus it has two extra coils,
which can be assumed to be at the C-terminal end as this region is more bent and has larger
helices. The LRR repeat of InlB is 22 residues rather than 24 in U2A0 and InlB has no a-helices
but has short stretches of 310, helix in each coil running anti-parallel to the main b-sheet. An
aspartate at position 14 of the repeat (counting the asparagine ladder as 10) may help stabilise this
helix. The resolution is sufficient to define a regular water structure of spines of waters with
relatively low temperature factors which bridge the extended regions of the structure. Thus in the
non-helical regions anti-parallel to ‘‘PB2’’, the main chains do not directly form hydrogen bonds
and are significantly further apart. The result of this and the 310 helices is that the LRR region is
bent by approximately 141 per coil.
InlB has a calcium binding N-terminal cap of residues 36–76 but no C-terminal cap. The
hydrophobic core of the cap is continuous with that of the LRR domain. The calciums do not
appear to be necessary for the folding of the structure and are rather weakly bound. However,
they may form bridges in the interaction of InlB with its ‘‘receptor’ or target.
Marino et al. (1999) proposed that the concave faces of the LRR structures make them
especially suitable to make protein–protein interactions. Marino et al. (2000) aligns the internalins
and begins the assignment of possible functional residues and binding sites.

2.8. Left-handed parallel b-helix structures

2.8.1. Left-handed parallel b-helix structures containing hexapeptide repeats


Five left-handed parallel b-helix structures containing hexapeptide repeats have been
determined: UDP-N-acetylglucosamine acyltransferase, LpxA, from E. coli (Raetz and Roderick,
1995), a carbonic anhydrase, Cam, from Methanosarcina thermophila (Kisker et al., 1996),
tetrahydrodipicolinate N-succinyltransferase, DapD, (Beaman et al., 1997, 1998a), a xenobiotic
acetyltransferase, PaXAT, from Pseudomonas aeruginosa (Beaman et al., 1998b) and the
bifunctional N-acetylglucosamine 1-phosphate uridyltransferase from E. coli (Brown et al., 1999;
Olsen and Roderick, 2001) and from Streptococcus pneumoniae (Sulzenbacher et al., 2001). These
are all members of a large family of enzymes with a hexapeptide repeat motif [LIV]-[GAED]-X2-
[STAV]-X (Vaara, 1992). Most function as acyl transferases, transferring acetate, succinate or R-
3-hydroxy fatty acids using either acyl-coenzyme A or acylated acyl carrier protein. The
differences between this area of mammalian and bacterial metabolism have inspired great interest
in finding inhibitors of these enzymes as potential antibiotics.
The left-handed parallel b-helix of these enzymes resembles a triangular prism with the flat but
pleated b-sheets forming the faces. The active form of all these enzymes is the trimer and the mode
of trimerization is generally conserved. Four of the five enzymes form trimers; with the axis of the
parallel b-helix aligned with the three-fold axis to within a few degrees. However, PaXAT has an
angle of 211 between the helix and the trimer axis. Beaman et al. (1998b) suggest that this
misalignment is possible because the parallel b-helix in PaXAT is the shortest of the known
structures. LpxA has 10 coils, DapD and Cam have 7 coils and PaXAT only 5 coils. The form of
N-acetylglucosamine 1-phosphate uridyltransferase used for the initial structure determination
(Brown et al., 1999) was truncated after residue 331 of the 456 residue enzyme because the crystals
from the complete molecule did not give satisfactory diffraction. This truncated molecule had the
pyrophosphorylase activity of the second step of the overall synthesis of Udp–GlcNAc, which is
148 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

associated with the N-terminal domain, but had lost its acetyl transferase activity and had only 4
complete coils and 13 of the 23 hexapeptide repeats. Subsequent structures (Kostrewa et al., 2001;
Olsen and Roderick, 2001; Sulzenbacher et al., 2001) revealed that N-acetylglucosamine 1-
phosphate uridyltransferase has the longest known left-handed parallel b-helix This left-handed
parallel b-helix has 10 complete coils without including an N-terminal helix, which plays a similar
role to that of the pectinases, and an extra b-strand. Then there is a remarkable ‘‘domain swap’’ in
which a C-terminal extension add an extra b-strand to extend the parallel b-helix of its symmetry
equivalent in the trimer. Thus we could generously argue that there are 11 complete coils.
However, the regularity of the hexapeptide repeat is less in this structure because there are several
single residue deletions and one large insertion at the T3 turn, which is the probable site of the
acetyltransferase activity. An interesting difference between the truncated and wild-type enzymes
from E. coli is that there is a disulphide bridge inside the b-helix only in the truncated enzyme,
which Olsen and Roderick (2001) suggest is due to the increased accessibility of these cysteines in
the truncated form.
The active sites of DapD and of PaXAT have been identified by binding substrates and
substrate analogues (Beaman et al., 1998a, b) and lie at a monomer–monomer interface. The
activity involves residues from both molecules. This is also the probable active site of LpxA
(Wyckoff and Raetz, 1999) and of the acetyltransferase activity of N-acetylglucosamine 1-
phosphate uridyltransferase, which Pompeo et al. (2001) have shown requires trimers for activity.
Similarly the zinc at the active site of Cam has three histidine ligands from two molecules at a
similar distance from the three-fold axis. Thus we would expect to find trimer formation to be
conserved in the evolution of all these enzymes. Mutants with altered activity have been made of
serine acetyltransferase (Wirtz et al., 2001) who also constructed a model from the known
structures and thus identified the same active site. Crystals of serine acetyltransferase had been
reported (Wigley et al., 1990) but only the symmetry of the structure has so far been reported
(Hindson et al., 2000).
The left-handed cross-over connections between the b-strands of these structure were
surprising. Richardson (1976) had noted that such connections are very rare and argued that
the inherent right-handed twist of extended polypeptides and of a-helical segments naturally folds
a protein into a right-handed coil, as the ends of these segments are brought together. In the left-
handed parallel b-helix this may not be important because the sheets are unusually flat and the
connections between adjacent b-strands are long. This leaves the question of why the chain folds
into a left-handed rather a right-handed helix.
Kisker et al. (1996) suggested that the origin of the chirality was the a–L turns. At first sight this
suggestion is immediately refuted by the prevalence of such turns in the right-handed folds.
However, the turns in the left-handed parallel b-helices are different from those generally found in
the right-handed folds (for example at T2) and in fact the hexapeptide repeat can be imagined as
encoding a single turn. The turns are tighter and sometimes resemble classical b-bends in forming
a 1–4 hydrogen bond. Good examples can be found in the region Y63–Q64–F65–A66 of LpxA
and the neighbouring coils (Fig. 6a). Residues 64 and 65 (escaping the confusion of defining i;
i þ 1; etc. from the hexapeptide repeat, the b-bend or the T2 turns of pectinases) are in polyproline
(proline is sometimes found) and a–L conformations, respectively. The turns are stacked with
reasonable hydrogen bonds between the oxygen of the residue in polyproline conformation and
the amide hydrogen of the residue in a–L conformation in the next coil. Thus the Wilmot and
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 149

Thornton (1990) definition of this turn would be bP gL (the classical type 11 of Richardson, 1981)
rather than bE gL : However, the turn does not form a b-hairpin and the strands diverge after
making one hydrogen bond.
The distortion away from a simple a–L conformation and formation of the 1–4 H-bond could
be significant, either as a cause of the chirality, or as a consequence of it. In the above example, the
rotation of the peptide 65–66 to make the hydrogen bond rotates the side chain of residue 66 to
the side of the side chain of residue 62, a choice which defines the overall chirality. However, this
argument may be weakened by comparing these turns with those at T3 towards the C-terminus of
pertactin. Residues 117–121 of LpxA has the sequence LMINA while residues 353–357 of
pertactin has the sequence LTGGA. Despite the GG pair the main-chain and side chains occupy
very similar positions. The five Ca atoms can be fitted with RMSD of 0.74 A. ( The alanines are
conserved not just in several turns of both proteins so that pertactin as A333, A357, A387 and
A404. Leucine also occurs at 329, 353 and 383 and these residues are stacked followed by
isoleucine 401. Similarly Leu 117 and Leu 135 are stacked in similar positions to Leu 353 and Leu
329 (note the inversion), respectively. Fig. 6 shows this comparison directly after superposing all
the main chain atoms of residues 116–123 of LpxA on residues 352–359 of pertactin with an
RMSD of 0.78 A. ( Thus it would seem that there may be a consensus sequence for this type of
turn, even in right-handed structures, but that it does not force a chirality on the protein.
Bateman et al. (1998) suggested that the most likely origin of the chirality is the side chain
interactions. The first residue of the motif is generally L, I or V although aromatic residues are
found. Three of these residues come together and intercalate their side chains, forming a chiral
packing at the centre of the parallel b-helix. When three stacks of mostly isoleucine side chains
(L44, 162, 192 and L117; V50, 168, 198 and 1123; and 156, 186 and V111, extending to V129 and
V147) come together at the centre of LpxA, the result is an attractive (to a crystallographer)
example of symmetric close packing (Fig. 11). The closest distances are between each Cd1 and two
Cg2 atoms. For example Cd1 of Ile 68 is 4.45 A ( from Cg2 of Ile 62 and 4.52 A ( from Cg2 of Ile 92.

Fig. 11. The packing of side chains in the left-handed parallel b-helix family. (a) Residues 83 to 99 of LpxA showing
most of a coil and the interaction of isoleucines 86, 92 and 98. Residues 93–96 form the characteristic turn of the left-
handed parallel b-helices. Carbons are drawn in light grey with oxygens and nitrogens in dark grey with some residue
numbers indicted to the right of the a-carbons. (b) Two coils of LpxA with residues 54–69 added in front of 83–99 in the
same orientation and colours as in Fig. 11a (with labels removed). Both aliphatic and polar stacks are shown. The view
is slightly off the axis of the parallel b-helix.
150 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

However, Cg2 of Ile 68 is less symmetrically placed 3.62 A ( from Cd1 of Ile 86 and 4.66 A
( from Cd1
of Ile 56. The imperfect 3-fold symmetry continues with Ile 86 Cg2 at 4.71 A ( from Ile 92 Cd1 and
Ile 86 Cd1 at 4.81 from Ile 98 Cg2: All the isoleucines are in rather similar conformations: Ile 68 is
slightly distorted with w1 ¼ 46; w2 ¼ 150; Ile 86 has w1 ¼ 68 and w2 ¼ þ155; Ile 62 has
w1 ¼ 66 and w2 ¼ 154: 156 has w1 ¼ 59 and w2 ¼ 177; 192 has w1 ¼ 56 and w2 ¼ þ161; 198
has w1 ¼ 50 and w2 ¼ þ171: Given the sequence and the positions of the main-chain it would be
possible to find this as an optimal or near optimal solution for side-chain packing. However, it is
more difficult to demonstrate that there is not a similar packing with the hand of the b-helix
reversed and to assess the importance of the chiral isoleucine side chain in defining the optimal
helix chirality.
The right-handed parallel b-helices do not have the approximate 3-fold symmetry of the left-
handed family and tend to be L-shaped. The interactions across the helix tend to involve side-
chains from only two b-sheets which are closer to a sandwich than to an equilateral triangle. The
first residue of PB1 often interacts with residues from PB2 and PB3 which are themselves in
contact with each other. The hand of the helix is defined by how these side chains interdigitate but
the altered main chain orientations and the lower symmetry make it hard to make a direct
comparison. The triangular region of pertactin seems more similar but the presence of aromatic
residues and lower symmetry again makes comparison difficult. However, residues Leu 417–Val
425–Leu 439 or Leu 444, Phe 450 and Leu 464 of pertactin could suggest that the isoleucines of
the left-handed family are critical. Isoleucines are found at the first position of the hexapeptide
repeat in all these left-handed structures but the packing is generally less symmetric than that of
LpxA.
It is interesting that the right-handed parallel b-helix proteins except pertactin have a-helices at
their N-termini while the left-handed parallel b-helix proteins have a-helices at their C-termini.
However, the N-acetylglucosamine 1-phosphate uridyltransferase shows that these helices are not
essential for establishing the overall chirality and also that more than half the helix can be
removed without changing the folding of the N-terminal region.
Stacking is even more obvious in the left-handed family than in the right-handed because of the
repetitive sequence but is mostly restricted to aliphatic residues. Polar residues are rare in the core
of left-handed parallel b-helices and there are no asparagine ladders. Aromatic residues are found
occasionally at position i; although there is not enough room for three large residues at position i:
Only a single two residue internal aromatic stack has been observed in PaXAT (Phe 71 and Phe
125). In the right-handed parallel b-helices aromatic stacks may be favoured by the twist of the
sheets as the rings must stack with sufficient offset so that the electron-rich centres do not repel
each other too strongly. However, there are external aromatic stacks in the left-handed parallel b-
helices, suggesting that either space limitations are critical or that aromatics do not favour the left-
handed fold.

