Full text of DOI:10.1110/ps.9.11.2151

Protein Science ~2000!, 9:2151–2160. Cambridge University Press. Printed in the USA.
Copyright © 2000 The Protein Society
Position effect of cross-strand side-chain
interactions on b-hairpin formation
CLARA M. SANTIVERI, MANUEL RICO, and M. ANGELES JIMÉNEZ
Instituto de Estructura de la Materia, Consejo Superior de Investigaciones Científicas, Serrano 119, 28006-Madrid, Spain
~Received January 3, 2000; Final Revision September 1, 2000; Accepted September 1, 2000!
Abstract
Previous conformational analysis of 10-residue linear peptides enabled us to identify some cross-strand side-chain
interactions that stabilize b-hairpin conformations. The stabilizing influence of these interactions appeared to be greatly
reduced when the interaction was located at the N- and C-termini of these 10-residue peptides. To investigate the effect
of the position relative to the turn of favorable interactions on b-hairpin formation, we have designed two 15-residue
b-hairpin forming peptides with the same residue composition and differing only in the location of two residues within
1
the strand region. The conformational properties of these two peptides in aqueous solution were studied by H and 13C
NMR. Differences in the conformational behavior of the two designed 15-residue peptides suggest that the influence of
stabilizing factors for b-hairpin formation, in particular, cross-strand side-chain interactions, depends on their proximity
to the turn. Residues adjacent to the turn are most efficient in that concern. This result agrees with the proposal that the
turn region acts as the driving force in b-hairpin folding.
Keywords: b-hairpin; b-turn; NMR; peptide design; protein folding; side-chain interactions
Understanding the forces that govern the formation and stability of
secondary structure is essential for the rational and successful de
novo design of proteins. Analysis of the conformational behavior
of protein fragments and designed peptides has provided a large
amount of information on the formation and stability of a-helices
~Scholtz & Baldwin, 1992; Lyu et al., 1993; Zhou et al., 1993;
Baldwin, 1995; Muñoz & Serrano, 1997; Aurora & Rose, 1998!.
However, experimental data on peptides adopting monomeric
b-sheet motifs in aqueous solution have been reported only in the
1990s ~see reviews by Smith & Regan, 1997; Blanco et al., 1998;
Gellman, 1998; Lacroix et al., 1999; and references therein!, prob-
ably because sequences with high b-sheet propensities tend to
aggregate. Since the publication of the first designed b-hairpin-
forming peptide ~Blanco et al., 1993!, the design of short linear
peptides able to form b-sheets has become a subject of intense
activity. Most of the b-sheet-forming peptides reported to date
adopt a b-hairpin motif, the simplest antiparallel b-sheet structure
consisting of two antiparallel hydrogen-bonded b-strands linked
by a turn region. b-Hairpin motifs differ in the length of the turn
region and are classified according to the number of turn residues
and the hydrogen bond pattern of residues flanking the turn ~Sibanda
& Thornton, 1985, 1991; Sibanda et al., 1989!. Several studies on
model peptides highlighted the importance of the turn region in
determining b-hairpin stability ~Ramírez-Alvarado et al., 1997;
Blanco et al., 1998; Gellman, 1998; Griffiths-Jones et al., 1999!.
Moreover, the NMR investigation of a series of model b-hairpin-
forming peptides ~de Alba et al., 1996, 1997a, 1999a!, showing
that they were able to adopt different b-hairpin structures when
differing only on their turn sequence, demonstrated that the turn
sequence determines the b-strand alignment in the b-hairpin. Fur-
ther evidence came from ubiquitin-derived peptides that showed
different strand register depending on the turn residues ~Searle
et al., 1995; Haque & Gellman, 1997!. Cross-strand side-chain
interactions are, at least in our studied peptide system, less relevant
for b-hairpin formation ~de Alba et al., 1997b!, though they cer-
tainly contribute to b-hairpin stability ~Ramírez-Alvarado et al.,
1996, 1999; Blanco et al., 1998; Gellman, 1998!. We have indeed
identified some favorable and unfavorable cross-strand side-chain
interactions from the conformational analysis of a series of deca-
peptides able to form b-hairpins ~de Alba et al., 1997b!. In that
study, the influence of cross-strand side-chain interactions on
b-hairpin stability appeared to vanish when the interaction was
placed at the N- and C-termini. To investigate how the location of
cross-strand side-chain interactions relative to the turn region af-
Reprint requests to: M. Angeles Jiménez, Instituto de Estructura de la
Materia, Consejo Superior de Investigaciones Científicas, Serrano 119,
28006-Madrid, Spain; e-mail: angeles@malika.iem.csic.es
Abbreviations: 1D, one-dimensional; 2D, two-dimensional; Dd
~ϭ dobserved Ϫ drandom coil!, conformational shift; COSY, homonuclear cor-
related spectroscopy; HSQC, heteronuclear single quantum coherence
spectroscopy; Mav, average molecular weight; Mth, theoretical molecular
weight; NOE, nuclear Overhauser enhancement; NOESY, nuclear Over-
hauser enhancement spectroscopy; ppm, parts per million; RMSD, root-
mean-square deviation; ROESY, rotating frame NOE spectroscopy; TOCSY,
total correlation spectroscopy; TSP, sodium @3-trimethylsilyl 2,2,3,3-2H#
propionate.
2151

2152
C.M. Santiveri et al.
Fig. 1. Peptide sequences. Residues that lengthen peptides 3–4 relative to peptides 1–2 are underlined and those located at different
positions in peptides 1 and 3 relative to peptides 2 and 4, respectively, are in bold.
fects their effect on b-hairpin formation, we have designed two
longer b-hairpin forming peptides. These peptides have the same
amino acid composition and differ only in the position of two
residues at the N-terminal strand. In this paper, we describe the
design and NMR conformational investigation of these two 15-
residue peptides. Analysis of the conformational properties of these
pentadecapeptides indicates that the contribution of cross-strand
side-chain interactions to b-hairpin formation depends on their
relative position to the turn, being more stabilizing, if favorable,
and more destabilizing, if unfavorable, when closer to the turn.
forming decapeptides 1 and 2 ~de Alba et al., 1997b; Fig. 1!. The
b-hairpin conformations adopted by these peptides are schemati-
cally shown in Figure 2. b-Sheet propensities and solubility crite-
ria were considered in selecting the five new residues, the sequences
SE and VTE added to the N- and C-end, respectively. They contain
two residues ~V and T! with high intrinsic b-sheet propensities
~Chou & Fasman, 1974; Kim & Berg, 1993; Minor & Kim, 1994;
Muñoz & Serrano, 1994; Smith et al., 1994; Swindells et al.,
1995!, an S residue that is the most hydrophilic residue with some
b-sheet propensity, and two E residues whose charges should im-
prove peptide solubility.
Results
Aggregation test
Peptide design
The absence of aggregation in peptides 3 and 4 in aqueous solution
was checked by recording 1D H NMR spectra of peptides 3 and
1
Pentadecapeptides 3 and 4 were designed by adding two and three
residues at the N- and C-termini, respectively, of the b-hairpin
4 at two different concentrations, 2 and 0.2 mM. The linewidths
Fig. 2. Schematic representation of the peptide backbone conformations of ~A! the 3:5 b-hairpin formed by peptides 1–2 ~left! and
expected for peptides 3–4 ~right!, and ~B! the 4:4 b-hairpin formed by peptide 1 ~left! and expected for peptides 3–4 ~right!. b-Sheet
hydrogen bonds are indicated by dotted lines and the expected long-range NOEs involving backbone protons by black arrows. A pair
of facing residues is in a hydrogen bonded site when their CO and NH are hydrogen bonded, and in a nonhydrogen bonded site when
they are not hydrogen bonded.

