SequenceâStructure and StructureâFunction Analysis ... - Biozentrum
SequenceâStructure and StructureâFunction Analysis ... - Biozentrum
SequenceâStructure and StructureâFunction Analysis ... - Biozentrum
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
doi:10.1016/j.jmb.2007.02.026 J. Mol. Biol. (2007) 368, 718–728<br />
Sequence–Structure <strong>and</strong> Structure–Function <strong>Analysis</strong><br />
in Cysteine-rich Domains Forming the<br />
Ultrastable Nematocyst Wall<br />
Sebastian Meier 1 ⁎, Pernille Rose Jensen 1 , Patrizia Adamczyk 2<br />
Hans Peter Bächinger 3 , Thomas W. Holstein 2 , Jürgen Engel 4<br />
Suat Özbek 2 <strong>and</strong> Stephan Grzesiek 1 ⁎<br />
1 Department of Structural<br />
Biology, <strong>Biozentrum</strong>,<br />
University of Basel,<br />
Klingelbergstrasse 70,<br />
CH-4056 Basel, Switzerl<strong>and</strong><br />
2 Institute of Zoology,<br />
Department for Molecular<br />
Evolution <strong>and</strong> Genomics,<br />
University of Heidelberg,<br />
Im Neuenheimer Feld 230,<br />
D-69120 Heidelberg, Germany<br />
3 Shriners Hospital for Children<br />
<strong>and</strong> Department of Biochemistry<br />
<strong>and</strong> Molecular Biology,<br />
Oregon Health <strong>and</strong> Science<br />
University, Portl<strong>and</strong>,<br />
OR 97239, USA<br />
4 Department of Biophysical<br />
Chemistry, <strong>Biozentrum</strong>,<br />
University of Basel,<br />
Klingelbergstrasse 70,<br />
CH-4056 Basel, Switzerl<strong>and</strong><br />
*Corresponding authors<br />
The nematocyst wall of cnidarians is a unique biomaterial that withst<strong>and</strong>s<br />
extreme osmotic pressures, allowing an ultrafast discharge of the nematocyst<br />
capsules. Assembly of the highly robust nematocyst wall is achieved by<br />
covalent linkage of cysteine-rich domains (CRDs) from two main protein<br />
components, minicollagens <strong>and</strong> nematocyst outer wall antigen (NOWA).<br />
The bipolar minicollagens have different disulfide patterns <strong>and</strong> topologies in<br />
their N <strong>and</strong> C-terminal CRDs. The functional significance of this polarity has<br />
been elusive. Here, we show by NMR structural analysis that all<br />
representative cysteine-rich domains of NOWA are structurally related to<br />
N-terminal minicollagen domains. Natural sequence insertions in NOWA<br />
CRDs have very little effect on the tightly knit domain structures, nor do they<br />
preclude the efficient folding to a single native conformation. The different<br />
folds in NOWA CRDs <strong>and</strong> the atypical C-terminal minicollagen domain on<br />
the other h<strong>and</strong> can be directly related to different conformational preferences<br />
in the reduced states. Ultrastructural analysis in conjunction with aggregation<br />
studies argues for an association between the similar NOWA <strong>and</strong><br />
N-terminal minicollagen domains in early stages of the nematocyst wall<br />
assembly, which is followed by the controlled association between the<br />
unusual structures of C-terminal minicollagen domains.<br />
© 2007 Elsevier Ltd. All rights reserved.<br />
Keywords: molecular evolution; conformational diversity; NMR structure;<br />
cysteine-rich; dihedral angle preference<br />
Present address: S. Meier, Institute of Molecular Biology <strong>and</strong> Physiology, August Krogh Building, University of<br />
Copenhagen, Universitetsparken 13, DK-2100 Copenhagen, Denmark.<br />
Abbreviations used: CRD, cysteine-rich domain; NOWA, nematocyst outer wall antigen; CROD,<br />
cysteine-rich octarepeat domain of NOWA; NW1r, first CRD of NOWA; NW6r, sixth CRD of NOWA; NW8r, eighth CRD<br />
of NOWA; Mcol1hN, N-terminal CRD of minicollagen 1 from Hydra; Mcol1hC, C-terminal CRD of minicollagen 1 from<br />
Hydra; RDC, residual dipolar coupling; NOE, nuclear Overhauser effect; TCEP,<br />
Tris(2-carboxyethyl)-phosphine; GSH, reduced glutathione; GSSG, oxidized glutathione; HSQC, heteronuclear single<br />
quantum correlation; indel, insertion/deletion.<br />
E-mail addresses of the corresponding authors: smeier@aki.ku.dk; stephan.grzesiek@unibas.ch<br />
0022-2836/$ - see front matter © 2007 Elsevier Ltd. All rights reserved.
Structure <strong>and</strong> Function of Nematocyst CRDs<br />
719<br />
Introduction<br />
highly sensitive to reducing agents, indicating that<br />
the wall proteins are involved in the formation of a<br />
Nematocysts are the characteristic explosive organelles<br />
of the phylum cnidaria. They comprise a soluble in the early stages of nematocyst develop-<br />
cysteine-linked network. 5 As minicollagens are still<br />
cylindrical capsule of about 10 μm diameter that ment 6 <strong>and</strong> their cysteine residues form intramolecular<br />
disulfide bridges, 7 it has been proposed that<br />
encloses an attached tubular coil structure <strong>and</strong><br />
toxins. Upon mechanical stimulation by a prey isomerization to intermolecular disulfide bridges<br />
organism or predator, a lid in the capsule wall would constitute the final maturation step in<br />
opens <strong>and</strong> the tube together with the stored toxins is morphogenesis, which involves wall compaction<br />
released in an ultrafast extrusion process. 1 This <strong>and</strong> polymerization. 8<br />
process is driven by the extreme osmotic pressure We have shown recently that recombinantly<br />
(150 bar; 1 bar=10 5 Pa) within the capsule. 2 The wall expressed NOWA as well as its cysteine-rich octarepeat<br />
domain (CROD) spontaneously form disulfide-<br />
structure of the Hydra nematocyst is a unique<br />
biological polymer formed mainly by two protein linked globular aggregates that resemble the globular<br />
building units of the nematocyst wall. 3 These<br />
species, nematocyst outer wall antigen (NOWA)<br />
<strong>and</strong> minicollagens, which share a common cysteine-rich<br />
motif. 3 These short cysteine-rich domains ranging from 15 nm to 45 nm in diameter <strong>and</strong> form<br />
NOWA globules are extremely heterogeneous in size<br />
(CRDs) contain a conserved pattern of six cysteine several densely packed layers along the wall profile.<br />
residues spaced by three residues. Minicollagens Ultrastructural analysis of the wall architecture<br />
exhibit a symmetrical domain organization with a revealed that the individual globules are interconnected<br />
via rod-like protrusions that probably repre-<br />
central collagen sequence flanked by polyproline<br />
stretches <strong>and</strong> a CRD at the N <strong>and</strong> C termini of the sent minicollagen. NOWA appears early during<br />
molecule (Figure 1(a)). 4 The best-characterized nematocyst morphogenesis <strong>and</strong> forms a thin layer at<br />
minicollagen in the nematocyst wall has been the inside of the nematocyst vesicle membrane. 9 Minicollagens,<br />
which are expressed at later stages,<br />
termed minicollagen 1, <strong>and</strong> serves as a model for<br />