2.8.2. The left-handed parallel b-helix antifreeze protein from spruce budworm
The structure of the spruce budworm antifreeze protein, SbwAFP, has been determined by
Graether et al. (2000) and is the first parallel b-helical protein structure determined by NMR. It
has a smaller coil than the hexapeptide repeat proteins of Section 2.8.1 and can be considered as a
‘‘pentapeptide repeat protein’’ with the caveat that the repeats are less regular. SbwAFP has 90
residues and the parallel b-helical region only extends to residue 72 and is followed by some anti-
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 151

parallel strands. There are only four coils making this the least repetitive parallel b-helix so far
observed. It is also one of the least regular parallel b-helices seen although stacking probably does
occur inside the b-helix with examples such as Val 32, Ile 47 and Ile 64. It is unusual in having a
variety of types of turn. It is triangular and the spacing between the b-sheets (one of which is
poorly defined as described below) is very similar to that seen in the other left-handed parallel b-
helices. Because of this it is possible to make a structural alignment of SbwAFP with one of the
hexapeptide repeat family. Clearly many alignments are possible but the simplest is to align the
first b-sheet of each protein, so that residues 7–72 of 1EWW (model 1) are aligned on 4–90 of
1LXA. Using O, this gives an RMSD of 1.85 A ( for 68 a-carbons. However, this is essentially only
an alignment of the b-sheets and the turns are generally very different with one less residue per
turn in SbwAFP. Thus the SbwAFP again argues that the turns do not enforce the chirality of the
parallel b-helix. There are isoleucines packing in the hydrophobic core of SbwAFP but there are
fewer than in the hexapeptide repeat family and they are less regularly packed with aromatic
residues and disulphides inside the parallel b-helix. In the alignment above, Ile 62 and Ile 68 of
1LXA are aligned with Ile 57 and Ile 64 of SbwAFP although Ile 56 of 1LXA is aligned with Ser
52 of SbwAFP. Formally, Ile 68 of SbwAFP then aligns with Ile 86 of 1LXA but the
conformations are quite different. However, there is another possibility that the chirality of this
fold is defined by the three disulphide bridges, Cys 25–Cys 37, Cys 62–Cys 85 and Cys 67–Cys 80
of SbwAFP.
Like other antifreeze proteins, SbwAFP probably functions by binding ice nucleii to prevent
their growth. This probably involves an array of threonine residues (5, 7, 21, 23, 36, 38, 51, 53 and
70) on one face of SbwAFP. In a regular parallel b-helix, many of these could form stacks. This
particular region of SbwAFP shows exchange broadening and thus the atomic positions on this
face are poorly defined. However, some 20–30% of the ensemble has some b-structure for this
region. It is possible that this region becomes a regular b-sheet and presents an array of ordered
stacks of threonines when the protein binds to ice. Certainly site directed mutagenesis shows that
T7L, T21L, T38L and T51L each show a 80–90% loss of activity.

2.9. Parallel b-rolls

The alkaline protease of Pseudomonas aeruginosa comprises an N-terminal zinc metalloprotease


domain and a C-terminal domain consisting of a 21-strand b-sandwich (Baumann et al., 1993).
Within the C-terminal domain the successive b-strands are wound into a right-handed superhelix
with calcium ions bound within the turns between a tandem repeated GGXGXDXLX. The two
layer b-sandwich architecture built by a succession of these sequence motifs is called the ‘parallel b
roll’ motif. The tightly bound internal calcium ions appear to lock the structure together. This
domain has no clear functional role although an involvement in secretion, export and receptor
binding has been suggested.
The Serratia marcescens mettalloprotease was shown to have a similar b-roll domain made of
tandem repeats with the four turns stabilised by five calcium ions (Baumann, 1994). The roll is
stabilised by three kinds of interactions: Firstly, by main-chain hydrogen bonds of the carbonyl
oxygen of residue i to the amide nitrogen of i þ 1; and the amide nitrogen of residue i and the
carbonyl oxygen atom of residue i þ 17; secondly, by hydrophobic interactions in the centre of the
sandwich, mediated mostly by interdigitating leucine residues at position 8 of the consensus
152 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

sequence (see above) and thirdly by electrostatic interactions between calcium ions and aspartic
acid residues at position 6 of the consensus sequence. Since the b-roll motif is made up of a
repeating nine-residue sequence motif, there is an almost exact repeat in the three-dimensional
structure after 18 residues with the exception of the edges of the b-helix (Baumann et al., 1993;
Miyatake et al., 1995; Hamada et al., 1996). In this respect the b-roll differs from the parallel b-
helix which has three parallel b-sheets and no calcium-ions bound within the domain architecture.

2.10. Spiral folds

The spiral fold is a family of repetitive structures with coiled or solenoid folds in which each coil
consists of two b-sheets and an a-helix and the overall sense of the fold is right-handed. The three
known structures are two coenzyme A dependent enzymes, 4-chlorobenzoyl coenzyme A
dehalogenase (Benning et al., 1996) and enoyl-coenzyme A hydratase (Engel et al., 1996), and
CIpP, an E. coli ATP-dependent protease (Wang et al., 1997). In these enzymes, the two parallel
b-sheets, A and B, approach each other at a similar angle to PB2 and PB3 of the parallel b-helix
proteins. The problem of packing the a-helices is not resolved by bending the b-sheets as in the
ribonuclease inhibitors but by altering the coils so that neighbouring a-helices are at different
distances from the b-sheets. This is one factor making these structures less regular than the
parallel b-helix proteins. However, aliphatic stacks do occur such as Ile 16–Ile 53 and Val 51–Val
103 of enoyl-coenzyme A hydratase or Ile 59–Val 87 of CIpP. CIpP also has an external aromatic
stack of Phe 30–Tyr 62. A second factor in making the spiral folds appear less regular is that the
two b-sheets cannot be joined by the short stacked a–L turns of the parallel b-helix proteins nor
by the a–R turn seen in some coils of pertactin. This is because the sheets are not ‘‘in phase’’ to
allow these turns so that if sheet A is aligned with an example of PB2, sheet B forms its hydrogen
bonds in the opposite direction to those of PB3. Viewed from the perspective of the parallel b-
helix proteins, the spiral folds importance is to illustrate that it is possible to fold a structure which
cannot make the a–L or a–R turns. This is another example of a fold arising by repeating a simple
theme and the limited stacking may reflect this evolution.

3. The prediction and design of parallel b-helix structures

General techniques for the identification of folds from amino acid sequence (Sippl and
Flockner, 1996; Jones et al., 1999) have been developed during the 1990s. The repetitive folds
seem to be a special case and it has been suggested by Yoder and Jurnak (Yoder and Jurnak,
1995) that this fold might be unusually simple to identify. Heffron et al. (1998) describe a method
based on searching with a sequence profile for a single coil of a parallel b-helix derived from
homologues of the pectate lyase PeIC. This method fails to find the known parallel b-helix
proteins, except for the pectate lyase family, because it demands that parallel b-helix proteins have
the asparagine ladder found in the pectate lyases at T2. It does find a number of protein families
but only the three-dimensional structure of internalin B has so far been determined (Marino et al.,
1999). Internalin A and B were the proteins most convincingly identified because these were found
with the most restrictive profile and the profile identified multiple hits as expected if these proteins
had several coils. Thus it is clear that this method does identify some repetitive structures but does
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 153

not distinguish the parallel b-helix fold from the LRR fold. However, before 1998 only long LRR
structures were known, so it was hard to fit short repeats to this fold. It may be possible in future
to identify the LRR folds by ‘‘threading’’ sequences onto the families of LRR folds now known.
Bateman et al. (1998) have proposed that a class of proteins containing the pentapeptide repeat
A(D/N)LXX fold into a right-handed parallel b-helix, with each repeat forming having the
conformation bbbba–L and three sheets forming a coil. This would form the small parallel b-helix
shown in Fig. 12. They note that a Fourier analysis of the sequences only reveals the pentapeptide
repeat, with no evidence for a 15 residue periodicity. However, Fourier analysis of the left-handed
b-helical hexapeptide repeat proteins does not reveal any periodicity higher than 6. As the less
conserved residues 4 and 5 were likeliest to point outwards, there were two choices for the turn.
Because residue 3 was assumed to point inwards, either residue 5 could be modelled as a–L, or
residue 4 as a–R. The a–L turn was preferred to avoid the side chain of residue 4 colliding with the
main chain of the next coil and preventing good hydrogen bonding of the 4–5 peptide. The choice
of a righthanded parallel b-helix was made to optimise the packing of the leucine side chains
(position 3) in the centre of the parallel b-helix. The recent structure of a left-handed parallel b-
helical antifreeze protein (Graether et al., 2000) made up of repeating pentapeptides will naturally
inspire speculation on the relationship between these structure.
Marino-Buslje et al. (1999) have reported their construction of a model of the insulin receptor, a
LRR protein, based on pectin lyase A and the phage P22 tailspike protein and compared it with
the X-ray structure of IGR (Garrett et al., 1998). In this case it was possible to identify the first
domain of the receptor as a parallel b-helix and to deduce that it had at least 4 and probably 5
coils, despite the differences between the receptor and the superfamily containing PnIA and TSP.
Again the problem of identifying a short LRR fold before 1998 is apparent.
An interesting link between the pectinases and the LRR proteins is supplied by the observation
that plant polygalacturonase inhibitor proteins, PGlPs, are LRR proteins. Leckie et al. (1999)
have predicted a model of the structure of PGIP from Phaseolus vulgaris based on that of
ribonuclease inhibitor and used it to interpret the effects of mutation. As the same team have also
grown crystals (Leech et al., 2000), it should soon be possible to compare this prediction with an
experimental structure.

Fig. 12. Stereo view of residues 16–46 (coils 2 and 3) of the right-handed parallel b-helix model proposed by Bateman
et al. (1998) for the pentapeptide repeat family. Residues of the second coil are numbered. Carbons are drawn in light
grey and polar nitrogens and oxygens in dark grey.
154 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

Collinson et al. (1999) have constructed models of AgfA, the insoluble fimbrial subunit of
Salmonella thin aggregative fimbriae and their preferred model is derived from the b-roll. The
sequence can be predicted to be mostly b-sheet and shows five clear repeats of 22–23 amino acids
except for the last coil with 18 residues. The model has internal stacks of both glutamine and
asparagine with their amides hydrogen bonding to the peptide backbone and the asparagine’s
conformation is rather similar to the asparagine ladder of the pectate lyases. However, the more
obvious similarity is to the asparagines sometimes found at the ends of the stacks of calcium
binding asparates in the b-roll structures (for example, Asn 347 of 1KAP). The model of AgfA
differs from the b-rolls in that the tight turn and very short distance between the b-sheets is only
seen at one end of the sheets with the asparagines and glutamines. The sheets are further apart and
the packing less interdigitated are the other end.
Lilie et al. (2000) briefly review the occurrence and possible function of the motif
GGXGXDX(LF/I)X in the RTX family of toxins. This motif occurs in all the proteins secreted
by the haemolysin pathway, including the alkaline protease of Pseudomonas aeruginosa where it
was shown to form the b-roll structure. They then report the design and properties of a 75-mer
with 8 identical repeats of the motif, NH2–WLS–[GGSGNDNLS]8–COOH. This sequence is a
soluble monomer in solution but the circular dichroism does not change on the addition of 6 M
guanidinium hydrochloride. Thus there is probably no defined structure present even when
calcium is added. Only on adding 100 mM calcium chloride and 25% (w/v) polyethylene glycol
8000 was a change to the spectrum expected for b-sheet observed and the peptide polymerised so
that no monomers or dimers were present. The circular dichroism spectrum and the very tight
binding of calcium suggest that the designed b-roll structure had been formed.
Liou et al. (1999) were very successful in predicting that 12 residue repeat sequences from the
antifreeze proteins of Tenebrio molitor were b-helical and that the conserved threonines formed an
ice-binding array on a parallel b-sheet. As described above, this prediction has been subsequently
verified. This may have been an unusually helpful case because the sequences differed by 12
residue insertions and deletions but shows that at least some b-helical structures can be predicted
from their sequences.
Graether and Jia (2001) have used the observed structure of the spruce budworm antifreeze
protein, SbwAFP, as a template to construct a model of an ice-nucleation protein from
Pseudomonas syringae, INP. The basic idea is that a short parallel b-helical protein can bind to the
prism face of ice via threonines and act as an antifreeze protein while a longer parallel b-helix can
form a nucleus and promote freezing. However, the sequence repeat of 16 in INP rather than 18
residues of the template forces the triangular structure to become more oval and the left-handed
chirality is enforced only by memory of the template.
The above fold identifications from the sequences have involved repetitive sequence motifs.
Initially it was much harder to identify the apparently nonrepetitive sequences of the enzymes
degrading polysaccharides. The obvious approach is classical sequence alignment assisted by
knowledge of conserved residues. Pissavin et al. (1998) align PelZ with PelC from Erwinia
chrysanthemi. The alignment gives 27% (96/353) identities with many gaps and is not obviously
significant as a single alignment. However, the overall significance is shown because the b-strands
tend to be conserved and most of the critical active site residues are also present in PelZ.
In a combination of direct sequence alignment with the identification of repetition, Finnie et al.
(1998) have identified a repeating motif N–(I/V)–X–(I/V)–(X)–(D/E)–N in two polysaccharide
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 155

glycanases, PlyA and PlyB from Rhizobium leguminosarum, and in SpsR from Sphingomonas
strain 888 (Yamazaki et al., 1996). PlyA and PlyB can be aligned with the extra-cellular pectate
lyase superfamily with approximately 20% identity, the sequences suggesting a high b-sheet
content with a predicted a-helix near the N-terminus. Finnie et al. proposed that PlyA and PlyB
are distant homologues of the pectate lyases. The motif occurs 10 times in PlyA and PlyB and
assuming that it forms a single b-sheet, suggested that these proteins have rather longer parallel b-
helix than the pectate lyases. It also seems likely that the structure needs some way of
accommodating the D/E residue in a stack. While the prediction of a parallel b-helix fold is
plausible, these enzymes have different specificities from the pectate lyases and the alignment given
only conserves one of the conserved active site residues of the pectate lyases.
Very recently, the results of the ‘‘critical assessment of fully automated structure prediction’’ or
CAFASP have become available at http://predictioncenter.IInl.gov and the results using both
pectin methylesterase and PeIl- suggest that advances in algorithms and the availability of several
b-helical structures have almost solved the problem of recognising these folds. However, while b-
helical structures are generally found as the most likely ‘‘guesses’’ for these structures, there is still
scope for progress in increasing the confidence level and improving the final alignments. It will be
extremely interesting to compare the models generated with human intervention in the parallel
CASP4 contest when these become available.
An alternate way of identifying the parallel b-helix fold was suggested by Sieber et al. (1995)
who argued that the circular dichroism, CD, from the parallel b-helix proteins is hard to represent
as a sum of the normal basis sets. CD recognises secondary rather than tertiary structure and thus
generally cannot distinguish parallel and anti-parallel b-structures (Johnson, 1999), presumably
because there is little overlap of orbitals except along the main chain. However, effects due to the
length of secondary structure elements and their end conformations have been reported (Pancoska
et al., 1999). Thus a plausible origin for a unique parallel b-helix CD spectrum is the frequent
occurrence of a–L residues at the ends of the short b-sheets. This suggests that if there is a unique
parallel b-helix spectrum, both left- and right-handed folds may have similar spectra if both
contain a–L bounded b-sheets, with the caveat that the turns are rather different as shown in
Figs. 4 and 6. Kamen et al. (2000) suggest an alternative explanation that the repetitive
orientation of the aromatic residues generates an unusual spectrum, although they also report that
the conventional interpretation of the CD spectrum gives a reasonable fit to their data.
Khurana et al. (2001) made the interesting observation that the pectate lyase PelC and the P22
tailspike endorhamnosidase bound Congo red and induced different bands but both with positive
ellipticity while the left-handed LpxA bound Congo red and induced two bands with negative
ellipticity. This raises the possibility that the chirality of the Congo red binding site can be directly
observed. However, Congo red binds to many proteins and may not always bind to b-helices.
Congo red did not bind to crystals of Erwinia polygalacturonase (RWP unpublished).
A very interesting article by Khurana and Fink (2000) discussed the Fourier transform infrared
(FTIR) spectrum of parallel b-helix proteins as well as reviewing many aspects of this architecture
and its possible relationship to amyloid fibrils. They concluded that FTIR could not reliably
distinguish parallel b-helix proteins from other proteins rich in b-sheet. However, they did note
that the percentage of b-sheet determined by FTIR was consistently higher than that calculated
from the crystal structures. They attribute this to the extra hydrogen bonds found both at turns
and in the asparagine ladders.
156 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

