Context-dependent effects of proline residues on the stability
and folding pathway of ubiquitin
Maria D. Crespo, Geoffrey W. Platt, Roger Bofill and Mark S. Searle
School of Chemistry, Centre for Biomolecular Sciences, University Park, Nottingham, UK
Substitution of trans-proline at three positions in ubiquitin
(residues 19, 37 and 38) produces significant context-
dependent effects on protein stability (both stabilizing
and destabilizing) that reflect changes to a combination
of parameters including backbone flexibility, hydrophobic
interactions, solvent accessibility to polar groups and
intrinsic backbone conformational preferences. Kinetic
analysis of the wild-type yeast protein reveals a predominant
fast-folding phase which c onforms to an apparent two-
state f olding model. Temperature-dependent studies of the
refolding rate reveal thermodynamic details of the nature of
the transition s tate fo r f olding consistent with hydrophobic
collapse providing the overall driving force. Brønsted
analysis of the refolding and unfolding rates of a family of
mutants w ith a variety o f side c hain substitutions for P 37 and
P38 reveals that the two prolines, which are located in a
surface l oop adjacent to the C terminus of the m ain a-helix
(residues 24–33), are not significantly structured in the
transition state for folding and appear to be consolidated
into the native structure only late in the folding process. We
draw a similar conclusion regarding position 19 in the loop
connecting the N-terminal b-hairpin to t he main a-helix. T he
proline residues of ubiquitin are passive spectators in the
folding process, but influence protein stability in a variety of
ways.
Keywords: folding kinetics; NMR structural analysis; proline
mutations; p rotein folding pathway; protein stability.

amino acid sequence) suggests that not all non-native cis
prolines result in slow folding p hases, and that cis–trans
isomerization in some structural contexts need not be rate
limiting [22–29]. More recent studies demonstrate that
nonprolyl cis-peptide bonds also contribute to the hetero-
geneous pool of unfolded molecules [18,30]. Although
individual cis-peptide bonds contribute little to the popu-
lation (% 0.15–0.5%) in the unfolded protein, their large
number generates a significant proportion of slow folding
molecules [18,30,31].
We report on the effects of proline on the stability and
folding kinetics of ubiquitin, a small model system of 76
residues that is uncomplicated by disulphide bonds and
bound cofactors [32]. Ubiquitin has been the subject of a
number of investigatio ns regarding i ts folding m echanism.
Early studies had suggested that the protein populates an
intermediate state identified on the basis of deviations of
kinetic data from linearity in the refolding arm of chevron
plots at low denaturant concentrations [33]. M ore recent
studies [13,14,34,35] report apparent two-state kinetics
under similar conditions, suggesting that t he roll-over effect
in the r efolding kinetics may b e a consequence o f either
transient aggregation that is exacerbated by the stabilizing
effects of inorganic salts [15,35], or due to data fitting at
rates near the instrumental limits where interference from
slower phases can decrease apparent folding rates resulting
Correspondence to M. S. Searle, School of Chemistry, Centre for
Biomolecular Sciences, University Park, Nottingham NG7 2RD, UK.
Tel.: +44 115 9513567, E-mail:
Abbreviations: TSE, transition state ensemble; GdmCl, guanidinium

cassette was inserted between the Eco RI and HindIII
restriction sites of pKK223-3, and the mutation confirmed
by DNA sequencing. Competent E. coli cells were trans-
formed with this construct. Expression and purification
were as described for the wild-type yielding typically
10–15 mgÆL
)1
of ubiquitin, as previously described [34].
NMR structural analysis
All NMR experiments were performed on a Bruker
Avance600 spectrometer. TOCSY and NOESY experi-
ments were used as p reviously described [34] on 1 -m
M
protein samples at pH 5.5. Spectra were referenced to
internal trimethylsilylpropionate. D ata were processed and
assigned using Bruker
XWINNMR
and
ANSIG
software [37].
Structural models were visualized using
MOLMOL
[38].
Equilibrium stability measurements
Protein stability was determined by fluorescence measure-
ments on 1 .5 l
M
solutions of protein i n 2 5 m
M
acetate

