Folding of the alphaII-spectrin SH3 domain under physiological salt conditions.

Abstract

The SH3 domain has often been used as a model for protein folding due to its typical two-state behaviour. However, recent experimental data at low pH as well as molecular dynamic simulations have indicated that the folding process of SH3 probably is more complicated, and may involve intermediate states. Using both kinetic and equilibrium measurements we have obtained evidence that under native-like conditions the folding of the spectrin SH3 domain does not follow a classic two-state behaviour. The curvature we observed in the Chevron plots is a strong indication of a non-linear activation energy relationship due to the presence of high-energy intermediates. In addition, circular dichroism measurements indicated that refolding after thermal denaturation did not follow the same pattern as thermal unfolding but rather implied less cooperativity and that the refolding transition increased with increasing protein concentration. Further, NMR experiments indicated that upon refolding the SH3 domain gave rise to more than one conformation. Therefore, our results suggest that the folding of the SH3 domain of alphaII-spectrin does not follow a classical two-state process under high-salt conditions and neutral pH. Heterogeneous folding pathways, which can include folding intermediates as well as misfolded intermediates, might give a more reasonable insight into the folding behaviour of the alphaII-spectrin SH3 domain.

Full text

Page 1

Folding of the aII-spectrin SH3 domain under physiological salt conditions
Katja Petzolda, Anders Öhmanb, Lars Backmanc,*
aDepartment of Medical Biochemistry and Biophysics, Umeå University, SE-901 87 Umeå, Sweden
bUmeå Centre for Molecular Pathogenesis, Umeå University, SE-901 87 Umeå, Sweden
cDepartment of Biochemistry, Umeå University, SE-901 87 Umeå, Sweden
a r t i c l ei n f o
Article history:
Received 20 December 2007
and in revised form 27 February 2008
Available online 7 March 2008
Keywords:
SH3
Spectrin
Protein folding
NMR
a b s t r a c t
The SH3 domain has often been used as a model for protein folding due to its typical two-state behav-
iour. However, recent experimental data at low pH as well as molecular dynamic simulations have indi-
cated that the folding process of SH3 probably is more complicated, and may involve intermediate
states. Using both kinetic and equilibrium measurements we have obtained evidence that under
native-like conditions the folding of the spectrin SH3 domain does not follow a classic two-state behav-
iour. The curvature we observed in the Chevron plots is a strong indication of a non-linear activation
energy relationship due to the presence of high-energy intermediates. In addition, circular dichroism
measurements indicated that refolding after thermal denaturation did not follow the same pattern
as thermal unfolding but rather implied less cooperativity and that the refolding transition increased
with increasing protein concentration. Further, NMR experiments indicated that upon refolding the
SH3 domain gave rise to more than one conformation. Therefore, our results suggest that the folding
of the SH3 domain of aII-spectrin does not follow a classical two-state process under high-salt condi-
tions and neutral pH. Heterogeneous folding pathways, which can include folding intermediates as well
as misfolded intermediates, might give a more reasonable insight into the folding behaviour of the aII-
spectrin SH3 domain.
? 2008 Elsevier Inc. All rights reserved.
The SH3 domain is a common structural motif, often found in
proteins involved in signal transduction as well as in proteins
localised to the plasma membrane [1–5]. This ?60–65 amino acids
long structural motif consists of five b-strands, which fold into two
antiparallel b-sheets, and a short 310-helix. The two b-sheets are al-
most perpendicular to each other forming a sandwich structure.
SH3 domains interact with other proteins through a hydrophobic
patch rich in aromatic residues. The protein ligands often interact
with the SH3 domains via a polyproline-rich segment with PXXP
as minimum consensus sequence [6]. The ligand-binding site is lo-
cated opposite to the region where the SH3 domain is connected to
the rest of the protein [2].
The SH3 domain has been very well characterized; the structure
of the SH3 domain of several different proteins, such as tyrosine ki-
nases [2,7–12], phosphoinositide kinases [13], the muscle protein
nebulin [14], amphiphysins [15], and the regulator protein Grb2
[16] as well as spectrin [7,17] have been determined by crystallog-
raphy as well as by NMR. Although the amino acid sequence can
differ considerably between SH3 domains of different proteins,
the overall native structure appears to be well conserved, espe-
cially in the so-called core region. The major structural differences
are found in the loops connecting the five b-strands [18].
Due to its role as a model system often used in folding studies
also the stability and folding behaviour is well characterized. Ear-
lier studies have indicated that the SH3 domain follows a classical
two-state folding process whereas in later studies this has usually
been taken for granted [19–22]. However, recent data have under-
mined this and instead implied a more complicated process. Based
on cross-linking and NMR diffusion experiment at high protein
concentrations, folding of spectrin and Fyn SH3 domains have been
suggested to follow a three-state process involving monomeric
[23] or oligomeric intermediate states [24].
The folding and stability properties of SH3 have mostly been
studied at low pH and low ionic strength, i.e. under non-native
conditions. Under those conditions the folding behaviour of SH3
is typical of a two-state folder. However, at neutral pH and high-
salt conditions we have observed atypical folding; in this case
the unfolding limb of the Chevron plots displayed a clear curvature
[25]. Therefore we wanted to extend previous studies to include
more native-like conditions, i.e. neutral pH and isotonic salt
conditions. We have therefore compared folding and structural
properties of the SH3 domain of non-erythroid human spectrin
(aII-spectrin) in high-salt conditions at neutral pHs (7.6 and 6,
respectively) with those at pH 4.
0003-9861/$ - see front matter ? 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.abb.2008.02.042
* Corresponding author. Fax: +46 90 786 7661.
E-mail address: lars.backman@chem.umu.se (L. Backman).
Archives of Biochemistry and Biophysics 474 (2008) 39–47
Contents lists available at ScienceDirect
Archives of Biochemistry and Biophysics
journal homepage: www.elsevier.com/locate/yabbi