the protein family. NOWA is a 90 kDa glycoprotein gradually attach to this layer in a pre-assembly<br />
with an N-terminal sperm-coating protein (SCP) process, which is followed by polymerization between<br />
domain, a central C-type lectin domain (CTLD) <strong>and</strong> the bipolar minicollagens. Thus, NOWA acts as a<br />
an eightfold CRD repeat at its C terminus (Figure positional organizer of the nematocyst superstructure,<br />
1(a)). The capsule structure has been shown to be providing a scaffold for minicollagen polymerization.<br />
It has been shown recently that the N <strong>and</strong><br />
C-terminal CRDs of minicollagen 1 from Hydra<br />
(Mcol1hN <strong>and</strong> Mcol1hC) form different disulfide<br />
links from identical cysteine patterns (Figure<br />
1(b)). 10,11 This clearly serves as a rare example for<br />
the divergent evolution of protein folds in closely<br />
related domains. We have demonstrated that the two<br />
folds overlap in sequence space <strong>and</strong> can be interconverted<br />
by changes of a few amino acids. 12 The<br />
ability of single cysteine-rich sequences to populate<br />
two different tertiary structures is reminiscent of the<br />
conformational plasticity of prion proteins <strong>and</strong><br />
amyloidogenic proteins. Moreover, this ability has<br />
given strong indications that complex features,<br />
including protein tertiary structures, can develop<br />
by smooth evolutionary transitions. While we have<br />
obtained strong indications that the structural<br />
polarity of N <strong>and</strong> C-terminal CRD folds is conserved<br />
in minicollagens, the functional significance of this<br />
polarity in the formation of the nematocyst wall has<br />
been elusive, mainly due to lack of structural<br />
information on the NOWA CROD.<br />
Here, we present the structure of all representative<br />
cysteine-rich modules in the NOWA CROD, thus<br />
Figure 1. Domain organization of NOWA <strong>and</strong> minicollagen<br />
1. (a) Schematic modular arrangement of signal NOWA, the key components of the nematocyst wall.<br />
completing the structural data on minicollagens <strong>and</strong><br />
peptide (SP), sperm-coating protein (SCP) domain, C-type Despite sequence insertions of varying length into<br />
lectin domain (CTLD) <strong>and</strong> cysteine-rich octarepeat<br />
the tightly knit disulfide rich domains, all NOWA<br />
domain (CROD) in NOWA; CRDs are indicated as circles<br />
<strong>and</strong> coloured according to sequence similarity; CRDs in CRDs form an identical, prototypical structure. This<br />
minicollagen1 are known to form different structures. (b) fold is similar to the Mcol1hN domain structure <strong>and</strong><br />
Sequence alignment of the eight CRDs in the NOWA different from the atypical Mcol1hC fold. NMR<br />
octarepeat domain <strong>and</strong> comparison to N <strong>and</strong> C-terminal analysis of isotope-enriched recombinant protein<br />
CRDs of Hydra minicollagen 1 (Mcol1hN <strong>and</strong> Mcol1hC). 10 shows that the different CRD folds in Mcol1hC <strong>and</strong>
720 Structure <strong>and</strong> Function of Nematocyst CRDs<br />
NOWA can be directly ascribed to different conformational<br />
sampling in the reduced states. We find<br />
that CRDs consistently show a rapid folding reaction<br />
in accordance with the fast cross-linking of the<br />
nematocyst wall. Distinct association properties of<br />
the different CRD domains in conjunction with<br />
electron microscopic analysis suggest that NOWA<br />
nanoparticles are deposited early in the nematocyst<br />
wall formation <strong>and</strong> are crosslinked by short minicollagen<br />
polymers in a controlled fashion to form a<br />
highly stable composite biomaterial.<br />
Results <strong>and</strong> Discussion<br />
Recombinant expression of cysteine-rich<br />
domains<br />
The pattern of six cysteine residues in little more<br />
than 20 amino acids of NOWA cysteine-rich<br />
domains resembles the pattern CXXXCXXXCXXXC-<br />
XXXCC in the closely related CRDs of minicollagen<br />
1(Figure 1). While the first repeat in the cysteine-rich<br />
octarepeat domain (NWr1) has the same length as<br />
the Mcol1hN <strong>and</strong> Mcol1hC domains, the repeats<br />
NWr2 to NWr7 carry an insertion of two amino acids<br />
between the first two cysteine residues <strong>and</strong> NWr8<br />
carries an insertion of three amino acids. For a<br />
modular study of the entire NOWA protein domains<br />
NWr1, NWr6 (representing the highly homologous<br />
repeats NWr2-NWr7), <strong>and</strong> NWr8 were recombinantly<br />
produced in Escherichia coli. The domains<br />
were expressed as fusion proteins C-terminal to the<br />
GB1 domain of protein G, 13 with a thrombin<br />
cleavage site in the linker between protein G <strong>and</strong><br />
the cysteine-rich domain. This resulted in good<br />
expression yields of about 20 mg of soluble fusion<br />
protein per litre of minimal medium. Cleavage,<br />
purification <strong>and</strong> oxidative refolding of the isolated,<br />
uniformly 15 N-labeled domains yielded >5 mg end<br />
product per litre of minimal medium for NWr1,<br />
NWr6 <strong>and</strong> NWr8. Mass spectrometry validated the<br />
chemical identity <strong>and</strong> purity of the CRDs, <strong>and</strong><br />
showed that the domains were fully oxidized with<br />
three disulfide bonds formed (Table 1).<br />
Structure of NOWA CRDs<br />
While six cysteine residues can be fully oxidized to<br />
15 different disulfide bond topologies, the 1 H- 15 N<br />
Table 1. Molecular mass of recombinantly expressed<br />
[U- 15 N]CRDs<br />
Molecular mass (Da)<br />
Mass spectroscopy<br />
Theoretical<br />
NWr1 3109.3 3109.6<br />
NWr6 2802.9 2802.8<br />
NWr8 2930.6 2931.0<br />
Theoretical values are given for fully 15 N-labelled, fully oxidized<br />
domains.<br />
heteronuclear single quantum correlation (HSQC)<br />
spectra show only one folded spectral species for<br />
each of the purified CRDs (Figure 2). Other<br />
prominent spectral species are absent, <strong>and</strong> any<br />
population of minor forms is smaller than 10% of<br />
the main structural species according to the HSQC<br />
peak intensities. Thus, the NOWA CRDs fold very<br />
effectively towards their native structure. Preliminary<br />
structural information can be obtained from the<br />
assigned HSQC spectra, as a central canonical βI<br />
turn between the third <strong>and</strong> fourth cysteine in<br />
Micol1hC (PDB accession number 1SP7) gives rise<br />
to a characteristic 15 N upfield shifts at turn position<br />
i+2. 14 This upfield shift is not observed in any of the<br />
NOWA CRDs at the respective positions E480, Q662<br />
<strong>and</strong> A738 (Figure 2), indicating that the fold of all<br />
NOWA domains is different from that of Mcol1hC.<br />
For a detailed analysis of the structural determinants<br />
of the different CRD folds, the structures of<br />
the three NOWA domains NWr1, NWr6 <strong>and</strong> NWr8<br />
