Direct evidence for CH...π interaction mediated stabilization of Pro-cisPro bond in peptides with Pro-Pro-aromatic motifs

Article abstract of DOI:10.1021/ja209334v

Although weak interactions play subtle but important roles in dictating protein structures, their experimental detection is nontrivial. From NOE experiments we provide direct evidence for the presence of CH...π interaction, operational between the C^α-H of the first Pro and the aromatic (Aro) side chain of Xaa, in a peptide series with the general sequence Ac-Pro-Pro-Xaa-NH₂. Indirect evidence of CH...π interaction is provided from ring current-induced upfield displacement of Pro(1) C^α-H chemical shifts and restriction of side-chain (χ1) rotation of Xaa. A consequence of this interaction is the enhanced stability of the Pro-cisPro conformer in Ac-Pro-Pro-Xaa-NH₂ when Xaa is aromatic. The free energies associated with trans to cis transformation of the Pro-Pro moiety are 0.35, 0.59, 0.64, and 0.82 kcal/mol when Xaa is Tyr, Trp, Phe, and His (pH of 8.4), respectively. In comparison, the corresponding free energy is ～1.55 kcal/mol when Xaa is nonaromatic. The observed population of Pro-cisPro-His and the pH-induced perturbation of electron density of the His side chain were correlated, providing further evidence for a direct role of CH...π interaction in modulating the stability of Pro-cisPro population in Ac-Pro-Pro-Aro-NH₂. Our study establishes Pro-Pro-Aro to be a new sequence motif that can stabilize Pro-cisPro peptide bonds. This study not only identifies a new structurally biased sequence motif but also directly demonstrates the role played by CH...π interactions in subtly altering conformational preferences of three-residue peptide sequences with implications on the role played by cis-peptide bonds in unfolded proteins.

Full text of DOI:10.1021/ja209334v

Article
pubs.acs.org/JACS
Direct Evidence for CH···π Interaction Mediated Stabilization of Pro-
cisPro Bond in Peptides with Pro-Pro-Aromatic motifs
Himal K. Ganguly,† Barun Majumder,† Sarbani Chattopadhyay,† Pinak Chakrabarti,‡ and Gautam Basu*,†
‡
†Department of Biophysics and Department of Biochemistry, Bose Institute, P-1/12 CIT Scheme VIIM, Kolkata 700054, India
S
* Supporting Information
ABSTRACT: Although weak interactions play subtle but important
roles in dictating protein structures, their experimental detection is
nontrivial. From NOE experiments we provide direct evidence for the
presence of CH···π interaction, operational between the Cα-H of the
first Pro and the aromatic (Aro) side chain of Xaa, in a peptide series
with the general sequence Ac-Pro-Pro-Xaa-NH2. Indirect evidence of
CH···π interaction is provided from ring current-induced upfield
displacement of Pro(1) Cα-H chemical shifts and restriction of side-
chain (χ1) rotation of Xaa. A consequence of this interaction is the
enhanced stability of the Pro-cisPro conformer in Ac-Pro-Pro-Xaa-NH2 when Xaa is aromatic. The free energies associated with
trans to cis transformation of the Pro-Pro moiety are 0.35, 0.59, 0.64, and 0.82 kcal/mol when Xaa is Tyr, Trp, Phe, and His (pH
of 8.4), respectively. In comparison, the corresponding free energy is ∼1.55 kcal/mol when Xaa is nonaromatic. The observed
population of Pro-cisPro-His and the pH-induced perturbation of electron density of the His side chain were correlated,
providing further evidence for a direct role of CH···π interaction in modulating the stability of Pro-cisPro population in Ac-Pro-
Pro-Aro-NH2. Our study establishes Pro-Pro-Aro to be a new sequence motif that can stabilize Pro-cisPro peptide bonds. This
study not only identifies a new structurally biased sequence motif but also directly demonstrates the role played by CH···π
interactions in subtly altering conformational preferences of three-residue peptide sequences with implications on the role played
by cis-peptide bonds in unfolded proteins.
from a number of bioinformatics studies indicate that selected
amino acids show significant propensity to occur around a cis
peptidyl-prolyl8 as well as cis nonprolyl amide bonds,8c
emphasizing the importance of local control of cis peptide
bonds.
INTRODUCTION
■
The three-dimensional structure of a protein results from the
net effect of a large number of weak interactions. A folded
protein is only marginally stable over the unfolded state,
underscoring the fact that disruption of only a few weak
interactions can unfold a protein. Among weak interactions,
CH···π interactions1 have been implicated to play a subtle but
important role. Although CH···π interactions in proteins are
typically assessed from high-resolution crystal structures using a
geometric criterion,2 a recent NMR study provided direct
evidence of CH···π interactions in proteins.3 CH···π
interactions have also been implicated to play a role in
stabilizing cis peptide bonds in short peptides in an earlier study
from this laboratory.4
Peptide bonds in proteins are rarely observed in cis
conformation5 due to unfavorable steric interaction between
sequence contiguous Cα atoms in the cis conformation.
However cis peptide bonds involving Pro are special. For
Xaa-Pro sequence motifs, the Xaa-transPro conformer also
acquires unfavorable Cα(Xaa)-Cδ(Pro) steric clash, decreasing
the cis−trans energy difference.6 This is consistent with the over
representation of Pro at positions that immediately follow a cis
peptide bond in proteins.5e The cis−trans energy difference is
further lowered when an aromatic (Aro) residue precedes the
Pro residue.7 Conservation of Pro in cis peptidyl-prolyl unit in
proteins indicates that the conformation of a peptide bond is in
a way encoded in the local sequence of a large protein.8a Results
Despite evidence in support of local control of cis peptide
bonds, subtle mismatch of local effects and global compulsions,
arising from a folded protein structure, decides the final
conformational fate of a peptide bond. Mutation of a cis prolyl
residue to a nonprolyl residue can result in the retention of the
cis bond,9a−g indicating a role of global tertiary interactions. It
can also result in a misfolded protein9f or a folded protein with
a trans conformation9f−h signifying the importance of the prolyl
residue and, in turn, to the role of local effects in stabilizing the
cis peptide bond in a protein. Although aromatic residues
neighboring cis prolyl residues can exert influence on
stabilization of the cis prolyl bond, to our knowledge, there
are no reports where an aromatic residue preceding a cis prolyl
residue has been mutated to a nonaromatic residue to observe
the effect on the protein structure.