Cross-strand side-chain interactions and b-hairpin folding
2153
and chemical shifts at the two concentrations were identical, indi-
cating the monomeric state of the two peptides. The similarity
between the average molecular weights obtained from sedimenta-
tion equilibrium experiments performed with the dilute samples of
the peptides and the molecular weights calculated on the basis of
the amino acid composition provided further confirmation of their
monomeric state ~peptide 3, Mth0Mav ϭ 1.09; peptide 4, Mth0Mav ϭ
0.97!.
peptide 4 ~Fig. 3! indicates the presence of a 3:5 b-hairpin. The
almost zero DdCaH value observed for residue N6 of peptide 3
~Fig. 3!, however, suggests that peptide 3 is adopting a 3:5 and
a 4:4 b-hairpin, since a population-weighted average DdCaH value
is expected in the presence of the two b-hairpins. The evidence
about the type of b-hairpin provided by the DdCaH profiles, in
particular when a mixture of different b-hairpins can coexist, is
not unambiguous, but it is useful to check their consistency with
other NMR parameters.
The formation of a 3:5 b-hairpin can explain the negative DdCaH
value observed for residue S5 in peptide 3 and I5 in peptide 4 ~see
above; Fig. 3!. These residues are contiguous to the aromatic res-
idue Y4 and in front of W11 in a 3:5 b-hairpin ~Fig. 2; Table 1!,
and hence, susceptible to their ring current effects. The ring current
effects on all the CaH protons of the 3:5 b-hairpin structure cal-
culated for peptide 4 ~see below! were calculated by the Johnson–
Bovey model using the MOLMOL program ~Koradi et al., 1996!.
As experimentally found, the CaH proton of I5 has a large upfield
shift, mainly due to W11. Residues facing W in reported 3:5
b-hairpin-forming peptides showed the same behavior ~de Alba
et al., 1996, 1997b, 1999a!. The positive DdCaH value found for
T12 in peptide 4 is very small, almost zero, and it can also be
accounted for their location adjacent to W11 and facing Y4 in a 3:5
b-hairpin.
CaH conformational shifts
The CaH conformational shift profiles ~DdCaH ϭ dobserved
Ϫ
drandom coil! of peptides 3 and 4 ~Fig. 3! present negative values
at residues 7–9 that are indicative of a chain-bend region, and
positive values at residues 2–6 and 10–14 as expected for
b-strands region ~Case et al., 1994; Wishart & Sykes, 1994;
Szilágyi, 1995!. The N- and C-terminal residues are excluded
due to charged end effects. Therefore, with the single exception
of the residue at position 5, the DdCaH profiles observed for
peptides 3 and 4 are characteristic of b-hairpins.
The DdCaH profiles expected for a 4:4 b-hairpin and for a 3:5
b-hairpin, the type of hairpins that peptides 3 and 4 will most
probably adopt, differ at a single position. The f and c angles
of the first turn residue, N6 in peptides 3 and 4, are in the
b-region of the Ramachandran map in a 3:5 b-hairpin and in the
aR region in a 4:4 b-hairpin ~Sibanda & Thornton, 1991! that
leads, respectively, to positive and negative DdCaH values. The
positive DdCaH value that is observed for the N6 residue in
13Ca conformational shifts
13
13
The Ca conformational shifts ~Dd
ϭ dobserved Ϫ drandom coil!
Ca
have been recognized as a useful and sensitive parameter to detect
and identify secondary structure ~Buckley et al., 1993; Gronenborn
& Clore, 1994; Yao et al., 1997; Guerois et al., 1998!. Since the
main factors affecting the 13Ca chemical shifts are the f and c
backbone dihedral angles ~Spera & Bax, 1991; de Dios et al., 1993;
Wishart & Sykes, 1994; Szilágyi, 1995; Iwadate et al., 1999!, they
are less dependent on ring current effects than the CaH chemical
shifts.
13
The Dd
profiles of peptides 3 and 4 ~Fig. 3! are negative at
C
a
residues 2–6 and 10–14 as expected for b-strands ~Spera & Bax,
1991; de Dios et al., 1993; Wishart & Sykes, 1994; Szilágyi, 1995;
Iwadate et al., 1999!. As in the case of DdCaH profiles, we do not
consider the N- and C-terminal residues. With regard to the turn
region, there is a large positive value for S7, the i ϩ 1 turn residue,
a negative value for D8, the i ϩ 2 residue, and a slightly positive
13
value for G9, the i ϩ 3 turn residue. This is the Dd
pattern
C
a
expected for 3:5 and 4:4 b-hairpins considering their characteristic
f and c angles ~Sibanda & Thornton, 1991! and the effect of
13
backbone dihedral angles on the d
value ~Spera & Bax, 1991;
C
a
13
Iwadate et al., 1999!. According to it, the ⌬␦ ␣ values at the first
C
turn residue, N6 in peptides 3 and 4, are expected to be negative in
3:5 b-hairpins and positive in 4:4 b-hairpins. Therefore, the neg-
13
ative Dd a and large ~in absolute value! found for the N6 residue
C
in peptide 4 ~Fig. 3! indicates the presence of a 3:5 b-hairpin. The
13
almost null Dd a value observed for N6 in peptide 3 is consistent
C
Fig. 3. ~A! CaH and ~B! 13Ca conformational shifts ~Dd ϭ d~obs!
Ϫ
with a mixture of 3:5 and 4:4 b-hairpins, because a population-
weighted average Dd
two b-hairpins, as in the case of DdCaH conformational shifts.
d~random coil!, ppm, where dCaH~random
were taken from Bundi and
from Wishart et al., 1995! as a
coil!
13
value is expected in the presence of the
C
a
13
Wüthrich ~1979! and d
C ~random coil!
a
function of sequence for peptides 3 ~in black! and 4 ~in grey! at pH 5.5
and 5 8C. For Gly, the DdCaH shown corresponds to the average value of
the two a-protons. Turn residues in the 3:5 b-hairpin formed by pep-
tides 3 and 4 are boxed. Inset shows the X3 and X5 amino acids cor-
13
NOE data
responding to peptides 3 and 4. The DdCaH and Dd
values shown for
C
a
NOE data provide the most conclusive NMR evidence for the
formation of b-hairpin structures. The set of observed NOEs per-
W11 in peptide 3 are at pH 5.5 and 20 8C because CaH signal is under
water resonance at 5 8C.

2154
C.M. Santiveri et al.
Table 1. Peptide sequences and populations of b-hairpin conformations formed by peptides 1-4 in D2Oa
Population
Population
estimated from
estimated from
Peptide
1c
4:4 b-hairpin strand
Turn
NOE intensityb
3:5 b-hairpin strand
Turn
NOE intensityb
I1
Y2
S3
||
N4 S5
I1
S1
I3
Y2
||
T10 W9
S3
N4
|
T8
S5
||
32%
D6
D6
D8
D8
27%
T10 W9 T8
G7 D6
N4 S5
G7
S5
2c
3d
4g
S1
Y2
I3
||
Y2
||
T10 W9
I3
N4
|
T8
||
Nondetected
6%e
47%
13%f
63%h
T10 W9 T8
G7 D6
N6 S7
G7
S7
S1
E2
I3
Y4
S5
||
S1
E2
||
Y4
||
S5
N6
|
||
||
E15 T14 V13 T12 W11 T10
G9 D8
N6 S7
E15 T14 V13 T12 W11 T10
G9
S7
S1
E2
S3
Y4
I5
S1
E2
S3
Y4
I5
N6
||
||
||
Nondetected
||
||
|
E15 T14 V13 T12 W11 T10
G9 D8
E15 T14 V13 T12 W11 T10
G9
aTurn residues are indicated in bold. Vertical lines indicate hydrogen bonds between strand residues. The differential cross-strand interactions between
the hairpins adopted by peptides 3 and 4 are underlined. Data for peptides 1 and 2 were taken from de Alba et al. ~1996, 1997b!.
bAs described in text.
cpH 6.3 and 2 8C.
dpH 4.3 and 20 8C.
eFrom CaHE2-CaHV13 NOE intensity.
f From CaHS5-CaHW11 NOE intensity.
gpH 5.5 and 5 8C.
hFrom the average of the populations obtained from the intensities of CaHS3-CaHV13 and CaHI5-CaHW11 NOEs.
mits the unambiguous determination of the strand register and the
type of b-hairpin adopted by a given peptide, since different types
of b-hairpin structures lead to different NOEs involving the back-
bone protons ~daa~i,j!, daN~i,j!, and dNN~i,j!; Fig. 2!. The NOE data
described herein come from joint analysis of NOESY and ROESY
spectra recorded under the same experimental conditions. ROESY
experiments permit one to check whether a particular cross peak
corresponds to a true NOE connectivity or to spin diffusion pro-
cesses. The same set of NOE cross peaks was observed in the
ROESY and NOESY spectra of peptides 3 and 4 under the same
experimental conditions.