were determined de novo with very high precision<br />
(Table 2). The RMSD of all heavy atoms in the<br />
folded core is less than 0.6 Å for the solution<br />
structures of NWr1, NWr6 <strong>and</strong> NWr8. All NOWA<br />
CRD structures have the same fold (Figure 3). This<br />
is in contrast to the structural polarity of minicollagens<br />
with different structures of Mcol1hN <strong>and</strong><br />
Mcol1hC domains, <strong>and</strong> indicates different assembly<br />
functions for NOWA <strong>and</strong> minicollagens. The<br />
cysteine pattern 1-4, 2-6 <strong>and</strong> 3-5 <strong>and</strong> a central cis<br />
proline turn between cysteine 3 <strong>and</strong> cysteine 4<br />
coincide with the Mcol1hN structure (PDB accession<br />
number 1ZPX). However, the overall lefth<strong>and</strong>ed<br />
fold of the NOWA domains <strong>and</strong> Mcol1hN<br />
differs completely from the Mcol1hC fold, which is<br />
right-h<strong>and</strong>ed. In particular, the NOWA cysteine<br />
pattern differs fundamentally from the Mcol1hC<br />
structure (1-5,2-4,3-6; Figures 3 <strong>and</strong> 4) <strong>and</strong> none of<br />
the NOWA turns, which bring the disulfidebonded<br />
cysteine side-chains into spatial proximity,<br />
coincides with the Mcol1hC domain.<br />
Due to the absence of long-range side-chain<br />
interactions apart from the disulfide bridges, the<br />
different structures of prototypical NOWA CRDs<br />
<strong>and</strong> the atypical Mcol1hC must be a consequence<br />
of different local turn propensities of the intercysteine<br />
residues. 12 The sequential arrangement of<br />
turns in the prototypical NWr1 domain starts with<br />
a canonical βII turn from residues 470–473<br />
between the first two cysteine residues, C469 <strong>and</strong><br />
C473, followed by a short 3 10 helix formed by two<br />
consecutive βIII turns (474–478), which contains<br />
the third cysteine, C477. The conserved proline<br />
residue 479 is in cis conformation <strong>and</strong> part of a<br />
βVIa turn from residues 478–481, which continues<br />
into a γ turn (480–482). Both turns position the<br />
fourth cysteine, C481. The chain terminates in a<br />
single α-helical turn (482–486) that contains the<br />
remaining cysteine residues C485 <strong>and</strong> C486. All<br />
hydrogen bonds <strong>and</strong> a scheme of the turn <strong>and</strong><br />
disulfide topology are given in Figure 4. The amino<br />
acid insertions between the first two cysteine<br />
residues in the NWr6 <strong>and</strong> NWr8 domains are
Structure <strong>and</strong> Function of Nematocyst CRDs<br />
721<br />
Figure 2. Assigned 1 H- 15 N HSQC spectra of the oxidized recombinant octarepeat domains NWr1, NWr6 <strong>and</strong> NWr8<br />
from NOWA. N-terminal artefactual serine residues of the thrombin cleavage sites are marked with an asterisk (*).<br />
incorporated into the NOWA fold in α-helical<br />
conformations directly following the first cysteine<br />
residue. As the α-helical pitch is small in comparison<br />
with an extended conformation, these turns<br />
can accommodate the increase in chain length<br />
without changing the overall structure.<br />
Thus, the NWr6 <strong>and</strong> NWr8 structures differ only<br />
locally from the NWr1 structure. Notably though,<br />
turns are not strictly conserved at homologous<br />
positions: while NWr8 retains the canonical βII<br />
turn before the second cysteine, NWr6 adopts a βI<br />
turn at this location. Presumably, the mutation G→S<br />
abolishes the high positional potential for the formation<br />
of a βII turn <strong>and</strong> induces a preference for the<br />
formation of a βI turn (Figure 1). 15<br />
Structural consequences of insertion, deletion<br />
<strong>and</strong> substitution events during the evolution of<br />
nematocyst CRDs<br />
The cysteine-rich domains from minicollagens <strong>and</strong><br />
NOWA form an attractive natural framework for the<br />
study of sequence–structure relationships as well as<br />
structural evolution resulting from insertion/deletion<br />
(indel) or substitution events. Despite the high<br />
level of similarity of NWr2–NWr7, which points to a<br />
rather recent gene duplication event, CRDs of<br />
different proteins have little sequence similarity<br />
due to extensive neutral drift. On the other h<strong>and</strong>,<br />
local sequence variations between the cysteine<br />
residues suffice to induce different folds in Mcol1hC<br />
Table 2. Statistics of the NWr1, NWr6 <strong>and</strong> NWr8 NMR structures<br />
NWr1 NWr6 NWr8<br />
RMSD from experimental distance constraints a (Å) 0.073±0.001 0.060±0.001 0.071±0.001<br />
RMSD from dihedral constraints b (deg.) 1.21±0.03 0.90±0.02 1.19±0.03<br />
RMSD from scalar coupling constraints c 0.59±0.02 0.97±0.02 0.32±0.01<br />
Deviation from the idealized covalent geometry<br />
Bond lengths (Å) 0.0086±0.001 0.0061±0.001 0.0048±0.001<br />
Bond angles (deg.) 1.373±0.01 0.775±0.01 0.665±0.01<br />
Impropers d (deg.) 1.348±0.01 0.660±0.01 0.539±0.01<br />
Coordinate precision e (Å)<br />
Backbone non-hydrogen atoms 0.15 0.12 0.20<br />
All non-hydrogen atoms 0.26 0.60 0.60<br />
Non-Gly, non-Pro residues in Ramach<strong>and</strong>ran plot f<br />
Core regions (%) 94.7 100 90.6<br />
Allowed regions (%) 5.3 0.0 9.4<br />
Generously allowed regions (%) 0.0 0.0 0.0<br />
Disallowed regions (%) 0.0 0.0 0.0<br />
The statistics were obtained for a subset of the ten lowest energy structures out of 100 calculated with a CNS 31 simulated annealing<br />
protocol. The number of constraints is given in the footnotes. Coordinate precision <strong>and</strong> Ramach<strong>and</strong>ran plot quality are reported for the<br />
core residues excluding flexible N <strong>and</strong> C termini.<br />
a NOEs comprise a non-redundant set of 77/61/75 intraresidual NOEs, 153/90/86 sequential NOEs (|i–j|=1), 148/46/45 short-range<br />
NOEs (15).<br />
b The dihedral angle constraints comprise 8/4/5 ϕ <strong>and</strong> 6/9/7 ψ angles obtained from 1 H, 15 N <strong>and</strong> 13 C chemical shift assignments as<br />
inputs for TALOS, 36 as well as 7/6/6 χ 1 angles obtained from ROESY peak intensities, mainly for the six cysteine residues.<br />
c 3 J HNHA scalar couplings were obtained with a quantitative J measurement to give (15/14/19) restraints. Additional dipolar coupling<br />
constraints were obtained for NWr1 <strong>and</strong> were included with the ISAC protocol, 37 resulting in an NMR quality factor of 0.16. 38 Only RDCs<br />
of the non-flexible NWr1 core residues were incorporated.<br />
d The improper torsion angle restraints serve to maintain planarity <strong>and</strong> chirality.<br />
e The coordinate precision is defined as the average RMS distance between the individual simulated annealing structures <strong>and</strong> the mean<br />
coordinates. Values are reported for core residues.<br />
f Values are calculated with the program PROCHECK-NMR 39 for core residues.