One way to delineate local versus global compulsions of a
peptide bond to be cis is to study a short peptide sequence, in
isolation, by NMR spectroscopy. A number of NMR studies
have been reported on short peptides with the general sequence
Received: October 4, 2011
Published: February 15, 2012
© 2012 American Chemical Society
4661
dx.doi.org/10.1021/ja209334v | J. Am. Chem. Soc. 2012, 134, 4661−4669

Journal of the American Chemical Society
Article
-Xaa(i − 1)-Pro(i)-Yaa(i + 1)-, where either Xaa(i − 1)7,10,11 or
Yaa(i + 1)4,11b,12 is aromatic, all focused on measuring the cis
content of the Xaa(i − 1)-Pro(i) bond. Enhanced cis content of
Xaa-Pro bond in these peptides was found to be associated with
Pro···Aro interaction. From NMR studies on Ac-Pro-Pro-Phe-
NH2, a designed peptide mimicking the CH···π interaction of
similar sequences in proteins, we had observed that Pro-cisPro
conformation is overpopulated in Ac-Pro-Pro-Phe-NH2 when
compared to Ac-Pro-Pro-Ala-NH2.4 It was suggested that this
originates from CH···π interaction between Pro1 and Phe.
To generalize our earlier observation on Pro-Pro-Phe,4 we
have studied the complete series of peptides with the sequence
Ac-Pro-Pro-Xaa-NH2, where Xaa is varied among all 20 amino
acids. The aim of this study is: (i) to confirm that Pro-Pro-Aro,
and not just Pro-Pro-Phe, stabilizes the Pro-cisPro motif in Pro-
Pro-Xaa peptide series; (ii) to derive a comparative
thermodynamic scale for trans−cis equilibrium of the Pro-Pro
bond in Pro-Pro-Xaa containing peptides as a function of Xaa;
(iii) to characterize NOE cross-peaks between the Aro side
chain and the CαH of the first Pro residue to provide direct
evidence for CαH···π interaction between them; and (iv) to
compare our results with a previous study on -Gly-Aro-Pro-Gly-
series of peptides for which only indirect experimental evidence
was provided for CαH···π interaction.11c We also revisit the
percent population of Xaa-cisPro motifs in proteins and discuss
its correlation with corresponding experimental values obtained
from peptides.7 Finally, significance and conservation of cis-
stabilizing Xaa-Pro sequence motifs in proteins are discussed.
spectroscopy (DQF-COSY). Resonance assignments were achieved
using TOCSY and DQF-COSY experiments, while sequential
assignments (Pro1 and Pro2) were performed from ROESY
experiments (mixing time: 250 ms).4,13 For each peptide, the four
isomers (Figure 1a) were detected from ROESY (Pro-cisPro- and Pro-
transPro- isomers were detected from Pro-Pro Cα-H/Cα-H and Cα-H/
Cδ-H cross-peaks, while Ac-cisPro- and Ac-transPro- isomers were
detected from Ac-Pro CH3/Cα-H and CH3/Cδ-H cross-peaks). The
3Jαβ coupling constants were measured from 1D 1H NMR spectra and
DQF-COSY spectra. Relative populations of the isomers were
estimated from the relative intensities of amide resonances, CH3
(Ala), or acetyl 13C (Xaa = Trp, His) resonances. For van’t Hoff
analysis, peptides were equilibrated for at least 30 min at a given
temperature prior to obtaining the NMR spectra.
Figure 1. (a) A schematic representation of the four conformational
isomers present in Ac-Pro-Pro- motif (tt, tc, ct, and cc). (b) Four amide
peaks (Tyr) corresponding to the four isomers in panel (a) in Ac-Pro-
Pro-Tyr-NH2.
pH Titrations. The pH dependence of the cis populations in Ac-
1
Pro-Pro-His-NH2 was followed using H NMR (pH < 6) and 13C
proton decoupled (pH > 6) NMR spectra (13C NMR signals showed
signal overlap below pH 6). The cis−trans population ratios obtained
from NMR spectra were fitted with the eq 1 (see Supporting
Information) to yield the trans to cis equilibrium constants, Kt0c and Kt+c,
corresponding to neutral and positively charged histidine side chains,
respectively.
MATERIALS AND METHODS
■
Peptide Synthesis and Purification. All peptides were
synthesized by a stepwise procedure in solid phase using Fmoc
chemistry and Rink amide MBHA resin (substitution 0.69 mmol/g;
Novabiochem) and Fmoc-protected amino acids (Novabiochem). The
Fmoc-amino acid derivatives (five times excess of the resin
substitution) were coupled with benzotriazol-1-yl-oxytripyrrolidino-
phosphonium hexafluorophosphate (PyBOP)/hydroxybenzotriazole
(HOBT)/diisopropylethylamine (DIPEA) (used as 5:5:10 times
excess of the resin, respectively), and Fmoc cleavage was performed
with 20% (v/v) piperidine in dimethylformamide (DMF). N-
acetylation was achieved using acetic anhydride/DIPEA (1:10 times
excess) in DMF. For Ac-Pro-Pro-Trp-NH2 and Ac-Pro-Pro-His-NH2,
the N-terminal acetyl carbon was labeled with 13C (13CH3COOH from
Sigma was coupled to H-Pro-Pro-Xaa- moiety using PyBOP/HOBT/
DIPEA). Peptides with aliphatic side chains were cleaved from the
resin with 95% trifluoroacetic acid (TFA) in water, and others with
side-chain functional group protections (Table S1, Supporting
Information) were cleaved using a cocktail (85% TFA, 5% H2O, 5%
phenol, 2.5% anisole, and 2.5% tri-isopropyl silane). All peptides were
kept in the cleavage cocktail for 1 h (for Xaa = Arg, the cleavage
reaction was carried on for 3 h). TFA was removed by evaporation in a
rotary evaporator, and the crude peptides were dissolved in methanol
and purified using reverse phase HPLC using a Phenomenix C18
column and CH3OH-H2O gradients (0−80% CH3OH in 40 min)
containing 0.1% TFA. The final products were characterized by NMR
and mass spectrometry (Tables S1 and S2, Supporting Information).
NMR Spectroscopy. NMR experiments were performed on a
Bruker DRX 500 spectrometer. All samples were prepared in 20 mM
phosphate buffer pH of 7 at 4 °C (unless otherwise stated), containing
10% 2H2O and the sodium salt of 3(trimethylsilyl) propionic-2,2,3,3-d4
+
⎛
⎞
1 + K
1 + K
+
0
tc
(pH−pK )
a
K
+ K ⎜
⎟10
tc
tc
0
⎝
⎠
[cis]
[trans]
tc
=
+
⎛
⎝
⎞
1 + K
1 + K
tc
(pH−pK )
a
1 + ⎜
⎟10
0
⎠
(1)
tc
Database Analyses. A representative list of nonredundant PDB14
structures were collected using the culling server PISCES15 with
sequence identity ≤25% and resolution of ≤2 Å. Sequences with Xaa-
Pro and Pro-Pro motifs were identified from the structures. A cis
peptide bond was defined as −90° ≤ ω ≤ +90°. Propensities of Xaa to
induce Xaa-cisPro motif in Xaa-Pro sequences were calculated as
[NX‑cP/(NX‑cP + NX‑tP)]/[NcP/(NcP + NtP)], and propensities of Xaa to
induce Pro-cisPro-Xaa motif in Pro-Pro-Xaa sequences were calculated
as [NP‑cP‑X/(NP‑cP‑X + NP‑tP‑X)]/[NP‑cP/(NP‑cP + NP‑tP)] where N, X, cP
and tP stand for number of occurrences, amino acid Xaa, cisPro, and
transPro, respecively.