Page 2

Kinetic measurements at neutral pHs, and to some extent also
at pH 4, showed a clear curvature in the unfolding limb at high
concentrations of the denaturant guanidinium chloride (GdmCl).1
Circular dichroism measurements indicated that refolding after ther-
mal denaturation did not follow the same pattern as unfolding but
rather implied less cooperativity. These measurements did also indi-
cate that the refolding transition increased with increasing protein
concentration. In addition, using NMR spectroscopy we identified
certain residues that gave rise to two or more resonances upon
refolding after thermal unfolding, indicating the presence of more
than one monomeric conformation. Therefore, our results suggest
that the folding of the SH3 domain of aII-spectrin does not follow
a classical two-state process under high-salt conditions but rather
involves folding intermediates that may be oligomeric forms of SH3.
Materials and methods
Protein expression and purification
aII-spectrin SH3 was expressed in BL21 (DE3) Escherichia coli cells, containing
the pET-TEV-aII-SH3 plasmid. Cells were grown at 37 ?C in Luria–Bertani media,
containing 100 lg/l carbencillin until mid-log phase (OD600? 0.5). Protein expres-
sion was induced by addition of isopropyl thio-b-D-galactoside to a final concentra-
tion of 0.5 mM. After over-night growth (15 h) at 23 ?C cells were harvested by
centrifugation, resuspended in 25 mM sodium phosphate buffer, pH 8.0, 150 mM
NaCl, and frozen at ?20 ?C until purification.15N-labelled SH3 was expressed from
the same plasmid and cell line grown in a modified M9-minimum medium (3 g/l
KH2PO4, 1.5 g/l Na2HPO4, 0.25 g/l NaCl, 0.5 g/l15NH4Cl, 1 g/l Na3-citrate, 0.4% glu-
cose, and 0.25 ml/l trace elements).
To purify fusion proteins, cells were thawed on ice, sonicated and clarified by
centrifugation. The His-tagged fusion protein was purified by affinity chromatogra-
phy on nickel-loaded NTA-Sepharose (Amersham Biosciences). Unbound material
was eluted by 25 mM sodium phosphate buffer, pH 7.6, 150 mM NaCl, 10 mM imi-
dazol. Bound material was eluted by an imidazol gradient in the same buffer. After
dialysis TEV protease was used to release the fusion protein. The His-tag and TEV
protease [26] were removed by passing the solution through a nickel-loaded
NTA-Sepharose column. The protein was dialysed against 25 mM NH4CO3, pH 7.3
and lyophilised. For determining the pH-dependency lyophilised protein was dis-
solved in PBS (25 mM sodium phosphate buffer, 150 mM NaCl, pH 6 or 7.6) or
Na-Ac (25 mM sodium acetate, 150 mM NaCl, pH 4).
aII-SH3 P54G and aII-SH3 P54A mutants were constructed by using the Quick-
Change mutagenesis kit (Stratagene). Mutants were expressed and purified as the
wildtype aII-SH3.
The purity of aII-SH3 and mutants were checked by 16% Tricine–SDS–PAGE [27].
Kinetic analysis
Intrinsic tryptophan fluorescence was used to measure the kinetics of unfolding
and refolding. The excitation wavelength was 280 nm, and emission was collected
using a 300-nm cut-off filter. All measurements were done at 20 ?C using a p*-180
or SX-17MV stopped-flow spectrometer (Applied Photophysics). The final concen-
tration of protein was 5 lM in PBS or Na-Ac. Guanidinium chloride (GdmCl) was
used as denaturant. The linear par of obtained data was fitted to:
?
where kobsis the observed folding rate depending on the denaturant concentration
[GdmCl], kf and ku are the rate constants of folding and unfolding, respectively,
kH2O
f
u
are the extrapolated rate constants for folding and unfolding, respec-
tively, in water. mfand muare the linear dependencies on the denaturant concentra-
tion. Only the linear part of the unfolding limb was included in the fitting process.
mfand muwere used to calculate the Tanford value (b), defined as
logkobs¼ logðkfþ kuÞ ¼ log10logkH2Oþmf?½GdmCl?
f
þ 10logkH2Oþmu?½GdmCl?
u
?
ð1Þ
and kH2O
b ¼
mf
mf—mu
ð2Þ
The Tanford value (b) indicates how close the transition state ensemble (TSE) is
to the native state. If b is close to 0, the TSE is similar to the denatured state,
whereas if b equals 1, the TSE is native-like [28].
Equilibrium analysis
Forequilibriummeasurements,intrinsicfluorescenceoftryptophanresidueswas
usedtofollowtheunfoldinginducedbyGdmCl.Tryptophanswereexcitedat280 nm,
and emission spectra were collected from 300 to 400 nm (SPEX fluorolog 112
equipped with Glan-Thompson polarizers). The final protein concentration was
55 lM in PBS or Na-Ac. Obtained spectra were analysed (using KaleidaGraph) and
the shift in wavelength or change in intensity were plotted against the concentration
ofdenaturant.Theresultingdenaturationcurveswerefittedtothefollowingequation
F ¼aNþ bN? ½Gdmcl? þ ðaDþ bD? ½Gdmcl?Þ þ eðDGu?meq?½Gdmcl?Þ=RT
1 þ eðDGu?meq?½Gdmcl?Þ=RT
where F is the observed shift in wavelength or the intensity at the indicated denatur-
ant concentration [GdmCl], aN+ bN[GdmCl] and aD+ bD?[GdmCl] are the baselines
for the folded and unfolded states, respectively, DGuis the free energy of unfolding,
meqis the equilibrium m-value [29].
ð3Þ
Circular dichroism analysis
A J-810 CD spectrometer (Jasco) was used for all circular dichroism measure-
ments. Temperature scans were measured between 198 and 250 nm. Three spectra
were usually collected using cuvettes with 0.1, 0.2 or 0.5 cm light path. The temper-
ature was increased in steps of 2 ?C. Protein concentration was varied between 7
and 166 lM. The ellipticity at 220 nm was used for the analysis and plotted against
temperature. Curves were normalised (SigmaPlot) and fitted to Eqs. (4a) and (4b)
(using Origin)
h ¼aNþ bN? T þ aDþ bD? T
ðÞ ? 10m? T?Tm
Þ
ðÞ
1 þ 10m? T?Tm
ð
ð4aÞ
h ¼aNþ bN? T þ aDþ bD? T
ðÞ ? e
1
Tm?1
DHm=R?
ð
Þ
1
Tm?1
ð
T
ÞÞ
1 þ e
DHm=R?
ð
T
ðÞ
ð4bÞ
where h is the observed ellipticity, aN+ bN?T and aD+ bD?T are the baselines for folded
and unfolded states, respectively. m is the linear dependencies of the temperature,
DHmis the enthalpy at the transition temperature, Tm(in Kelvin) and R is the gas
constant [30,31].
Tmvalues determined by fitting obtained unfolding data to the two equations
were identical within ±0.1 ?C. When refolding data were analysed using Eq. (4b),
the fitting process did not converge.
Nuclear magnetic resonance
A Bruker AMX-500 with a z-gradient TXI probe was used to collect spectra at pH
4 whereas spectra at pH 6 and 7.6 were recorded using a Bruker DRX-600 equipped
with a TXI z-gradient cryo-probe. Concentrations of samples were 300 lM (pH 4),
143 lM (pH 6) and 119 lM (pH 7.6) in water containing 10% D2O and sodium trim-
ethylsilylpropan sulphonic acid as internal standard. Heteronuclear spectra (2D
15N-HSQC) and homonuclear spectra (2D-TOCSY and 2D-NOESY) were recorded
at all pHs and at 298 K. 3D-15N-edited NOESY-HSQC and DIPSI-HSQC experiments
were recorded at pH 7.6 and 298 K. Thermal unfolding and refolding of the SH3 do-
main at pH 4 and 7.6 were followed by recording15N-HSQC spectra at temperatures
between 298 and 358 K. Data was processed with XwinNMR (Bruker Biospin) and
analysed by standard procedures in Ansig [32]. The change in chemical shift of
the backbone moieties were calculated as described elsewhere [33]
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
DdðavÞ ¼
1
Dd
1HN
????2
þ
1
5Dd15N
ðÞ
??2
!
v
u
u
t
ð5Þ
where Dd(av) is the difference in chemical shifts at different pHs.
Pdb file 1AEY (chicken brain a-spectrin) was used to model the SH3 domain
using VMD [34].
Chemical cross-linking
Glutaraldehyde (1% (v/v) final concentration) was used for cross-linking in the
presence of varied concentrations of GdmCl. Fifty micromolar SH3 was incubated
with the cross-linker and GdmCl for 1–60 min. Excess cross-linker was reduced
with 50 mM NaBH4 for 20 min, dialysed against PBS and analysed by 16% Tri-
cine–SDS–PAGE.
Results
Kinetic analysis of the folding and unfolding of SH3
Stopped-flow fluorescence was used to determine the kinetic
properties of the SH3 domain. Under high-salt conditions, at pH
1Abbreviations used: GdmCl, guanidinium chloride; TSE, transition state ensemble;
SDS–PAGE, sodium dodecyl sulphate polyacrylamide-gel electrophoresis; CD, circular
dichroism; HSQC, heteronuclear single quantum correlation; TOCSY, total correlation
spectroscopy; NOESY, nuclear Overhauser enhancement spectroscopy; DIPSI-HSQC,
decoupling in the presence of scalar interactions HSQC; DGu, Gibbs free energy of
unfolding.
40
K. Petzold et al./Archives of Biochemistry and Biophysics 474 (2008) 39–47