722 Structure <strong>and</strong> Function of Nematocyst CRDs<br />
Figure 3. Solution structures of cysteine-rich domains. The structures of the first (PDB 2HM3), sixth (PDB 2NX6) <strong>and</strong><br />
eighth (PDB 2NX7) cysteine-rich domains of NOWA are shown in comparison to the sequence-related C-terminal<br />
cysteine-rich domain of minicollagen 1 (PDB 1SP7). The ten lowest energy conformers out of 100 structures calculated in<br />
CNS 31 are depicted with cystine shown in yellow. The topology <strong>and</strong> disulfide pattern are conserved upon natural amino<br />
acid insertions between the first two cysteine residues in NOWA domains. The closely related Mcol1hC domain differs in<br />
cysteine pattern, topology <strong>and</strong> one proline cis-trans isomerization.<br />
relative to the NOWA domains <strong>and</strong> Mcol1hN. 12<br />
While indels are often major contributors to protein<br />
evolution expected to promote more drastic structural<br />
changes than substitutions, 16,17 the sequence<br />
insertions into a tightly knit disulfide-rich structure<br />
in longer NOWA domains have very little effect on<br />
the structure, nor do they preclude the efficient<br />
folding to a single native conformation (Figure 2). As<br />
these insertions into CRDs assume no obvious<br />
structural role, they may result from neutral evolutionary<br />
drift, rather than performing central functions<br />
in the assembly of the nematocyst wall. Such<br />
structurally neutral indels may, however, have a<br />
compensatory function in relaxing structural tension<br />
upon an accumulation of amino acid substitutions in<br />
order to retain the original NOWA CRD fold. 18<br />
C-terminal of the insertion between cysteine residues<br />
1 <strong>and</strong> 2, the structures of the NOWA repeats are<br />
virtually identical (Figure 3). This lack of consequences<br />
for the overall structure <strong>and</strong> disulfide<br />
Figure 4. A stereo representation<br />
of the NOWA CRD domain<br />
structures. 12 The structure, disulfide<br />
pattern <strong>and</strong> turn topology is<br />
conserved in the NOWA fold upon<br />
natural insertions into the structure<br />
at the indicated location.
Structure <strong>and</strong> Function of Nematocyst CRDs<br />
723<br />
pattern is highly unexpected, given the high dependence<br />
of disulfide loop closing kinetics <strong>and</strong> stability on<br />
the number of intercysteine residues. 19 As sequence<br />
insertion into other loops of nematocyst CRDs is not<br />
known, insertions may be tolerated only directly C-<br />
terminal of the first cysteine residue, even before the<br />
first turn of the NWr1 structure. This indicates that<br />
thefirstcysteineisnotpartofthecoreofthedomain<br />
<strong>and</strong> suggests a special role of the disulfide bond 1–4,<br />
which will be particularly accessible in the N-terminal<br />
domain of minicollagens.<br />
Determinant residues in the folding of<br />
nematocyst CRDs to different structures<br />
The predominance of the NOWA fold relative to<br />
the Mcol1hC fold indicates that particular mutations<br />
have been used in evolution to destabilize <strong>and</strong><br />
“design out” the more common NOWA structure<br />
in C-terminal minicollagen domains, thus populating<br />
a novel target structure. 12,20 Residual dipolar<br />
couplings (RDCs) are very well suited to monitor<br />
directly the different conformational propensities in<br />
largely unfolded reduced CRD sequences that<br />
ultimately give rise to different structures for<br />
NOWA <strong>and</strong> Mcol1hC cysteine-rich domains. RDCs<br />
report on the time-average <strong>and</strong> ensemble-average<br />
orientation of internuclear bond vectors relative to<br />
the magnetic field in weakly aligned NMR samples.<br />
R<strong>and</strong>om coil polymers yield non-zero RDCs due to<br />
the anisotropy of the linear chain, but are expected to<br />
show a smooth profile of RDCs along the chain in the<br />
absence of sequence-specific interactions. 21 Likewise,<br />
abrupt changes of RDCs between sequential<br />
amino acids in largely unfolded states must arise<br />
from sequence-specific interactions resulting in<br />
specific conformational preferences, since the persistence<br />
length of polypeptide chains is about five to<br />
seven amino acid residues.<br />
To monitor different conformational preferences<br />
that determine the different structures in the oxidized<br />
domains, we compared 1 H- 15 N RDCs recorded<br />
on recombinantly expressed 15 N-labeled<br />
Mcol1hC <strong>and</strong> NWr1 in the reduced state (Figure 5).<br />
While the 1 H- 15 N HSQC spectra of reduced CRDs<br />
exhibit a small signal dispersion characteristic for<br />
unfolded states (Figure 5(a)), RDCs confirm the<br />
persistence of local conformational preferences in the<br />
reduced CRDs (Figure 5(b)). The sequential RDC<br />
pattern <strong>and</strong> thus the structural sampling in reduced<br />
NWr1 <strong>and</strong> Mcol1hC differ markedly between the<br />
first two cysteine residues <strong>and</strong> between the fourth<br />
<strong>and</strong> fifth cysteine residues, thus rationalizing the<br />
importance of these sites for the structural differences<br />
in the native, oxidized form. In agreement with<br />
this direct conformational analysis, a mutational<br />
approach showed that only two central substitutions<br />
in the NWr1 sequence are sufficient to induce the<br />
formation of the Mcol1hC instead of the NOWA<br />
domain structure: 12 one mutation (G11V in the<br />
numbering used in Figures 4 <strong>and</strong> 5) that disfavours<br />
the native βII-turn conformation between the first<br />
<strong>and</strong> second cysteine residues as well as a proline<br />
Figure 5. NMR spectroscopic analysis of residual<br />
structure in reduced CRD domains. (a) Assigned 1 H- 15 N<br />
HSQC spectra of reduced Mcol1HC <strong>and</strong> NWr1. (b)<br />
Distinct conformational preferences of NWr1 <strong>and</strong><br />