MD Simulations. Using the GROMACS 4.0 package,16 two 20 ns
molecular dynamics (MD) simulations were performed on Ac-Pro-
cisPro-Phe-NH2 and Ac-Phe-cisPro-NH2 (OPLSAA17 force field)
starting from the pdb structures 1vky (residues 277−279A) and
1wbh (134−135A), respectively. Both starting structures were
associated with CH···π interaction. The peptides were placed in a
cubic box with the box edge adjusted approximately 7.5 Å from the
peptides’ periphery. The box was filled with suitable number of pre-
equilibrated SPC water molecules. After initial energy minimization
and pre-equilibrium run, the simulations were carried out at constant
temperature (27 °C) and pressure (1 atm) using the Berendsen
coupling method.18 Time constants for temperature and pressure
coupling were 0.1 and 0.5 ps, respectively. All bonds were constrained
1
acid (internal standard). Water signals in 1D H NMR spectra were
suppressed by standard excitation sculpting procedure. For 2D 1H
NMR experiments, the water signal was suppressed by WATERGATE
pulse sequence total correlation spectroscopy and rotating frame
overhause effect spectroscopy (TOCSY and ROESY) or by
presaturation of the water signal double-quantum filtered-correlated
4662
dx.doi.org/10.1021/ja209334v | J. Am. Chem. Soc. 2012, 134, 4661−4669

Journal of the American Chemical Society
Article
using LINCS algorithm.19 A cutoff of 10 Å was used for nonbonded
interactions, and the pairwise neighbor list was updated every 10 steps
(integration step: 2 fs). The trajectories were analyzed with in-house
FORTRAN programs and visualized with CHIMERA.20
Table 1. Populations of cc, tc, ct and tt Isomers in PPX
Peptide Series
b
percent population
ΔG0 (kcal/mol)
a
peptide
cc
tc
ct
tt
tt → tc
tt → ct
RESULTS
PPF
PPY
PPW
18.6
21.5
28.5
11.0
3.0
17.1
24.3
16.6
14.3
4.8
5.0
3.6
3.4
4.5
3.0
7.9
4.4
5.1
4.8
7.0
3.8
5.5
1.1
1.3
0.9
9.5
8.7
54.8
45.5
48.4
63.4
77.8
79.0
79.5
79.8
76.6
80.5
74.1
78.9
77.3
74.3
76.6
80.6
76.0
79.0
81.6
80.0
0.64
0.35
0.59
0.82
1.54
1.52
1.71
1.74
1.56
1.81
1.23
1.59
1.50
1.51
1.32
1.68
1.45
2.36
2.28
2.47
0.97
0.91
1.11
0.95
0.93
0.90
0.91
0.90
0.82
0.92
0.84
0.95
0.88
0.88
0.95
1.00
0.85
0.79
0.88
0.80
■
Populations of cc, ct, tt, and tc in Ac-Pro-Pro-Xaa-NH2.
In an earlier report we demonstrated that the cis isomer of the
Pro-Pro peptide unit in Ac-Pro-Pro-Phe-NH2 shows enhanced
stability compared to that observed in Ac-Pro-Pro-Ala-NH2.4
The origin of this stability was suggested to arise from CH···π
interaction between the Cα-H of the first Pro and the aromatic
side chain of Phe (upfield chemical shift changes in Cα-H and
the restriction of side chain χ1 of Phe). It was further suggested
that such CH···π interactions might be a general phenomenon
in peptides with Pro-Pro-Aromatic motifs. To fully explore the
sequence space of such peptides, we synthesized a series of
peptides with the sequence Ac-Pro-Pro-Xaa-NH2 (PPX) and,
taking advantage of the slow cis−trans isomerization rate,
measured the populations of the cis/trans isomers by
6.5
c
PPH
11.3
14.4
15.5
15.2
15.6
17.2
15.3
16.0
14.0
15.7
15.0
13.7
13.2
16.2
19.0
16.6
18.8
d
PPH
e
PPA
PPS
PPT
PPI
<0.5
1.7
1.2
1.7
PPV
PPL
PPG
PPE
PPD
PPN
PPC
PPM
PPQ
PPK
PPR
1.2
2.0
2.7
1.9
5.9
2.7
integrating appropriate H- /13C NMR resonances. As shown
in Figure 1, the inherent propensities of Ac-Pro and Pro-Pro
moieties to stabilize the cis isomer distribute the total
population of PPX among four isomers: cc, tc, tt, and ct. For
1
2.4
2.3
e
e
e
e
e
e
0.9
0.5
1
0.3
most PPX peptides, the amide H NMR spectra showed four
a
b
PPP not studied. All ΔG0 values correspond to 4 °C and pH = 7
unless otherwise stated. pH = 8.4. pH = 3.5. Unreliable data (signal
∼ noise).
well resolved peaks (for PPY, see Figure 1b) that could be
assigned as one of the four categories from 2D-NMR
experiments. For PPW the tt amide signal overlapped with
the aromatic signals, and for PPH the amide signals were too
broad at high pH. The cis/trans populations for these peptides
were measured by integrating the 13C signals of isotope-labeled
peptides 13CH3CO-Pro-Pro-Xaa-NH2.
c
d
e
The NMR-derived relative populations of cc, tc, tt, and ct
isomers of PPX series of peptides measured at 4 °C in aqueous
buffer of pH 7.0 (unless otherwise stated) are shown in Table
1. All amino acids with an aromatic side chain (Phe, Tyr and
Trp) exhibited enhanced stability of the tc state compared to
the tt state. In addition, PPH also exhibited this enhanced
stability at pH 8.4 (neutral His side chain) but not at pH 3.5
(positively charged His side chain). Populations of the cis
conformers (tc and cc) were the lowest (≤1%) for Arg, Lys, and
Gln, almost at the border of reliable population measurement
by NMR. To emphasize the observed tendency of Phe, Tyr,
Trp, and His (pH 8.4) to be outliers, we computed the trans to
cis free energy differences for the Ac-Pro bond, ΔGt0t/ct, and for
the Pro-Pro bond, ΔGt0t/tc, using the relation ΔG0 = −RT ln K
(Table 1). As shown in Figure 2, the distribution of ΔGt0t/ct for
the peptide series shows a Gaussian distribution (mean: 0.91
0.07 kcal/mol) with no outliers, indicating that the nature of X
in Ac-PPX-NH2 does not influence the free energy difference
between the Ac-cisPro and Ac-transPro isomers. In addition to
the PPX series of peptides, we also synthesized the peptide Ac-
Pro-Ala-NH2 and measured the ΔG0 for the trans to cis
isomerization of the Ac-Pro bond. The value of ΔG0 obtained,
0.88 kcal/mol, is very close to the mean of the ΔGt0t/ct
distribution, indicating that ΔGt0t/ct in PPX remains unaltered
even when X is absent.