4. Are amyloid fibrils related to parallel b-helices?

The formation of amyloid fibrils is a feature, and possibly an important link in the causation, of
a range of human diseases including Alzheimer’s disease., where the role of fibrils and protofibrils
in pathogenesis has been recently discussed by Selkoe (1999) and Lansbury (1999). Deposition of
amyloid fibrils of many different proteins occurs in various diseases (amyloidoses) but the fibrils
all appear to be 50–130 A ( wide, rigid and either non-branching or showing infrequent branches.
All give a characteristic green birefringence with the dye Congo Red under cross-polarised light.
Fibrils grow from a wide range of peptides, which normally either have a biological function or
arise by degradation of a larger functional molecule, or else from covalently intact proteins. In
Alzheimer’s disease the fibrils are formed from a peptide fragment, typically 40–42 amino acids
long, of a much larger membrane bound protein. Fibril formation depends on the concentration
of the peptide both in vivo and in vitro. Kinetic studies suggest a nucleation-growth mechanism
(Jarrett and Lansbury, 1993) and a clear distinction between nucleation and further deposition
(Esler et al., 1996a, b). The monomer is reported to be the species involved in adding to the
plaques (Tseng et al., 1999; Esler et al., 2000) and adds reversible followed by irreversible binding.
Interestingly, the kinetic analysis of Esler et al. (2000) suggests that nucleation should require ‘‘at
least thousands of years to occur spontaneously’’. Islet amyloid polypeptide or IAPP, a 37 amino
acid peptide which is normally secreted together with insulin and whose deposition occurs in non-
insulin-dependent diabetes mellitus, also shows similar nucleationgrowth kinetics of fibril
formation (Kayed et al., 1999).
Recently, a number of proteins not normally associated with any disease have been shown to
form fibrils resembling classical amyloid fibrils. For example, Guijarro et al. (1998), Chiti et al.
(1999), Alexandrescu and Rathgeb-Szabo (1999) and Gross et al. (1999) describe the formation of
fibrils from an SH3 domain of the p85a subunit of phosphatidylinositol 3-kinase, acylphospho-
sphatase and the cold shock proteins CspA and CspB (see also Section 2.2 for the case of the
phage P22 tailspike endorhamnosidase). This suggests that there is an ‘‘amyloid state’’ accessible
to many proteins, especially in the presence of low concentrations of denaturants. Proteins have
apparently evolved to fold rapidly under ‘‘natural’’ conditions without generating significant
concentrations of partially folded intermediates and there may often also be a significant energy
barrier to the formation of the amyloid state (Kusumoto et al., 1998) at least under native
conditions. The mutations known to be associated with hereditary amyloidoses seem to act by
destabilising the native state (Kelly, 1998; Dobson, 1999; Radford and Dobson, 1999) and thus
favouring partially folded conformations. The refolding of recombinant proteins from inclusion
bodies offers an obvious parallel. Another approach to the requirements for amyloid formation is
via combinatorial peptide libraries (West et al., 1999), which shows that many sequences with an
alternating pattern of hydrophobic and hydrophilic amino acids can form fibrils. West et al. also
found in a survey of heptapeptide sequences in native proteins with four hydrophilic and three
hydrophobic residues that the alternating pattern occurred least frequently of the 35 possibilities.
Kalberg et al. (2001) argue that an important predictor of a protein’s ability to convert to amyloid
is that the folded structure contains an a-helix which could be predicted from the sequence as b-
sheet. However, it may be important to notice that not all conversions to b-sheet lead to amyloid
fibrils. Takahashi et al. (2000) show that quite similar peptides may produce or not produce
amyloid on conversion and that the kinetics of conversion are very sensitive to mutation.
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 157

The classic method for the analysis of the structure of amyloid fibrils has been X-ray fibre
diffraction. Models have been proposed by Blake and Serpell (1996) and Blake et al. (1996) from
analysis of fibrils from the Val30Met mutant of human transthyretin, which gave relatively high
resolution. The diffraction from transthyretin is compared with that from amyloid from several
sources by Sunde et al. (1997) and in a more general review by Sunde and Blake (1997). They
concluded that all the fibrils are formed from protofilaments of 25–60 A ( in diameter. The fibrils
are predominantly composed of b-structure although the three-dimensional structures of the
native proteins are sometimes predominantly a-helical. The X-ray diffraction pattern is the
characteristic cross b-structure, implying that the b-sheets lie approximately perpendicular to
the fibril axis. The diffraction pattern and thus the model does not constrain whether the b-strands
are parallel, anti-parallel or mixed. The degree of twist between adjacent strands is also not directly
accessible but Blake and Serpell’s model uses the average twist seen in globular protein structures
(about 151) to account for a 115 A ( repeat distance seen along the fibre axis as a 24 strand repeat.
Malinchik et al. (1998) have analysed fibre diffraction and fibril formation from Alzheimer
b-peptides (Ab below) and compared Ab (1–40) with shorter peptides such as Ab (1–28), Ab (11–
28), Ab (13–28), Ab (11–25), Ab (9–28), Ab (19–28), Ab (34–42), Ab (22–35) and Ab (25–35).
Their indexing of the diffraction pattern is different but their model is generally similar to Blake
and Serpell’s model and the conclusion that fibrils are formed from 3–5 protofilaments, in this
case approximately 30 A ( in diameter, is also similar. However, Malinchik et al. conclude that the
hydrogen bonds of the b-sheet are not exactly perpendicular to the fibril’s axis. Distinct twisted
and flat fibrils were identified by electron microscopy using negative stain but both gave similar
diffraction spacings.
Cryo-electron microscopy was used by Jimenez et al. (1999) to examine the fibrils from the SH3
domain at 25 A ( resolution. These fibrils are twisted with two pairs of protofilaments wound as a
double helix around a hollow core. The helical crossover repeat was B600 A ( and there was an
( ( (
axial repeat of B27 A. The protofilaments were 20 A by 40 A and thus flatter than the native SH3
domain. Thus even an all b-protein has extensively refolded in forming the fibril. It is argued that
the protofilaments are so flat that they cannot be included more than a b-sandwich of two
essentially flat sheets (> 21 twist between strands).
The identification of protofilaments as components of fibrils can be contrasted with the
identification by Walsh et al. (1997, 1999) and Harper et al. (1997, 1999) of independently existing
protofibrils of the Ab peptide, which are probably intermediates in fibril formation. Protofibrils
grow from Ab (1–40) after aggregates have been removed by either size exclusion chromatography
or dissolution in DMSO and filtration. The dimensions of these flexible rods were determined by
atomic force microscopy, AFM (Harper et al., 1999) or electron microscopy and quasi-elastic
light scattering (Walsh et al., 1997) with reasonable agreement. Both groups suggested that it may
be the protofibrils which are critical for Alzheimer’s Disease pathology. AFM has also been used
to characterise the formation of fibrils of immunoglobin light-chain amyloid by coiling of two or
three protofibrils (Ionescu-Zanetti et al., 1999), suggesting that protofibrils are intermediates in
the formation of most amyloid fibrils. The review of Rochet and Lansbury (2000) gives an
excellent account of this rapidly developing research.
Fourier transform infrared spectroscopy (FTIR), especially when combined with isotopic
labeling (Anderson et al., 1996), can distinguish parallel and antiparallel b-structure and give
detailed information on the orientation of amide carbonyls. Lansbury et al. (1995) combined
158 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

FTIR with solid state NMR to analyze the fibrils formed from the C-terminal Ab (34–42) peptide,
using 13C labelling. The 13C NMR line widths suggest that the environment of the peptide in the
fibril is less uniform than a crystalline solid, but more uniform than a typical amorphous material.
As the conformations do not exchange on the NMR time scale, most of the peptide may be
inaccessible to water. By diluting the 13C labelled peptide by unlabelled peptide, it was possible to
obtain intramolecular 13C–13C distances and these were used to generate models of the monomer.
These were all highly pleated extended conformations and hairpin folds could be excluded.
Lansbury et al. conclude that the b-sheets are intermolecular and that the intermolecular distances
show that two different interactions occur. In a subsequent study of the same peptide Costa et al.
(1997) eliminated the possibility that a cis peptide occurs between residues Gly 37–Gly 38, leaving
models with two regions of highly pleated anti-parallel b-structure on either side of an unknown
conformation for the glycines. The central residue 20–29 peptide of the islet amyloid polypeptide
(IAPP) has been shown to form highly pleated b-structure by solid state NMR (Griffiths et al.,
1995). The intermolecular packing was constrained by an intermolecular rotational resonance
between the Ca of Ala25 and the C=O of Ile26.
Benzinger et al. (1998) studied the peptide Ab (10–35)NH2 by solid state NMR. Ab (10–35)
NH2 was more soluble than the complete peptide Ab (1–42) but is able to bind to existing amyloid
plaques unlike Ab (1–28) (Lee et al., 1995) and includes residues found by substitution to alter
fibril formation (Fraser et al., 1994; Soto et al., 1995; Esler et al., 1996a, b). Transamination
followed by SDS-PAGE suggested that residues GIn15 and Lys16 from different monomers were
close in space. Thus initially peptides labelled with 13C at the carbonyls of residues 15 and 16 were
synthesised. Using a DRAWS (Gregory et al., 1995, 1998) pulse sequence to selectively recouple
these nuclei gave distances of approximately 5 A ( for each peptide and the variation of intensity
13
with mixing time suggested that each C had two (or more) equidistant neighbours. This implies
the completely unexpected result that this peptide forms parallel b-sheet with the residues in exact
register. This was checked by synthesising peptides labelled at Leu 17 and Val 18, which also gave
similar distances. There remains the question of how these b-sheets interact via their side chains
and what is the ‘‘tertiary’’ fold of the peptide. The many problems of working with insoluble
peptides were elegantly side-stepped by synthesising Ab (10–35) with a 3 kDa polyoxyethylene C-
terminal extension (Burkoth et al., 1998). This species forms aggregates which resemble those of
the peptide in forming b-sheet above pH 5.6, binds Congo Red and forms similar ladders of
oligomers on transamination. However, the aggregation processes is reversible and can be studied
in solution by small angle neutron scattering, SANS, and by electron microscopy (Burkoth et al.,
1999). The SANS data suggests that the peptide forms rods with a 45 A ( radius, so that the
diameter is the length of one 26 aminoacid peptide in b-sheet conformation. This is slightly larger
than the protofibrils studied by electron microscopy or AFM (Walsh et al., 1999; Harper et al.,
1999). Recently, a more complete analysis of data from both the Ab (10–35) peptide and the PEG
derivative (Burkoth et al., 2000) was published. 12 carbonyl–carbonyl DRAWS distances
distributed along the peptide established that it was almost entirely parallel b-sheet and a model
was proposed of 6 stacked single b-sheets to a slightly twisted fibril (or protofilament) which might
then interact with a second fibril to form a double helix.
Recently Antzutkin et al. (2000) reported that multiple quantum 13C NMR measurements on
the Ab (1–40) peptide also strongly favour an in-register structure with parallel b-sheet, in
apparent disagreement with conclusions drawn from FTIR spectra. These authors do not argue
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 159

that all amyloid consists of parallel b-sheet, noting that they observed antiparallel b-sheet in the
short Ab (16–22) peptide (Balbach et al., 2000).
The peptide Ab (11–25) has been used by Serpell and Smith (2000) to form fibres which were
examined by cryo-electron microscopy. The result was that (astonishingly) that it was possible to
directly observe a spacing of 4.7–4.8 A, ( corresponding to the spacing between b-sheets. The b-
strands run perpendicular to the fibre axis and there was no evidence of twisting of the fibres.
Serpell and Smith concluded that ‘‘the b-sheets are arranged in a cylinder or tube, where the
strands are in exact register’’. Fibres formed from the Ab (1–42) peptide showed lower contrast,
poorly defined edges and did not appear so straight or rigid as those formed by the Ab (11–25)
peptide.
Lansbury et al. (1995) first suggested a relationship between the parallel b-helix architecture and
amyloid fibrils noting both the repetitive nature of the parallel b-helix fold and that the b-sheets
lay approximately perpendicular to the helix axis. Gay et al. (1991) had earlier shown that a single
leucine-rich repeat of 23 amino acids forms fibrils, suggesting a relationship to the folded LRR
structures (which were unknown at that time). Symmons et al. (1997) explicitly suggested that
these filaments are composed of parallel b-sheet and that the asparagine in their peptide formed
interactions resembling the asparagine ladder. Lazo and Downing (1997) studied the synthetic
peptides Ac–KLKLKLELELELG–NH2 and Ac–ELELELELELELG–NH2 and suggested from
their CD spectra that these peptides formed parallel b-helix structures. However, Khurana and
Fink (2000) concluded from their FTIR measurements that the conformation of Ac–
KLKLKLELELELG–NH2 was different from that of the parallel b-helix proteins and was
probably a fully extended conformation without turns.
Lazo and Downing (1998) proposed a model for amyloid fibrils inspired by the b-helical
architecture but with alternating coils running anti-parallel to accommodate predominantly anti-
parallel b-sheet. In fact the model is closer to the lectin fold of concanavalin A (Chothia et al.,
1997; Deacon et al., 1997) but has features in common with parallel b-helices such as the short flat
sheets with turn regions and thus has the advantage that many sequences can be accommodated.
However, stacking of similar residues is excluded because alternate residues would have strongly
disallowed w1 angles and glycine or alanine residues might be required if there are tight turns
between the sheets with alternating left- and right-handed a-helical conformations.
Several alternative models of amyloid are briefly reviewed by Li et al. (1999) and compared with
a detailed model of Ab (12–42) that they derived from Blake and Serpell’s model, starting from a
hairpin of transthyretin. However, their molecular dynamics simulations sharply reduced the twist
of the b-sheet, which might be expected as twisted b-sheets can only form all the potential
hydrogen bonds if they are coiled. This is inconsistent with the hydrogen bonds being along the
fibril, as suggested by Blake and Serpell (1996). Also the anti-parallel packing of two sheets would
normally be associated with untwisted sheets (Chothia et al., 1997). It is possible, as suggested by
Rochet and Lansbury (2000) that the initial intermediate is untwisted but that twisting occurs on
fibril formation. Twisting of a flat structure on fibril formation was observed by Ionescu-Zanetti
et al. (1999) at lower resolution by AFM.
A possible model of fibres formed by small peptides might be a single sheet as suggested for Ac–
KLKLKLELELELG–NH2 (Khurana and Fink, 2000). Models of fibrils formed from peptides
which form two or more b-strands can either have a b-meander (up-down) fold so that one
peptide is entirely in one b-sheet, or the peptide can take part in two or more b-sheets. Clearly b-
160 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