Mcol1HC in their reduced states are monitored via<br />
residual dipolar couplings in weakly aligned samples.<br />
substitution of a hydrogen bond donor (K21P)<br />
between the fourth <strong>and</strong> fifth cysteine residues.<br />
A central, conserved proline (P18 in Figures 4 <strong>and</strong><br />
5) between the third <strong>and</strong> fourth cysteine induces a<br />
βI turn in the Mcol1hC structure but assumes a cis<br />
conformation in NOWA <strong>and</strong> Mcol1hN CRDs.<br />
Previous studies had suggested that different<br />
propensities for the formation of cis-peptide bonds<br />
at P18 explain the structural differences between the<br />
different CRD folds. Notably though, favouring the<br />
trans over the cis proline conformation with a (4R)-<br />
fluoroproline analogue did not suffice to switch the<br />
structure of the prototypical domain structure, 22<br />
most likely due to the only weakly increased<br />
stabilization of the trans form by 0.2 kcal/mol<br />
relative to unsubstituted proline. Inspection of CRD<br />
sequences further suggests that sequences assuming<br />
the prototypical CRD fold are not optimized for an<br />
especially high Xaa-Pro cis prolyl propensity. The<br />
Ala-Pro sequences found in NWr2-8 <strong>and</strong> Mcol1N in<br />
fact have a cis prolyl propensity near or below the<br />
average both in model peptides <strong>and</strong> in the<br />
Brookhaven Protein Data Base. 23 Despite a substantially<br />
higher propensity for cis-peptide bond<br />
formation between Tyr17 <strong>and</strong> Pro18, the NWr1<br />
peptide structure has been transformed to the<br />
atypical Mcol1C fold with a trans-peptide bond. 12
724 Structure <strong>and</strong> Function of Nematocyst CRDs<br />
In naturally evolved C-terminal minicollagen<br />
CRDs, Pro18 is usually preceded by β-branched<br />
amino acids, 12 which are known to have particularly<br />
low propensities for the formation of<br />
cis-peptide bonds N-terminal of proline in the<br />
PDB. 23 Thus, it would seem that the cis prolyl<br />
bond is not enforced in the prototypical CRDs but is<br />
rather disfavoured in the atypical C-terminal CRDs.<br />
The cis proline conformation in oxidized NWr1,<br />
NWr6 <strong>and</strong> NWr8 is evidenced from the absence<br />
of strong sequential Hα i–1 -Hδ i nuclear Overhauser<br />
effects (NOEs) <strong>and</strong> the presence of strong sequential<br />
Hα i–1 -Hα i NOEs between the proline <strong>and</strong> its<br />
preceding residue. This central proline is not<br />
conserved in the second NOWA repeat NWr2, thus<br />
presumably destabilizing the folded state <strong>and</strong><br />
indicating a special role of the domain within<br />
NOWA. Accordingly, our attempts to refold <strong>and</strong><br />
purify this domain in high yields after recombinant<br />
expression have failed, but have resulted in a<br />
complex mixture of several spectral species (not<br />
shown), presumably due to the stochastic formation<br />
of disulfide bonds.<br />
A temperature series of oxidative refolding reactions<br />
on NWr1 (Figure 6) was recorded to monitor<br />
the role of the cis proline βVIa turn in the formation<br />
of the NOWA CRD domain structure. NWr1 was<br />
fully reduced with Tris(2-carboxyethyl)-phosphine<br />
(TCEP) <strong>and</strong> oxidative refolding was started by the<br />
addition of oxidized glutathione (GSSG) to a final<br />
concentration of 40 mM at pH 7.5. One-dimensional<br />
proton NMR spectra were recorded to follow the<br />
refolding reaction in real time <strong>and</strong> averaged relative<br />
intensities of well-separated backbone resonances<br />
were obtained (Figure 6(a)). At low temperatures<br />
(
Structure <strong>and</strong> Function of Nematocyst CRDs<br />
725<br />
accelerated upon oxidative collapse, which thus may<br />
function as a built-in catalyst in the oxidative folding<br />
of prototypical CRD domains.<br />
Implications for the formation of the nematocyst<br />
wall superstructure<br />
The cooperativity of unfolding <strong>and</strong> the redox<br />
stability of disulfide bonds were investigated with<br />
an equilibrium reductive unfolding analysis on<br />
NWr1 at pH 7.0 after incubation with increasing<br />
ratios [GSH] 2 / [GSSG] for 12 h in airtight Shigemi<br />
NMR tubes (GSH is reduced glutathione). The<br />
unfolding reaction was followed by 1D proton<br />
NMR spectroscopy. Signals of the native <strong>and</strong><br />
unfolded state were integrated to yield the ratio<br />
[NWr1 unfold ]/ [NWr1 fold ]. The overall transition<br />
midpoint [NWr1 unfold ]/ [NWr1 fold ]=1 is found at<br />
[GSH] 2 / [GSSG] =156 mM. Using a st<strong>and</strong>ard redox<br />
potential for glutathione E' o (glutathione) of −230 mV<br />
at 25 °C 28 the Nernst relation gives:<br />
EVðNOWAÞ¼E o<br />
VðglutathioneÞ o<br />
RT=2F lnð½GSHŠŠ 2<br />
=½GSSGŠŠÞ midpoint<br />
¼ 206 mV<br />
Thus, the disulfide bonds in the NWr1 domain are<br />
only slightly more stable towards the reductant than<br />
the disulfide bonds in the Mcol1hC structure (E' o<br />
(Mcol1hC) =−185 mV). 14 A Hill analysis of the<br />
unfolding curve (see Materials <strong>and</strong> Methods for<br />
details) showed that unfolding results from an<br />
opening of all three disulfide bonds at [GSH] 2 /<br />
[GSSG]>200 mM, but that unfolding is not cooperative<br />
at lower [GSH] 2 / [GSSG] (Figure 6(d)). The<br />
presence of one less stable disulfide bond is identical<br />
with findings for Mcol1hC 14 <strong>and</strong> may be a prerequisite<br />
for the reshuffling of CRD disulfide bridges<br />
from intra- to intermolecular during the formation<br />
of the nematocyst wall.<br />
Rapid re-oxidation of NWr1 by 1% (v/v) H 2 O 2<br />
<strong>and</strong> subsequent mass spectrometric analysis shows<br />
the presence of dimers <strong>and</strong> gives indications for<br />
higher-order aggregates (Figure 7). Individual<br />
NOWA cysteine-rich domains thus bear an intrinsic<br />
Figure 7. Mass spectrometric evidence for oligomerization<br />
of NWr1 domains. (a) <strong>and</strong> (b) Mass spectrum of<br />