Figure 2. Distributions of free energies for trans to cis isomerization,
ΔGt0t/ct (tt → ct) and ΔGt0t/tc (tt → tc), for peptides with sequence Ac-
PPX-NH2. X = R/K/Q (data not shown) and X = aromatic (Y, W, F,
and H (pH 8.4), all highlighted by arrows) data points are excluded
from the ΔGt0t/ct (tt → tc) distribution.
ΔGt0t/tc values exhibited by all aromatic side chains (see Figure
2) are significantly lower than the rest by several standard
deviations (0.82, 0.64, 0.59, and 0.35 kcal/mol for His (pH
8.4), Phe, Trp, and Tyr, respectively). The data are comparable
with an earlier NMR study performed on a series of water-
soluble peptides with the sequence motif -(Xaa-Pro)-.7 In that
study, while the mean ΔGt0/c for nonaromatic Xaa side chains
was 1.21 kcal/mol, the ΔGt0/c values for aromatic Xaa side
chains were lower, 0.98, 0.71, 0.68, and 0.29 kcal/mol for His
(pH 8.4), Phe, Tyr, and Trp, respectively).
pH Dependence of cis/trans Equilibrium of Pro-Pro-
His. The cis/trans isomerization observed for PPH is pH
dependent (Table 1). At low pH, when the imidazole side chain
is protonated, the relative cis content is less than what is
observed at high pH, when the imidazole side chain is neutral.
This is consistent with a CH···π interaction-mediated
stabilization of the cis conformer, since a neutral imidazole is
Except for two groups, X = {R, K, Q} and X = aromatic,
ΔGt0t/tc values of all PPX peptides clustered around 1.55 0.16
kcal/mol (Figure 2 and Table 1). The first outlier group (Lys,
Arg, and Gln) is not considered further, since the data is less
reliable due to low sensitivity of NMR resonances of the cis
populations in these peptides (Table 1). On the other hand, the
4663
dx.doi.org/10.1021/ja209334v | J. Am. Chem. Soc. 2012, 134, 4661−4669

Journal of the American Chemical Society
Article
a better π-donor than the protonated species. If this is indeed
true, then the cis/trans populations should be modulated by pH
in such a manner that their variation match with the His
protonation equilibrium characterized by the His side-chain
pKa. Populations of all the four isomers of PPH were estimated
as a function of pH using NMR spectroscopy (1H amide and
13C acetyl resonances). Variation of cis/trans population ratios
(tc/tt for the Pro-Pro and ct/tt for the Ac-Pro bond) as a
function of pH is shown in Figure 3. Strikingly, the ct/tt ratio is
pH-independent, while the tc/tt ratio shows the expected pH-
dependence.
Figure 4. Cα-H (Pro1) chemical shift differences, (Δδ = δxx − δtt; xx =
tt, ct, tc, and cc) in Ac-Pro-Pro-Xaa-NH2 for all Xaa with aromatic side
chains (red) and a selected set of Xaa with nonaromatic side chains
(blue).
trend was similar to PPAro peptides, both at low and high pH,
indicating that at low pH, even with a protonated side chain,
the His chain is in close proximity to Cα-H of Pro1; although
the stabilization of the CH···π interaction is less than that at
high pH, as was shown in the pH dependent study.
Restriction of Aro χ1 Angle in Pro-Pro-Aro Peptides.
In addition to the upfield shifts of Cα-H of Pro1, we also
3
observed nondegenerate Jαβ coupling constants in aromatic
residues for the Pro-Pro-Aro peptides. As shown in Figure 5,
Figure 3. NMR-detected cis/trans population ratios for Ac-Pro-Pro-
His-NH2 as a function of pH at 4 °C. The [ct]/[tt] ratio (blue) is
unaffected by pH, while the [tc]/[tt] ratio (red) is strongly modulated
by pH. The solid line is the best fit (eq 1) to experimental [tc]/[tt]
ratios.
The pH-dependent [tc]/[tt] population ratio yielded an
0
0
excellent fit with eq 1 with pKa = 6.81 0.22, Ktc = [tc]PPH
/
0
+
0
[tt]PPH = 0.21 0.01 (ΔGtc0 = 0.86 kcal/mol at 4 °C) and Ktc
0
+
+
= [tc]PPH /[tt]PPH = 0.047 0.01 (ΔGtc+ = 1.68 kcal/mol at 4
°C). The pKa obtained from the fit matches well with the His
side-chain pKa, confirming that indeed the cis/trans isomer-
ization of the Pro-Pro bond is modulated by the protonation
state of the imidazole side chain (which in turn is expected to
modulate the CH···π interaction). The net extra stability that
PPH gains going from a protonated to an unprotonated His,
0.82 kcal/mol (ΔΔG0 = ΔGt0c+ − ΔGt0c0), arises due to CH···π
interaction. It is the best estimate for the free energy associated
with CH···π interaction, since the two interconverting
molecules being considered, PPH+ and PPH0, have identical
internal degrees of freedom.
3
3
Figure 5. Absolute differences between Jαβ1 and Jαβ2 in Aro, as
observed in the four isomers of Ac-Pro-Pro-Aro-NH2 (PPY: filled
cricle, PPW: open circle, PPF: filled triangle, PPH (pH 3.5): open
square; PPH (8.0): filled square). The Pro-transPro and the Pro-cisPro
isomers are highlighted in blue and red, respectively.
3
the nondegenerate Jαβ coupling constants were observed only
for Pro-cisPro isomers (tc and cc), with an absolute difference of
Upfield Shift of Cα-H of Pro1 in Pro-Pro-Aro Peptides.
CH···π interaction, if operative in the cis isomer, is expected to
leave its signature in the NMR spectra due to the ring current
effect of the π-system on protons that are in its vicinity. In
Figure 4, Pro1 Cα-H proton chemical shifts of the four isomers
are shown for six peptides, three with aromatic side chains
(PPW, PPY, and PPF) and three without (PPA, PPD, and PPI).