meanders such as the model of Li et al. must be anti-parallel. Only parallel or anti-parallel packing
(with translations) is possible in the third direction perpendicular to both the peptide and the fibril
in these model, using hydrophobic interactions. As the X-ray fibre diffraction and cryo-electron
microscopy data suggests that all the b-strands must be approximately perpendicular to the fibril
axis, other models with hydrogen bonds along the fibril axis will tend to resemble parallel b-helices
or b-rolls but can use either parallel or anti-parallel folding (as can also models of short peptides).
These folds do allow anti-parallel packing of side chains as discussed in Section 1.2.5.
Khurana et al. (2001) discuss the binding of Congo Red to a variety of proteins, including fibrils
and b-helical proteins, and suggest that it binds to many different types of structure. However, as
Congo Red has its own twofold axis, its binding may suggest that the fibrils and putative
intermediates which bind the dye (Fraser et al., 1994) have a two-fold axis, either because of the
same residues forming anti-parallel b-sheet or by the anti-parallel interaction of protofilaments.
Turnell and Finch (1992) experimentally shows how Congo Red can bind by intercalation into
anti-parallel b-sheet in the native insulin structure. By contrast, Li et al. (1999) proposed a Congo
Red binding site using only side-chains of their model.
The identification of parallel b-sheet with stacked residues in both Ab (10–35)NH2 and Ab (1–
40) and the observation of very flat sheets in the SH3 derived amyloid (Jimenez et al., 1999) and
for the peptide Ab (11–25) studied by Serpell and Smith (2000), suggests that amyloid fibrils, can
have a close relationship to parallel b-helix structures. Flat parallel b-sheets can be stabilised by
stacking of similar residues while the only flat anti-parallel sheets known from silk have rather
special sequences with alternating glycines although Perutz et al. (1994) predicted a flat
polyglutamine structure. Thus the family 28 hydrolyases of Section 2.1.2 may be the best available
structural models of a flat b-sheet, irrespective of its topology. Parallel b-helix structures show
that short flat b-sheets can form coils with less regular turn regions and that these turn regions
may also be stabilised by stacking similar conformations. Whilst making parallel b-helix
structures hard to predict, it allows many peptides to adopt a related structure. The ‘‘advantage’’
of the parallel structure is not that stacking is ‘‘better’’ than the interactions across an anti-parallel
sheet as there are many interactions giving excellent packing across anti-parallel sheets. However,
a peptide can always align with itself in a parallel b-sheet so that valines stack on valines,
phenylalanines on phenylalanines and asparagines on asparagines. By contrast, anti-parallel
packing with residue i aligned demands that the sequence has residues i  N and i þ N which can
pack in a satisfactory manner. Antzutkin et al. (2000) argue that the critical issue is the symmetry
of the hydrophobic regions under the 2-fold axis of an anti-parallel b-sheet. When the
hydrophobic regions can interact across a 2-fold axis, anti-parallel b-sheet may be favoured.
Clearly in the case of a peptide such as Ab (10–35)NH2 there would be no distinction between
left- and right-handed parallel b-helix connections. There is no obvious reason to prefer any of the
parallel b-helix or b-roll structures as a starting point for model building although the cryo-EM
derived thickness of the SH3 amyloid might suggest a closer relationship to the b-roll, especially
that seen in pertactin rather than the even thinner structure seen in the alkaline proteases which
required rather special sequences of glycines and calcium binding carboxylates (see http://
www.cryst.bbk.ac.uk/Bubcg16r/amyloid.html for a possible model from Jimenez and Saibil).
The size of a parallel b-helix can be measured from the known structures for comparison with
low resolution images from EM or AFM. However, PB1 and PB2 essentially form a sandwich,
which is rather similar in dimensions to most other all b-structures. The outward facing b-carbons
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 161

of PB1 and PB2 are about 17 A ( apart compared to 15–17 A ( for the b-roll region of pertactin. The
unusually narrow b-rolls of the proteases are significantly smaller. The length of the ‘‘sandwich’’
in the known parallel b-helix structures is harder to define because of the L shape (Fig. 1).
Measuring from the a-carbons of T1 to those at the end of PB3 gives 28 A ( but gives only 20 A(
(
measuring from the a-carbons of T2 to T3. Using b-carbons again adds 3 A to the distances but
clearly large side chains such as tyrosine can increase these measurements by 10 A ( or more even
before we add loops. The left-handed hexapeptide repeat structures are triangular prisms of 19 A (
(
(using a-carbons) or 21.5 A (using b-carbons) along each side and the left-handed antifreeze
structure is slightly smaller.
Construction of a detailed model requires data on the role of the side chains interactions to
bury hydrophobic surfaces. In the absence of a detailed model it is difficult to predict the effects of
substitutions. Prolines are found to inhibit amyloid formation in IAPP (Moriarty and Raleigh,
1999) and would tend to destabilise b-sheets, although pectin methylesterase and the receptor for
insulin-like growth factor show that prolines can be accommodated within parallel b-helix
structures. Neighbouring hydrophilic residues might also destabilise hydrophobic packing of b-
sheets but could define the positions of turns. Neighbouring residues with the same charge,
especially aspartates, would be even harder to accommodate unless these residues are at turns and
bind cations or the pH is reduced as suggested by the appearance of fibrils at pH 2.5 in a
glutamate substituted peptide (Lazo and Downing, 1999). The effect of the ‘Dutch’ mutation of
E22Q (Watson et al., 1999) is thus consistent with a parallel b-helix model but the increased fibril
formation from the A21G as well as the E22Q mutants (El Agnaf et al., 1998) could arise by
destabilisation of a soluble a-helical form. There is a further complexity in cases where patients are
heterozygous for such mutations. Clearly cases such as the E22K mutant described by Bugiani
et al. (1998) would allow an alternating pattern of EKEK across a flat b-sheet.
The arguments for a parallel b-helix structure for amyloid fibrils from longer proteins are
weaker because the stacking would not be automatically available. The parallel b-helix
architecture claim to being almost independent of the local sequence is that (1) small or large
insertions can be introduced every 4–6 residues and (2) that no special motif is necessary except a
tendency to form b-structure. Experience with protein sequence alignment suggests that frequent
gaps with low penalties can give apparently good alignments of unrelated proteins. It is thus not
impossible, but highly speculative, that a large protein could incorporate more than half of its
sequence in the stacked flat b-sheets of a parallel b-helix like fibril. However, it should be stressed
that two short peptides, Ab (16–22) and the C-terminal Ab (34–42) form fibrils with an anti-
parallel b-sheet. Thus ‘‘similar’ fibrils can apparently form from either parallel or anti-parallel
sheet.

5. Conclusions

5.1. Evolutionary relationships

The parallel b-helix fold differs from many enzyme architectures in that it is fundamentally
simple. We can imagine that a gene could be created by copying many times the DNA coding for a
relatively short peptide. This is almost certainly the origin of the penta- and hexapeptide repeat
162 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

families of proteins and is the probable origin of all the repetitive folds. If the sequence has a high
enough propensity to form b-sheet, stacking of identical residues would favour formation of one
of the folds discussed above, especially as most other folds apparently do not form satisfactory
stacks. Similarly high concentrations of a peptide may form amyloid with related b-structures.
Once a parallel b-helix fold has arisen, we can expect it to shrink or expand by duplicating the
DNA coding for an entire coil. Again, if there was good stacking in the ancestor, the new coil can
be easily inserted. Thus neither the length of a parallel b-helix nor the distance between features
such as an N-terminal a-helix and an active site are ideal tests for homology. We can even
speculate without evidence that adding a further coil could be useful in binding a longer polymeric
substrate and might be especially likely in proteins binding polysaccharide as the rise per coil
( is comparable to the repeat in a polysaccharide. Clearly the expanded protein must
(4.8 A)
subsequently evolve to avoid folding with a coil omitted from the helix and this may be one of the
main roles of the N- and C-terminal extensions, which occur in most of the proteins which do not
form trimers. This will exert an evolutionary pressure to move away from perfect repetition.
The creation of new repetitive folds has occurred several times during evolution and thus there
is no reason to expect that all the proteins with these folds will be homologous. However, there are
some cases where proteins have evolved from a common ancestor (and some more debatable
cases). The left-handed parallel b-helix or hexapeptide repeat family shows homology through (1)
close structural similarity; (2) conserved sequence pattern; (3) conserved trimer formation; (4)
active sites at similar positions; and (5) in most enzymes a conserved function as an acyl
transferase. There are clear sequence similarities enabling us to form seven families within the
class of right-handed parallel b-helix proteins with known L or kidney shaped structures. These
are (1) the extra-cellular pectate lyase family; (2) the polygalacturonases and rhamnogalactur-
onase A; (3) the pectin methylesterases; (4) pectate lyases homologous to PelL; (5) the phage P22
tailspike endorhamnosidases; (6) chondroitinase B; and (7) pertactin. The large number of
similarities shared by these families arise either from common ancestry or because they are
required for folding or function. These include (a) the a–L turns (especially at T2 but also at the
start of PB2); (b) the a-helix at the N-terminus of the parallel b-helix forming a coil with a strand
of PB2; (c) the a–R turn at the start of PB1; (d) the binding site for substrate is formed by the T3–
PB1–T1 region and the polysaccharide substrates bind almost parallel to the parallel b-helix; (e)
the long T3 loops tend to be at the N-terminal end of the parallel b-helix and the long T1 loops
tend to be at the C-terminal end of the parallel b-helix; (f) all the enzymes degrade
polysaccharides, which are closely related except for the chrondroitinase B substrate; and (g)
all the identified catalytic residues are carboxylates, arginines and lysines. Some of these
similarities are considered in detail below.
(a) T2 turns
The T2 turns of the parallel b-helix proteins have two unusual features: the angle between the b-
sheets and the detailed form of the turn as discussed in Section 1.2.4. During 1997 a survey of a
large non-redundant database (Oliva et al., 1997) found no other examples of this type of turn
between b-sheets although a–R turns through similar angles between sheets can be found (for
example residue 107 of concanavalin A (Deacon et al., 1997)). The subsequent occurrence of
similar turns in the structure of the insulin-like growth factor receptor (Garrett et al., 1998)
suggests that this type of turn may arise by convergent evolution (unless the insulin-like growth
factor receptor is homologous to the pectinases). The only known a–R turns with an external side
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 163

chain in a parallel b-helix are the stack of four in pertactin at T3. However, there is also the
possibility that the sheets approach as in the spiral folds requiring a more complex turn. As they
can stack with or without hydrogen bonds along the helix axis, the a–L (gbE) turns may be an
optimal solution to the problem of turning a b-sheet through 801 and will probably occur in far
more than a quarter of independent folds containing two sheets meeting approximately at right
angles. Thus their occurrence only weakly supports the homology of the right-handed parallel
b-helices.
(b) The N-terminal a-helix
The N-terminal a-helix is not required in a right-handed parallel b-helix fold as it does not
occur in pertactin. However, it does solve the problem of capping the parallel b-helix. It may also
be easier to fit it at the N-terminus rather than insert it after a PB1 strand in place of PB2. Again
the leucine-rich repeat structures show that this type of structure can evolve independently, so that
its occurrence only slightly increases our confidence that all the families except pertactin are
homologous.
(c) The substrate binding site
The a–R turn at the start of PB1, the substrate binding sites and the position of long T3 and T1
loops are not completely independent. The a–R turn makes a groove which forms the centre of the
active site. The T3 and T1 loops assist in constructing the active site and as PB1 has a twist, it is
simpler to have the long T3 loops at the N-terminal end extending the twist of the b-sheet to form
a valley (and similarly with the T1 loops at the C-terminal end of PB1). The shape of the coils (or
in a case such as pertactin some of the coils) is sufficiently similar that they can be readily
superposed. This defines one groove as an obvious enzyme active site and binding of substrates or
products to this site has been observed directly in the extra-cellular pectate lyases, the tailspike
endorhamnosidase and in chrondroitinase B. Polygalacturonases judged by sequence conserva-
tion and the effects of site directed mutagenesis and pectin methylesterases simply from sequence
conservation probably use the same site. The pectate lyase PelL also has a conserved cluster of
residues at the same site. Finally, pertactin has no known enzymatic function and the equivalent
site is not readily accessible.
(f) The substrates
All the right-handed parallel b-helical enzymes degrade polysaccharide and four of the six
enzymes degrade pectin. In fact all four of the sequence families of pectin degrading enzymes with
known structures fold to parallel b-helices. The substrate for the tailspike endorhamnosidase
contains a-linked galactose and rhamnose suggesting some similarity to pectin. Only the b-linked
substrate of chrondroitinase B is less closely related and this substrate appears to bind to the same
grove but in the opposite direction. The similarity of the substrates combines with the other
similarities to argue strongly for the homology of all six enzyme families although the observed
binding of substrate to the pectate lyase PelC via calcium bridges warns us that the details of the
enzyme–polysaccharide interactions are not likely to be conserved.
The case of pertactin is open and will remain open until we understand how many different
shapes are possible for a parallel b-helix or find proteins clearly related to both pertactin and the
polysaccharide degrading enzymes. If pertactin does not have a common ancestor with pectin
methylesterase, its N-terminal coils represent an extreme example of convergent evolution.
However, the similarity of the turns at T3 to those of the left-handed family seems almost as
marked even if it is only local.
164 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

Assuming that right-handed parallel b-helix proteins are homologous to each other (note that
the case of the glutamate synthase domain cannot be discussed until the coordinates are released),
the repetitive structure and unusual relationships between the sheets joined by a–L turns may
make the homology unusually obvious. This is because a-helices or b-sheets forming a sandwich
can undergo relative movement during evolution (Chothia and Lesk, 1986) but the sheets in a
parallel b-helix are linked together so that the angle at the turn is determined by local packing
while changes in the size of a side chain in a single coil might require changes in neighbouring coils
if good hydrogen bonding is to be retained. In fact, it is clear from the studies on the tailspike
endorhamnosidase that quite large changes can be introduced by mutagenesis inside the parallel
b-helix without significant changes in the positions of the hydrogen bonded main-chain atoms
(Schuler et al., 2000).
More distant relationships between the right-handed parallel b-helix proteins and the other
folds are possible but do not seem well supported. There may be an evolutionary path from the
leucine-rich repeat proteins to the right-handed parallel b-helix enzymes but we do not need to
postulate one if stacking can easily generate new repetitive folds when DNA is duplicated. The
spiral folds are apparently unrelated to right-handed parallel b-helix enzymes because it is
impossible to simultaneous superpose the two sheets despite the angle between them being similar.