Page 3

7.6 as well as 4, the refolding limbs of the Chevron plot were con-
sistent with a two-state process, as can be seen in Fig. 1A, and as
has been shown previously [22]. In contrast, the unfolding limbs
displayed a curvature in the Chevron plot under high destabilizing
conditions (more than 5–6 M GdmCl). Even at pH 4, we observed
the curvature in the unfolding limb although it was much smaller.
Several theories have been presented to explain the observed
curvature in the unfolding limb of Chevron plots. It has been sug-
gested that the curvature is due unfolding intermediates, to the
presence of a second transition state involving the formation of
oligomers [35] or to a moving transition state [36,37].
We also observed that the rate of refolding was pH-dependent;
the rate increased about twofold when pH was lowered from 7.6 or
6 to 4. At the same time the reduction in pH caused a small shift in
the midpoint of the Chevron plot, from 2.4 to 2.8 M GdmCl, indicat-
ing an increased stability. This shift in midpoint corresponds to a
small increase in DG from 13.3 to 14.3 kJ/mol (Table 1).
From the obtained mfand muvalues, the Tanford value (mf/
(mf? mu) at each pH was calculated; independent on pH, a value
of ?0.82 was obtained. This is slightly higher, but agrees reason-
able well with previously obtained values (0.66–0.76) in 50 mM
phosphate buffer at pH 7.0 [38,39].
In the native spectrin SH3 the two prolines, Pro20 and Pro54, at-
tain a trans conformation. In order to determine the role, if any, of a
cis/trans isomerisation in the folding process Pro54 was mutated to
glycine (aII-SH3 P54G) or alanine (aII-H3 P54A). Due to the re-
duced stability of both mutants, it was only possible to analyse
the unfolding kinetics. The unfolding limbs of both mutants were
very similar, and like the wildtype aII-H3, displayed a clear curva-
ture. However, both mutants unfolded ca 100 times faster than the
wildtype aII-H3 (Fig. 1B).
Chemical denaturation followed by fluorescence
One of the tryptophans in the sequence of a-spectrin SH3 is
completely buried in the hydrophobic core (Trp42) whereas the
other (Trp41) is part of a b-heet and the side-chain points out from
the core [7,17]. Since both quantum yield and emission energy is
higher in a hydrophobic environment, unfolding of the hydropho-
bic core should expose the tryptophans and consequently lead to a
decreased intensity as well as a red shift. This is exactly what we
observed. When the fluorescence red shifts or changes in intensity
induced by the denaturant were plotted for each condition, the
patterns at all pHs were similar (Fig. 2). Although the red shift usu-
ally is not linearly proportional to the concentration of folded (or
unfolded) state [40], Eq. (3) was used to estimate the midpoints
of both sets of these curves. The midpoints determined from the
intensity change were ?0.2 M lower than those determined from
the red shift. However, in both cases the amount of GdmCl required
to reach the midpoint value was somewhat higher at pH 4 than at
the other pHs; the midpoints determined by the intensity change
were ?2.1 M GdmCl at pH 4 compared to ?1.9 M at pH 6 or 7.6
(Table 2). The midpoints determined from both the red shift and
the intensity changes were lower than those determined from
the kinetic measurements. A similar difference was also apparent
when comparing kinetic and equilibrium data from urea denatur-
ation of SH3 [41]. The short unfolding limb makes it difficult to
determine the midpoint accurately from the published data; the
midpoint from the kinetic results was estimated to be around
5.5 M urea compared to ca 4.6 M urea in the equilibrium results.
Although parameters (i.e. meqand DGu) determined from equi-
librium measurements are less reliable, DGu-values determined
from the intensity changes agreed reasonable well with those ob-
tained from the kinetic measurements; at pH 4, DGuwas estimated
to ?15.1 kJ/mol compared to the kinetically determined value of
?14.3 kJ/mol.
Thermal and chemical stability analysed by circular dichroism
When followed by CD, thermal unfolding of the spectrin SH3
domain was highly cooperative whereas the refolding process ap-
peared less cooperative (Fig. 3A). This behaviour was most notice-
able at neutral pH (7.6) where the midpoint temperature for
unfolding was 66.2 ?C compared to 60.3 ?C for refolding (Table 2).
It was also obvious that the refolding was dependent on the con-
centration; higher protein concentration led to more cooperativity.
At pH 4 unfolding and refolding followed more or less the same
pattern (Fig. 3A). At pH 7.6 and the lowest concentrations used
(6.9 lM) the initial CD signal was not fully recovered upon refold-
ing, indicating an incomplete regain of the secondary structure
(Fig. 3B).
The unfolding data was also fitted to Eq. (4b) to obtain esti-
mates of DHm. At pH 4, we obtained a value of 197 kJ/mol which
agrees very well with previous determinations [22]. At pH 6 and
7.6, the values of DHmwere 188 and 185 kJ/mole, respectively.
Using a Dcp,U value of 3.36 kJ mol?1K?1[38], we estimate the
enthalpic change at 25 ?C and pH 7.6 to 40.5 kJ?mol?1.
The melting curves were the same independent of the time al-
lowed between scans for temperature equilibration. Therefore we
[GdmCl] M
02468
log k
-1.0
0.0
1.0
2.0
[GdmCl] M
02468
log k
-1.0
-0.5
0.0
0.5
A
B
Fig. 1. Chevron plots. (A) Folding and unfolding kinetics of aII-SH3. The rate con-
stants of folding and unfolding at pH 4.0 (s), pH 6.0 (N) and pH 7.6 (h), at different
denaturant concentrations were determined by stopped-flow fluorescence spec-
troscopy. Data were fitted to a two-state folding process (lines). The concentration
of aII-SH3 was 5 lM. Due to the curvature at high denaturant concentrations only
data up to 5 M GdmCl was included in the fitting procedure. Best fitting values are
reported in Table 1. (B) Unfolding kinetics of aII-SH3 P54A (j) and aII-SH3 P54G
(D). The rates of unfolding at pH 7.6 of the mutants were compared to that of
wildtype aII-SH3 (s).
K. Petzold et al./Archives of Biochemistry and Biophysics 474 (2008) 39–47
41