The NOESY spectrum of peptide 4 in aqueous solution shows
long-range NOEs involving CaH ~CaHS3-CaHV13 and CaHI5-
CaHW11; Fig. 4B! and amide protons ~NHY4-NHT12! that are
compatible with the formation of a single b-hairpin structure, the
3:5 b-hairpin ~Fig. 2!. The NOEs involving side-chain protons
include NOEs between residues S3-V13, Y4-T12, and I5-W11 that
face each other in the b-hairpin 3:5 and between S3-W11, Y4-N6,
Y4-T14, W11-V13, and V13-E15 whose side chains are on the
same side of the b-sheet ~Table SM4 in Supplementary material in
the Electronic Appendix!. This provides further support for the
formation of a 3:5 b-hairpin. No NOE indicative of even a small
population of a 4:4 b-hairpin was detected.
not with a 3:5 b-hairpin ~Table SM3 in Supplementary material in
the Electronic Appendix!.
Estimation of b-hairpin populations
Peptide 3 in aqueous solution at pH 4.3 and 20 8C adopts about 6%
of a 4:4 b-hairpin and 13% of a 3:5 b-hairpin, as estimated from
the CaHE2-CaHV13 and the CaHS5-CaHW11 NOE intensities
~Table 1; see Materials and methods!, respectively. These confor-
mations coexist in equilibrium with the random coil. The other
CaHi-CaHj NOEs characteristic of 4:4 and 3:5 b-hairpins ~Fig. 2!
could not be observed in peptide 3 due to signal overlapping. In
aqueous solution at pH 5.5 and 5 8C, the fraction of peptide 4
forming a 3:5 b-hairpin is 63% ~Table 1! on the basis of the
CaHI5-CaHW11 and CaHS3-CaHV13 NOE intensities. The pop-
ulation estimated from the intensity of the CaHI5-CaHW11 NOE
corresponding to the inner residues ~73%! is slightly higher than
that evaluated from the intensity of the CaHS3-CaHV13 NOE
involving the residues at the middle of the strands ~54%!. This
difference, though small, suggests the existence of an increasing
flexibility at the strands when approaching the chain ends. This
effect is analogous to the fraying observed in a-helical conforma-
tions ~Baldwin, 1995!, and it was previously observed in one
b-hairpin forming peptide ~de Alba et al., 1997a!. Unfortunately,
we could not measure the hairpin populations adopted by peptides
3 and 4 at identical conditions of pH and temperature due to
overlap of the CaH of W11 and the water signal ~see below!.
We have also evaluated the overall b-hairpin populations for the
two peptides from the CaH and the 13Ca conformational shifts
observed in aqueous solution at pH 5.5 and 5 8C ~Fig. 3!, as de-
scribed in Materials and methods. The evaluated values for peptide
4 are 62 and 51% of b-hairpin, as estimated from the CaH and the
In contrast, the pattern of NOEs observed for peptide 3 in aque-
ous solution is not compatible with a single preferred structure.
The observed set of NOEs includes NHE2-NHT14 and CaHS5-
CaHW11 NOEs indicative of a 3:5 b-hairpin, as well as a CaHE2-
CaHV13 NOE characteristic of a 4:4 b-hairpin ~Figs. 2, 4A;
Table SM3 in Supplementary material in the Electronic Appendix!.
In addition, this peptide shows NOEs involving side-chain protons
that can arise from a 3:5 b-hairpin, but are incompatible with a 4:4
b-hairpin, as well as NOEs compatible with a 4:4 b-hairpin, but

Cross-strand side-chain interactions and b-hairpin folding
2155
Fig. 5. Stereoscopic view of the superposition of all heavy atoms of the
best 20 structures calculated for the 3:5 b-hairpin formed by peptide 4.
Backbone atoms are shown in black and side-chain atoms in red.
lation of b-hairpin higher at pH 5.3 than at pH 4.3 ~de Alba et al.,
1995, 1996!.
Structure calculation
The high population of b-hairpin adopted by peptide 4 makes
meaningful to perform a structure calculation. To this end, we used
the distance restraints provided by the observed nonsequential NOEs
~Table SM4 in Supplementary material in the Electronic Appen-
3
dix!, and the dihedral constraints coming from the large JCaH-NH
coupling constant measured for residues E2, Y4, T10, W11, T12,
and V13 ~Ͼ8 Hz! at the b-strands and the small 3JCaH-NH found for
residue S7 ~Ͻ5 Hz! at the turn. Intraresidue and sequential NOEs
were excluded because random coil conformations contribute to
their intensities. The resulting structure is well defined. The RMSD
of the best 20 calculated structures is 0.9 6 0.2 Å for backbone
atoms and 1.8 6 0.4 Å for all heavy atoms ~Fig. 5!.
No structure calculation was performed for peptide 3 because of
the co-existence of two b-hairpins in aqueous solution.
Fig. 4. Regions of NOESY spectra of ~A! peptide 3 ~2 mM, D2O, pH 4.3,
20 8C, 200 ms mixing time! and ~B! peptide 4 ~2 mM, D2O, pH 5.5, 5 8C,
200 ms mixing time!. Nonsequential CaH-CaH NOE cross peaks are boxed.
Discussion
The NMR conformational analysis of the peptides 3 and 4, whose
sequences differ only in the position of two strand amino acids
~Fig. 1!, demonstrates their different abilities to form 3:5 and 4:4
b-hairpins ~Table 1!. Before going into the analysis of the factors
contributing to the distinct conformational behavior of these two
peptides, it is convenient to analyze the accuracy of the population
estimates and the applicability of the two-state model to our pep-
tide system.
13Ca conformational shifts, respectively, which is in good agree-
ment with the population of 3:5 b-hairpin obtained from the NOE
intensities ~Table 1!. The corresponding values for peptide 3 at
pH 5.5 and 5 8C are 44 and 44%, as evaluated from the CaH and
the 13Ca conformational shifts, respectively. Taking into account
that peptide 3 adopts a mixture of two b-hairpins ~see above!, these
values correspond to the sum of the populations of 4:4 and 3:5
b-hairpins and are higher than the sum of the populations esti-
mated from NOE intensities in spectra taken at 20 8C and pH 4.3
~Table 1!. This result was expected since b-hairpin populations on
one side increase at low temperature, and on the other the forma-
tion of b-hairpin is pH-dependent in peptides containing the NXDG
sequence, where X is P or S, at the turn region, being the popu-
Estimation of b-hairpin populations and validity
of the two-state transition approximation
There is no well-established method to quantify the population
of b-hairpin adopted by peptides ~de Alba et al., 1996, 1997a,
1997b, 1999b; Ramírez-Alvarado et al., 1997, 1999; Blanco et al.,

2156
C.M. Santiveri et al.