NWr1 (theoretical mass =3109.6 Da, deconvoluted in (a))<br />
upon reduction at 2 mM peptide <strong>and</strong> fast re-oxidation<br />
with 1% (v/v) H 2 O 2 . (b) The isotope ladder in the raw<br />
spectrum shows an NWr1 dimer with z=3 <strong>and</strong> m=6218<br />
Da. The background between signals in (b) further<br />
indicates the presence of higher aggregates.<br />
ability for oligomerization, which is in agreement<br />
with the observation of spontaneous crosslinking of<br />
the NOWA CROD. 3 The homogeneity of CRD structures<br />
in NOWA <strong>and</strong> the ability of the NOWA cysteine-rich<br />
octarepeat domain to self-aggregate<br />
further argue for a homo-association mechanism<br />
between similar CRD folds during the formation of<br />
the nematocyst wall superstructure. As the structures<br />
of NOWA domains <strong>and</strong> Mcol1hN coincide, a<br />
primordial appearance of this domain structure<br />
during evolution is likely. The Mcol1hC domain<br />
structure may have appeared later. As compared to<br />
the NWr1 dimerization, Mcol1hC homodimerization<br />
as well as NWr1-Mcol1hC heterodimerization<br />
tendencies are reduced. 12 This lower association<br />
tendency of Mcol1hC may provide the basis for a<br />
controlled <strong>and</strong> most likely catalysed aggregation<br />
step during nematocyst wall assembly.<br />
Presumably, the initial wall assembly occurs in the<br />
absence of minicollagens as NOWA <strong>and</strong> minicollagens<br />
are secreted by separate pathways. 9 The interaction<br />
between NOWA <strong>and</strong> minicollagens thus is<br />
constricted to the nematocyst membrane, to which<br />
NOWA is directed by the association of its basic C<br />
terminus with acidic lipids (data not shown). The<br />
fact that there is only one known isoform of NOWA<br />
supports a primordial function in the assembly of<br />
the capsule wall. Variations in minicollagen sequences<br />
might then have induced different capsule<br />
morphologies <strong>and</strong> functions. 12 Future work will<br />
focus on the molecular pattern of the resulting<br />
heteropolymers (Figure 8) <strong>and</strong> the structure of<br />
dimeric CRD complexes.<br />
Materials <strong>and</strong> Methods<br />
Protein synthesis<br />
The 15 N-labeled NWr1 (NOWA cysteine-rich domain<br />
repeat 1), NWr6 <strong>and</strong> NWr8 proteins were expressed at<br />
30 °C as C-terminal fusions to the protein G B1<br />
domain 13 in M9 medium containing 15 NH 4 Cl as a sole<br />
nitrogen source. The fusion proteins were purified by<br />
affinity chromatography on an IgG Sepharose fast-flow<br />
column (Pharmacia) <strong>and</strong> cleaved by thrombin (Pharmacia).<br />
The cleaved CRD peptides were purified on a C8<br />
reverse-phase HPLC column (Vydac; 10 μm film<br />
thickness; 250 mm×22 mm). Subsequently, the cleaved<br />
peptides were subjected to oxidative folding in refolding<br />
buffer as described, 10 <strong>and</strong> purified by preparative<br />
HPLC. The final yield was 5–10 mg of CRD peptides<br />
per litre of M9 medium. The recombinantly expressed<br />
CRDs have sequences<br />
NWr1: GSTGTCP 470 SGCSGDCYPE 480<br />
CKPGCCGQVN 490 LN<br />
NWr6: GSSSCP 650 QFPSCSPSCA 660<br />
PQCSQQCCQQ 670 P<br />
NWr8: GSA 720 QNPCSLQQPG 730 CSSACAPACR 740<br />
LSCCSLG<br />
where the first two residues (GS) are cloning artifacts<br />
from the thrombin cleavage site. HPLC, mass spectroscopy<br />
<strong>and</strong> NMR spectra of both the synthetic peptide
726 Structure <strong>and</strong> Function of Nematocyst CRDs<br />
Figure 8. Model for the assembly of a NOWA/minicollagen network to form the nematocyst wall superstructure. The<br />
insert shows an electron micrograph of the nematocyst wall surface with NOWA particles interconnected by rod-like<br />
protrusions formed probably by minicollagen (the scale bar represents 40 nm). NOWA is deposited at the membrane via<br />
basic C-terminal residues (not shown). A pre-assembly between NOWA CRDs <strong>and</strong> minicollagen N-terminal CRDs (red<br />
squares) is indicated by the high homo-oligomerization propensity of the NOWA <strong>and</strong> minicollagen N-terminal CRD fold<br />
only. The invention of a novel fold in minicollagen C-terminal CRDs may allow a controlled step, which leads to a<br />
catalyzed intercysteine linkage between C-terminal CRDs (blue squares) <strong>and</strong> to final wall maturation.<br />
<strong>and</strong> the expressed peptides indicate an estimated purity<br />
above 85 % in all cases. Mass spectroscopy <strong>and</strong> NMR<br />
spectra show the presence of three intact disulfide bonds.<br />
Isotope-labelled CRDs were lyophilized <strong>and</strong> dissolved in<br />
5 mM phosphate (pH 5.5), 2 mM NaN 3 , 95% H 2 O/ 5%<br />
2 H 2 O, to a concentration of 0.7 mM peptide. In addition,<br />
solid phase synthesis of unlabeled NWr1 peptide of the<br />
sequence<br />
TCP 470 SGCSGDCYPE 480 CKPGCCGQVN 490 LN<br />
was carried out by an N-(9-fluorenyl)methoxycarbonyl<br />
(Fmoc) strategy. The peptide was purified by preparative<br />
HPLC both before <strong>and</strong> after oxidative folding as described.<br />
10 Unlabelled NWr1 was dissolved to a final<br />
concentration of 3.5 mM in 5 mM phosphate (pH 5.5),<br />
2 mM NaN 3 , 95% H 2 O/ 5% 2 H 2 O,.<br />
Mass spectrometry<br />
Recombinant peptides were characterized by electrospray<br />
ionization-mass spectrometry using a Bruker<br />
microTOF mass spectrometer. Experimental <strong>and</strong> theoretical<br />
masses for uniformly 15 N-labeled, fully oxidized<br />
domains are given in Table 1.<br />
NMR spectroscopy<br />
The different uniformly 15 N-labelled cysteine-rich<br />
domains were assigned using 2D as well as 3D- 15 N<br />
separated NOE spectroscopy (NOESY) <strong>and</strong> total correlated<br />
spectroscopy (TOCSY) in conjunction with 1 H- 15 N HSQC<br />
spectra <strong>and</strong> 1 H- 13 C HSQC spectra at natural 13 C isotope<br />
enrichment. 3 J HNHA scalar couplings were measured with<br />