Page 4

can exclude that there was any time dependency during thermal
unfolding and refolding, at least not in the time range of 5–15 min.
Unfortunately, the CD spectrum in the far-UV range of the
a-spectrin SH3 domain is very weak, due to the almost exclusive
content of b-sheet [7,17]. Since GdmCl gives rise to a substantial
background in the far-UV range, especially at high concentrations,
in combination with the low solubility of SH3 at pH 7.6 it was not
possible to determine the molar ellipticity in the presence of
GdmCl.
NMR analysis
The sequence-specific resonance assignments for the spectrin
SH3 domain at different pHs were determined from the recorded
NMR-spectra by applying standard analysis techniques [42] to-
gether with a previously published assignment of an identical
SH3 domain at pH 3.5 [17]. Judged from the weighted average of
the variation in backbone1H and
corded15N-HSQC spectra certain regions of the protein were more
affected by pH than others (Fig. 4A). The most prominent changes
in chemical shift were observed in residues: Leu8, Glu17, Ser19,
Val23, Thr24, Lys26, Asp29, Asp40, Glu45, Asp48, Arg49, Gln50,
Gly51, Leu61 and Asp62. These residues are mapped onto our mod-
el of the spectrin SH3 domain in Fig. 4B. Also the side chain Gln50
was strongly influenced by the change in pH, in contrast to most
other side chains with exchangeable protons. Interestingly, the
NOE cross-peak patterns at the different pHs were virtually over-
lapping (not shown), suggesting minor variations of the tertiary
structure.
The effect of thermal unfolding and refolding on the SH3 do-
main at pH 4 and 7.6 were investigated on a residue-specific level
15N chemical shifts in the re-
by using15N-HSQC experiments. At pH 4 and 25 ?C the spectrum
before unfolding was identical with the spectrum after refolding.
After refolding at pH 7.6 and 25 ?C, a second and third set of peaks
were visible for approximately one-fourth of the residues, suggest-
ing the presence of at least two minor conformations. The occu-
pancy of these conformations was around 20% as calculated from
the intensity of the resonances. Due to resonance overlap and poor
signal intensity only part of these resonances were assigned with
certainty. Additional resonances were found in the N-terminal
(Glu3, Thr4, Lys6, Glu7, Leu8 and Val9) and in the vicinity of
Pro54 (Gln50, Gly51, Val53, Ala56 and Tyr57). Fig. 5 shows the
multiple resonances observed for Val9, Gly51 and Val53 and the
single set of resonances for Gly5 and Asp48.
It is important to note that the aII-spectrin SH3 domain was
monomeric, even after refolding based on NMR diffusion and T2-
relaxation experiments as well as gel filtration in PBS (data not
shown). Therefore, it is unlikely that the observed additional reso-
nances did appear due to aggregation.
Presence of additional resonances for residues close to a proline
residue indicated that a cis/trans isomerisation could be the cause.
To investigate this further we mutated Pro54, into either a glycine
or alanine residue. Unfortunately, in both cases the stability of the
SH3 mutant (aII-SH3 P54G or aII-SH3 P54A) was reduced signifi-
cantly. The15N-HSQC spectrum of aII-SH3 P54A was typical of a
partially unfolded protein (not shown). It was possible to obtain
for certain resolved resonances single or and multiple conforma-
tions. Residues adopting multiple conformers in the wildtype after
thermal denaturation could also be observed in the P54A mutant,
suggesting that similar conformations were present.
Chemical cross-linking
Cross-linking was used in an attempt to trap unfolded or par-
tially unfolded oligomerised intermediates. Therefore SH3 was
first denatured by GdmCl, and then any formed oligomeric spe-
cies were cross-linked by glutaraldehyde. Excess glutaraldehyde
was reduced with sodium borohydride and the protein solution
was dialysed against PBS before analysis by SDS–PAGE. As Fig. 6
shows, no oligomers could be detected in samples that had not
been treated with the denaturant. However, in the presence of
denaturant, dimers, tetramers and even higher oligomers were
detected after cross-linking. The oligomerisation seemed most
prominent around 2 M GdmCl, which turned out to be at the
denaturing midpoint as determined by equilibrium and kinetic
measurements.
Discussion
When the thermally unfolded spectrin SH3 domain was re-
folded at neutral pH, the NMR spectra we collected showed that
some residues displayed two or three sets of peaks. These residues
were mainly located to the N-terminal (e.g. Glu3, Thr4, Lys6, Glu7,
Leu8 and Val9) and around one of the prolines (e.g. Gln50, Gly51,
Val53, Ala56 and Tyr57). This demonstrated that a fraction of the
Table 1
Kinetic parameters of the aII-SH3 domain at 20 ?C
Assay conditiona
kH2O
f
(s?1)
kH2O
u
(s?1)
mf(kJ mol?1M?1)
mu(kJ mol?1M?1)
DGH2O
D—N
b(kJ mol?1) Tanford valuecb
Chevron plotd[GdmCl] (M)
pY 4.0
pY 6.0
pH 7.6
8.0
4.3
4.4
0.023
0.018
0.016
?0.93
?1.03
?1.03
0.21
0.23
0.24
?14.3
?13.3
?13.7
0.82
0.82
0.81
2.8
2.4
2.4
aTwenty five millimolar sodium phosphate buffer, 150 mM NaCl, pH 7.6, or 6. 25 mM sodium acetate buffer, 150 mM NaCl, pH 4.
bDGH2O
cThe Tanford value, b, was calculated from determined mfand mu.
dThe guanidinium chloride concentration [GmdCl] at the mid-point was determined from Chevron plots.
D—Nwas calculated from rate constants extrapolated to water (kH2O
f
and kH2O
u
).
[GdmCl] M
012345
Normalized shift in λmax or intensity
0.0
0.2
0.4
0.6
0.8
1.0
Fig. 2. Chemical stability of aII-SH3. The stability of aII-SH3 at different GdmCl
concentrations was monitored by fluorescence spectroscopy at pH 4.0 (d, s), pH
6.0 (N, D) and pH 7.6 (j, h). Filled symbols represent the relative change in max-
imal intensity and open symbols shift in wavelength with maximal intensity (red
shift). Data was fitted to Eq. (3) and the parameters are reported in Table 2.
42
K. Petzold et al./Archives of Biochemistry and Biophysics 474 (2008) 39–47