D
is the
measured fluorescence at a g iven [D] a nd f
U
and f
N
are
the limiting values for the unfolded and native states,
respectively. The mid-point of the unfolding transition
[D]
50%
for each mutant was determined by nonlinear least
squares fitting to the expression:
F
f
¼ exp½mð½DÀ½D
50%
Þ=RT=ð1 þ exp½mð½D
À½D
50%
Þ=RTÞ ð1Þ
The e quilibrium stability DG
eq
was d etermined from the
expression DG
eq
¼ –m[D]
50%
,wherem forasetofmutants
is assumed constant (10.9 ± 0.23 kJÆmol

final protein concentration of 1.36 l
M
. For unfolding
experiments, a buffered solution of native protein was
unfolded by a 1 : 10 dilution to yield final concentrations
of GdmCl near or above the midpoint of the equilibrium
unfolding transition (concentrations of GdmCl in the
range 3.7–7.3
M
). Kinetic measurements for both unfold-
ing and refolding reactions were averaged four to six
times at each GdmCl concentration. In all cases, the
GdmCl c oncentration w as determined using a refracto-
meter [ 40].
Analysis of kinetic data
The kinetic traces were analysed using a multiexponential
fitting procedure (two o r three components). The kinetic
data wer e analysed assuming an apparent two-state
model using standard equations described in detail by
others [41,43,44]. T he observed rate constant k
obs
is the
sum of t he folding and unfolding rates, k
obs
¼ k
fold
+
k
unfold
where k

M
GdmCl
and 1.81 l
M
protein and the data fitted according to the
following expressions [30]:
Ó FEBS 2004 Proline residues in ubiquitin stability and folding (Eur. J. Biochem. 271) 4475
lnk
obs
¼ lnk
o
À DGz =RT ð3Þ
where k
o
is the t emperature independent pre-exponential
factor (% 10
8
), and the temperature dependence of the
activation free energy DGà is given by:
DGz¼DHzþDC
p
zðT À 298Þ
À T½DSzþDC
p
z lnðT=298Þ ð4Þ
with DHà, DC
p
à and DSà representing the change in
activation enthalpy, heat capacity and entropy of formation
of the TSE for folding (U-à). Reported errors reflect the

GdmCl (WT*) to 2.18
M
GdmCl.
This equates to a reduction in stability o f 4 .5 ± 0.6 kJÆ
mol
)1
. In contrast, the P38A mutation re sults in a significant
increase in stability of )4.6±0.6kJÆmol
)1
.TheA37A38
double mutant is slightly less stable than WT* ( 1 . 1 ± 0. 6 k J Æ
mol
)1
), showing that the contributions from P37A and
P38A are approximately additive.
We also examined the effects of substituting a proline
residue at position 19 in the loop region connecting the
N-terminal b-hairpin to t he main a-helix (Fig. 1). Proline is
highly conserved at t his site in many s pecies; however,
in yeast ubiquitin residue 19 is serine. The mutation
S19P produces a significant increase in stability o f
)5.3 ± 0.7 kJÆmol
)1
. Thus, the P19S, P37A a nd P38A
mutations produce contrasting effects that do not appear to
simply relate to entropic factors concerning changes in
backbone flexibility.
Structural analysis of the proline mutants by NMR
NMR structural analysis was used to establish whether the
substitutions of P37 and P38 are substantially perturbing

analysis to the short helix (residues 38–40), the strong
sequential NH–NH NOEs from D39 through to Q41 are
preserved in all mutants. The P38A mutation appears to
extend the helical turn by one residue with Ala38 having a
3
J
NH–Ha
value < 6 Hz with evidence of i,i+3 NOEs to
Gln41. NOE contacts from Ala38 protons to the side chains
of Lys27 and Gln31 in the main a-helix are also evident and
confirm t hat the Ala38 methyl g roup occupies the same
hydrophobic pocket as the side chain of Pro38. Mod elling
the structure with Ala substitutions imposed on the
backbone conformation of WT* shows that the pattern of
P37
P38
P19
W45
Fig. 1. Ribbon structure modelled on the X-ray structure of human
ubiquitin [32]. The position and orientation of the side chains of Pro19,
Pro37 and Pro38 are highlighted along with the F45W mutation
(drawn using
MOLMOL
[38]). The sequences o f human and yeast
ubiquitin differ at the f ollo wing positions: P19S, E24D and A28S.
4476 M. D. Crespo et al.(Eur. J. Biochem. 271) Ó FEBS 2004
NOEs is entire ly consistent with native-like /,w angles.
Thus, w e conclude that the Pro to Ala substitutions are not
significantly perturbing the backbone conformation and
dynamics of the protein around the mutation sites and in the