For the tt isomer, Pro1 Cα-H chemical shifts are close to the
‘random coil’ chemical shift value for Pro Cα-H (4.44 ppm) for
all peptides (PPY: 4.65; PPW: 4.57; PPF: 4.67; PPI: 4.72; PPD:
4.71; PPA: 4.71). This is true for all the other three isomers in
peptides lacking an aromatic side chain. However, for peptides
containing an aromatic side chain, the Pro Cα-H exhibits a
marked upfield shift (1−2 ppm) for the tc and the cc isomers,
both containing the Pro-cisPro motif and not the ct isomer,
containing the Ac-cisPro motif only. This strongly suggests that
the Pro-cisPro, in PPY, PPW, and PPF, are associated with a
conformation where Cα-H of Pro1 and the aromatic side chain
are proximal, consistent with CH···π interaction. For PPH, the
∼6−9 Hz between 3Jαβ1 and 3Jαβ2. Nondegenerate 3Jαβ coupling
constants (3Jαβ1 > 5 Hz and 3Jαβ2 < 10 Hz or vice versa; Δ3Jαβ
>
5 Hz) can arise only when the χ1 dihedral angle is restricted to
the g- (χ1 = −60°) or the t- (χ1 = 180°) rotamer; on the other
hand, degenerate 3Jαβ values can either indicate the presence of
the g+ (χ1 = +60°) rotamer (Δ3Jαβ1 ∼ Δ3Jαβ2 < 5 Hz) or fast
rotation along χ1 (3Jαβ1 ∼ Jαβ2 ∼ 7 Hz).13 In the Pro-Pro-Aro
3
peptide series, 3Jαβ1 and 3Jαβ2 values were similar (7 1 Hz) for
the tt and the ct conformations, indicating fast rotation of the
aromatic side chain. On the other hand for the tc and the cc
3
isomers, the observed spread in nondegenerate Jαβ values
indicated the presence of restricted rotation along χ1 (either
180° or −60°). Analyzing the relative ROE cross-peak
intensities (α-βA, α-βB, N-βA, N-βB)13 it was further found
(Table 2) that the preferred χ1 angle is −60° (g-) and not 180°
(t). Restriction of side-chain dihedral angle χ1 of a terminal
amino acid in a three-residue peptide is remarkable and is fully
compatible with a model where the side chains of the aromatic
4664
dx.doi.org/10.1021/ja209334v | J. Am. Chem. Soc. 2012, 134, 4661−4669

Journal of the American Chemical Society
Article
a
shown in Figure 7c, cross-peaks between Pro1 Cα-H and Phe
Cδ-H were observed for the cc as well as the tc isomer in PPF.
The acetyl methyl protons also exhibited a cross peak with the
aromatic side chain (either Cε-H or Cζ-H but not Cδ-H) for the
cc isomer. For PPW (Figure 7d) all aromatic protons of the cc
isomer exhibited ROE cross-peaks with Pro1 Cα-H. This is true
for the methyl protons of the acetyl group as well. For the tc
isomer the Cα-H of Pro1 (3.40 ppm) overlaps with Cβ-H (3.45
and 3.31 ppm). Thus even if ROE cross-peaks were present
between Pro1 Cα-H and Trp aromatic protons, they would be
buried under a strong Cβ-H cross-peak.
Table 2. Relative ROE Intensities of Aro β-Protons
3Jαβ (Hz)
relative ROE intensities
peptides
α-βA/α-βB
α-βA
α-βB
N-βA
N-βB
PPF (cc)
PPF (tc)
PPY (cc)
PPY (tc)
PPW (cc)
PPW (tc)
?/12.50
3.95/12.13
3.38/12.84
3.85/12.83
3.75/12.69
?/?
?
?
+++
+++
+++
+++
+++
?
+++
+++
+++
+++
?
+
+
+
+
?
?
a
βA/βB: downfield/upfield-shifted β-protons; ?: overlap; +++: strong
ROE; +: weak ROE; : weak/absent ROE.
DISCUSSION
■
residues are ‘frozen’ due to their participation in CH···π
interaction.
CH···π Interaction and the Pro-Pro-Aro Motif. Weak
interactions can play an important role in dictating protein
structures. Among them the CH···π interaction is notable.1
First proposed by Perutz,21 CH···π interactions have been
extensively studied in crystal structures of proteins2 and
peptides.22 In some proteins, even the disruption of a single
CH···π interaction can alter the stability and functional
robustness of a protein.23 However, due to its weak nature, a
direct experimental observation of CH···π interaction is tricky.
A recent report describes the direct detection of CH···π
interactions in proteins from carefully designed NMR experi-
ments.3 In this work we report the direct observation of CH···π
interactions in a series of three-residue peptides with the
general sequence Pro-Pro-Aro, operating between the Cα-H of
the first Pro residue and the π-electrons from the side chain of
the third aromatic residue. The CH···π interaction stabilizes the
cis conformation of the tertiary amide (tamide) bond between
the first two Pro residues.
Thermodynamic Analysis. In our earlier work on PPF and
PPA,4 van’t Hoff analyses of temperature-dependent NMR
spectra yielded entropic and enthalpic contributions for the tt
→ tc transition. Both PPF and PPA yielded unfavorable
entropies, while the associated enthalpy was favorable for PPF
and unfavorable for PPA. Similar analyses were performed on a
larger set of PPX peptides with aromatic (W and Y) as well as
nonaromatic (I, E, and N) X. As shown in Figure 6, for all
As demonstrated earlier by a number of studies on
peptides,7,10−12,24 the amide bond (Xaa-Pro) can be stabilized
t
by the presence of aromatic residues in the vicinity of Pro in the
amino acid sequence. Inspired by the observation of local
CH···π interactions in protein structures associated with the cis
tamide bond,5e we had also demonstrated that the presence of
Phe instead of Ala, following a Pro-Pro fragment, increases the
Figure 6. Thermodynamic parameters (ΔH° and ΔS°) associated with
the equilibrium tt → tc (Figure 1) for a selected set of peptides in the
PPX series where X is aromatic (Y, W, and F) and nonaromatic (I, E,
N, and A). The ΔH° and ΔS° values were obtained from van’t Hoff
analysis of temperature-dependent NMR experiments (integrals of
amide/acetyl resonances).
4
t
cis population of the amide bond. None of these studies,
including ours, could provide direct evidence for the Pro···Aro
interaction, although indirect evidence from chemical shift
anomaly and restricted rotation of the aromatic side chain was
provided in some of the studies.4,10,11c,12 In another study, not
related to cis peptide bonds, direct evidence of Cε-H(Lys)···π
interaction was provided from NOE crosspeaks in a designed β-
hairpin peptide.25 Interestingly, an early NMR study of cyclic
somatostatin analog by Kessler et al.,26 where evidence was
provided for interaction between sequence contiguous Phe
(aromatic protons) and D-Pro (Cδ-H), but the nature of the
interaction not explored, might have been one of the earliest
observation on the manifestation of CH···π interaction in
peptides.
peptides the entropy change associated with the tt → tc process
is unfavorable, from ∼5 cal/deg/mol (PPW and PPY) to ∼3.5
cal/deg/mol (I, E and N). The associated enthalpy change is
favorable only when X is aromatic (−0.45,4 −1.7, and −1.1
kcal/mol for F, W, and Y, respectively). The observed trend
indicates that tt → tc transition for PPX is enthalpy driven only
when X is aromatic.
Direct Evidence for CH···π Interaction from ROE. The
indirect evidence for CH···π interaction operative in the Pro-
cisPro conformations in PPX series where X is aromatic was
further confirmed from direct experimental evidence. Direct
evidence for CH···π interaction was obtained from ROE cross
peaks between Pro1 Cα-H and Tyr side-chain protons in PPY.
As shown in Figure 7b, the ROESY spectrum of PPY shows
clear cross-peaks between all the aromatic protons of Tyr (Cε-
H and Cδ-H) and Pro1 Cα-H for the Pro-cisPro isomers (cc and
tc) only. Models of PPY in cc and in tc conformations are
shown in Figure 7a for clarity. For PPY, ROESY cross-peaks
were also observed between the methyl protons of the acetyl
group and the Tyr side chain (Cε-H/Cδ-H) for the cc isomer.