Page 5

protein refolded into one or more minor conformations. These ex-
tra peaks did not disappear with time, therefore it was also appar-
ent that the minor conformation(s) was stable over time. From the
peak intensities we estimated the occupancy of these conforma-
tions to ?20% or less. Thus a small but not insignificant fraction
of the unfolded protein did not refold into its native structure,
not even after prolonged incubation. However, at acidic conditions
(i.e. pH 4) the native structure was regained completely as all res-
idues displayed single resonances.
Since the refolded SH3 domain was monomeric, the additional
resonances did not arise due to aggregation but rather was the re-
sult of the appearance of a different conformation.
There are two proline residues (Pro20 and Pro54) in the spec-
trin SH3 sequence and both of these probably lose the native
trans configuration to a certain extent upon unfolding. To regain
the native structure the folding process would therefore involve
a cis-to-trans isomerisation. Since the rate of folding is much fas-
ter than proline isomerisation, it is possible that a certain frac-
tion of the protein is trapped in the cis conformation during
the folding process. Due to the high activation barrier for cis/
trans isomerisation in native or native-like states [43,44], a cis-
proline would be stable and probably not isomerise into the
trans conformation.
In such a case, it can be expected that residues in the vicinity of
the cis-proline would experience a different surrounding compared
to that close to a trans-proline. Consequently, each affected residue
would give rise to at least two resonances and peak intensities
would depend of the cis/trans equilibrium ratio. It is generally as-
sumed that at neutral pH the cis/trans ratio is about 1:4 [45], which
agrees well with our NMR data.
Based on studies of model peptides and dimethylacetamide, it
has been suggested that the rate constant for the cis/trans isomeri-
sation is increased at acidic conditions [46,47]. Later it has been
suggested that if a histidine residue is preceding the proline the
rate acceleration does not depend on a specific fold [48]. In addi-
tion, at pH 3.5 the fraction of the cis isomer of chicken spectrin
SH3 was estimated to ca 4% [22]. Thus, the reduced fraction of
the cis-proline configuration at low pH may explain why we did
not observe any extra resonances after refolding under acidic
conditions.
The spectra of the P54A mutant of aII-SH3 indicated a partially
unfolded protein (not shown). In spite of this, certain resonances
were possible to recover which suggests, that some secondary
structures are still intact although the tertiary structure has been
strongly destabilised. This could mean, that some secondary struc-
tures fold independently of each other and would so represent a
separate folding domain, which is in agreement with previous pub-
lished data [49]. Interesting but in retrospect not surprising is, that
resonances assignable showed several conformations as found in
the refolded wildtype and demonstrated the importance of Pro54
as a energy border for maintaining/stabilising the tertiary structure
of the SH3 domain [50].
When the folding process of wildtype aII-SH3 was followed by
far-UV CD, the unfolding pattern did not coincide with the refold-
ing pattern. Similar to the NMR results, this observation also indi-
cate that several conformations are formed in the refolding
process. In addition, we observed that the cooperativity was
dependent on the protein concentration as well as on pH. Surpris-
ingly, the cooperativity increased with increasing protein concen-
Table 2
Chemical and thermal denaturation of the aII-SH3 domain
Assay conditionsa
Fluorescenceb
[GdmCl] (M)
DGuc
(kJ mol?1)
Thermal unfoldingd
(?C)
Thermal refoldingd
(?C)
DHme
(kJ mol?1)
Red shift Intensity
pH 4
pH 6
pH 7.6
2.3
2.1
2.1
2.1
1.9
1.9
15.1
14.3
18.3
66.9
66.1
66.2
68.0
65.8
60.3
197
188
185
aTwenty-five millimolar Sodium phosphate buffer, 150 mM NaCl, pH 7.6 or 6. 25 mM sodium acetate buffer, 150 mM NaCl, pH 4.
bFluorescence spectroscopy was used to determine the wavelength with maximum intensity. From plots of kmaxor maximum intensity versus guanidinium chloride
concentrations [GmdCl], the concentration at the mid-point was determined. Eq. (3) was used to calculate the GdmCl concentration at the mid-point.
cDGuwas determined from the normalised changes in intensity.
dThermal unfolding and refolding were followed by circular dichroism, and the mid-point temperature was obtained from plots of ellipticity at 220 nm versus
temperature.
eDHmwas determined from thermal unfolding data obtained by circular dichroism.
Temperature (°C)
020 406080100
fraction unfolded
0.0
0.2
0.4
0.6
0.8
1.0
A
Temperature (°C)
0 2040 6080 100
fraction unfolded
0.0
0.2
0.4
0.6
0.8
1.0
B
Fig. 3. Effect of pH and concentration on thermal stability. (A) Effect of pH on
thermal stability. Fraction unfolded aII-SH3 was determined from CD spectra at
220 nm obtained at pH 4.0 (d, s), pH 6.0 (N, D) and pH 7.6 (j, h), during unfolding
(filled symbols) and refolding (open symbols). (B) Effect of protein concentration on
thermal stability. Fraction unfolded aII-SH3 was determined from CD spectra at
220 nm obtained at 6.9 lM (N, D), 20.7 lM (j, h) and 83 lM (?, }), during unfol-
ding (filled symbols) and refolding (open symbols). Data was fitted to Eq. (4a) and
the parameters are reported in Table 2.
K. Petzold et al./Archives of Biochemistry and Biophysics 474 (2008) 39–47
43