1998; Gellman, 1998; Lacroix et al., 1999!. The most widely
used procedure is based on the intensity of the CaHi-CaHj NOEs,
where i and j are facing residues in a nonhydrogen-bonded site
~Searle et al., 1995; de Alba et al., 1996, 1997a, 1997b; Ramírez-
Alvarado et al., 1996; Maynard et al., 1998!. Apart from the
experimental error in the NOE intensity measurement, the main
problem of this method lies in the fact that small differences in
the actual CaHi-CaHj distance relative to the reference value for
the folded state ~taken as the averaged value in antiparallel
b-sheets in proteins! translate into large variations in population
because NOE intensities are inversely proportional to the sixth
power of the interproton distance. This dependence, however,
makes correct the assumption that the CaHi-CaHj distance in the
unfolded state is large enough to make negligible the NOE in-
tensity. Other error source is the potential existence of different
local correlation times along the peptide chain and0or between
the folded and the unfolded states. The overall b-hairpin popu-
lations can be independently evaluated on the basis of the CaH
and the 13Ca conformational shifts averaged over the b-strand
residues since these shifts are dependent on the distinct environ-
ment corresponding to the different elements of secondary struc-
ture. The problems with the d-based quantification methods arise
from the lack of good reference values for both the random coil
and the folded state ~Ramírez-Alvarado et al., 1997, 1999; Gell-
man, 1998!. Model peptides that are specific of the correspond-
ing b-hairpin system ~Ramírez-Alvarado et al., 1997; Syud et al.,
1999! have been used as references. In analogy to the quantifi-
cation of a-helical conformations ~Jiménez et al., 1993, 1994;
Muñoz et al., 1995; Lee & Cao, 1996!, we have used as refer-
ence for the folded state the averaged CaH conformational shifts
observed in protein b-sheets ~Wishart et al., 1991!, and the
d-values measured in random coil tetrapeptides ~Wüthrich, 1986!
for the unfolded state. The use of the d-values reported for other
two different sets of nonstructured peptides ~Merutka et al., 1995;
Wishart et al., 1995! as reference for the unfolded state hardly
varies the b-hairpin population estimates. Thus, the range ob-
tained for the b-hairpin population evaluated for peptide 3 at
pH 5.5 and 5 8C using the different random coil references is
44–50% and for peptide 4 under identical conditions 62–65%.
13Ca conformational shifts have been previously used to quan-
tify b-sheet populations ~de Alba et al., 1999b!, taking as ran-
dom coil values those reported by Wishart et al. ~1995! and as
reference for 100% b-hairpin the averaged 13Ca conformational
shift observed in protein b-sheets ~Spera & Bax, 1991!. In the
case of peptides that adopt a mixture of b-hairpins, as peptide 3,
the populations estimated on the basis of d values correspond to
the total hairpin population, since the d-values are linear aver-
ages over all the conformers in fast exchange. However, the
characteristic NOEs of each type of hairpin are unique probes to
separately quantify the corresponding population.
the appropriate protons in the strands! strongly suggests a two-
state behavior. Partially folded structures with, for example, one
13
strand formed but no hairpin would lead to large Dd
confor-
C
a
mational shifts in absolute value and zero NOE intensity. Further
evidence about two-state behavior would be provided by the prac-
tical coincidence of the individual residue estimates for the folded
populations obtained by any quantification method. However, the
lack of appropriate reference values for the folded and unfolded
states ~see above! is more problematic when considering individ-
ual residues. To circumvent this problem, the ratio between the
CaH conformational shifts at two different experimental condi-
tions, what gives the ratio of populations at the two conditions,
may be used ~Maynard et al., 1998; Griffiths-Jones et al., 1999;
Syud et al., 1999!. b-Hairpin folding can be considered two-state
if the obtained ratio is the same for all residues and equal to the
ratio of the populations obtained by other quantification method. In
that way, the reference for the folded state is not needed and only
the errors in the reference for the unfolded state still affect the
estimates of population. In peptide 4 in aqueous solution at pH 5.5,
the overall b-hairpin population evaluated from the CaH confor-
mational shifts at the strands at 5 8C ~62%, see above! is 1.9 times
that at 35 8C ~33%; see Materials and methods!, and the ratio
DdCaH~58C!0DdCaH~358C! for each residue is within the range 1.5–
2.7 ~average ratio 2.1 6 0.6!. The exceptions are residues S1, D8,
W11, T12, and E15, with a ratio close to 1, what is due to their
CaH d-values being almost invariant with temperature, and T10,
with a ratio of 3.7, which is affected by a large error due to its
almost zero value for DdCaH~358C!, as found in other b-hairpin
systems ~Maynard et al., 1998; Griffiths-Jones et al., 1999!. This
result reinforces the idea that b-hairpin folding in peptide 4 ap-
proximates to a two-state transition.
Contributions to 4:4 and 3:5 b-hairpin formation
in peptides 3 and 4
To investigate the effect of the position of stabilizing interactions
relative to the turn in b-hairpin formation, we carried out the NMR
conformational analysis of two peptides, 3 and 4, whose sequences
differ only in the position of two strand amino acids ~Fig. 1!.
Taking into account the turn sequence of these peptides, they may
adopt two different types of b-hairpins, a 4:4 and a 3:5 ~Fig. 2!.
The analysis of several NMR parameters, nonsequential NOEs,
13Ca conformational shifts, and CaH conformational shifts, dem-
onstrates that, in aqueous solution, peptide 3 adopts two folded
structures, a 4:4 and a 3:5 b-hairpin, and peptide 4 only forms a 3:5
b-hairpin. The 3:5 b-hairpins formed by the peptides 3 and 4 are
populated at a quite different extent ~Table 1!. A change in the
position of two amino acids at the N-terminal strand greatly affects
the conformational properties of the peptides. It has already been
shown that the geometry of the type I ϩ G1 b-bulge turn present
in the 3:5 b-hairpin is more appropriate for b-hairpin formation
than the topology of the type I b-turn present in 4:4 b-hairpins ~de
Alba et al., 1997b!. Therefore, it is reasonable that, as experimen-
tally found, peptides 3 and 4 having the same sequence in the turn
region and the same amino acid composition in the strands are both
able to adopt a 3:5 b-hairpin, whereas only peptide 3 forms a 4:4
b-hairpin. The fact that there was no correlation between an in-
crease in the population of 4:4 b-hairpin and a decrease in the
population of 3:5 b-hairpin in a series of 10-residue peptides ~de
Alba et al., 1997b! suggested that the formation of 4:4 and 3:5
b-hairpins is, to a first approach, independent of each other. Mak-
The d-values of 13Ca carbons report on the local backbone
conformation through their dependence on f and c angles ~Spera
& Bax, 1991; de Dios et al., 1993; Case et al., 1994; Wishart &
Sykes, 1994; Szilágyi, 1995; Iwadate et al., 1999!, and any pref-
erentially populated conformer will contribute to the observed
d-values. In contrast, the long-range NOEs, such as the CaHi-CaHj
NOE, arise only from those conformations, like the b-hairpins,
with a short distance between the two protons. The agreement in
the overall b-hairpin population evaluated for peptide 4 from 13Ca
d-values ~probes for extended strands! with the one calculated
from long-range CaHi-CaHj NOEs ~probes for proximity between

Cross-strand side-chain interactions and b-hairpin folding
2157
ing this assumption, it is feasible to analyze the factors affecting
the formation of each hairpin independently.
stabilizing interaction, such as S-T, can be larger when adjacent to
the turn.
Among the several factors that contribute to the overall stability
of 4:4 b-hairpins, the intrinsic b-sheet, and b-turn propensities, the
hydrogen-bonding network and cross-strand side chain–side chain
interactions, though the location of two of them differs, are iden-
tical for peptides 3 and 4. In the case of 3:5 b-hairpins, the cross-
strand side chain–side chain interactions are different for peptides
3 and 4. The packing of side chains on one face of the 4:4 b-hairpin,
which were composed of residues 1, 3, 5, 10, 12, and 14, and on
one face of the 3:5 b-hairpin, which was formed by residues 1, 3,
5, 11, 13, and 15, is different in peptides 3 and 4 ~Fig. 2!. This, in
turn, could lead to differences in buried hydrophobic surface areas.