ROESY cross-peaks were observed between Pro1 Cα-H and
aromatic side-chain protons for PPF and PPW as well. As
The presence of ROE cross-peaks between Pro1 Cα-H of and
aromatic protons in Ac-Pro-cisPro-Aro-NH2 (Aro = Tyr, Phe,
Trp) is a direct evidence CH···π interaction between Pro1 and
Aro3, separated by Pro2. This is supported by the observed
restriction of Aro χ1 angle and upfield shift of Pro1 Cα-H. The
corresponding Ac-Pro-transPro-Aro-NH2 conformers did not
exhibit any evidence for CH···π interaction. An enhanced
stability of the -Pro-cisPro-Xaa- unit (compared to -Pro-
transPro-Xaa) was associated with this CH···π interaction.
When Xaa changes from nonaromatic to aromatic, the
corresponding ΔG0 (tt → tc) also becomes less unfavorable
4665
dx.doi.org/10.1021/ja209334v | J. Am. Chem. Soc. 2012, 134, 4661−4669

Journal of the American Chemical Society
Article
Figure 7. (a) Molecular models of the tc (Ac-tPro-cPro) and the cc (Ac-cPro-tPro) isomers in Ac-Pro-Pro-Tyr-NH2. ROE cross peaks between Pro1
Cα-H/acetyl methyl and aromatic ring protons in: (b) Ac-Pro-Pro-Tyr-NH2, (c) Ac-Pro-Pro-Phe-NH2, and (d) Ac-Pro-Pro-Trp-NH2. Relevant cross
peaks are indicated by arrows and colored according to isomer type (cc: red; tc: blue).
(1.55 0.16 to 0.82, 0.64, 0.59, and 0.35 kcal/mol for His (pH
8.4), Phe, Trp, and Tyr, respectively). Therefore, the CH···π
interaction stabilizes the Pro-cisPro-Aro conformer by about
0.95 kcal/mol. The pH-dependent modulation of the Pro-
cisPro conformation in Pro-Pro-His yielded a ΔG0 value of 0.82
kcal/mol for the trans/cis isomerization of the Pro-Pro bond,
demonstrating how protonating the His side chain, which
directly affects the CH···π interaction, can make Pro-Pro-His
behave like Pro-Pro-Xaa (Xaa not aromatic) series of peptides.
Our results establish Pro-Pro-Aro to be a new sequence motif
that can stabilize Pro-cisPro peptide bonds along with a Xaa-
dependent thermodynamic scale of the cis/trans isomerization
of Pro-Pro-Xaa motif (Table 1).
The thermodynamic scale could be useful in fine-tuning the
design of water-soluble peptides. Balaram and co-workers27
studied the effect of N-terminal diproline templates in helix
nucleation and had observed significant cis population of Pro-
Pro-Xaa moiety from NMR studies in nonaqueous solvents.
Hilser and co-worker28 probed the denatured state of
polyproline II conformation by studying the binding of SH3
domain of SEM-5 to a series of peptides with Pro-Pro-Xaa
motif. In both cases, Xaa-mediated cis/trans isomerization can
potentially affect the conformational ensemble of the designed
peptides. To account for such effect, a comprehensive
thermodynamic scale as in Table 1 will be helpful.
End-Capping Effects. The peptide series we studied,
capped by an acetyl and an amide group at the two termini,
were only three-residues long, essentially eliminating end-
capping effects that could arise from the presence of full amino
acid residues at the termini. A preliminary study of end-capping
effects was undertaken with two peptides with an aromatic and
a nonaromatic residue following the Pro-Pro unit, PPY and
PPN. Four peptides corresponding to PPY and PPN were
synthesized, each capped by an Ala residue at the either
termini: Ac-APPY-NH2, Ac-APPN-NH2, Ac-PPYA-NH2, and
Ac-PPNA-NH2. NMR investigation (data not shown)
performed on the peptides showed that the cis/trans population
ratios of APPY and PPYA were comparable to PPY, while that
in APPN and PPNA were comparable to PPN, demonstrating
that the enhanced preference for the Pro-cisPro conformation
in Pro-Pro-Aro is not affected by Ala capping of the two
termini. Although only a thorough investigation with a variety
of amino acids capping the Pro-Pro-Xaa motif can result in a
more robust conclusion about end-capping effects on
modulating the cis/trans equilibrium of Pro-Pro-Xaa sequence
motif, an earlier published result shows that the PPY motif’s
preference for the Pro-cisPro remains unaffected even when
placed in a longer peptide. Kemmink and Creighton studied
two peptides with the general sequence RPDFSLE-(PPX)-
TGPSK (X = A and Y).10 The populations of the Pro-cisPro
isomer in their ‘peptide-capped’ PPA and PPY peptides were
5% and 20%, respectively, matching well with our ‘uncapped’
PPA and PPY peptides (Table 1).
CH···π Interaction in Pro-cisPro-Aro and Aro-cisPro.
Another characteristic of the trans/cis isomerization of the Pro-
Pro motif in Pro-Pro-Xaa series of peptides is: (i) unfavorable
ΔS0 (3.5−5 cal/mol/deg) independent of the nature of Xaa
and (ii) favorable ΔH0 only when Xaa is aromatic. Wu and
Raleigh have reported ΔH0 and ΔS0 values (van’t Hoff analysis)
associated with trans → cis isomerization of a related peptide
series, containing the -Aro-Pro- motif: Gly-Aro-Pro-Gly (Aro =
Phe, Tyr, Trp).11c The ΔS0 values for the Aro-Pro series of
peptides are near zero, while for the Pro-Pro-Xaa series it is
unfavorable. This difference is understandable since only one
(Aro-ψ) backbone and two (Aro-χ1, Aro-χ2) side-chain
dihedral angles get restricted in the Aro-cisPro conformer,
while three (Pro1-ψ, Pro2-ψ, Aro-ϕ) backbone and two (Aro-
χ1, Aro-χ2) side-chain dihedral angles get restricted in the Pro-
cisPro isomer (assuming the Pro-ϕ angles are constrained). The
ΔH0 values reported for the Aro-Pro series of peptides are all
unfavorable, 0.53, 0.65, and 0.75 kcal/mol for Phe, Tyr, and
Trp, respectively. The ΔH0 values strongly contrast with
favorable ΔH0 values for the same (trans → cis) isomerization
process in our peptide series (−0.45,4 −1.7, and −1.1 kcal/mol
4666
dx.doi.org/10.1021/ja209334v | J. Am. Chem. Soc. 2012, 134, 4661−4669

Journal of the American Chemical Society
Article
for PPF, PPW, and PPY, respectively). Notwithstanding the
enthalpy difference trend, there was indirect evidence for
Aro1···Pro2/Pro1···Aro3 interaction in the cis isomer for both
the Aro-Pro as well as the Pro-Pro-Aro series of peptides: (i)
upfield shifted Pro Cα-H resonances in NMR and (ii)
restriction of Aro χ1 angle. Yet, the trans isomer is enthalpically
favored (compared to the cis isomer) for the Aro-Pro series,
while the cis isomer is enthalpically favored (compared to the
trans isomer) for the Pro-Pro-Aro series. The observed ΔH0 is
the enthalpy difference between the cis and the trans states.