Page 6

trations. This behaviour suggests that the folding process in this
case includes a new equilibrium involving cooperative transitions
with defined aggregation states. The effect of concentration on
cooperativity was not seen at pH 4, contrary to the observation
of Casares et al. [24]. In their case, they observed decreased coop-
erativity with increasing SH3 concentrations. It is possible that
these disparate observations were due to the use of different pHs
(pH 4 versus pH 3) or concentrations. All our experiments were
done at lower protein concentrations, which could explain that
we did not notice any hysteresis at pH 4.
Previously it has been shown that partial unfolding of colla-
gen by heat leads to formation of oligomers that can be trapped
by glutaraldehyde cross-linking [51]. We observed a similar
behaviour upon GdmCl-induced unfolding of spectrin’s SH3 do-
main. At 2 M GdmCl, which is close to the denaturation mid-
point, dimers and tetramers as well as higher oligomers were
detected after cross-linking. It was also evident that higher as
well as lower denaturant concentrations reduced the fraction
of oligomers. Higher concentration of denaturant prevents the
protein from aggregation and at lower concentration not enough
protein will be present in an unfolded form to be able to form
oligomers. Although it can be argued that glutaraldehyde is a
promiscuous cross-linker, it seems possible that under certain
conditions the spectrin SH3 may form oligomers, at least when
fully or partially unfolded.
Due to mass action, it is expected that any tendency to form
aggregates or oligomers will be more pronounced at higher protein
concentrations. Therefore, if an oligomeric species, such as a dimer
or tetramer, would accelerate the initial folding and change the
rate-limiting step, this could explain the increased cooperativity
we observed at high protein concentrations. If correct, this would
also imply that there are more than one pathway for folding; one
that is a direct conversion to the native state and one that involves
intermediate oligomeric states. In such a case, the pathway includ-
ing oligomeric intermediates would be favoured at higher protein
concentrations.
Further, the curvature we observed in the unfolding limb of
the Chevron plots indicates a non-linear activation energy rela-
tionship. Such curvature has been suggested to be due to the
presence of high-energy unfolding intermediates [35]. It has also
Change in chemical shift (Δδ(av))
0.00
0.10
0.20
0.30
0.40
GS D E T GK E L V L A L Y D YQE K S P RE V TMKKGD I L T L L N S T NKDWWKV E V NDRQG F V P AA Y V KK L D
0.00
0.01
0.02
0.03
0.04
D62
L61
L8
D29
E45
D48
R49
Q50
G51
T24
V23
K26
E17
S19
P54
D40
A
B
Fig. 4. Structural changes due to pH. (A) The change in chemical shift (Dd(av)) due to a change in pH was calculated from NMR data using Eq. (5). (Top) Chemical shifts for
backbone amides at pH 7.6 compared to pH 4.0. (Bottom) Chemical shifts for backbone amides for backbone amides at pH 7.6 compared to pH 6.0. The dashed lines represent
0.05 ppm and 0.006 ppm, respectively. Notice the different scales in the two panels. (B) Model of aII-SH3 (based on the NMR structure of chicken a-spectrin SH3, pdb file:
1AEY). Amino acid residues undergoing change in chemical shifts larger than 0.05 ppm are marked.
44
K. Petzold et al./Archives of Biochemistry and Biophysics 474 (2008) 39–47

Page 7

been suggested that curvature and non-linear Chevron plots can
be explained by the presence of a moving transition state
[36,37].
Taken together, our observations imply that the folding of the
spectrin SH3 domain does not follow a typical two-state behav-
iour, at least not under physiologically relevant conditions (i.e.
neutral pH and isotonic saline). This view is supported by molec-
ular dynamic simulations [52] and relaxation dispersion NMR
experiments [23] that have indicated that the folding process
of SH3 is more complicated, and may involve intermediate
states.
Based on these results a thermodynamic description of the
folding process of spectrin SH3 must encompass native (N) and
unfolded (U) states as well as an intermediate (I) state. As most
but not all of the unfolded protein refold into the native state
there must be at least one additional state (N*), which we be-
lieve is native-like. This state appears to be stable as well as
monomeric according to our NMR data. Therefore it must be
formed from the unfolded state (U). If it is the cis-proline con-
taining fraction of the SH3 protein that fold into N*, it follows
that this folding pathway is different from that leading to the
native state. The high Tanford values (mf/(mf+ mu)) indicate that
the structure of the transition state ensemble is very similar to
the native structure. Therefore it seems unlikely that the
pathways leading to the N and N*states would share a common
intermediate with the proline in the native trans confor-
mation.
In addition, the concentration-dependent cooperativity sug-
gests that oligomeric species (In) are formed during the refolding
process. Also the curvature in the unfolding limb of the Chevron
plots may indicate the presence of a certain fraction of an
intermediate.
G51
V53
V9
G5
D48
G28
T4
D40
T37
Y57
V58
Y13
sQ16
sQ50
V23
V23
sN47
T24
S19
A56
sN38
N47
N35
R21
T32
D14
Q50
K59
E45
F52
Y15
E3
K39
R49
I30
K6
E22
E17
W41
L8
M25
V44
K27
A55
Q16
D62
L61
L33
sW42
sW41
L12
L31
A11
L34
N47
K60
W42
K26
V46
K43
L10
K18
D29
N38 E7
15N (ppm)
130
128
126
124
122
120
118
116
114
112
110
108
10.0 9.59.0 8.5 8.07.5 7.0
H (ppm)
1
Fig. 5. NMR spectra of aII-SH3 Contour plots of15N-HSQC spectra before (red) and after (blue) thermal unfolding followed by refolding of a 0.12-mM15N-labelled sample of
aII-SH3. Spectra were recorded at pH 7.6 and 25 ?C. Assignments are indicated.
K. Petzold et al./Archives of Biochemistry and Biophysics 474 (2008) 39–47
45