acetate
buffer at 298 K and was mo nitored by
tryptophan fluorescence. Stability d ata are
shown in T able 1.
Table 1. Equilibrium stability dat a for ubiquitin mutants (pH 5. 0,
25 m
M
acetate b uffer, 298 K) determined by GdmCl denaturation
monitored b y changes i n tryptophan fluorescence.
Mutant
m
eq
a
(kJÆmol
)1
Æ
M
)1
) [D]
50%
b
DG
eq
c
(kJÆmol
)1
)
WT* 11.3 2.62 )28.6 (± 0.6)
P37A 11.9 2.21 )24.1 (± 0.5)
P38A 10.2 3.05 )33.2 (± 0.7)

helix (residues 38–40) and fourth strand of
b-sheet (re sidues 4 2–46) on the C-terminal side
of the mutation sites (Fig. 1). Differences in
chemical shifts with respect to r ando m coil
values [46,47] are plotte d against sequence
position.
Ó FEBS 2004 Proline residues in ubiquitin stability and folding (Eur. J. Biochem. 271) 4477
k
3
¼ 0.14 s
)1
, and relative amplitudes o f 1 1% and 2 %,
respectively. The k
2
and k
3
processes, also identified for
human ubiquitin [13,33], have previously been attributed to
slow rate-limiting cis–trans prolyl isomerization reactions.
However, we have shown using double-jump (interrupted
unfolding) experiments (data not shown) that k
2
is a direct
refolding event whose amplitude is unaffected by the
equilibration time of the dou ble-jump experiment. I n an
isomerization-limited process, the pop ulation of t he non-
native cis-isomer w ould be expected to build up only slowly
in the unfolded state (rate constant < 2 s
)1
[30]). While k

4
,doweseeany
evidence for deviations from a two-state model. Under
these conditions rollover effects are now apparent in the
refolding data at low denaturant concentrations (Fig. 4B),
together with burst-phase changes in the fluorescence
intensity (Fig. 5B) [33,35]. We conclude that the data
collected for yeast ubiquitin at protein concentrations
<2 l
M
are adequately described in terms of a two-state
folding model in concurrence with recent detailed studies
of human ubiquitin [13,14,35].
Kinetic experiments on the Pro mutants reveal that the
changes in protein stability associated with the P ro substi-
tutions are largely manifested in effects on the unfolding
rather than refolding kinetics (Table 2). The chevron plot
analysis shown in Fig. 6 reveals little change in the m-values
for either the refold ing or unfolding phas es, indicating that
the TSE is not significantly perturbed by the mutations, nor
do we see any evidence for deviation from the two-state
folding model using the criteria described above.
Tolerance to substitutions at the P37P38 site
Kinetic studies with other systems, aimed at probing the
nature of the TSE for folding, have focused primarily on
nondisruptive Ala or Gly s ubstitutions, a rguing that more
sterically demanding substitutions have the potential to
shift the position of the TSE along the folding pathway or
even stabilize intermediate s tates [48,49]. We have exam-
ined the robustness of the TSE for folding in the current

¼ 1604 ± 88 JÆmol
)1
Æ
M
)1
and m
unfold
¼
2919 ± 4 3 J mol
)1
Æ
M
)1
. (B) Refolding and unfolding data f or WT* as
in (A) and in the presence of 0.4
M
Na
2
SO
4
. The data for the latter
were fitted to a three-state on-pathway model (U«I«N) in which the
intermediate state is significantly populated with an equilib-
rium constant K
UI
¼ 204. Rate constants and m-values are as
follows: m
UI
¼6992 ± 250 JÆmol
)1