Therefore, either the nature of Aro···Pro interaction in Aro-
cisPro and Pro-cisPro-Aro is different or one of the two trans
conformers is characterized by some unique interaction absent
in the other trans state that gets disrupted in the cis form.
If differences in the Pro-cisPro and Aro-cisPro isomers are
responsible for the different ΔH0 trends in the two peptide
series, the underlying cause can either be the different nature of
CH···Aro interaction or the different residence times of
CH···Aro interactions in the two cis isomers (or both). The
NMR-derived dominant Aro χ1 angle for the cis conformer is
different in both t- (χ1 = 180°) in Aro-cisPro and g- (χ1 =
−60°) in Pro-cisPro-Aro. Analyses of high-resolution protein
structures from pdb also showed that the Aro···Pro interacting
geometries are different for the two cases (Figure 8). For Pro-
Aro]. K0 is given by K0 = KNAR/KARO − 1. Using KNAR (tt → tc
of Pro-Pro-Ala) = 0.06, K0 values (ratio of tc isomers with and
without CH···Aro interaction) were calculated for the Pro-Pro-
Aro series. Similarly, using KNAR (ct → cc of Pro-Pro-Ser) =
0.11, K0 values (ratio of cc isomers with and without CH···Aro
interaction) were calculated for the Pro-Pro-Aro series. The
ratios were translated into percent Pro-cisPro populations
(Table 3). About 90−95% of the cc isomers and 70−85% of the
Table 3. Percent cis Population Associated with CH···π
Interaction
percent population
peptides
cc isomer
tc isomer
a
PPF
PPY
94 (91)
78 (87)
88 (94)
81 (-)
96 (94)
97 (93)
PPW
PPH (pH 8.0)
89 (77)
71 (77)
a
3
Numbers within parentheses are Jαβ-based populations.
tc isomers were found to exhibit the CH···Aro interaction.
These values are larger by ∼20−30% than the corresponding cis
populations in the Aro-Pro series that exhibited the CH···Aro
interaction,11c estimated by an identical method (Trp-cisPro:
76%, Tyr-cisPro: 67%, Phe-cisPro: 60%).
The residence times of the side-chain rotamers (χ1) of the
Aro moiety in Xaa-Pro series have also been reported for the
Aro-Pro series of peptides using Jαβ values.11c The reported
3
values are not much different than those obtained from
equilibrium constant-based values. We also calculated NMR-
derived residence times of Aro side-chain rotamers in the Pro-
cisPro-Aro peptide series using a methodology suggested by
Pachler.28 The estimated values for the most populated rotamer
(Pro-cisPro-Aro) are shown in Table 3. The NMR-derived
populations of the most populated rotamers are consistent with
corresponding estimates from equilibrium constants, confirm-
ing that Pro-cisPro-Aro is characterized by a very tight CH···Aro
interaction. This was also confirmed from MD simulations.
While the Pro1-Phe3 interaction in Pro-cisPro-Aro remained
intact for the entire 20 ns trajectory, Phe1-Pro2 interaction in
the Phe-cisPro peptide survived for only about 50% of the 20 ns
trajectory.
Pro-cisPro-Xaa Populations in Peptides and Proteins.
Using a database of nonredundant protein structures (25 588
Xaa-Pro and 1039 Pro-Pro units), we recalculated propensities
of Pro-cisPro as a function of Xaa. Qualitatively the propensity
values (Table S3, Supporting Information) matched with an
earlier report that used a different database.4 Tyr showed the
highest propensity (3.5), followed by Phe (2.8). The high
propensity (1.6) for Trp is not reliable due to sparse data.
Overall the trend, that aromatic side chains show a preference
to occur following a Pro-cisPro unit, correlated with the NMR
results.
Figure 8. Side-chain χ1 dihedral angle distributions of Aro side chain
in a) Aro-Pro and b) Pro-Pro-Aro sequence motifs in high-resolution
protein structures as a function of cis/trans isomeric state of the Aro-
t
Pro or the Pro-Pro amide bond.
cisPro-Aro motifs, the most populated χ1 angle in Aro is −60°.
On the other hand, for Aro-cisPro motifs, the most populated
χ1 value is 180°. These two χ1 angles correspond to favorable
CH···Aro interaction in the two peptides. We compared the
nature of CH···π interactions in the two peptides by analyzing
20 ns MD trajectories of Ac-Pro-cisPro-Phe-NH2 and Ac-Phe-
cisPro-NH2 in water (starting from two pdb coordinates that
exhibited CH···Aro interaction, 1vky and 1wbh). Distances
between Pro1 Cα-H and Phe ring center and the angle between
Pro1 Cα, Hα and Phe ring center were calculated for those
structures for which χ1 was 180° (Phe-Pro) or −60° (Pro-Pro-
Phe). These values were similar for the two peptides: 4.0 0.3
Å and 137 13° for Phe-Pro and 4.1 0.3 Å and 132 9°,
both compatible with favorable CH···π interaction geometry. In
other words, the two cis conformations are very similar as far as
CH···π interaction is concerned.
Next we investigated the residence times of CH···π
interactions in the two cis conformations. Using a methodology
suggested by Kemmink and Creighton,10 relative populations of
Pro-cisPro and Pro-cisPro were calculated where Pro-cisPro and
Pro-cisPro are two types of cis isomers, with and without the
CH···Aro interaction, respectively. Let KARO and KNAR be the
trans → cis equilibrium constants for two peptides with
sequence Pro-Pro-Xaa, where Xaa is aromatic (KARO) and
nonaromatic (KNAR), and K0 = [Pro-cisPro-Aro]/[Pro-cisPro-
Reimer et al.7 had estimated percent Xaa-cisPro units in
proteins and compared them with percent cis population of
Xaa-Pro moiety in a synthetic peptide series Ac-Ala-Xaa-Pro-
Ala-Lys-NH2 (blue symbols in Figure 9). Xaa = Pro was an
outlier in their data in the sense that the NMR-derived cis
content was markedly lower than the corresponding protein cis
content, when compared to other amino acids. This prompted
us to hypothesize that an isolated -Pro-Pro- motif does not
truly reflect major local factors responsible for the observed
-Pro-cisPro- population in proteins. We designed and studied
4667
dx.doi.org/10.1021/ja209334v | J. Am. Chem. Soc. 2012, 134, 4661−4669

Journal of the American Chemical Society
Article
structure is cis rather than when it is trans. For example, for
tyrosinase (pdb code: 1wxc; containing 168Pro-cisPro-Tyr170),
alignment of homologous sequences (up to 37% sequence
identity) obtained from a BLAST30 search showed Pro-Pro-Tyr
to be conserved with the general sequence motif Xaa-Pro-Aro
(Figure 10a). As shown in Figure 10b, NMR spectra of Ac-Ala-
Figure 9. Correlation between percent cis population of Xaa-Pro
sequence motifs in peptides and proteins. The peptide-based data are
from NMR studies on synthetic peptides by Reimer et al.7 The
protein-based data are from analysis of pdb structures, one set from
Reimer et al.7 (blue) and the other from this study (red); differences
between the two sets are shown in the inset. For Pro, the third point
(P, shown in black) corresponds to protein cis population from Reimer
et al.7 and an adjusted peptide cis population based on this study (see
text for details). The solid line is the linear fit for a subset (F, Y, and
W), and the dotted line represents equal populations of Xaa-cisPro
conformation in peptides and proteins.