Lower stability would be expected for the b-hairpin that buries less
hydrophobic surface area upon formation. The burial of hydropho-
bic surface area has been proposed to contribute to the 2:2 b-hairpin
stability in other model peptides ~Ramírez-Alvarado et al., 1996;
Maynard et al., 1998; Griffiths-Jones et al., 1999!. To check if, in
our peptide system, the hydrophobic effect is responsible for the
stability differences observed between the two 4:4 b-hairpins and
the two 3:5 b-hairpins, we have built four sets of model structures,
corresponding to the ideal 4:4 and 3:5 b-hairpins for the sequences
of peptides 3 and 4 ~see Materials and methods!. The hydrophobic
surface areas buried for the two sets of 4:4 b-hairpins modelled as
well as for the sets of 3:5 b-hairpins are very similar and within the
experimental error. They are also very close to the hydrophobic
surface area buried upon formation of the 3:5 b-hairpin structure
adopted by peptide 4, as calculated on the basis of the NMR
experimental restraints. Therefore, although these results do not
constitute definitive evidence, they suggest that the hydrophobic
effect does not account either for the distinctive abilities of pep-
tides 3 and 4 to adopt a 4:4 b-hairpin or for the differences in 3:5
b-hairpin stability between peptides 3 and 4. If the hydrophobic
effect contribution to the experimentally observed differences in
b-hairpins stability is negligible, the remaining distinctive factors
that can play a role in b-hairpin formation are the nature and
location of cross-strand interactions.
As a consequence of our design strategy, the potential 4:4
b-hairpin for peptide 4 differs from that adopted by peptide 3 only
in the position of S-T and I-T interactions relative to the turn
~Table 1; Fig. 2!. Both interactions are at hydrogen-bonded sites.
When the favorable S-T interaction ~de Alba et al., 1997b! is
adjacent to the turn, as occurs in peptide 3, the 4:4 b-hairpin is
present in aqueous solution. In contrast, no population of 4:4
b-hairpin is detected for peptide 4 where the favorable S-T inter-
action is shifted to the middle of the strands and the unfavorable
I-T interaction ~de Alba et al., 1997b! is adjacent to the turn. This
result parallels the conformational behavior previously found for
peptides 1 and 2 relative to 4:4 b-hairpin formation ~Table 1; de
Alba et al., 1997b!. In these shorter peptides, the terminal location
of the stabilizing S-T interaction in the 4:4 b-hairpin expected for
peptide 2 suggests that fraying is the main reason that the inter-
action is inefficient at stabilizing the hairpin. However, the fraying
effect is minimized in peptide 4 where the stabilizing S-T inter-
action is at the middle of the strands ~Table 1!. Thus, the differ-
ential ability of peptides 3 and 4 to adopt a 4:4 b-hairpin suggests
that the effects of cross-strand side-chain interactions on b-hairpin
formation do depend on their proximity to the turn. A destabilizing
interaction, such as I-T, adjacent to the turn can hinder b-hairpin
formation, if the turn topology is not very appropriate for b-hairpin
formation as occurs in 4:4 b-hairpins, while the efficiency of a
Analysis of cross-strand interactions in the adopted 3:5 b-hairpins
is not straightforward, because the position swap between S and I
leads to different interactions and not just to a change in their
location relative to the turn. The 3:5 b-hairpin formed by peptide
3 contains the I-V and S-W interactions in a nonhydrogen-bonded
site and peptide 4, S-V, and I-W ~Table 1!. Considering statistical
analysis of pairwise interactions in antiparallel b-sheets ~Wouters
& Curmi, 1995; Hutchinson et al., 1998!, I-V in a nonhydrogen-
bonded site is favorable and S-V unfavorable. Data for S-W and
I-W in a nonhydrogen-bonded site, which are not statistically sig-
nificant, suggest an almost equal and null effect of these inter-
actions on b-sheet stability. According to this, S-W and I-W, the
interactions closer to the turn, will contribute equally to b-hairpin
formation. Then, the population of 3:5 b-hairpin should be higher
in peptide 3, which contains the favorable I-V interaction than in
peptide 4 with the unfavorable S-V, but the opposite result is ex-
perimentally found ~Table 1!. A reasonable explanation comes
from the fact that the I residue, which has higher intrinsic b-sheet
propensity than S, is closer to the turn region in peptide 4 than
peptide 3, suggesting that residues with higher intrinsic b-sheet
propensities are more effective when closer to the turn. This is in
analogy to the conclusions about the larger effectiveness of the
cross-strand interactions when closer to the turn previously de-
duced from the 4:4 b-hairpins. It might also be possible that the
interaction I-W is more stabilizing than S-W, and that this could not
be inferred as a consequence of shortage of data in the statistical
analysis. If so, the effect of cross-strand side-chain interactions on
b-hairpin formation would depend on their proximity to the turn.
The effectiveness of the interaction I-W in stabilizing the 3:5
b-hairpin formed by peptide 4 could be related to the fact that I and
W are the two largest hydrophobic side chains. In any case, the
stabilizing contributions from the closer residues to the turn appear
to be more important in b-hairpin formation.
Considering the clear-cut results on 4:4 b-hairpins and those on
3:5 b-hairpins, we suggest that the contribution to b-hairpin for-
mation of stabilizing and destabilizing factors, i.e., cross-strand
side chain–side chain interactions and intrinsic b-sheet propensi-
ties, is likely more important when closer to the turn region. This
is in agreement with the essential role of the turn sequence in
nucleating b-hairpin folding as supported by experimental results
~de Alba et al., 1997a, 1997b, 1999a, 1999b; Griffiths-Jones et al.,
1999! and proposed by theoretical studies ~Muñoz et al., 1997,
1998!. If the turn sequence directs b-hairpin formation by predis-
posing the orientation of the two b-strands, it is reasonable that a
stabilizing interaction between the two residues adjacent to the
turn favors the formation of that b-hairpin. Our conclusion could
also explain the different mutational tolerance displayed by two
exposed residue pairs belonging to two b-hairpins of the b-barrel
protein CspA ~Zaremba & Gregoret, 1999!, since, as we would
expect, the less tolerant pair was that closer to the turn region.
Nevertheless, the fact that this pair belongs to a nonhydrogen-
bonded site and the other to a hydrogen-bonded site must also be
considered, since it may contribute to the differential behavior of
the two pairs, as suggested by Zaremba and Gregoret ~1999!. Our
proposal also agrees with the amino acid b-sheet intrinsic propen-
sities being context-dependent as indicated by the discrepancies
found among different experimental determinations ~Kim & Berg,
1993; Minor & Kim, 1994; Smith et al., 1994!. Statistical analysis
of pairwise interactions in antiparallel b-sheets have distinguished

2158
C.M. Santiveri et al.
between hydrogen- and nonhydrogen-bonded positions, with no
consideration about the linker between the two antiparallel strands
~Wouters & Curmi, 1995; Hutchinson et al., 1998!. Moreover, the
only reported survey of side-chain interactions in b-hairpins ~Gu-
nasekaran et al., 1997! does not take into account any positional
effect. Statistical analysis of pairwise interactions in b-hairpins of
known protein structures that consider the position dependence of
cross-strand interactions, as suggested by our results, will be useful
to test the general validity of our conclusion.