an HNHA experiment. All spectra were recorded on a<br />
Bruker DRX600 spectrometer equipped with a TXI probe<br />
or on a Bruker DRX800 spectrometer equipped with a TCI<br />
cryogenic probe. Spectra were processed with NMRPipe 29<br />
<strong>and</strong> analysed with PIPP. 30 Structures were calculated in<br />
CNS 31 from the experimental data summarized in Table 1.<br />
All structure representations were generated with the<br />
program MOLMOL. 32<br />
Structural study on reduced CRDs<br />
Mcol1hC <strong>and</strong> NWr1 were reduced by the addition of<br />
5 mM TCEP <strong>and</strong> were assigned with 15 N separated<br />
NOESY <strong>and</strong> TOCSY spectra after adjusting the pH to 5.5.<br />
RDCs were measured in strained (10 %, w/v) polyacrylamide<br />
gels 33,34 in the presence of 5 mM TCEP at 200 μM<br />
peptide in 5 mM sodium phosphate buffer (pH 5.5).<br />
Equilibrium unfolding <strong>and</strong> oxidative refolding<br />
For oxidative refolding in the NMR spectrometer,<br />
300 μM unlabelled NWr1 peptide in 20 mM phosphate<br />
buffer (pH 7.5), was fully reduced with 3 mM TCEP at<br />
pH 7.5. Oxidative refolding was initiated by adding<br />
GSSG in 20 mM phosphate buffer (pH 7.5) to a final<br />
[GSSG]= 40 mM. Oxidative refolding was carried out at<br />
280 K, 295 K, 308 K <strong>and</strong> 320 K. An equilibrium unfolding<br />
titration was performed on synthetic NWr1 by titrating<br />
various amounts of a stock solution of 300 mM GSH,<br />
20 mM phosphate buffer (pH 7.0) to a solution of 300 μM<br />
CRD peptide, 8 mM GSSG, 20 mM phosphate buffer (pH<br />
7.0). After equilibrating for 12 h at 25 °C in airtight<br />
Shigemi tubes, the relative amount of folded <strong>and</strong><br />
unfolded CRD was determined from the loss of native<br />
signal <strong>and</strong> the increase of reduced CRD in 1D proton<br />
NMR spectra. Similarly, the ratio of reduced <strong>and</strong><br />
oxidized glutathione was determined from the respective<br />
signals to yield the redox potential of the solution. The<br />
unfolding reaction:<br />
NWr1 ox þ nð2GSHÞ X NWr1 red þ nGSSG<br />
proceeds with an opening of n disulfide bonds upon<br />
titration with GSH. The equilibrium of the form K eq =<br />
[NWr1 red ]/[ NWr1 ox ]·( [GSSG]/[GSH] 2 ) n yields the cooperativity<br />
of reduction n as the slope of a Hill plot of log<br />
([NWr1 red ]/[NWr1 ox ]) versus log([GSH] 2 /[GSSG]). 35<br />
Protein Data Bank accession number<br />
The atomic coordinates of the ten lowest energy CNS<br />
conformers of NW1r, NW6r <strong>and</strong> NW8r have been
Structure <strong>and</strong> Function of Nematocyst CRDs<br />
727<br />
deposited at the RCSB Protein Data Bank (www.rcsb.org)<br />
under PDB accession numbers 2HM3, 2NX6 <strong>and</strong> 2NX7,<br />
respectively.<br />
Acknowledgements<br />
We thank M. Rogowski for the acquisition of<br />
mass spectra. This work was supported by SNF<br />
grant 31-109712 to S.G. by a grant from the German<br />
Science Foundation (DFG) to T.W.H. <strong>and</strong> S.Ö. <strong>and</strong><br />
by a grant from the Benzon foundation to S.M.<br />
References<br />
1. Nüchter, T., Benoit, M., Engel, U., Özbek, S. &<br />
Holstein, T. W. (2006). Nanosecond-scale kinetics of<br />
nematocyst discharge. Curr. Biol. 16, R316–R318.<br />
2. Holstein, T. & Tardent, P. (1984). An ultrahigh-speed<br />
analysis of exocytosis - nematocyst discharge. Science,<br />
223, 830–833.<br />
3. Özbek, S., Pokidysheva, E., Schwager, M., Schulthess,<br />
T., Tariq, N., Barth, D. et al. (2004). The glycoprotein<br />
NOWA <strong>and</strong> minicollagens are part of a disulfidelinked<br />
polymer that forms the cnidarian nematocyst<br />
wall. J. Biol. Chem. 279, 52016–52023.<br />
4. Kurz, E. M., Holstein, T. W., Petri, B. M., Engel, J. &<br />
David, C. N. (1991). Mini-collagens in Hydra nematocytes.<br />
J. Cell. Biol. 115, 1159–1169.<br />
5. Blanquet, R. & Lenhoff, H. M. (1966). A disulfidelinked<br />
collagenous protein of nematocyst capsules.<br />
Science, 154, 152–153.<br />
6. Engel, U., Pertz, O., Fauser, C., Engel, J., David, C. N.<br />
& Holstein, T. W. (2001). A switch in disulfide linkage<br />
during minicollagen assembly in Hydra nematocysts.<br />
EMBO J. 20, 3063–3073.<br />
7. Özbek, S., Pertz, O., Schwager, M., Lustig, A.,<br />
Holstein, T. & Engel, J. (2002). Structure/function<br />
relationships in the minicollagen of Hydra nematocysts.<br />
J. Biol. Chem. 277, 49200–49204.<br />
8. Özbek, S., Engel, U. & Engel, J. (2002). A switch in<br />
disulfide linkage during minicollagen assembly in<br />
Hydra nematocysts or how to assemble a 150-barresistant<br />
structure. J. Struct. Biol. 137, 11–14.<br />
9. Engel, U., Özbek, S., Streitwolf-Engel, R., Petri, B.,<br />
Lottspeich, F. & Holstein, T. W. (2002). Nowa, a novel<br />
protein with minicollagen Cys-rich domains, is<br />
involved in nematocyst formation in Hydra. J. Cell.<br />
Sci. 115, 3923–3934.<br />
10. Pokidysheva, E., Milbradt, A. G., Meier, S., Renner, C.,<br />
Häussinger, D., Bächinger, H. P. et al. (2004). The<br />
structure of the Cys-rich terminal domain of Hydra<br />
minicollagen, which is involved in disulfide networks<br />
of the nematocyst wall. J. Biol. Chem. 279, 30395–30401.<br />
11. Milbradt, A. G., Boulegue, C., Moroder, L. & Renner, C.<br />
(2005). The two cysteine-rich head domains of minicollagen<br />
from Hydra nematocysts differ in their cystine<br />
framework <strong>and</strong> overall fold despite an identical<br />
cysteine sequence pattern. J. Mol. Biol. 354, 591–600.<br />
12. Meier, S., Jensen, P. R., David, C. N., Chapman, J.,<br />
Holstein, T. W., Grzesiek, S. & Özbek, S. (2007).<br />
Continuous molecular evolution of protein-domain<br />
structures by single amino acid changes. Curr. Biol. 17,<br />
173–178.<br />
13. Huth, J. R., Bewley, C. A., Jackson, B. M., Hinnebusch,<br />
A. G., Clore, G. M. & Gronenborn, A. M. (1997).<br />
Design of an expression system for detecting folded<br />
protein domains <strong>and</strong> mapping macromolecular interactions<br />