these side chains. NMR analysis o f H a chemical shifts for
the SQ and QL mutants, in line with structural studies
described above, confirms that only relatively small local
perturbations to the structure have taken place. Detailed
kinetic analysis shows that the reduction in stability of these
mutants is largely manifested in perturbations to the
unfolding rates with the degree of compactness of the
TSE (a
D
) and linearity of the chevron plots very similar to
WT* (Fig. 6).
The analysis o f m ultiple mutations at a common site
(P37/P38) is conveniently expressed in terms of a Brønsted
plot, allowing the r elationship to be e xamined between the
logarithm o f the re folding and unfolding r ates and the
effect on protein stability [50]. Such a relationship should
enable us to assess the extent to which P37 and P38 are
involved in native-like contacts in the TSE. Linear
Brønsted p lots have been interpreted as indicating that
the r esidues a t the mutation site give rise to the same
degree of partial structure in the transition s tate as in WT*,
and that the substitutions are not significantly perturbing
the position o f the TS E along the folding pathway [51,52].
We have con sidered th e P37/P38 mutations simultaneously
and constructed the Brønsted plot shown in Fig. 7 on the
basis of the following:
lnk
fold
¼ lnk
fold

unfold
vs. DDG/RT (both
DDG
eq
/RT and DDG
kin
/RT; Fig. 7) are linear demonstra-
ting that all mutants show the same degree of structure
formation in the TSE, which appears to be tolerant to the
variety o f changes introdu ced. Values of b
f
¼ 1 have been
interpreted as evidence that residues at the mutation site
occupy a highly native-like environment in the TSE,
whereas much smaller values (close to zero) suggest that
these residues are largely unstructured in the rate-limiting
step for folding. The linear plots in Fig. 7 indicate a b
f
-value
of 0.09 supporting the latter model. We see that the proline
mutations produce very small effects on the folding rate of
ubiqutin with only a two-fold difference between the fastest
and slowest folding mutants. In contrast, we see a 26-fold
range in the rate of unfolding.
This trend i s also r eflected i n t he effects o f t he S19P
mutation on the kinetics. The significant stabilizing effect
of this mutat ion ()5.3 kJÆmol
)1
) is also manifested largely
in a deceleration of the unfolding rate. By a nalogy with the

flexible in the TSE, with native-like contacts and back-
bone F,w angles becoming consolidated at a late stage in
the folding process.
Fig. 5. Amplitude of the raw fluorescence signal for the refolding of
WT* ubiquitin. In the absence (A) and presence of 0.4
M
Na
2
SO
4
(B) a t
298 K in 25 m
M
acetate buffer, pH 5 .0. The b lac k dots and solid line
are the fit to the refolding data enabling a two-state equilibrium
unfolding curve to be constructed. The dashed line (circles) is a linear
fit in (A) t o the denaturant dependence of the fluorescence signal of the
unfolded state. In (B), in the presence of stabilizing salt, the fluores-
cence s ignal of th e unfolded state (dashed line, circles) shows deviations
from a linear extrapolation, providing evidence for a burst phase
around 1
M
GdmCl w here the fluorescence intensity increases signifi-
cantly as the collapsed state is destabilized by t he denaturant. This is
consistent w ith the curvature observed in the corresponding chevron
plot in Fig. 4B and formation of an intermediate co llap sed state at low
denaturant concentrations.
Ó FEBS 2004 Proline residues in ubiquitin stability and folding (Eur. J. Biochem. 271) 4479
Activation parameters for folding
The temperature-dependence of the refolding kinetics were

Æmol
)1
, respectively) reflects a small f avour-
able stabilization of the TS, however, t he enthalpy te rm
(66 ± 2 and 67 ± 2 kJÆmol
)1
, respectively) is highly unfa-
vourable to folding and dominates the size of t he activation
barrier, DGà [9].
Discussion
Context-dependent effects of proline residues
on protein stability
Ubiquitin i s highly conserved across species with the y east
and human forms differing in on ly three r esidues (S19P,
E24D and A28S). The first of these is located in a loop
region which connects the N-terminal b-hairpin sequence
(residues 1–17) to the main a-helix (residues 24–33) (Fig. 1).
The E24D and A28S substitutions lie within the main
a-h elix. Both structures have conserved prolines (P37 and
P38) in adjacent po sitions at the N terminus of a short
a-h elix (residues 38–40) in an otherwise extended loop
region connecting the C terminus of the main a-helix to
subsequent strands of b-sheet (Fig. 1). We have investigated
the context-dependent effects of mutations at these sites on
Fig. 6. Chevron plot analysis of the logarithm
of the r efolding and unfolding rates v s. concen-
tration of denaturant (GdmCl). Da ta shown for
WT* and all ubiquitin m utants studied ( 298 K
in 25 m
M