5.2. Future directions

The most obvious future developments in the structural biology of the parallel b-helix enzymes
will be the detailed characterisation of their substrate binding as this defines specificity, which in
turn should be at least partially understood if these enzymes are to be used in biotechnology. For
example, the properties of pectin both in plant tissue and in food depend, not only on its degree of
esterification, but also on the pattern of esterification. The patterns of esterification produced by
the various pectin methylesterases are different, which will only be understood once we know the
molecular details of substrate binding to pectin methylesterases. Comparison of the substrate
binding sites of the different enzyme families may also suggest how the righthanded parallel b-
helix enzymes have evolved from a common ancestor.
The second development will be the determination of further structures with parallel b-helix
folds, some of which will have no obvious sequence relationships to the known structures.
Experimental tests of the predicted structures of the pentapeptide repeat family (Bateman et al.,
1998) and the Pseudomonas syringae ice-nucleation protein (Graether and Jia, 2001) are eagerly
awaited but each new structure will help us to understand the fold and eventually to predict if a
sequence will fold to either a right- or left-handed parallel b-helix.
In addition to calculations using the packing of side chains to determine which sequences will
favour the different chiralities, it may be possible to test experimentally by protein engineering
whether a left-handed parallel b-helix can be converted into a right-handed fold. One approach
might be to replace the chiral isoleucines in a left-handed structure by leucines and methionines.
However, if an active enzyme is to be designed it will also be necessary to reverse the order of the
coils in the sequence and then search for mutants restoring folding and activity.
Structure prediction is clearly advancing rapidly using mostly using ‘‘threading’’ (Sippl and
Flockner, 1996; Jones et al., 1999). The increased number of known structures allows the use of
each as a search model but this approach might not recognise new folds which formed parallel b-
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 165

helices such as is suggested for the pentapeptide repeats. However, it will certainly now be possible
to use the short leucine-rich repeat structures as alternatives to the parallel b-helix proteins as
models for proteins with 20–24 residues per coil. It might also be possible to use our current and
developing understanding of both folds as a screen after a possible model has been built. For
example, examination of the known structures suggests that any parallel b-helix model without
extensive stacking of residues is likely to be wrong.
The prediction of new repetitive folds is likely to advance by incorporating more explicitly
the repetitive nature of the parallel b-helix. Examination of the recently determined
structures of several right-handed parallel b-helix proteins without obvious sequence
homology suggests that no single sequence profile is likely to find all these structures. For
example amongst the pectinases, aromatic stacks are found on PB3 in the pectate and pectin
lyases, PB2 in the esterases and PB1 in PelL while the hydrolases have few aromatic stacks.
Thus one possibility is to search for the occurrence of stacks of similar residues in
successive coils rather than for a single pattern. The variable lengths of the turn regions has
so far hindered the development of such prediction methods and the observation by
Bateman et al. (1998) that there is no obvious periodic signal in the sequences of the
left-handed parallel b-helix family corresponding to a coil also shows the problems
needing to be overcome. However, the identification of the antifreeze protein from
Tenebrid molitor as b-helical (Liou et al., 1999) and the recent identification of two
pectinases as b-helical by automatic servers shows that prediction of b-helical structures from
the sequence is frequently possible.

Acknowledgements

We thank Alex Bateman, Alex Drake, Curtis Johnson, Peter Lansbury and Bill Turnell for
helpful discussions, Paul Emsley and Neil Isaacs, Fran Jurnak, Leping Li, Lee Darden, Bill Kay
and Peter Davies for sending coordinates before their general release, and Bostran Kobe, Mirek
Cygler, Peter Lansbury and Jean-Christophe Rochet for preprints of their articles. The figures
above were made with the programs MOLSCRIPT (Kraulis, 1991) or MOLMOL (Koradi et al.,
1996). This work was funded by the BBSRC (UK) and EU projects AIR2-CT94-1345 and BIO4-
CT96-0685.

References

Akita, M., Suzuki, A., Kobayashi, T., Ito, S., Yamane, T., 2000. Crystallization and preliminary X-ray analysis of high-
alkaline pectate lyase. Acta Crystallogr. D56, 749–750.
Alexandrescu, A.T., Rathgeb-Szabo, K., 1999. An NMR investigation of solution aggregation reactions preceding the
misassembly of acid-denatured cold shock protein A into fibrils. J. Mol. Biol. 291, 1191–1206.
Anderson, T.S., Hellgeth, J., Lansbury, P.T., 1996. Isotope-edited infrared linear dichroism: determination of amide
orientational relationships. J. Am. Chem. Soc. 118, 6540–6546.
Antzutkin, O.N., Balbach, J.J., Leapman, R.D., Rizzo, N.W., Reed, J., Tycko, R., 2000. Multiple quantum solid state
NMR indicates a parallel, not antiparallel, organization of b-sheets in Alzheimer’s b-amyloid fibrils. Proc. NatI.
Acad. Sci. 97, 13045.
166 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

Armand, S., Wagmaker, W.J.M., S!anchez-Torres, P., Kester, H.C.M., van Santen, Y., Dijkstra, B.W., Visser, J.,
Benen, J.A.E., 2000. The active site topology of Aspergillus niger endopolygalacturonase II as studied by site-
directed mutagenesis. J. Biol. Chem. 275, 691–696.
Balbach, J.J., Ishii, Y., Antzutkin, O.N., Leapman, R.D., Rizzo, N.W., Dyda, F., Reed, J., Tycko, R., 2000. Amyloid
fibril formation by Ab (16–22), a seven-residue fragment of the Alzheimer’s b-amyloid peptide, and structural
characterization by solid state NMR. Biochemistry 39 (45), 13748–13759.
Bateman, A., Murzin, A.G., Teichmann, S.A., 1998. Structure and distribution of pentapeptide repeats in bacteria.
Protein Sci. 7, 1477–1480.
Baumann, U., 1994. Crystal structure of the 50 kDa metalloprotease from serratia marcescens. J. Mol. Biol. 242,
244–251.
Baumann, U., Wu, S., Flaherty, K.M., McKay, D.B., 1993. Three-dimensional structure of the alkaline
protease of Pseudomonas aeroginosa: a two-domain protein with a calcium binding b-roll motif. EMBO J. 12,
3357–3364.
Baxa, U., Steinbacher, S., Miller, S., Weintraub, A., Huber, R., Seckler, R., 1996. Interactions of phage P22 tails with
their cellular receptor, salmonella o-antigen polysaccharide. Biophys. J. 71 (4), 2040–2048.
Baxa, U., Steinbacher, S., Weintraub, A., Huber, R., Seckler, R., 1999. Mutations improving the folding of phage P22
tailspike protein affect its receptor binding. J. Mol. Biol. 293, 693.
Beaman, T.W., Binder, D.A., Blanchard, J.S., Roderick, S.L., 1997. Three-dimensional structure of tetrahydrodipi-
colinate N-succinyltransferase. Biochemistry 36, 489–494.
Beaman, T.W., Blanchard, J.S., Roderick, S.L., 1998a. The conformational change and active site structure of
tetrahydrodipicolinate N-succinyltransferase. Biochemistry 37, 10363–10369.
Beaman, T.W., Sugantino, M., Roderick, S.L., 1998b. Structure of the hexapeptide xenobiotic acetyltransferase from
Pseudomonas aeruginosa. Biochemistry 37, 6689–6696.
Beissinger, M., Lee, S.C., Steinbacher, S., Reinemer, P., Huber, R., Yu, M.H., Seckler, R., 1995. Mutations that
stabilize folding intermediates of phage P22 tailspike proteinFfolding in-vivo and in-vitro, stability, and structural
context. J. Mol. Biol. 249, 185–194.
Benning, M.M., Taylor, K.L., Liu, R.-Q., Yang, H., Wesenberg, G., Dunaway-Mariano, D., Holden, H.M., 1996.
Structure of 4-chlorobenzoyl coenzyme A dehalogenase determined to 1.8 A ( resolution: an enzyme catalyst
generated via adaptive mutation. Biochemistry 35, 8103–8109.
Benzinger, T.L.S., Gregory, D.M., Burkoth, T.S., Miller-Auer, H., Lynn, D.G., Botto, R.E., Meredith, S.C., 1998.
Propagating structure of Alzheimer’s beta-amyloid (10–35) is parallel beta-sheet with residues in exact register. Proc.
Natl. Acad. Sci. USA 95, 13407–13412.
Betts, S., King, J., 1999. There’s a right way and a wrong way: in vivo and in vitro folding, misfolding and subunit
assembly of the P22 tailspike. Struct. Folding Des. 7, R131–139.
Betts, S., Haase- Pettingell, C., King, J., 1997. Mutational effects on inclusion body formation. Adv. Protein. Chem. 50,
243–264.
Biely, P., Benen, J., Heinrichova, K., Kester, H.C.M., Visser, J., 1996. Inversion of configuration during
hydrolysis of a-(1,4)-galacturonidic linkage by three Aspergillus polygalacturonases. FEBS Lett. 382,
249–255.
Binda, C., Bossi, R.T., Wakatsuki, S., Arzt, S., Coda, A., Curti, B., Vanoni, M.A., Mattevi, A., 2000. Cross-talk and
ammonia channelling between active centers in the unexpected domain arrangement of glutamate synthase.
Structure 8 (12), 1299–1308.
Blake, C., Serpell, L., 1996. Synchrotron X-ray studies suggest that the core of the transthyretin amyloid fibril is a
continuous beta-sheet helix. Structure 4, 989–998.
Blake, C.C.F., Serpell, L.C., Sunde, M., Sandgren, O., Lundgren, E., Kirschner, Pepys, Sipe, Westermark, Kelly,
Goldgaber, Merlini (1996) A molecular model of the amyloid fibril. CIBA Found. Symp. 199, 6–21.
Brown, K., Pompeo, F., Dixon, S., Mengin-Lecreuix, D., Cambillau, C., Bourne, Y., 1999. Crystal structure of the
bifunctional N-acetylglucosamine 1-phosphate uridyltransferase from Escherichia coli: a paradigm for the related
pyrophosphorylase superfamily. EMBO J. 18, 4096–4107.
Buchanan, S.G.S., Gay, N.J., 1996. Structural and functional diversity in the leucine rich repeat family of proteins.
Prog. Biophys. Mol. Biol. 65, 1–44.
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 167