Page 8

The folding of spectrin SH3 can be described by the following
speculative model
Since we did not observe that unfolding was dependent on pro-
tein concentration, the major unfolding pathway does probably not
include the Instate. At low pH and protein concentration, the mod-
el reduces to a simple two-state process as neither N*nor Inappear
to be populated.
As mentioned above, an acidic pH may speed up cis/trans-pro-
line isomerisation, particularly if there is a histidine preceding
the proline. In such a case, the imidazole side chain must be pro-
tonated [48]. A protonated arginine in the same position does
not accelerate the isomerisation process indicating that the pro-
tonated group must fulfil certain spatial requirements. If the cis/
trans isomerisation is accelerated by intramolecular general acid
catalysis, as has been suggested [48], any protonated group in
the right position could be involved in the reaction mechanism.
There are 11 acidic residues (six aspartates and five glutamates)
in aII-SH3. All of these residues are surface exposed and may be-
come protonated even in the folded state when the pH is lowered.
Since lowering the pH from 7.6 or 6 to 4 did not influence the
NOE cross-peak pattern significantly the overall structure of the
SH3 domain seems not to be influenced by pH. However, a number
of residues showed differences of their chemical shifts in the15N-
HSQC spectra. These affected residues were located in the RT- and
distal loops; in the RT-loop the resonances of Glu17, Ser19, Val23,
Thr24 and Lys26 (Fig. 4B) were affected, and in the distal loop
Glu45, Asp48, Arg49, Gln50 and Gly51 showed changes in the
chemical shifts. Observed changes may reflect an inherent pH-
dependent structural variability and flexibility of these regions that
are known to be important for binding specificity [53,54]. The RT-
loop has also been suggested to be involved in SH3 fibril formation
via an inter-molecular domain-swapping mechanism [55]. In con-
trast to other SH3 domains, there is no evidence of fibril formation
of spectrin SH3 in the present study and has not previously been
reported, the only exception being one engineered mutant [56].
This behaviour has been attributed to the KK motif found in the
RT-loop, a motif that can prevent protein aggregation and fibril for-
mation [57,58].
The chemical shifts of residues at the N- (Leu8) and C-termini
(Leu61 and Asp62) were also influenced by pH. These C-terminal
aminoacidsareinvolvedinahydrogenbondnetworkwhichalsoin-
cludesLeu8from the N-terminus [17]. Protonation of the carboxylic
group of Asp62 would probably reduce the hydrogen bond network
and thereby increase the flexibility of the C-terminus of the protein.
However, these residues are not directly connected with the rest of
the influenced residues around the distal and the RT-loops.
Calculationsofthe electrostatic surface potentialof SH3showeda
considerablymorepositivesurfacepotentialatlowpHthanatneutral
pH (not shown). Some of these acidic residues are involved in salt
bridgesaswellashydrogenbonds[59].Thereforeitseemshighlypos-
siblethatprotonationofcarboxylicgroupscouldleadtoreducedelec-
trostatic interactions as well as to influence hydrogen bonding
networks. Together these changed bonding patterns may decrease
the activation energy required for folding. A change in pH from 7.6
to4willnotonlychangethesignofthenetchargeofSH3(theisoelec-
tricpointisca.5.3)butalsomakethesurfacepotentialmorepositive.
ThereforethischangeinsurfacepotentialwhenloweringthepHmay
increasetherepulsionbetweentheSH3moleculesandwouldexplain
the higher solubility of aII-spectrin SH3 in acidic solution.
The region around the distal loop and the 310-helix has been de-
scribed as a subdomain in the folding of the a-spectrin SH3 that in
contrast to the rest of the protein is partially folded in the transition
state[60].Inaddition,U-valueanalysishaveindicatedthatthedistal
loop (residues 41–53) and probably also contiguous residues form
early in the folding process [61,62]. This may explain the reduced
stability and increased unfolding rate observed in the aII-SH3
P54A and aII-SH3 P54 G mutants. Pro54 is located at the end of the
distal loop and the beginning of the 310-helix. It can be imaging that
the steric constraints imposed by a proline residue is advantageous
forthefoldingprocess,particularlyifthe310-helixformsearly.Inthe
mutantsthis steric constraintwould be lost and the denaturedstate
would be stabilised due to entropic gain.
Conclusions
Using CD and stopped-flow fluorescence measurements as well
as NMR spectroscopy and chemical cross-linking, we have shown
that folding of aII-spectrin SH3 does not follow a classical two-
state process but rather involves an intermediate state that may
be populated by oligomers. The results also indicate that the fold-
ing may include trajectories that are not present during unfolding,
in that oligomeric intermediates are formed in the folding process.
Lowering the pH not only stabilise the aII-spectrin SH3 domain but
also increases the folding rate. These changes are due to the differ-
ent influence of the pH on the different regions in the SH3 domain,
as shown in the HSQC spectra. Therefore it is possible that there is
a subdomain involving the distal loop and 310-helix that folds dif-
ferently in the transition state ensemble depending on the pH and
thereby might cause differences in oligomerisation.
Acknowledgments
His-TEV(S219V)-Arg was a kind gift from Dr. David Waugh. We
appreciate the help from Dr Jürgen Schleucher with NMR measure-
ments. This work was supported by grants from Magn. Bergvalls
Stiftelse, Carl Tryggers Stiftelse and Sven och Lily Lawskis Fond.
References
[1] L. Buday, Biochim. Biophys. Acta 1422 (1999) 187–204.
[2] A. Musacchio, M. Wilmanns, M. Saraste, Prog. Biophys. Mol. Biol. 61 (1994)
283–297.
Fig. 6. Chemical cross-linking aII-SH3 was cross-linked at pH 7.6 and the indicated
concentration of GdmCl using 1% glutaraldehyde. After reduction and dialysis, cr-
oss-linked samples were separated on 16% PAGE using a Tricine buffer system.
Right lane contains molecular weight standard (6.5, 14.4, 21.5, 31, 45, 66.2, 97.4,
116.3 and 200 kDa). The positions of monomeric, dimeric and tetrameric aII-SH3
are indicated.
46
K. Petzold et al./Archives of Biochemistry and Biophysics 474 (2008) 39–47