by NMR. Protein Sci. 6, 2359–2364.<br />
14. Meier, S., Häussinger, D., Pokidysheva, E., Bächinger,<br />
H. P. & Grzesiek, S. (2004). Determination of a highprecision<br />
NMR structure of the minicollagen cysteine<br />
rich domain from Hydra <strong>and</strong> characterization of its<br />
disulfide bond formation. FEBS Letters, 569, 112–116.<br />
15. Hutchinson, E. G. & Thornton, J. M. (1994). A revised<br />
set of potentials for beta-turn formation in proteins.<br />
Protein Sci. 3, 2207–2216.<br />
16. Ferraro, D. M., Ferraro, D. J., Ramaswamy, S. &<br />
Robertson, A. D. (2006). Structures of ubiquitin insertion<br />
mutants support site-specific reflex response to<br />
insertions hypothesis. J. Mol. Biol. 359, 390–402.<br />
17. Ferraro, D. M., Hope, E. K. & Robertson, A. D. (2005).<br />
Site-specific reflex response of ubiquitin to loop<br />
insertions. J. Mol. Biol. 352, 575–584.<br />
18. Grishin, N. V. (2001). Fold change in evolution of<br />
protein structures. J. Struct. Biol. 134, 167–185.<br />
19. Zhang, R. M. & Snyder, G. H. (1989). Dependence of<br />
formation of small disulfide loops in two-cysteine<br />
peptides on the number <strong>and</strong> types of intervening<br />
amino acids. J. Biol. Chem. 264, 18472–18479.<br />
20. Bornberg-Bauer, E. & Chan, H. S. (1999). Modeling<br />
evolutionary l<strong>and</strong>scapes: mutational stability, topology,<br />
<strong>and</strong> superfunnels in sequence space. Proc. Natl<br />
Acad. Sci. USA, 96, 10689–10694.<br />
21. Louhivuori, M., Paakkonen, K., Fredriksson, K.,<br />
Permi, P., Lounila, J. & Annila, A. (2003). On the<br />
origin of residual dipolar couplings from denatured<br />
proteins. J. Am. Chem. Soc. 125, 15647–15650.<br />
22. Boulegue, C., Milbradt, A. G., Renner, C. & Moroder, L.<br />
(2006). Single proline residues can dictate the oxidative<br />
folding pathways of cysteine-rich peptides. J. Mol. Biol.<br />
358, 846–856.<br />
23. Reimer, U., Scherer, G., Drewello, M., Kruber, S.,<br />
Schutkowski, M. & Fischer, G. (1998). Side-chain effects<br />
on peptidyl-prolyl cis/trans isomerisation. J. Mol. Biol.<br />
279, 449–460.<br />
24. Renner, C., Alefelder, S., Bae, J. H., Budisa, N., Huber,<br />
R. & Moroder, L. (2001). Fluoroprolines as tools for<br />
protein design <strong>and</strong> engineering. Angew. Chem. Int. Ed.<br />
Engl. 40, 923–925.<br />
25. Shi, T., Spain, S. M. & Rabenstein, D. L. (2006). A<br />
striking periodicity of the cis/trans isomerization of<br />
proline imide bonds in cyclic disulfide-bridged peptides.<br />
Angew. Chem. Int. Ed. Engl. 45, 1780–1783.<br />
26. Br<strong>and</strong>ts, J. F., Halvorson, H. R. & Brennan, M. (1975).<br />
Consideration of possibility that slow step in protein<br />
denaturation reactions is due to cis-trans isomerism of<br />
proline residues. Biochemistry, 14, 4953–4963.<br />
27. Fischer, S., Dunbrack, R. L. & Karplus, M. (1994). Cistrans<br />
imide isomerization of the proline dipeptide.<br />
J. Am. Chem. Soc. 116, 11931–11937.<br />
28. Fasman,G.D.(1975).CRC H<strong>and</strong>book of Biochemistry <strong>and</strong><br />
Molecular Biology, 3rd edit., CRC Press, Clevel<strong>and</strong> OH.<br />
29. Delaglio, F., Grzesiek, S., Vuister, G. W., Zhu, G.,<br />
Pfeifer, J. & Bax, A. (1995). Nmrpipe - a multidimensional<br />
spectral processing system based on<br />
Unix Pipes. J. Biomol. NMR, 6, 277–293.<br />
30. Garrett, D. S., Powers, R., Gronenborn, A. M. & Clore,<br />
G. M. (1991). A common-sense approach to peak<br />
picking in 2-dimensional, 3- dimensional, <strong>and</strong> 4-dimensional<br />
spectra using automatic computer-analysis<br />
of contour diagrams. J. Magn. Reson. 95, 214–220.<br />
31. Brunger,A.T.,Adams,P.D.,Clore,G.M.,DeLano,W.L.,<br />
Gros, P., Grosse-Kunstleve, R. W. et al. (1998).
728 Structure <strong>and</strong> Function of Nematocyst CRDs<br />
Crystallography <strong>and</strong> NMR system: a new software<br />
suite for macromolecular structure determination.<br />
Acta Crystallog. sect. D, 54, 905–921.<br />
32. Koradi, R., Billeter, M. & Wüthrich, K. (1996).<br />
MOLMOL: a program for display <strong>and</strong> analysis of<br />
macromolecular structures. J. Mol. Graph. 14, 51–55.<br />
33. Sass, H. J., Musco, G., Stahl, S. J., Wingfield, P. T. &<br />
Grzesiek, S. (2000). Solution NMR of proteins within<br />
polyacrylamide gels: diffusional properties <strong>and</strong> residual<br />
alignment by mechanical stress or embedding<br />
of oriented purple membranes. J. Biomol. NMR, 18,<br />
303–309.<br />
34. Tycko, R., Blanco, F. J. & Ishii, Y. (2000). Alignment of<br />
biopolymers in strained gels: a new way to create<br />
detectable dipole-dipole couplings in high-resolution<br />
biomolecular NMR. J. Am. Chem. Soc. 122, 9340–9341.<br />
35. Hawkins, H. C., de Nardi, M. & Freedman, R. B.<br />
(1991). Redox properties <strong>and</strong> cross-linking of the<br />
dithiol/disulphide active sites of mammalian<br />
protein disulphide-isomerase. Biochem. J. 275,<br />
341–348.<br />
36. Cornilescu, G., Delaglio, F. & Bax, A. (1999). Protein<br />
backbone angle restraints from searching a database<br />
for chemical shift <strong>and</strong> sequence homology. J. Biomol.<br />
NMR, 13, 289–302.<br />
37. Sass, H. J., Musco, G., Stahl, S. J., Wingfield, P. T. &<br />
Grzesiek, S. (2001). An easy way to include weak<br />
alignment constraints into NMR structure calculations.<br />
J. Biomol. NMR, 21, 275–280.<br />
38. Cornilescu, G., Marquardt, J. L., Ottiger, M. & Bax, A.<br />
(1998). Validation of protein structure from anisotropic<br />
carbonyl chemical shifts in a dilute liquid crystalline<br />
phase. J. Am. Chem. Soc. 120, 6836–6837.<br />
39. Laskowski, R. A., Rullmann, J. A. C., MacArthur,<br />
M. W., Kaptein, R. & Thornton, J. M. (1996). AQUA<br />
<strong>and</strong> PROCHECK-NMR: programs for checking the<br />
quality of protein structures solved by NMR. J. Biomol.<br />
NMR, 8, 477–486.<br />
Edited by M. F. Summers<br />
(Received 1 December 2006; received in revised form 30 January 2007; accepted 8 February 2007)<br />
Available online 24 February 2007