Bugiani, O., Padovani, A., Magoni, M., Andora, G., Sgarzi, M., Saoiardo, M., Bizzi, A., Giaccione, G., Rossi, G.,
Tagliavini, F., 1998. Neurobiol. Aging 19, S238.
Burkoth, T.S., Benzinger, T.L.S., Jones, D.N.M., Hallenga, K., Meredith, S.C., Lynn, D.G., 1998. C-terminal peg
blocks the irreversible step in b-amyloid (10–35) fibrillogenesis. J. Am. Chem. Soc. 120, 7655–7656.
Burkoth, T.S., Benzinger, T.L.S., Volker, U., Lynn, D.G., Meredith, S.C., Thiyagarajan, P., 1999. Self-assembly of Ab
(10–35)-PEG copolymer fibrils. J. Am. Chem. Soc. 121, 7429–7430.
Burkoth, T.S., Benzinger, T.L.S., Volker, U., Lynn, D.G., Meredith, S.C., Thiyagarajan, P., 2000. Structure of the b-
amyloid (10–35) fibril. J. Am. Chem. Soc. 122, 7883–7889.
Chan, A.W.E., Hutchinson, E.G., Harris, D., Thornton, J.M., 1993. Identification, classification, and analysis of b-
bulges in proteins. Protein Sci. 2, 1574–1590.
Chavagnat, F., Duez, C., Guinand, M., Potin, P., Barbeyron, T., Henrissat, B., Wallach, J., Ghuysen, J.-M., 1996.
Cloning, sequencing and overexpression in Escherichia coli of the alginatelyase-encoding aly gene of Pseudomonas
alginovora: identification of three classes of alginate lyases. Biochem. J. 319, 575–583.
Chiti, F., Webster, P., Taddei, N., Clark, A., Stefani, M., Ramponi, G., Dobson, C.M., 1999. Designing conditions for
in vitro formation of amyloid protofilaments and fibrils. Proc. Natl. Acad. Sci. USA 96, 3590–3594.
Chothia, C., Lesk, A.M., 1986. The relationship between the divergence of sequence and structure in proteins. EMBO.
J. 5, 823–826.
Chothia, C., Murzin, A.G., 1993. New folds for all-b proteins. Structure 1, 217–222.
Chothia, C., Hubbard, T., Brenner, S., Barns, H., Murzin, A., 1997. Protein folds in the all-b and all-a classes. Annu.
Rev. Biomol. Struct. 26, 597–627.
Collinson, S.K., Parker, J.M.R., Hodges, R.S., Kay, W.W., 1999. Structural predictions of AgfA, the insoluble fimbrial
subunit of Salmonella thin aggregative fimbriae. J. Mol. Biol. 290 (3), 741–756.
Costa, P.R., Kocisko, D.A., Sun, B.Q., Lansbury, P.T., Griffin, R.G., 1997. Determination of peptide amide
configuration in a model amyloid fibril by solid-state NMR. J. Am. Chem. Soc. 119, 10487–10493.
Danner, M., Fuchs, A., Miller, S., Seckler, R., 1993. Folding and assembly of phage P22 tailspike endorhamnosidase
lacking the N-terminal, head-binding domain. Eur. J. Biochem. 215 (3), 653–661.
Deacon, A., Gleichmann, T., Kalb, A.J., Price, H., Raftery, J., Bradbrook, G., Yariv, J., Helliwell, J.R., 1997. The
structure of concanavalin A and its bound solvent determined with small-molecule accuracy at 0.94 A ( resolution. J.
Chem. Soc. Faraday Trans. 93, 4305–4312.
Doan, C.N., Caughron, M.K., Myers, J.C., Breakfield, N.W, Oliver, R.L., Yoder, M.D., 2000. Purification,
crystallization and X-ray analysis of crystals of pectate lyase A from Erwinia chrysanthemi. Acta Crystallogr. D56,
351–353.
Dobson, C.M., 1999. Protein misfolding, evolution and disease. Trends Biochem. Sci. 24, 329–332.
El Agnaf, O.M.A., Guthrie, D.J.S., Walsh, D.M., Irvine, G.B., 1998. The influence of the central region containing
residues 19–25 on the aggregation properties and secondary structure of Alzheimer’s beta-amyloid peptide. Eur. J.
Biochem. 256, 560–569.
Emsley, P., Charles, I.G., Fairweather, N.F., Isaacs, N.W., 1996. Structure of Bordetella pertussis virulence factor P.69
pertactin. Nature 381, 90–92.
Engel, C.K., Mathieu, M., Zeelen, J.P., Hiltunen, J.K., Wierenga, R.K., 1996. Crystal structure of enoyl-coenzyme A
(CoA) hydratase at 2.5 A ( resolution: a spiral fold defines the CoA-binding pocket. EMBO J. 15, 5135–5145.
Engel, C.K., Kiema, T.R., Hiltunen, J.K., Wierenga, R.X., 1998. The crystal structure of enoyl-CoA hydratase
complexed with octanoyl-CoA reveals the structural adaptations required for binding of a long chain fatty acid-CoA
molecule. J. Mol. Biol. 275, 847.
Esler, W.P., Stimson, E.R., Ghilardi, J.R., Lu, Y.-A., Felix, A.M., Vinters, H.V., Mantyh, P.W., Lee, J.P., Maggio,
J.E., 1996a. Point substitution in the central hydrophobic cluster of a human b-amyloid congener disrupts peptide
folding and abolishes plaque competence. Biochemistry 35, 13914–13921.
Esler, W.P., Stimson, E.R., Ghilardi, J.R., Vinters, H.V., Lee, J.P., Mantyh, P.W., Maggio, J.E., 1996b. In vitro growth
of Alzheimer’s disease beta-amyloid plaques displays first-order kinetics. Biochemistry 35, 749–757.
Esler, W.P., Stimson, E.R., Jennings, J.M., Vinters, H.V., Ghilardi, J.R., Lee, J.P., Mantyh, P.W., Maggio, J.E., 2000.
Alzheimer’s disease amyloid propagation by a template-dependent dock–lock mechanism. Biochemistry 39 (21),
6288–6295.
168 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

Everest, P., Li, J., Douce, G., Charles, I., De Azavedo, J., Chatfield, S., Dougan, G., Roberts, M., 1996. Role of the
Bordetella pertussis P.69/pertactin protein and the P.69/pertactin RGD motif in the adherence to and invasion of
mammalian cells. Microbiology 142, 3261–3268.
Federici, L., Mattei, B., Caprari, C., Savino, C., Cervone, F., Tsernoglou, D., 1999. Crystallization and preliminary
X-ray diffraction study of the endopolygalacturonase from Fusarium moniliforme. Acta Crystallogr. 55 (Pt 7),
1359–1361.
Finnie, C., Zorreguieta, A., Hartley, N.M., Downie, J.A., 1998. Characterization of Rhizobium leguminosarum
exopolysaccharide glycanases that are secreted via a type I exporter and have a novel heptapeptide repeat motif.
J. Bacteriol. 180, 1691–1699.
Foguel, D., Robinson, C.R., deSousa, P.C., Silva, J.L., Robinson, A.S., 1999. Hydrostatic pressure rescues native
protein from aggregates biotechnol. Bioeng. 63, 552–558.
Fraser, P.E., McLachlan, D.R., Surewicz, W.K., Mizzen, C.A., Snow, A.D., Nguyen, J.T., Kirschner, D.A., 1994.
Conformation and fibrillogenesis of Alzheimer A-beta peptides with selected substitution of charged residues.
J. Mol. Biol. 244, 64–73.
. V., Jones, D.T, 1999. b-Propellers: structural rigidity and functional diversity. Curr. Opinion Struct. Biol. 9,
. op,
Full
715–721.
Gacesa, P., 1987. Alginate-modifying enzymesFa proposed unified mechanism of action for the lyases and epimerases.
FEBS Lett. 212, 199–202.
Garrett, T.P.J., McKern, N.M., Lou, M.Z., Frenkel, M.J., Bentley, J.D., Lovrecz, G.O., Elleman, T.C., Cosgrove, L.J.,
Ward, C.W., 1998. Crystal structure of the first three domains of the type-1 insulin-like growth factor receptor.
Nature 394, 395–399.
Gay, N.J., Packman, L.C., Weldon, M.A., Barna, J.C.J., 1991. A leucinerich repeat peptide derived from the Drosophila
toll receptor forms extended filaments with a b-sheet structure. FEBS Lett. 291, 87–91.
Gerlt, J.A., Gassman, P.G., 1992. Understanding enzyme-catalyzed proton abstraction from carbon acidsFdetails of
stepwise mechanisms for b-elimination reactions. J. Am. Chem. Soc. 114, 5928–5934.
Goldenberg, D., King, J., 1982. Trimeric intermediates in the in vivo folding and subunit assembly pathways of the
tailspike endorhamnosidase of bacteriophage P22. Proc. Natl. Acad. Sci. USA 79, 3403–3407.
Graether, S.P., Jia, Z.C., 2001. Modeling Pseudomonas syringae icenucleation protein as a b-helical protein. Biophys.
J. 80 (3), 1169–1173.
Graether, S.P., Kuiper, M.J., Gagn!e, S.M., Walker, V.K., Jia, Z., Sykes, B.D., Davies, P.L., 2000. b-helix structure and
ice-binding properties of a hyperactive antifreeze protein from an insect. Nature 406, 325–328.
Gregory, D.M., Mitchell, D.J., Stringer, J.A., Kiihne, S., Shiels, J.C., Callahan, J., Mehta, M.A., Drobny, G.P., 1995.
Windowless dipolar recouplingFthe detection of weak dipolar couplings between spin-1/2 nuclei with large
chemical-shift anisotropies. Chem. Phys. Lett. 246, 654–663.
Gregory, D.M., Benzinger, T.L.S., Burkoth, T.S., Miller-Auer, H., Lynn, D.G., Meredith, S.C., Botto, R.E., 1998.
Dipolar recoupling NMR of biomolecular self-assemblies: determining inter- and intrastrand distances in fibrilized
Alzheimer’s beta-amyloid peptide. Solid State Nucl. Magn. Reson. 13, 149–166.
Griffiths, J.M., Ashburn, T.T., Auger, M., Costa, P.R., Griffin, R.G., Lansbury, P.T., 1995. Rotational resonance solid-
state NMR elucidates a structural model of pancreatic amyloid. J. Am. Chem. Soc. 117, 3539–3546.
Gross, M., Wilkins, D.K., Pitkeathly, M.C., Chung, E.W., Higham, C., Clark, A., Dobson, C.M., 1999.
Formation of amyloid fibrils by peptides derived from the bacterial cold shock protein cspB. Protein Sci. 8,
1350–1357.
Guijarro, J.I., Sunde, M., Jones, J.A., Campbell, I.D., Dobson, C.M., 1998. Amyloid fibril formation by an SH3
domain. Proc. Natl. Acad. Sci. USA 95, 4224–4228.
Guthrie, J.P., Kluger, R., 1993. Electrostatic stabilization can explain the unexpected acidity of carbon acids in enzyme-
catalyzed reactions. J. Am. Chem. Soc. 115, 11569–11572.
Haase-Pettingell, C., King, J., 1997. Prevalence of temperature sensitive folding mutations in the parallel beta coil
domain of the phage P22 tailspike endorhamnosidase. J. Mol. Biol. 267 (1), 88–102.
Hamada, K., Hata, Y., Katsuya, Y., Hiramatsu, H., Fujiwara, T., Katsube, Y., 1996. Crystal structure of serratia
protease, a zinc-dependent proteinase from Serratia sp. E-15 containing A beta-sheet coil motif at 2.0 angstrom
resolution. J. Biochem. 119, 844–851.
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 169

Harper, J.D., Wong, S.S., Lieber, C.M., Lansbury, P.T., 1997. Observation of metastable A beta amyloid protofibrils
by atomic force microscopy. Chem. Biol. 4, 119–125.
Harper, J.D., Wong, S.S., Lieber, C.M., Lansbury, P.T., 1999. Assembly of A beta amyloid protofibrils: an in vitro
model for a possible early event in Alzheimer’s disease. Biochemistry 38, 8972–8980.
Heffron, S., Moe, G.R., Sieber, V., Mengaud, J., Cossart, P., Vitali, J., Jurnak, F., 1998. Sequence profile of the parallel
b-helix in the pectate lyase superfamily. J. Struct. Biol. 122 (1–2), 223–235.
Henrissat, B., Bairoch, A., 1996. Updating the sequence-based classification of glycosyl hydrolases. Biochem. J. 316 (Pt
2), 695–696.
Henrissat, B., Heffron, S.E., Yoder, M.D., Lietzke, S.E., Jurnak, F., 1995. Functional implications of
structure-based sequence alignment of proteins in the extracellular pectate lyase superfamily. Plant Physiol. 107,
963–976.
Herron, S.R., Benen, J.A.E., Scavetta, R.D., Visser, J., Jurnak, F., 2000. Structure and function of pectic enzymes:
virulence factors of plant pathogens. Proc. Natl. Acad. Sci. USA 97, 8762–8769.
Hillig, R.C., Renault, L., Vetter, I.R., Drell, T., Wittinghofer, A., Becker, J., 1999. The crystal structure of rna1p: a new
fold for a GTPase-activating protein. Mol. Cell 3, 781–791.
Hindson, V.J., Moody, P.C.E., Rowe, A.J., Shaw, W.V., 2000. Serine acetyltransferase from E. coli is a dimer of
trimers. J. Biol. Chem. 275 (1), 461–466.
Huang, W., Matte, A., Li, Y., Kim, Y.S., Linhardt, R.J., Su, H., Cygler, M., 1999. Crystal structure of chondroitinase
B from Flavobacterium heperinum and its complex with a disaccharide product at 1.7 A ( resolution. J. Mol. Biol. 294,
1257–1269.
Hunter, C.A., Singh, J., Thornton, J.M., 1991. p–p interactionsFthe geometry and energetics of phenylalanine
phenylalanine interactions in proteins. J. Mol. Biol. 218, 837–846.
Hutchinson, E.G., Thornton, J.M., 1996. PROMOTIFFa program to identify and analyze structural motifs in
proteins. Protein Sci. 5, 212–220.
Ionescu-Zanetti, C., Khurana, R., Gillespie, J.R., Petrick, J.S., Trabachino, L.C., Minert, L.J., Carter, S.A., Fink, A.L.,
1999. Monitoring the assembly of lg light-chain amyloid fibrils by atomic force microscopy. Proc. Natl. Acad. Sci.
USA 96, 13175–13179.
Isupov, M.N., Fleming, T.M., Dalby, A.R., Crowhurst, G.S., Bourne, P.C., Littlechild, J.A., 1999. Crystal structure of
the glyceraldehyde-3-phosphate dehydrogenase from the hyperthermophilic archaeon Sulfolobus solfataricus. J.
Mol. Biol. 291, 651–660.
Jarrett, J.T., Lansbury, P.T., 1993. Seeding one-dimensional crystallization of amyloidFa pathogenic mechanism in
Alzheimers-disease and scrapie. Cell 73, 1055–1058.
Jenkins, J., Mayans, O., Pickersgill, R., 1998. The structure and evolution of parallel b-helix proteins. J. Struct. Biol.
122, 236–246.
Jenkins, J., Mayans, O., Smith, D., Worboys, K., Pickersgill, R., 2001. Three-dimensional structure of Erwinia
chrysanthemi pectin methylesterase reveals a novel esterase active site. J. Mol. Biol. 305, 951–960.
Jimenez, J.L., Guijarro, J.L., Orlova, E., Zurdo, J., Dobson, C.M., Sunde, M., Saibil, H.R., 1999. Cryo-electron
microscopy structure of an SH3 amyloid fibril and model of the molecular packing. EMBO J. 18, 815–821.
Johnson, W.C., 1999. Analyzing protein circular dichroism spectra for accurate secondary structures. Proteins-
Structure Function Genetics 35 (3), 307–312.
Jones, D.T., Tress, M., Bryson, K., Hadley, C., 1999. Successful recognition of protein folds using threading
methods biased by sequence similarity and predicted secondary structure. Proteins–Struct. Function Genet. 35,
104–111.
Jones, T.A., Zou, J.-T., Cowan, S.W., Kjeldgaard, M., 1991. Improved methods for building protein models in
electron-density maps and the location of errors in these models. Acta Crystallogr. A47, 110–119.
Jurnak, F., Yoder, M.D., Pickersgill, R., Jenkins, J., 1994. Parallel b-domains: a new fold in protein structures. Curr.
Opin. Struct. Biol. 4, 802–806.
Jurnak, F., Kita, N., Garrett, M., Heffron, S.E., Scavetta, R., Boyd, C., Keen, N., 1996. Functional implications of the
three-dimensional structures of pectatelyases. In: Visser, J., Voragen, A.G.J. (Eds.), Pectins & Pectinases. Elsevier,
Amsterdam, pp. 295–308.
Kajava, A.N., 1998. Structural diversity of leucine-rich repeat proteins. J. Mol. Biol. 277, 519–527.
170 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