U fi N
(s
)1
)
m
U fi N
(JÆmol
)1
Æ
M
)1
) a
D
WT* 0.0090 (± 0.0008) 2876 (± 44) 304 (± 11) 5934 (± 58) 0.67
P38A 0.0036 (± 0.0009) 2992 (± 113) 243 (± 15) 5236 (± 94) 0.64
P37A 0.042 (± 0.004) 2383 (± 51) 161 (± 11) 5881 (± 125) 0.71
AA 0.0204 (± 0.002) 2614 (± 54) 250 (± 13) 5725 (± 89) 0.69
SQ 0.066 (± 0.007) 2370 (± 61) 142 (± 16) 5904 (± 205) 0.71
QL 0.092 (± 0.008) 2555 (± 47) 271 (± 27) 6000 (± 184) 0.70
VV 0.060 (± 0.003) 2428 (± 31) 228 (± 11) 6133 (± 93) 0.71
S19P 0.0038 (± 0.0005) 2953 (± 67) 501 (± 22) 5761 (± 62) 0.66
4480 M. D. Crespo et al.(Eur. J. Biochem. 271) Ó FEBS 2004
protein stability, and their involvement in the folding
pathway from studies of refolding/unfolding kinetics. While
the single point mutation P37A is destabilizing b y
4.5 kJÆmol
)1
, in contrast the P38A mutation produces
an equal and opposite en hancement of stability of
)4.6 kJÆmol

with the side chain of Met1, which becomes more solvent
accessible when P ro is replaced with Ser. There may also be
solvation implications for the Ser hydroxyl group, which
may also contribute a small destabilizing effect. The
contrasting effects of the S19P, P37A and P38A mutations
on stability appear to reflect a c omplex balance between
entropic factors relating to changes in backbone flexibility,
changes in hydrophobic surface burial, effects on solvent
accessibility t o other polar group s and changes in intrinsic
backbone conformational preferences. These observations
are consistent with those of others that proline residues play
a variety of context-dependent roles in modulating protein
stability [10–12,16,19].
Apparent two-state model for folding of ubiquitin
There have been conflicting reports as to whether ubiquitin
folds via an apparent two-state model o r via a m ore
complex process involving a significantly populated inter-
mediate, which forms rapidly in t he dead-time of the
stopped-flow experiment [13,14,33]. In the case of the yeast
protein d escribed here, the linear dependence o f the folding
and unfolding rates on denaturant concentration ( both
GdmCl and urea), and the lack of a burst phase change in
fluorescence intensity at low denaturant concentrations, is
indicative of an apparent two-state model in which any
intermediate state is too high in energy to be significantly
populated [34,35]. However, k inetic experiments a t low
temperature, using multiple probes including CD and
SAXS, suggest rapid formation of a c ompact ensemble
which is invisible by fluorescence [55]. All of the mutants
studied here by fluorescence conform to the t wo-state

and unfolding) v s. change in stability (DDG/RT) for th e family of P 37 /
P38 mutants. DDG values were estimated from both equilibrium (cir-
cles) and kinetic data (squares). D ata were fitted t o the lin ear corre-
lations represented by equations 6 an d 7. A b
f
value of 0.09 indicates
that the loop r egion containing the two adjacent proline residues is
largely u nstructured in the rate-limiting s tep for folding.
Ó FEBS 2004 Proline residues in ubiquitin stability and folding (Eur. J. Biochem. 271) 4481
A description of the TSE for f olding of ubiquitin, at the
level of a detailed F-value analysis to map out interactions
present in the TSE, has not yet been reported. However,
human ubiquitin has been studied by Krantz et al .[56]using
a combination of w-value analysis and protein engineering
methods to introduce bis-His metal coordination sites to
identify native noncovalent interactions involved in the
folding TSE [57]. This approach, through metal complex-
ation, enables the degree of partial structure formation at
specific sites to be continuously varied over a wide range o f
relative populations such that the effects on the rate-limiting
step can be determined. The conclusions of this novel
approach are that ubiquitin folds through a native-like TSE
with a common nucleus but with heterogeneous structural
features populated according to their relative stability. A
broad TSE, a nd pathway diversity, reflects the variable
degrees of structure formation which appears to b e formed
around a common folding nucleus consisting of part of the
major helix docked against native-like b-strand structure.
Previously, HX exchange studies have suggested that the
formation of hydrogen bonded structure (and hence pro-