2D correlated spectroscopy ~COSY! ~Aue et al., 1976!, total cor-
related spectroscopy ~TOCSY! ~Rance, 1987!, NOE spectros-
copy ~NOESY! ~Jeener et al., 1979; Kumar et al., 1980! and
rotating frame NOE spectroscopy ~ROESY! ~Bothner-By et al.,
1984; Braunschweiler & Ernst, 1983! spectra were recorded by
standard techniques using presaturation of the water signal and the
time-proportional phase incrementation mode ~Redfield & Kuntz,
1975!. NOESY and ROESY mixing times were 200 and 120 ms,
respectively. TOCSY spectra were recorded using 80 ms MLEV17
with z filter spin-lock sequence ~Rance, 1987!. The 1H-13C het-
eronuclear single quantum coherence spectra ~HSQC! ~Boden-
hausen & Ruben, 1980! at natural 13C abundance were recorded in
2 mM peptide samples in D2O. Acquisition data matrices were
defined by 2,018 ϫ 512 points in t2 and t1, respectively. Data were
processed using the standard XWIN-NMR Bruker program on a
Silicon Graphics computer. The 2D data matrix was multiplied by
a square-sine-bell window function with the corresponding shift
optimized for every spectrum and zero-filled to a 2 ϫ 1K complex
matrix prior to Fourier transformation. Baseline correction was
applied in both dimensions. The 0 ppm 13C d was obtained indi-
rectly by multiplying the spectrometer frequency that corresponds
Conclusion
In brief, the conformational behavior of two designed 15-residue
b-hairpin-forming linear peptides suggests that the contribution of
stabilizing and destabilizing factors, such as cross-strand side-
chain interactions, to b-hairpin formation depends on the relative
position of the pair with respect to the turn. The strand residues
adjacent to the turn appear to be the most relevant for b-hairpin
formation, which is in agreement with previous results indicating
the essential role of the turn region for b-hairpin formation. This
positioning effect might be contributing to the context dependence
found for amino acid intrinsic b-sheet propensities.
1
to 0 ppm in the H spectrum, assigned to internal TSP reference,
by 0.25144954 ~Bax & Subramanian, 1986; Spera & Bax, 1991!.
Materials and methods
NMR assignment
The 1H NMR signals of peptides 3 and 4 in aqueous solution
were readily assigned by standard sequential assignment methods
~Wüthrich et al., 1984; Wüthrich, 1986!. Then, the 13C resonances
were straightforwardly assigned on the basis of the cross-correlations
observed in the HSQC spectra between the proton and the carbon
to which it is bonded. The 1H and 13C d-values of the two peptides
are available in Supplementary material in the Electronic Appen-
dix ~Tables SM1 and SM2! and have been deposited at the PESCA-
DOR database ~http:00ucmb.ulb.ac.be0Pescador0!.
Peptide synthesis and purification.
Peptides 3 and 4 were chemically synthesized by the Peptide Syn-
thesis Facility at the Department of Organic Chemistry ~University
of Barcelona, Spain!.
Sedimentation equilibrium
Sedimentation equilibrium experiments were performed to obtain
the average molecular weight of peptide samples at ;0.2 mM
concentration in aqueous solution at pH 5.5 and containing 150 mM
NaCl to screen nonideal effects involving charged residues at high
peptide concentrations. Peptide samples ~70 mL! were centrifuged
at 60,000 rpm at 278 K in 12 mm triple-sector Epon charcoal
centrepieces, using a Beckman Optima XL-A ultracentrifuge with
a Ti60 rotor. Radial scans were taken at different wavelengths
every 2 h until equilibrium conditions were reached. The data were
analyzed using the program XLAEQ from Beckman. The partial
specific volumes of the peptides were calculated on the basis of
their amino acid composition and corrected for temperature ~Laue
et al., 1992!.
Estimation of b-hairpin populations
The b-hairpin populations were estimated from the ratio of inten-
sities of the CaHi-CaHj NOE characteristic of each hairpin to that
'
of the conformationally independent CaH-Ca H Gly NOE, which
was used with calibration purposes ~Searle et al., 1995; de Alba
et al., 1996, 1997a, 1997b; Ramírez-Alvarado et al., 1996; May-
nard et al., 1998!. The intensity ratio corresponding to a 100%
'
b-hairpin was taken as @dCaH -CaH # #
Ϫ60@dCaH-Ca H Gly Ϫ6, where
j
i
dCaH -CaH is the average CaHi-CaHj distance in antiparallel
i
j
b-sheets in proteins ~2.3 Å; Wüthrich, 1986!, where i and j are
'
facing residues in a nonhydrogen-bonded site, and dCaH-Ca H Gly is
'
the CaH-Ca H distance characteristic of Gly residues ~1.78 Å!.
NMR spectra
NOE intensities were measured by volume integration in NOESY
spectra ~200 ms mixing time! recorded in pure D2O samples.
The overall b-hairpin populations were independently evaluated
Peptide samples for NMR experiments were prepared in 0.5 mL of
H2O0D2O ~9:1 ratio by volume! or in pure D2O. Peptide concen-
trations were about 2 mM. pH was measured with a glass micro
electrode and was not corrected for isotope effects. The tempera-
ture of the NMR probe was calibrated using a methanol sample.
Sodium @3-trimethylsilyl 2,2,3,3-2H# propionate ~TSP! was used
as an internal reference. The 1H-NMR spectra were acquired on
a Bruker AMX-600 pulse spectrometer operating at a proton
frequency of 600.13 MHz. One-dimensional spectra were
acquired using 32K data points, which were zero-filled to 64K data
points before performing the Fourier transformation. Phase-sensitive
13
by using the DdCaH ~dCaH~observed! Ϫ dCaH~random coil!! and Dd
C
a
13
13
~␦
Ϫ d C ~random coil!! conformational shifts averaged
C
~observed!
␣
a
over all the strand residues in the adopted hairpins, except for the
N- and C-terminal residues that would present charge end effects.
The DdCaH conformational shift of residue 5 of peptides 3 and 4
was not included because the d value of its CaH proton is strongly
affected for the aromatic ring of its facing W side chain in the 3:5
b-hairpin. The reference values for each residue in the random coil
state were taken from Wüthrich ~1986! for proton chemical shifts

Cross-strand side-chain interactions and b-hairpin folding
2159
imentation equilibrium experiments. We are also grateful to Dr. D.V. Lau-
rents and Dr. F.J. Blanco for their critical reading of the manuscript. This
work was supported by the Spanish DGICYT project no. PB98-0677 and
European Project no. CEE B104-97-2086. C.M.S. is recipient of a pre-
doctoral fellowship from Autonomous Community of Madrid, Spain.
~dCaH~random coil!! and from Wishart et al. ~1995! for carbon chem-
13
ical shifts ~d C ~random coil!!. The reference values for 100%
a
b-hairpin were the mean conformational shifts reported for b-sheet
proteins, 0.40 ppm for DdCaH ~Wishart et al., 1991! and Ϫ1.48 ppm
13
for Dd
~Spera & Bax, 1991!.
C
a
The ratio of the b-hairpin populations for each residue of pep-
tide 4 was computed as the ratio DdCaH~58C!0DdCaH~358C!, where
DdCaH~58C! and DdCaH~358C! are the DdCaH conformational shifts
measured for each residue at 5 and at 358C, respectively.
References
Aue WP, Bertholdi E, Ernst RR. 1976. Two-dimensional spectroscopy. Appli-
cation to NMR. J Chem Phys 64:2229–2246.
Aurora R, Rose GD. 1998. Helix capping. Protein Sci 7:21–38.
Baldwin RL. 1995. The a-helix formation by peptides of defined sequence.
Biophys Chem 55:127–135.
Structure calculation
Bax A, Subramanian J. 1986. Sensitivity-enhanced two-dimensional heteronu-
clear shift correlation NMR spectroscopy. J Magn Reson 67:565–570.
Blanco FJ, Jiménez MA, Herranz J, Rico M, Santoro J, Nieto JL. 1993. NMR
evidence of a short linear peptide that folds into a b-hairpin in aqueous
solution. J Am Chem Soc 115:5887–5888.
Blanco FJ, Ramírez-Alvarado M, Serrano L. 1998. Formation and stability of
b-hairpin structures in polypeptides. Curr Opin Struct Biol 8:107–111.
Bodenhausen G, Ruben DJ. 1980. Natural abundance nitrogen-15 NMR by
enhanced heteronuclear spectroscopy. Chem Phys Lett 69:185–189.
Bothner-By AA, Stephens RL, Lee JM, Warren CD, Jeanloz RW. 1984. Struc-
ture determination of a tetrasaccharide: Transient nuclear Overhauser effects
in the rotating frame. J Am Chem Soc 106:811–813.