Kallberg, Y., Gustafsson, M., Persson, B., Thyberg, J., Johansson, J., 2001. Prediction of amyloid fibril-forming
proteins. J. Biol. Chem. 276 (16), 12945–12950.
Kayed, R., Bernhagen, J., Greenfield, N., Sweimeh, K., Brunner, H., Voelter, W., Kapurniotu, A., 1999.
Conformational transitions of islet amyloid polypeptide (IAPP) in amyloid formation in vitro. J. Mol. Biol. 287,
781–796.
Kamen, D.E., Griko, Y., Woody, R.W., 2000. The stability, structural organization, and denaturation of pectatelyase
C, a parallel b-helix protein. Biochemistry 39 (51), 15932–15943.
Kelly, J.W., 1998. The alternative conformations of amyloidogenic proteins and their multi-step assembly pathways.
Curr. Opin. Struct. Biol. 8, 101–106.
Khurana, R., Fink, A.L., 2000. Do parallel b-helix proteins have a unique Fourier transform infrared spectrum?
Biophys. J. 78, 994–1000.
Khurana, R., Uversky, V.N., Nielsen, L., Fink, A.L., 2001. Is Congo Red an amyloid-specific dye? J. Biol. Chem. 276,
22715–22721.
King, J., Haase-Pettingell, C., Robinson, A.S., Speed, M., Mitraki, A., 1996. Thermolabile folding intermediates:
inclusion body precursors and chaperonin substrates. FASEB J. 10 (1), 57–66.
Kisker, C., Schindelin, H., Alber, B.E., Ferry, J.G., Rees, D.C., 1996. A left-handed b-helix revealed by
the crystal structure of a carbonic anhydrase from the archaeon methanosarcina thermophila. EMBO J. 15,
2323–2330.
Kita, N., Boyd, C.M., Garrett, M.R., Jurnak, F., Keen, N.T., 1996. Differential effect of site-directed mutations in Pelc
on pectatelyase activity, plant-tissue maceration, and elicitor activity. J. Biol. Chem. 271, 26529–26535.
Kobe, B., 1996. Leucines on a roll. Nature Struct. Biol. 3, 977–980.
Kobe, B., Deisenhofer, J., 1993. Crystal structure of porcine ribonuclease inhibitor, a protein with leucine-rich repeats.
Nature 366, 751–756.
Kobe, B., Deisenhofer, J., 1995a. A structural basis of the interactions between leucine-rich repeats and protein ligands.
Nature 374, 183–186.
Kobe, B., Deisenhofer, J., 1995b. Proteins with leucine-rich repeats. Curr. Opin. Struct. Biol. 5, 409–416.
Kobe, B., Kajava, A.V., 2000. When protein folding is simplified to protein coiling: the continuum of solenoid protein
structures. TIBS 25, 509–515.
Kobe, B., Gleichmann, T., Horne, J., Jennings, I.G., Scotney, P.D., The, T., 1999. Turn up the HEAT. Structure 7,
R91–R97.
.
Koradi, R., Billeter, M., Wuthrich, K., 1996. MOLMOL: a program for display and analysis of macromolecular
structures. J Mol. Graphics 14, 51–55.
Kostrewa, D., D’Arcy, A., Takacs, B., Kamber, N., 2001. Crystal structures of Streptococcus pneumoniae
N-acetylglucosamine-1-phosphate uridyltransfe rase, GlmU, in apo form at 2.33 A ( resolution and in complex with
2+ (
UDP-N-acetylglucosamine and mg at 1.96 A resolution. J. Mol. Biol. 305 (2), 279–289.
Kraulis, P.J., 1991. MOLSCRIPT; a program to produce both detailed and schematic plots of proteins. J. Appl.
Crystallogr. 24, 946–950.
Kreisberg, Kreisberg, J.F., Betts, S.D., King, J., 2000. b-helix core packing within the triple-stranded oligomerization
domain of the P22 tailspike. Protein Sci. 9 (12), 2338–2343.
Kusumoto, Y., Lomakin, A., Teplow, D.B., Benedek, G.B., 1998. Temperature dependence of amyloid beta-protein
fibrillization. Proc. Natl. Acad. Sci. USA 95 (21), 12277–12282.
Langs, D.A., 1988. Three-dimensional structure at 0.86 A ( of the uncomplexed form of the transmembrane ion channel
peptide gramicidin A. Science 241, 188–191.
Lansbury, P.T., 1999. Evolution of amyloid: what normal protein folding may tell us about fibrillogenesis and disease.
Proc. Natl. Acad. Sci. USA 96, 3342–3344.
Lansbury, P.T., Costa, P.R., Griffiths, J.M., Simon, E.J., Auger, M., Halverson, K.J., Kocisko, D.A., Hendsch, Z.S.,
Ashburn, T.T., Spencer, R.G.S., Tidor, B., Griffin, 1995. Structural model for the beta-amyloid fibril based
on interstrand alignment of an antiparallel-sheet comprising a C-terminal peptide. Nature Struct. Biol. 2,
990–998.
Laskowski, R.A., MacArthur, M.W., Moss, D.S., Thornton, J.M., 1993. PROCHECKFa program to check the
stereochemical quality of protein structures. J. Appl. Crystallogr. 26, 283–291.
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 171

Laurent, F., Kotoujansky, A., Labesse, G., Bertheau, Y., 1993. Characterization and overexpression of the pern gene
encoding pectin methylesterase of Erwinia chrysanthemi strain-3937. Gene 131, 17–25.
Lazo, N.D., Downing, D.T., 1997. b-helical fibrils from a model peptide. Biochem. Biophys. Res. Commun. 235,
675–679.
Lazo, N.D., Downing, D.T., 1998. Amyloid fibrils may be assembled from b-helical protofibrils. Biochemistry 37,
1731–1735.
Lazo, N.D., Downing, D.T., 1999. Fibril formation by amyloid-beta proteins may involve b-helical protofibrils. J.
Peptide Res. 53, 633–640.
Leckie, F., Mattei, B., Capodicasa, C., Hemmings, A., Nuss, L., Aracri, B., DeLorenzo, G., Cervone, F., 1999. The
specificity of polygalacturonase-inhibiting protein (PGIP): a single amino acid substitution in the solvent-exposed
beta-strand/beta-turn region of the leucine-rich repeats (LRRs) confers a new recognition capability. EMBO J. 18,
2352–2363.
Lee, J.P., Stimson, E.R., Ghilardi, J.R., Mantyh, P.W., Lu, Y.-A., Felix, A.M., Llanos, W., Behbin, A., Cummings, M.,
Van Criekinge, M., Timms, W., Maggio, J.E., 1995. 1H NMR of Ab amyloid peptide congeners in water solution.
Conformation changes correlate with plaque competence. Biochemistry 34, 5191–5200.
Leech, A., Mattei, B., Federici, L., De Lorenzo, G., Hemmings, A.M., 2000. Preliminary X-ray crystallographic
analysis of a plant defence protein, the polygalacturonase inhibiting protein from Phaseolus vulgaris. Acta
Crystallogr. D56, 98–100.
Leininger, E., Roberts, M., Kenimer, F.G., Charles, I.G., Fairweather, N., Novotny, P., Brennan, M.J., 1991.
Pertactin, an Arg–Gly–Asp-containing Bordetella pertussis surface protein that promotes adherence of mammalian
cells. Proc. Natl. Acad. Sci. USA 88, 345–349.
Li, L., Darden, T.A., Bartolotti, L., Kominos, D., Pedersen, L.G., 1999. An atomic model for the pleated b-sheet
structure of Ab amyloid protofilaments. Biophys. J. 76, 2871–2878.
Lietzke, S.E., Yoder, M.D., Keen, N.T., Jurnak, F., 1994. The three-dimensional structure of pectate lyase-E, a plant
virulence factor from Erwinia chrysanthemi. Plant Physiol. 106, 849–862.
Lietzke, S.E., Scavetta, R.D., Yoder, M.D., Jurnak, F., 1996. The refined three-dimensional structure of pectate lyase E
from Erwinia chrysanthemi at 2.2 A ( resolution. Plant Physiol. 111, 73–92.
Liker, E., Fernandez, E., Izaurralde, E., Conti, E. (2000) The structure of the mRNA export factor TAP reveals a cis
arrangement of a non-canonical RNP domain and an LRR domain. EMBO J. 19, 21, 5587–5598.
Lilie, H., Haehnel, W., Rudolph, R., Baumann, U., 2000. Folding of a synthetic parallel b-roll protein. FEBS Lett. 470,
173–177.
Liou, Y.C., Thibault, P., Walker, V.K., Davies, P.L., Graham, L.A., 1999. A complex family of highly heterogeneous
and internally repetitive hyperactive antifreeze proteins from the beetle Tenebrio molitor. Biochemistry 38,
11415–11424.
Liou, Y.-C., Tocilj, A., Davies, P.L., Jia, Z., 2000. Mimicry of ice structure by surface hydroxyls and water of a b-helix
antifreeze protein. Nature 406, 322–324.
Lojkowska, E., Masclaux, C., Boccara, M., Robert-Baudouy, J., Hugouvieux-Cotte-Pattat, N., 1995. Mol. Microbiol.
16, 1183–1195.
Lu, C.T., Tada, T., Nakamura, Y., Wada, K., Nishimura, K., Katsuya, Y., Sawada, M., Takao, M., Sakai, T., 2000.
Crystallization and preliminary X-ray analysis of endopolygalacturonase SE1 from Trichosporon penicillaturn. Acta
Crystallogr. D56, 1668–1669.
Maki, H., Atsutoshi, M., Fujiyama, K., Kinoshita, S., Yoshida, T., 1993. Cloning, sequence analysis and
expression in E. coli of a gene encoding an alginate lyase from Pseudomonas sp. OS-ALG-9. J. Gen. Microbiol.
139, 987–993.
Malinchik, S.B., Inouye, H., Szumowski, K.E., Kirschner, D.A., 1998. Structural analysis of Alzheimer’s beta(1–40)
amyloid: protofilament assembly of tubular fibrils. Biophys. J. 74 (1), 537–545.
Malissard, M., Duez, C., Guinand, M., Vacheron, M.-J., Michel, G., Marty, N., Joris, B., Thamn, I., Ghuysen, J.-M.,
1993. Sequence of a gene encoding a (poly ManA) alginate lyase active on Pseudomonas aeruginosa alginate. FEMS
Microbiol. Lett. 110, 101–106.
Marino, M., Braun, L., Cossart, P., Ghosh, P., 1999. Structure of the InIB Leucine-rich repeats, a domain that triggers
host cell invasion by the bacterial pathogen L. monocytogenes. Mol. Cell 4, 1063–1072.
172 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

Marino, M., Braun, L., Cossart, P., Ghosh, P., 2000. A framework for interpreting the leucine-rich repeats of the
Listeria internalins. Proc. Natl. Acad. Sci. USA 97, 8784–8788.
Marino-Buslje, C., Martin-Martinez, M., Mizuguchi, K., Siddle, K., Blundell, T.L., 1999. The insulin receptor: from
protein sequence to structure. Biochem. Soc. Trans. 27, 715–726.
Markovic, O., Jornvall, H., 1992. Disulfide bridges in tomato pectinesteraseFvariations from pectinesterases of other
speciesFconservation of possible active-site segments. Protein Sci. 1, 1288–1292.
Mayans, O., Scott, M., Connerton, I., Gravesen, T., Benen, J., Visser, J., Pickersgill, R., Jenkins, J., 1997. Two crystal
structures of pectinlyase A from Aspergillus niger reveal a pH driven conformational change and striking divergence
in the substrate-binding clefts of pectin and pectate lyases. Structure 5, 677–689.
Miller, S., 1995. Deletions- und Punktmutanten des P22-Tailspike-Proteins: Klonierung, Reinigung, Kristallisation und
Physikalisch-Biochernische Analyse, Thesis, University of Regensburg
Miller, S., Schuler, B., Seckler, R., 1998a. Phage P22 tailspike protein: removal of head-binding domain unmasks effects
of folding mutations on native-state thermal stability. Protein Sci. 7 (10), 2223–2232.
Miller, S., Schuler, B., Seckler, R., 1998b. A reversibly unfolding fragment of P22 tailspike protein with native
structure: the isolated b-helix domain. Biochemistry 37 (25), 9160–9168.
Miyatake, H., Hata, Y., Fujii, T., Hamada, K., Morihara, K., Katsube, Y., 1995. Crystal-structure of the unliganded
alkaline protease from Pseudomonas aeruginosa IFO3080 and its conformational-changes on ligand-binding.
J. Biochem. 118, 474–479.
Moriarty, D.F, Raleigh, D.P., 1999. Effects of sequential proline substitutions on amyloid formation by human amylin
(20–29). Biochemistry 38, 1811–1818.
Oliva, B., Bates, P.A., Querol, E., Aviles, F.X., Sternberg, M.J.E., 1997. An automated classification of the structure of
protein loops. J. Mol. Biol. 266, 814–830.
Olsen, L.R., Roderick, S.L., 2001. Structure of the Escherichia coli GlmU pyrophosphorylase and acetyltransferase
active sites. Biochemistry 40(7), 1913–1921.
Pag"es, S., Heijne, W.H.M., Kester, H.C.M., Visser, J., Benen, J.A.E., 2000. Subsite mapping of Aspergillus niger
endopolygalacturonase II by sitedirected mutagenesis. J. Biol. Chem. 275, 29348–29353.
Pancoska, P., Janota, V., Keiderling, T.A., 1999. Novel matrix descriptor for secondary structure segments in proteins:
demonstration of predictability from circular dichroism spectra. Anal. Biochem. 267, 72–83.
Papageorgiou, A.C., Shapiro, R., Acharya, K.R., 1997. Molecular recognition of human angiogenin by placental
ribonuclease inhibitor-an X-ray crystallographic study at 2.2 A ( resolution. EMBO J. 16, 5162–5177.
Perutz, M.F., Johnson, T., Suzuki, M., Finch, J.T., 1994. Glutamine repeats as polar zippers: their possible role in
inherited neurodegenerative diseases. Proc. Natl. Acad. Sci. USA 91, 5355–5358.
Peters, J.W., Stowell, H.B., Rees, D.C., 1996. A leucine-rich repeat variant with a novel repetitive protein structural
motif. Nat. Struct. Biol. 3, 991–994.
Petersen, T.N., Kauppinen, S., Larsen, S., 1997. The crystal structure of rhamnogalacturonase A from Aspergillus
aculeatus: a right-handed parallel b-helix. Structure 5, 533–544.
Pickersgill, R., Jenkins, J., Harris, G.B., Nasser, W., Robert-Baudouy, J., 1994. The structure of Bacillus subtilis pectate
lyase in complex with calcium. Nat. Struct. Biol. 1, 717–723.
Pickersgill, R., Smith, D., Worboys, K., Jenkins, J., 1998. Crystal structure of polygalacturonase from Erwinia
carotovora ssp. carotovora. J. Biol. Chem. 273, 24660–24664.
Pissavin, C., Robert-Baudouy, J., Hugouvieux-Cotte-Pattat, N., 1998. Biochemical characterization of the pectate lyase
PeIZ of Erwinia chrysanthemi 3937. Biochim. Biophys. Acta 1383, 188–196.
Pitk.anen, K., Heikinheimo, R., Pakkanen, R., 1992. Purification and characterization of Erwinia chrysanthemi B374
pectin methylesterase produced in Bacillus subtilis. Enzyme. Microbiol. Technol. 14, 832–836.
Pitson, S.M., Mutter, M., van den Broek, L.A.M., Voragen, A.G.J., Beldman, G., 1998. Stereochemical course of
hydrolysis catalysed by alpha-l-rhamnosyl and alpha-d-galacturonosyl hydrolases from Aspergillus aculeatus.
Biochem. Biophys. Res. Commun. 242, 552–559.
Pompeo, F., Bourne, Y., van Heijenoort, J., Fassy, F., Mengin-Lecreuix, D., 2001. Dissection of the bifunctional
E. coli N-acetylglucosamine-1-phosphate uridyltransferase enzyme into autonomously functional domains and
evidence that trimerization is absolutely required for glucosamine-1-phosphate acetyltransferase activity and cell
growth. J. Biol. Chem. 276 (6), 3833–3839.
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 173