fold
is
characteristic of a change in heat capacity associated with
burial of hydrophobic surface area. The a
D
values d erived
from the denaturant dependence of k
fold
and k
unfold
,
namely from the m
fold
and m
unfold
values, a re consistent
with a compact TSE (a
D
in the range 0.66–0.71). T he
temperature dependence of the refolding r ate enables us to
estimate a DC
p
à of )2.1 (± 0 .3) to )2.4 (± 0.5) k JÆK
)1
Æ
mol
)1
for W T and the A37A38 double mutant. Despite
the small fitting errors, the estimated DC
p

p
à and large
a
D
are all consistent with hydrophobic surface burial
driving t he folding polypeptide chain o ver the transition
state energy b arrier. We have shown that the proline
residues play a passive role in the apparent two-state
folding of ubiquitin, forming native-like contacts at a late
stage in the folding process, despite the observation that
mutations produce significant and highly context-depend-
ent effects on protein stability.
Acknowledgements
MDC thanks the University of Nottin gham, Astex Technology Ltd.
and Roche Products Ltd. for funding, GWP thanks the EPSRC and
GlaxoSmithKline for financial support, and RB acknowledges the EU
for a Ma rie-Curie individual research f ellowship.
References
1. Ge llman, S.H. (1998) Minimal mod el sy stems fo r b-sheet
secondary structure in proteins. Curr. Opin. Chem. Biol. 2, 717–
725.
2. Pal, D. & Chakrabarti, P. (1999) Cis-peptide bonds in proteins:
residues involved, their conformations, interactions and locations.
J. Mol. Biol. 294, 271–288.
3. Forood, B., Feliciano, E.J. & Nambiar, K.P. ( 1993) Stabilisation
of a-helical structure in short peptides by end capping. Proc. Natl.
Acad. Sci. USA 90, 838–842.
4. Prieto, J . & Serrano, L. (1997) C-capping and helix stability: the
Pro C-capping motif. J. M ol. Biol. 274, 276–288.
5. Gunasekaran, K., N agarajaram, H.A., Ram akrishnan, C. &

state and three-state models for ub iquitin folding. Biochemistry 39 ,
11696–11701.
14. Krantz, B.A., Mayne, L., Rumbley, J., Englander, S.W. & S os-
nick, T.R. (2002) Fast and slow intermediate accumulation and
the initial barrier mechanism in protein folding. J. Mol. Biol. 324,
359–371.
15. Silow, M. & Oliveberg, M. (1997) Transient aggregates in protein
folding are easily mistaken for fo lding i nterm ediates. Proc. Natl.
Acad. Sc i. USA 94 , 6084–6086.
16. Kamagata, K., Sawano, Y., T anokura, M. & Kuwajima, K .
(2003) Multiple parallel-pathway folding of proline-free staphy-
loccal nuclease. J. Mol. Biol. 33 2, 1143–1153.
17. Osvath, S. & Gruebele, M. (2003) Proline can have opposite effects
on fast and slow p rotein foldin g phases. Biophys. J. 85, 1215–1222.
18. Odefey, C., Mayr, L.M. & Schmid, F.X. (1995) Non-prolyl cis-
trans peptide bond isomerisation as a rate-determining step in
protein unfolding and refolding. J. M ol. Biol. 245, 69–78.
19. Agah, S., Larson, J.D. & Henzl, M.T. (2003) Impact of proline
residues on p arvalbumin stability. Biochemistry 42, 10886–10895.
20. Sanchez, I.E. & Kiefhaber, T. (2003) Origin of u nusual /-values in
protein folding: evidence against specific nucleation evidents.
J. Mol. Biol. 33 4, 1077–1085.
21. Kiefhaber, T. (1995) Kinetic traps i n lysozyme folding. Proc. Natl.
Acad. Sc i. USA 92 , 9029–9033.
22. Levitt, M. (1981) Effect of proline residues on protein folding.
J. Mol. Biol. 14 5, 251–263.
23. Jullien, M. & Baldwin, R.L. (1981)Theroleofprolineresiduesin
the folding kinetics of bovine pancreatic t rypsin i nhibitor deriva-
tive RCAM (14–38). J. Mol. Biol. 145, 2 65–280.
24. Kreb s, H., Schmid, F.X. & J a enicke, R. (1983) Folding of