Intensities of medium- and long-range NOEs were evaluated qual-
itatively and used to obtain upper limit distant constraints: strong
~3.0 Å!, intermediate between strong and medium ~3.5 Å!, medium
~4.0 Å!, intermediate between medium and weak ~4.5 Å!, weak
~5.0 Å!, and very weak ~5.5 Å!. Pseudo atom corrections were
added where necessary. f Angles were constrained to the range
Ϫ180 to 08 except for Asn and Gly. For those residues with
3
J
CaH-NH Ͼ 8.0 Hz and 3JCaH-NH Ͻ 5.0 Hz, f angles were restricted
to the range Ϫ160 to Ϫ808, and Ϫ100 to Ϫ208, respectively.
Structures were calculated on a Silicon Graphics computer using
the program DIANA ~Güntert et al., 1991!.
Braunschweiler L, Ernst RR. 1983. Coherence transfer by isotropic mixing:
Application to proton correlation spectroscopy. J Magn Reson 53:521–528.
Buckley P, Edison AS, Kemple MD, Prendergast FG. 1993. The 13Ca-NMR
assignments of melittin in methanol, and chemical shift correlations with
secondary structure. J Biomol NMR 3:639–652.
Estimation of hydrophobic surface area buried
upon b-hairpin formation
Bundi A, Wüthrich K. 1979. 1H NMR parameters of the common amino acid
residues measured in aqueous solution of linear tetrapeptides H-Gly-Gly-
X-Ala-OH. Biopolymers 18:285–297.
To model the 4:4 b-hairpins formed by peptide 3 and expected for
peptide 4, a structure calculation was performed by using the
DYANA program ~Güntert et al., 1997! and introducing as distance
and dihedral angle constraints those expected for an ideal 4:4
b-hairpin. For each sequence, the set of the 30 structures with the
lowest target function was taken as model for the 4:4 b-hairpin.
The same procedure was used to model the 3:5 b-hairpin adopted
by peptides 3 and 4, but taking as distance and dihedral angle
constraints those expected for an ideal 3:5 b-hairpin. The solvent-
accessible nonpolar area for each structure was calculated using
the program VADAR ~Wishart et al., 1996!. The hydrophobic sur-
face area buried upon b-hairpin formation was computed as the
difference between the solvent-accessible nonpolar area averaged
over the 30 modeled structures and that corresponding to a com-
pletely extended structure. To obtain the hydrophobic surface area
buried upon formation of the 3:5 b-hairpin formed by peptide 4,
we followed the same procedure but using the 20 best structures
calculated from experimental restraints ~Fig. 5!. The hydrophobic
surface area buried upon formation of a 4:4 b-hairpin is 524 6
66 Å2 and 524 6 62 Å2 for peptides 3 and 4, respectively. That
buried upon formation of the modeled 3:5 b-hairpins is 509 6
40 Å2 for peptide 3 and 499 6 40 Å2, for peptide 4, and upon
formation of the 3:5 b-hairpin experimentally found for peptide 4,
545 6 62 Å2.
Case DA, Dyson HJ, Wright PE. 1994. Use of chemical shifts and coupling
constants in nuclear magnetic resonance structural studies on peptides and
proteins. Methods Enzymol 239:392–416.
Chou PY, Fasman GD. 1974. Conformational parameters for amino acids in
helical, b-sheet, and random coil regions calculated from proteins. Biochem-
istry 13:211–222.
de Alba E, Blanco FJ, Jiménez MA, Rico M, Nieto JL. 1995. Interactions
responsible for the pH dependence of the b-hairpin conformational popu-
lation formed by a designed linear peptide. Eur J Biochem 233:283–292.
de Alba E, Jiménez MA, Rico M. 1997a. Turn residue sequence determines
b-hairpin conformation in designed peptides. J Am Chem Soc 119:175–183.
de Alba E, Jiménez MA, Rico M, Nieto JL. 1996. 1H NMR conformational
investigation of designed short linear peptides able to fold into b-hairpin
structures in aqueous solution. Fold Des 1:133–144.
de Alba E, Rico M, Jiménez MA. 1997b. Cross-strand side-chain interactions
versus turn conformation in b-hairpins. Protein Sci 6:2548–2560.
de Alba E, Rico M, Jiménez MA. 1999a. The turn sequence directs b-strand
alignment in b-hairpins. Protein Sci 8:2234–2244.
de Alba E, Rico M, Jiménez MA. 1999b. De novo design of a monomeric
three-stranded antiparallel b-sheet. Protein Sci 8:854–865.
de Dios AC, Pearson JG, Oldfield E. 1993. Secondary and tertiary structural
effects on protein NMR chemical shifts: An ab initio approach. Science
260:1491–1496.
Gellman SH. 1998. Minimal model systems for b-sheet secondary structure in
proteins. Curr Opin Chem Biol 2:717–725.
Griffiths-Jones SR, Maynard AJ, Searle MS. 1999. Dissecting the stability of a
b-hairpin peptide that folds in water: NMR and molecular dynamics analy-
sis of the b-turn and b-strand contributions to folding. J Mol Biol 292:1051–
1069.
Gronenborn AM, Clore GM. 1994. Identification of N-terminal helix capping
boxes by means of 13C chemical shifts. J Biomol NMR 4:455–458.
Guerois R, Cordier-Oschsenbein F, Baleux F, Huynh-Dinh T, Neumann J-M,
Sanson A. 1998. A conformational equilibrium in a protein fragment caused
by two consecutive capping boxes: 1H-, 13C-NMR, and mutational analysis.
Protein Sci 7:1506–1515.
Gunasekaran K, Ramakrishnan C, Balaram P. 1997. b-Hairpins in proteins
revisited: Lessons for de novo design. Protein Eng 10:1131–1141.
Güntert P, Braun W, Wüthrich K. 1991. Efficient computation of three-dimensional
protein structures in solution from nuclear magnetic resonance data using
the program DIANA and the supporting programs CALIBA, HABAS, and
GLOMSA. J Mol Biol 217:517–530.
Supplementary material in the Electronic Appendix
Two tables listing the 1H and 13C d values of peptides 3 and 4
~Tables SM1, SM2! and two tables listing the nonsequential NOEs
observed for peptides 3 and 4 ~Tables SM3, SM4!.
Acknowledgments
We thank Mrs. C. López, Mr. A. Gómez, and Mr. L. de la Vega for technical
assistance, and Drs. G. Rivas and C. Alfonso for their help with the sed-
Güntert P, Mumenthaler C, Wüthrich K. 1997. Torsion angle dynamics for NMR
structure calculation with the new program DYANA. J Mol Biol 273:283–298.

2160
C.M. Santiveri et al.
Haque TS, Gellman SH. 1997. Insights into b-hairpin stability in aqueous so-
lution from peptides with enforced type I9 and type II9 b-turns. J Am Chem
Soc 119:2303–2304.
Hutchinson EG, Sessions RB, Thornton JM, Woolfson DN. 1998. Determinants
of strand register in antiparallel b-sheets: The differences between backbone
hydrogen-bonded and non-hydrogen-bonded residue pairs. Protein Sci
7:2287–2300.
Iwadate M, Asakura T, Williamson MP. 1999. Ca and Cb carbon-13 chemical
shifts in proteins from an empirical database. J Biomol NMR 13:199–211.
Jeener J, Meier BH, Bachmann P, Ernst RA. 1979. Investigation of exchange
processes by two-dimensional NMR spectroscopy. J Chem Phys 71:4546–
4553.
Jiménez MA, Bruix M, González C, Blanco FJ, Nieto JL, Herranz J, Rico M.
1993. CD and 1H-NMR studies on the conformational properties of peptide
fragments from the C-terminal domain of thermolysin. Eur J Biochem
211:569–581.
Jiménez MA, Muñoz V, Rico M, Serrano L. 1994. Helix stop and start signals
in peptides and proteins. The capping box does not necessarily prevent helix
elongation. J Mol Biol 242:487–496.