Page 9

[3] A. Musacchio, Adv. Protein Chem. 61 (2002) 211–268.
[4] V. Agrawal, K.V. Kishan, Protein Pept. Lett. 9 (2002) 185–193.
[5] B.J. Mayer, J. Cell Sci. 114 (2001) 1253–1263.
[6] T. Pawson, Nature 373 (1995) 573–580.
[7] A. Musacchio, M. Noble, R. Pauptit, R. Wierenga, M. Saraste, Nature 359 (1992)
851–855.
[8] S.R. Tzeng, Y.C. Lou, M.T. Pai, M.L. Jain, J.W. Cheng, J. Biomol. NMR 16 (2000)
303–312.
[9] C.J. Morton, D.J.R. Pugh, E.L.J. Brown, J.D. Kahmann, D.A.C. Renzoni, I.D.
Campbell, Structure 4 (1996) 705–714.
[10] Y.Q. Gosser, J. Zheng, M. Overduin, B.J. Mayer, D. Cowburn, Structure 3 (1995)
1075–1086.
[11] S.E. Pursglove, T.D. Mulhern, J.P. Mackay, M.G. Hinds, G.W. Booker, J. Biol.
Chem. 277 (2002) 755–762.
[12] M.E.M. Noble, A. Musacchio, M. Saraste, S.A. Courtneidge, R.K. Wierenga, EMBO
J. 12 (1993) 2617–2624.
[13] S. Koyama, H. Yu, D.C. Dalgarno, T.B. Shin, L.D. Zydowsky, S.L. Schreiber, Cell 72
(1993) 945–952.
[14] A.S. Politou, S. Millevoi, M. Gautel, B. Kolmerer, A. Pastore, J. Mol. Biol. 276
(1998) 189–202.
[15] D.J. Owen, P. Wigge, Y. Vallis, J.D. Moore, P.R. Evans, H.T. McMahon, EMBO J. 17
(1998) 5273–5285.
[16] L. Guruprasad, V. Dhanaraj, D. Timm, T.L. Blundell, I. Gout, M.D. Waterfield, J.
Mol. Biol. 248 (1995) 856–866.
[17] F.J. Blanco, A.R. Ortiz, L. Serrano, J. Biomol. NMR 9 (1997) 347–357.
[18] S.M. Larson, A.R. Davidson, Protein Sci. 9 (2000) 2170–2180.
[19] E.S. Cobos, V.V. Filimonov, M.C. Vega, P.L. Mateo, L. Serrano, J.C. Martinez, J.
Mol. Biol. 328 (2003) 221–233.
[20] M. Sadqi, S. Casares, O. Lopez-Mayorga, J.C. Martinez, F. Conejero-Lara, FEBS
Lett. 514 (2002) 295–299.
[21] D.K. Klimov, D. Thirumalai, J. Mol. Biol. 317 (2002) 721–737.
[22] A.R. Viguera, J.C. Martinez, V.V. Filimonov, P.L. Mateo, L. Serrano, Biochemistry
33 (1994) 2142–2150.
[23] D.M. Korzhnev, X. Salvatella, M. Vendruscolo, A.A. Di Nardo, A.R. Davidson,
C.M. Dobson, L.E. Kay, Nature 430 (2004) 586–590.
[24] S. Casares, M. Sadqi, O. Lopez-Mayorga, F. Conejero-Lara, N.A. van Nuland,
Biophys. J. 86 (2004) 2403–2413.
[25] J. Robertsson, K. Petzold, L. Lofvenberg, L. Backman, Cell. Mol. Biol. Lett. 10
(2005) 595–612.
[26] R.B. Kapust, J. Tozser, J.D. Fox, D.E. Anderson, S. Cherry, T.D. Copeland, D.S.
Waugh, Protein Eng. 14 (2001) 993–1000.
[27] H. Schägger, G. von Jagow, Anal. Biochem. 166 (1987) 368–379.
[28] C. Tanford, Adv. Protein Chem. 24 (1970) 1–95.
[29] K.L.Maxwell,D. Wildes,A.Zarrine-Afsar,M.A. DeLosRios, A.G.Brown,C.T.Friel,
L. Hedberg, J.C. Horng, D. Bona, E.J. Miller, A. Vallee-Belisle, E.R. Main, F.
Bemporad, L. Qiu, K. Teilum, N.D. Vu, A.M. Edwards, I. Ruczinski, F.M. Poulsen,
B.B.Kragelund,S.W.Michnick,F.Chiti,Y.Bai,S.J.Hagen,L.Serrano,M.Oliveberg,
D.P. Raleigh, P. Wittung-Stafshede, S.E. Radford, S.E. Jackson, T.R. Sosnick, S.
Marqusee, A.R. Davidson, K.W. Plaxco, Protein Sci. 14 (2005) 602–616.
[30] L. Swint, A.D. Robertson, Protein Sci. 2 (1993) 2037–2049.
[31] S. Knapp, P.T. Mattson, P. Christova, K.D. Berndt, A. Karshikoff, M. Vihinen, C.I.
Smith, R. Ladenstein, Proteins 31 (1998) 309–319.
[32] M. Helgstrand, P. Kraulis, P. Allard, T. Härd, J. Biomol. NMR 18 (2000) 329–336.
[33] M. Pellecchia, P. Sebbel, U. Hermanns, K. Wüthrich, R. Glockshuber, Nat. Struct.
Biol. 6 (1999) 336–339.
[34] W. Humphrey, A. Dalke, K. Schulten, J. Mol. Graph. 14 (1996) 33–38.
[35] I.E. Sánchez, T. Kiefhaber, J. Mol. Biol. 325 (2003) 367–376.
[36] M. Oliveberg, Curr. Opin. Struct. Biol. 11 (2001) 94–100.
[37] D.E. Otzen, O. Kristensen, M. Proctor, M. Oliveberg, Biochemistry 38 (1999)
6499–6511.
[38] J. Prieto, M. Wilmans, M.A. Jimenez, M. Rico, L. Serrano, J. Mol. Biol. 268 (1997)
760–778.
[39] L. Spagnolo, S. Ventura, L. Serrano, Protein Sci. 12 (2003) 1473–1482.
[40] M.R. Eftink, Biophys. J. 66 (1994) 482–501.
[41] S. Ventura, M. Cristina Vega, E. Lacroix, I. Angrand, L. Spagnolo, L. Serrano, Nat.
Struct. Biol. 9 (2002) 485–493.
[42] K. Wüthrich, NMR of Proteins and Nucleic Acids, John Wiley and Sons, New
York, 1986.
[43] J.F. Brandts, H.R. Halvorson, M. Brennan, Biochemistry 14 (1975) 4953–4963.
[44] S.E. Jackson, A.R. Fersht, Biochemistry 30 (1991) 10436–10443.
[45] V.P. Grantcharova, D. Baker, Biochemistry 36 (1997) 15685–15692.
[46] R.L. Stein, Adv. Protein Chem. 44 (1993) 1–24.
[47] A. Berger, A. Loewensein, S. Meiboom, J. Am. Chem. Soc. 81 (1959) 62–67.
[48] U. Reimer, N. el Mokdad, M. Schutkowski, G. Fischer, Biochemistry 36 (1997)
13802–13808.
[49] J.M. Borreguero, N.V. Dokholyan, S.V. Buldyrev, E.I. Shakhnovich, H.E. Stanley, J.
Mol. Biol. 318 (2002) 863–876.
[50] A. Steward, S. Adhya, J. Clarke, J. Mol. Biol. 318 (2002) 935–940.
[51] J. Engel, H.P. Bachinger, Matrix Biol. 19 (2000) 235–244.
[52] J.M. Borreguero, F. Ding, S.V. Buldyrev, H.E. Stanley, N.V. Dokholyan, Biophys. J.
87 (2004) 521–533.
[53] S. Arold, R. O’Brien, P. Franken, M.P. Strub, F. Hoh, C. Dumas, J.E. Ladbury,
Biochemistry 37 (1998) 14683–14691.
[54] J.M. Martin-Garcia, I. Luque, P.L. Mateo, J. Ruiz-Sanz, A. Camara-Artigas, FEBS
Lett. 581 (2007) 1701–1706.
[55] F. Ding, N.V. Dokholyan, S.V. Buldyrev, H.E. Stanley, E.I. Shakhnovich, J. Mol.
Biol. 324 (2002) 851–857.
[56] B. Morel, S. Casares, F. Conejero-Lara, J. Mol. Biol. 356 (2006) 453–468.
[57] S. Ventura, J. Zurdo, S. Narayanan, M. Parreno, R. Mangues, B. Reif, F. Chiti, E.
Giannoni, C.M. Dobson, F.X. Aviles, L. Serrano, Proc. Natl. Acad. Sci. USA 101
(2004) 7258–7263.
[58] I. Liepina, S. Ventura, C. Czaplewski, A. Liwo, J. Pept. Sci. 12 (2006) 780–789.
[59] S. Casares, M. Sadqi, O. Lopez-Mayorga, J.C. Martinez, F. Conejero-Lara, FEBS
Lett. 539 (2003) 125–130.
[60] J.C. Martínez, M.T. Pisabarro, L. Serrano, Nat. Struct. Biol. 5 (1998) 721–729.
[61] J.C. Martinez, L. Serrano, Nat. Struct. Biol. 6 (1999) 1010–1016.
[62] T. Kortemme, M.J.S. Kelly, L.E. Kay, J. Forman-Kay, L. Serrano, J. Mol. Biol. 297
(2000) 1217–1229.
K. Petzold et al./Archives of Biochemistry and Biophysics 474 (2008) 39–47
47