Hatfield, P.M. & Cook, W.J. (1987) Comparison of the three-
dimensional structures of human, yeast and oat ubiquitin. J. Biol.
Chem. 262, 6396–6399.
33. Khorasanizadeh, S., Peters, I.D. & Roder, H. (1996) Evidence for
a three state-model of protein folding from kinetic analysis of
ubiquitin variants with altered core residues. Nat. Struct. Biol. 3,
193–205.
34. Platt, G .W., Simpson, S.A., Layfield, R. & Searle, M.S. (2003)
Stability and folding kinetics o f a ubiquitin mutant with a strong
propensity for non-native b-hairpin conformation in the unfolded
state. Biochemistry 42, 13762–13771.
35. Went, H.M., Benitez-Cardoza, C.G. & Jackson, S.E. (2004) Is an
intermediate state populated on the folding pathway of ubiquitin?
FEBS Lett. 567, 333–338.
36. Gladwin, S.T. & Evans, P.A. (1996) Structure of the very early
protein folding intermediates: new insights through a variant of
hydrogen exchange labelling. Fold. Des. 1, 407–417.
37. Kraulis, P.J. (1989) ANSIG: a program for the assignment of
protein
1
H 2D NMR spectra by interactive computer graphics.
J. Magn. Res. 84, 627–633.
38. Koradi,R.,Billeter,M.&Wu
¨
thrich, K. (1996) MOLMOL: a
program for display and analysis of macromolecular structures.
J. Mol. Graph. 14, 51–55.
39. Santoro, M.M. & Bolen, D .W. (1988) Unfolding free e ne rgy
changes determined by the linear extrapolation method. 1.
Unfolding of phenylmethane sulfonyl a-chymotrypsin using

substitutions to investigate the struc ture o f the SH3 do main
folding transition state. J. Mol. Biol. 320, 389– 402.
49. Matouschek, A. & Fersht, A.R. (1991) Protein engineering in
analysis of protein folding pathways and stability. Methods
Enzymol. 202, 8 2–112.
Ó FEBS 2004 Proline residues in ubiquitin stability and folding (Eur. J. Biochem. 271) 4483
50. Fersht, A.R., Itzhaki, L.S., el Masry, N.F., Matthews, J.M. &
Otzen, D.E. (1994) Single versus parallel pathways of protein
folding and fractional formation of structure in the transition
state. Proc. Natl A cad. Sci. USA 91, 10426–10429.
51. Villegas, V., Martinez, J.C., Aviles, F.X. & Serrano, L. (1998)
Structure of the transition state i n t he foldin g p roce ss o f hum an
procarboxypeptidase A2 activation domain. J. Mol. Biol. 283,
1027–1036.
52. Milla, M.E., Brown, B.M., Waldburger, C.D. & Sauer, R.T.
(1995) P22 Arc repressor: transition state properties inferred from
mutational effects on the rates of protein unfolding and refolding.
Biochemistry 34 , 13914–13919.
53. Matthews, C.R. (1987) Effe cts of point mutations on t he folding of
globular proteins. Methods Enzymol. 154, 498 –511.
54. Fersht, A.R., Matouschek, A. & Serrano, L. (1992) The folding of
an enzyme. I . T heory o f p rotein en gin eering a nalysis o f s tability
and pathway of protein folding. J. Mol. Biol. 224, 771–782.
55.Larios,E.,Li,J.S.,Schulten,K.,Kihara,H.&Gruebele,M.
(2004) Multiple probes reveal a native-like intermediate during
low temperature refo lding of u biquitin. J. Mo l. Bio l. 340, 115–
125.
56. Krantz, B.A., Dothager, R.S. & Sosnick, T.R. (2004) Discerning
the s tructure and energy of multiple t ransition states in protein
folding using w-analysis. J. Mol. Biol. 337, 463–475.