Kim CA, Berg JM. 1993. Thermodynamic b-sheet propensities measured using
a zinc-finger host peptide. Nature 362:267–270.
Koradi R, Billeter M, Wüthrich K. 1996. MOLMOL: A program for display and
analysis of macromolecular structures. J Mol Graphics 14:51–55.
Kumar A, Ernst RR, Wüthrich K. 1980. A two-dimensional nuclear Overhauser
enhancement ~2D NOE! experiment for the elucidation of complete proton-
proton cross-relaxation networks in biological macromolecules. Biochem
Biophys Res Commun 95:1–6.
Lacroix E, Kortemme T, López de la Paz M, Serrano L. 1999. The design of
linear peptides that fold as monomeric b-sheet structures. Curr Opin Struct
Biol 9:487–493.
Ramírez-Alvarado M, Blanco FJ, Serrano L. 1996. De novo design and struc-
tural analysis of a model b-hairpin peptide system. Nat Struct Biol 3:604–612.
Ramírez-Alvarado M, Kortemme T, Blanco FJ, Serrano L. 1999. Beta-hairpin
and beta-sheet formation in designed linear peptides. Bioorg Med Chem
7:107–111.
Rance M. 1987. Improved techniques for homonuclear rotating-frame and iso-
tropic mixing experiments. J Magn Reson 74:557–564.
Redfield AG, Kuntz SD. 1975. Quadrature Fourier NMR detection: Simple
multiplex for dual detection. J Magn Reson 19:250–254.
Scholtz JM, Baldwin RL. 1992. The mechanism of a-helix formation by pep-
tides. Annu Rev Biophys Biomol Struct 21:95–118.
Searle MS, Williams DM, Packman LC. 1995. A short linear peptide derived
from the N-terminal sequence of ubiquitin folds into a water-stable non-
native b-hairpin. Nat Struct Biol 2:999–1006.
Sibanda BL, Blundell TL, Thornton JM. 1989. Conformation of b-hairpins in
protein structures. A systematic classification with applications to modelling
by homology, electron density fitting and protein engineering. J Mol Biol
206:759–777.
Sibanda BL, Thornton JM. 1985. b-Hairpin families in globular proteins. Nature
316:170–174.
Sibanda BL, Thornton JM. 1991. Conformation b-hairpins in protein structures.
Classification and diversity in homologous structures. Methods Enzymol
202:59–82.
Smith CK, Regan L. 1997. Construction and design of b-sheets. Acc Chem Res
30:153–161.
Smith CK, Withka JM, Regan L. 1994. A thermodynamic scale for the b-sheet
forming tendencies of amino acids. Biochemistry 33:5510–5517.
Spera S, Bax A. 1991. Empirical correlation between protein backbone confor-
mation and Ca and Cb 13C NMR chemical shifts. J Am Chem Soc 113:5490–
5492.
Laue TM, Shak BD, Ridgeway TM, Pelletier SL. 1992. Computer-aided inter-
pretation of analytical sedimentation data for proteins. In: Harding SE,
Rowe AJ, Horton JC, eds. Analytical ultracentrifugation in biochemistry
and polymer science. Cambridge, UK: Royal Society of Chemistry. pp
90–125.
Lee MS, Cao B. 1996. Nuclear magnetic resonance chemical shift: Comparison
of estimated secondary structures in peptides by nuclear magnetic resonance
and circular dichroism. Protein Eng 9:15–25.
Lyu PC, Wemmer DE, Zhou HX, Pinker RJ, Kallenbach NR. 1993. Capping
interactions in isolated a-helices: Position-dependent substitution effects
and structure of a serine-capped peptide helix. Biochemistry 32:421–425.
Maynard AJ, Sharman GJ, Searle MS. 1998. Origin of b-hairpin stability in
solution: Structural and thermodynamic analysis of the folding of a model
peptide supports hydrophobic stabilization in water. J Am Chem Soc
120:1996–2007.
Swindells MB, MacArthur MW, Thornton JM. 1995. Intrinsic f, c propensities
of amino acids, derived from the coil regions of known structures. Nat
Struct Biol 2:596–603.
Syud FA, Espinosa JF, Gellman SH. 1999. NMR-based quantification of b-sheet
populations in aqueous solution through use of reference peptides for the
folded and unfolded states. J Am Chem Soc 121:11578–11579.
Szilágyi L. 1995. Chemical shifts in proteins come of age. Prog Nucl Magn
Reson Spectros 27:325–443.
Wishart DS, Bigam CG, Holm A, Hodges RS, Sykes BD. 1995. 1H, 13C and 15
N
random coil NMR chemical shifts of the common amino acids. I. Investi-
gations of nearest neighbor effects. J Biomol NMR 5:67–81.
Wishart DS, Sykes BD. 1994. Chemical shifts as a tool for structure determi-
nation. Methods Enzymol 239:363–392.
Wishart DS, Sykes BD, Richards FM. 1991. Relationship between nnuclear
magnetic resonance chemical shift and protein secondary structure. J Mol
Biol 222:311–333.
Wishart DS, Willard L, Sykes BD. 1996. VADAR. Volume angles define s area,
version 1.3. Edmonton: University of Alberta. Protein Engineering Network
of Centres of Excellence.
Wouters MA, Curmi PM. 1995. An analysis of side chain interactions and pair
correlations within antiparallel b-sheets: The differences between backbone
hydrogen-bonded and non-hydrogen-bonded residue pairs. Proteins 22:119–
131.
Merutka G, Dyson HJ, Wright PE. 1995. “Random coil” 1H chemical shifts
obtained as a function of temperature and trifluoroethanol concentration for
the peptide series GGXGG. J Biomol NMR 5:14–24.
Minor DL Jr, Kim PS. 1994. Measurement of the b-sheet-forming propensities
of amino acids. Nature 367:660–663.
Muñoz V, Henry ER, Hofrichter J, Eaton WA. 1998. A statistical mechanical
model for b-hairpin kinetics. Proc Natl Acad Sci USA 95:5872–5879.
Muñoz V, Serrano L. 1994. Intrinsic secondary structure propensities of the
amino acids, using statistical f-c matrices: Comparison with experimental
scales. Proteins Struct Funct Genet 20:301–311.
Wüthrich K. 1986. NMR of proteins and nucleic acids. New York: J. Wiley &
Sons.
Muñoz V, Serrano L. 1997. Development of the multiple sequence approxima-
tion within the AGADIR model of a-helix formation: Comparison with
Zimm-Bragg and Lifson-Roig formalisms. Biopolymers 41:495–509.
Muñoz V, Serrano L, Jiménez MA, Rico M. 1995. Structural analysis of pep-
tides encompassing all a-helices of three a0b parallel proteins: Che-Y,
Flavodoxin, and P21-Ras: Implications for a-helix stability and the folding
of a0b parallel proteins. J Mol Biol 247:648–669.
Muñoz V, Thompson PA, Hofrichter JA, Eaton WA. 1997. Folding dynamics and
mechanism of b-hairpin formation. Nature 390:196–199.
Ramírez-Alvarado M, Blanco FJ, Niemann H, Serrano L. 1997. Role of b-turn
residues in b-hairpin formation and stability in designed peptides. J Mol
Biol 273:898–912.
Wüthrich K, Billeter M, Braun W. 1984. Polypeptide secondary structure de-
termination by nuclear magnetic resonance observation of short proton-
proton distances. J Mol Biol 180:715–740.
Yao J, Dyson HJ, Wright PE. 1997. Chemical shift dispersion and secondary
structure prediction in unfolded and partly folded proteins. FEBS Lett
419:285–289.
Zaremba SM, Gregoret LM. 1999. Context-dependence of amino acid residue
pairing in antiparallel b-sheets. J Mol Biol 291:463–479.
Zhou NE, Kay CM, Sykes BD, Hodges RS. 1993. A single-stranded amphi-
pathic alpha-helix in aqueous solution: Design, structural characterization,
and its application for determining alpha-helical propensities of amino acids.
Biochemistry 32:6190–6197.