Price, S.R., Evans, P.R., Nagai, K., 1998. Crystal structure of the spliceosomal U2B00 –U2A0 protein complex bound to
a fragment of U2 small nuclear RNA. Nature 394, 645–650.
Radford, S.E., Dobson, C.M., 1999. From computer simulations to human disease: emerging themes in protein folding.
Cell 97, 291–298.
Raetz, C.R.H., Roderick, S.L., 1995. A left-handed parallel b helix in the structure of UDP-N-acetylglucosamine
acyltransferase. Science 270, 997–1000.
Richardson, J.S., 1976. Handedness of crossover connections in b-sheets. Proc. Natl. Acad. Sci. SA 73, 2619–2623.
Richardson, J.S., 1981. The anatomy and taxonomy of protein structure. Adv. Protein Chem. 34, 167–339.
Richardson, J.S., Getzoff, E.D., Richardson, D.C., 1978. The b-bulge: a common small unit of nonrepetative protein
structure. Proc. Nati. Acad. Sci. USA 75, 2574–2578.
Robinson, A.S., King, J., 1997. Disulphide-bonded intermediate on the folding and assembly pathway of a non-
disulphide bonded protein. Nat. Struct. Biol. 4 (6), 450–455.
Rochet, J.-C., Lansbury, P.T., 2000. Amyloid fibrillogenesis: themes and variations. Curr. Opinion Struct. Biol.
10, 60–68.
Rouard, M., Bass, J., Grigorescu, F., Garrett, T.P.J., Ward, C.W., Lipkind, G., Jaffiole, C., Steiner, D.F., Bell, G.I.,
1999. Congenital insulin resistance associated with a conformational alteration in a conserved b-sheet in the insulin
receptor L1 domain. J. Biol. Chem. 274, 18487–18491.
Rozwarski, D.A., Dodaiko, T., Almo, S.C., 1999. Crystal structure of the actin binding domain of cyclase associated
protein reveals a dimer of b helices. Mol. Biol. Cell 10 (SS), 903.
Scavetta, R.D., Herron, S.R., Hotchkiss, A.T., Kita, N., Keen, N.T., Benen, J.A., Kester, H.C., Visser, J., Jurnak, F.,
1999. Structure of a plant cell wall fragment complexed to pectate lyase C. Plant Cell 11 (6), 1081–1092.
Schuler, B., Seckler, R., 1998. P22 tailspike folding mutants revisited: effects on the thermodynamic stability of the
isolated b-helix domain. J. Mol. Biol. 281 (2), 227–234.
Schuler, B., Rachel, R., Seckler, R., 1999. Formation of fibrous aggregates from a non-native intermediate: the isolated
P22 tailspike b-helix domain. J. Biol. Chem. 274 (26), 18589–18596.
.
Schuler, B., Furst, F., Osterroth, F., Steinbacher, S., Huber, R., Seckler, R., 2000. Plasticity and steric strain in a
parallel b-helix: rational mutations in the P22 tailspike protein. Proteins 39, 89–101.
Seckler, R., 1998. Folding and function of repetitive structure in the homotrimeric phage P22 tailspike protein. J.
Struct. Biol. 122 (1–2), 216–222.
Serpell, L.C., Smith, J.M., 2000. Direct visualisation of the b-sheet structure of synthetic Alzheimer’s amyloid. J. Mol.
Biol. 299 (1), 225–231.
Selkoe, D.J., 1999. Translating cell biology into therapeutic advances in Alzheimer’s disease. Nature 399, A23–A31.
Sieber, V., Jurnak, F., Moe, G.R., 1995. Circular dichroism of the parallel beta helical proteins pectate lyase C and E.
Proteins. 23 (1), 32–37.
Sippl, M.J., Flockner, H., 1996. Threading thrills and threats. Structure 4, 15–19.
Soto, C., Castano, E.M., Frangione, B., Inestrosa, N.C., 1995. The alphahelical to beta-strand transition in the
amino-terminal fragment of the amyloid beta-peptide modulates amyloid formation. J. Biol. Chem. 270,
3063–3067.
Speed, M.A., Wang, D.I., King, J., 1996. Specific aggregation of partially folded polypeptide chains: the molecular basis
of inclusion body composition. Nat. Biotechnol. 14 (10), 1283–1287.
Speed, M.A., King, J., Wang, D.I., 1997. Polymerization mechanism of polypeptide chain aggregation. Biotechnol.
Bioengin. 54, 333–343.
Steinbacher, S., Seckler, R., Miller, S., Steipe, B., Huber, R., Reinemer, 1994. Crystal structure of P22 tailspike protein:
interdigitated subunits in a thermostable trimer. Science 265, 383–386.
Steinbacher, S., Baxa, U., Miller, S., Weintraub, A., Seckler, R., Huber, R., 1996. Crystal-structure of
phage-p22 tailspike protein complexed with Salmonella sp. o-antigen receptors. Proc. Natl. Acad Sci. USA 93,
10584–10588.
Steinbacher, S., Miller, S., Baxa, U., Budisa, N., Weintraub, A., Seckler, R., Huber, R., 1997. Phage P22
tailspike protein: crystal structure of the head-binding domain at 2.3 A, ( fully refined structure of the
(
endorhamnosidase at 1.56 A resolution, and the molecular basis of o-antigen recognition and cleavage. J. Mol.
Biol. 267, 865–880.
174 J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175

Sulzenbacher, G., Gal, L., Peneff, C., Fassy, F., Bourne, Y., 2001. Crystal structure of Streptococcus pneumoniae
N-acetyl-glucosamine-1-phosphate uridyltransferase bound to acetyl-coenzyme A reveals a novel active site
architecture. J. Biol. Chem. 276 (15), 11844–11851.
Sunde, M., Blake, C., 1997. The structure of amyloid fibrils by electron microscopy and X-ray diffraction. Adv. Protein
Chem. 50, 123–159.
Sunde, M., Serpell, L.C., Bartlam, M., Fraser, P.E., Pepys, M.B., Blake, C.C.F., 1997. Common core structure of
amyloid fibrils by synchrotron X-ray diffraction. J. Mol. Biol. 273, 729–739.
Symmons, M.F., Buchanan, S.G.S., Clarke, D.T., Jones, G., Gay, N.J., 1997. X-ray diffraction and far-UV studies of
filaments formed by a leucinerich repeat peptide: structural similarity to the amyloid fibrils of prions and
Alzheimer’s disease b-protein. FEBS Lett. 412, 397–403.
Takahashi, Y., Ueno, A., Mihara, H., 2000. Mutational analysis of designed peptides that undergo structural transition
from a helix to b sheet and amyloid fibril formation. Structure 8, 915–925.
Tseng, B.P., Esler, W.P., Clish, C.B., Stimson, E.R., Ghilardi, J.R., Vinters, H.V., Mantyh, P.W., Lee, J.P., Maggio,
J.E., 1999. Deposition of monomeric, not oligomeric, Ab mediates growth of Alzheimer’s disease amyloid plaques in
human brain preparations. Biochemistry 38, 10424–10431.
Turnell, W.G., Finch, J.T., 1992. Binding of the dye congo red to the amyloid protein pig insulin reveals a novel
homology amongst amyloidforming peptide sequences. J. Mol. Biol. 227, 1205–1223.
Vaara, M., 1992. Eight bacterial proteins, including UDP-N-acetylglucosamine acyltransferase (LpxA) and three other
transferases of E. coli, consist of a six-residue periodicity theme. FEMS Microbiol. Lett. 76 (3), 249–254.
van den Berg, B.M., Beekhuizen, H., Willems, R.J., Mooi, F.R., van Furth, R., 1999. Role of Bordetella pertussis
virulence factors in adherence to epithelial cell lines derived from the human respiratory tract. Infect. Immun. 67,
1056–1062.
Van Santen, Y., Benen, J.A.E., Schroter, K-H., Kalk, K.H., Armand, S., Visser, J., Dijksta, B.W., 1999. 1.68-A ( crystal
structure of endopolygalacturonase II from Aspergillus niger and identification of active site residues by site-directed
mutagenesis. J. Biol. Chem. 274, 30474–30480.
Vitali, J., Schick, B., Kester, H.C.M., Visser, J., Jurnak, F., 1998. The three-dimensional structure of Aspergillus niger
pectinlyase B at 1.7 A ( resolution. Plant Physiol. 116, 69–80.
Wallace, B.A., Ravikumar, K., 1988. The gramicidin pore: crystal structure of the caesium complex. Science 241,
182–187.
Walsh, D.M., Lomakin, A., Benedek, G.B., Condron, M.M., Teplow, D.B., 1997. Amyloid beta-protein fibrillogenesis.
Detection of a protofibrillar intermediate. J. Biol. Chem. 272, 22364–22372.
Walsh, D.M., Hartley, D.M., Kusumoto, Y., Fezoui, Y., Condron, M.M., Lomakin, A., Benedek, G.B., Condron,
M.M., Benedek, G.B., Selkoe, D.J., Teplow, D.B., 1999. Amyloid beta-protein fibrillogenesisFstructure and
biological activity of protofibrillar intermediates. J. Biol. Chem. 274, 25945–25952.
Wang, J., Hartling, J.A., Flanagan, J.M., 1997. The structure of ClpP at 2.3 A ( resolution suggests a model for ATP-
dependent proteolysis. Cell 91 (4), 447–456.
Watson, D.J., Selkoe, D.J., Teplow, D.B., 1999. Effects of the amyloid precursor protein Glu(693)-Gln ‘Dutch’
mutation on the production and stability of amyloid beta-protein. Biochem. J. 340, 703–709.
West, M.W., Wang, W., Patterson, J., Mancias, J.D., Beasley, J.R., Hecht, M.H., 1999. De novo amyloid proteins from
designed combinatorial libraries. Proc. Natl. Acad. Sci. USA 96, 11211–11216.
Wigley, D.B., Derrick, J.P., Shaw, W.W., 1990. The serine acetyltransferase from E. coliFover-expression, purification
and preliminary crystallographic analysis. FEBS Left. 277 (1–2), 267–271.
Wilmot, C.M., Thornton, J., 1990. b-turns and their distortionsFa proposed new nomenclature. Protein Eng. 3,
479–493.
Wirtz, M., Berkowitz, O., Droux, M., Hell, R. (2001) The cysteine synthase complex from plantsFmitochondrial serine
acetyltransferase from Arabidopsis thaliana carries a bifunctional domain for catalysis and protein–protein
interaction. Eur. J. Biochem. 268, ISS 3, 686–693.
Wyckoff, T.J.O., Raetz, C.R.H., 1999. The active site of E. coli UDP-N-acetylglucosamine acyltransferaseFchemical
modification and site directed mutagenesis. J. Biol. Chem. 274, 27047–27055.
Yamazaki, M., Thorne, L., Mikolajczak, M., Armentrout, R.W., Pollock, T.J., 1996. Linkage of genes essential for
synthesis of a polysaccharide capsule in Sphingomonas strain S88. J. Bacteriol. 178, 2676–2687.
J. Jenkins, R. Pickersgill / Progress in Biophysics & Molecular Biology 77 (2001) 111–175 175

Yoder, M.D., Jurnak, F., 1995. The parallel b-helix and other coiled folds. FASEB J. 9, 335–342.
Yoder, M.D., Schell, M.A., 1995. X-ray-analysis of crystals of polygalacturonase A from Pseudomonas solanacearum.
Acta Crystallogr. D51, 1097–1098.
Yoder, M.D., Keen, N.T., Jumak, F., 1993a. New domain motifFthe structure of pectate lyase C, a secreted plant
virulence factor. Science 260, 1503–1507.
Yoder, M.D., Lietzke, S.E., Jurnak, F., 1993b. Unusual structural features in the parallel b-helix in pectate lyases.
Structure 1, 241–251.
Zhang, H., Seabra, M.C., Deisenhofer, J., 2000. Crystal structure of rab geranylgeranyltransferase at 2.0 angstrom
resolution. Structure 8 (3), 241–251.

Vous aimerez peut-être aussi