Figure 10. (a) Sequence logo plot31 of a three-residue fragment of
tyrosinase (168Pro-cisPro-Tyr170; pdb code: 1wxc) and its sequence
homologues (up to 37% sequence identity). Major sequence motifs are
1
listed separately. (b) H NMR subspectrum of Ac-Ala-Pro-Trp-NH2
depicting the methyl (Ala) resonances in Ac-Ala-transPro-Trp-NH2
and Ac-Ala-cisPro-Trp-NH2 isomers.
the Pro-Pro-Xaa series of peptides to investigate if the nature of
Xaa affects the Pro-cisPro content in Pro-Pro-Xaa. Indeed Xaa
was found to affect the cis content of Pro-Pro-Xaa. As a follow
up, the corrected NMR cis content for Pro in Figure 9 was
updated as: [%cis]Pro‑Pro = ∑Xaa(FXaa × YXaa), where FXaa is the
fraction of Pro-cisPro-Xaa in all Pro-Pro motifs in proteins and
Pro-Trp-NH2 (Xaa = Ala; Aro = Trp) showed the population
of Ala-cisPro to be about 21%, compatible with the cis content
of Pro-Pro-Aro (Table 1), while the corresponding population
of Ac-Ala-cisPro-Ala-NH2 was only 5%, also compatible with cis
content of Pro-Pro-nonAro (Table 1). It has been previously
shown that the Ala-cisPro conformation is about 20% populated
in Ac-Ala-Pro-Tyr-NHCH3, another peptide with Ala-Pro-Aro
Y
Xaa is the percent cis content of the Pro-Pro-Xaa peptide (cc +
tc in Table 1). As expected, the new value, shown in Figure 9
(black letter P, indicated by a vertical arrow), increased the
NMR-derived cis content of Pro-Pro (from 6% to 15.4%) and
Pro was no more an outlier.
motif.12b In other words, the Xaa-cisPro amide bond is
t
conserved, sometimes as Pro-Pro-Aro but more generally as
Xaa-Pro-Aro. It is not that the Pro-Pro-Aro sequence motif is
conserved (as Pro-Pro-Aro or as Xaa-Pro-Aro) only in proteins
We also recalculated the percent cis populations of Xaa-Pro
motifs in proteins using our database (Table S4, Supporting
Information; red alphabets in Figure 9) to update the values
reported by Reimer et al.7 Eleven amino acid types showed
more than 1% variation than that reported by Reimer et al.7
(inset to Figure 9). All amino acids except Pro registered an
increase; with the new protein Pro-Pro cis content (horizontal
arrow, Figure 9), the Pro-Pro data point moved closer to the
line representing the line correlating peptide and protein cis
contents of Phe, Tyr and Trp. However, the new Gly-Pro data
seems to be the new exception. Interestingly, analysis of -Gly-
cisPro- fragments in protein showed an unusually high number
of Gly-cisPro-Tyr fragments. Possibly Gly-Pro-Aro is a
sequence motif that is responsible for the unusually high
percent cis population of Gly-Pro in proteins. Only a series of
peptides with the general sequence Gly-Pro-Xaa (with Xaa
being aromatic and nonaromatic) can resolve the issue.
Conservation of Pro-cisPro-Xaa in Proteins. Cis prolyl
residues are conserved more than trans prolyl residues in
proteins, the former sometimes maintained in proteins with
sequence identity as low as 20%.8 We looked at the sequence
conservation of Pro-Pro-Aro motifs in our structural data set.
Conservation of Pro-Pro-Aro sequence motifs in sequence
homologous proteins of a representative set of protein
structures with Pro-cisPro-Aro and Pro-transPro-Aro fragments
were examined. Preliminary analysis showed that Pro-Pro-Aro
motifs are more conserved (either as Pro-Pro-Aro or Xaa-Pro-
Aro, where Xaa is any amino acid) in sequence homologous
t
containing Pro-cisPro amide bonds. Even when the Pro-Pro is
trans, the Pro-Pro-Aro sequence motif is sometimes conserved
among homologous proteins, probably implying either a
functional role of the sequence motif or that the cis conformer
is accessed by the protein transiently for proper folding or
function.
CONCLUSION
■
From a systematic analysis of a peptide series with the general
sequence Ac-Pro-Pro-Xaa-NH2, we show that the populations
of the Pro-cisPro structural motif are significantly higher in
peptides when Xaa is aromatic. This establishes Pro-Pro-Aro as
a sequence motif that can stabilize the Pro-cisPro conformation.
As supported by direct (NOE) as well as indirect (ring-current
effect on Pro1 Cα-H chemical shifts and χ1 restriction)
evidence, the origin of this stabilization arises due to interaction
between Cα-H of the first proline and the side chain of the third
aromatic residue. The nature of the interaction is CH···π type,
as evident from the pH-dependent variation of cis/trans
population ratio in the peptide Ac-Pro-Pro-His-NH2. When
compared to another sequence motif that also stabilizes cisPro,
Aro-cisPro, the CH···π interaction in Pro-Pro-cisAro sequence
motif was shown to be tighter, allowing direct observation of
NOE cross peaks between Cα-H of the first proline residue and
the side-chain atoms of the third aromatic residue. A complete
thermodynamic scale, quantifying the effect of Xaa in stabilizing
t
proteins when the Pro-Pro amide bond in the original protein
4668
dx.doi.org/10.1021/ja209334v | J. Am. Chem. Soc. 2012, 134, 4661−4669

Journal of the American Chemical Society
Article
(13) Wuthrich, K. NMR of Protiens and Nucleic Acids; Wiley-
Interscience Publication: New York, 1986.
̈
Pro-cisPro conformation in Pro-Pro-Xaa sequence motifs, is
presented, and the relevance of this sequence motif in
stabilizing cisPro bonds in proteins is explored.
(14) Berman, H. M.; Westbrook, J.; Feng, Z.; Gilliland, G.; Bhat, T.
N.; Weissig, H.; Shindyalov, I. N.; Bourne, P. E. Nucleic Acids Res.
2000, 28, 235−242.
(15) Wang, G.; Dunbrack, R. L. Jr. Nucleic Acids Res. 2005, 33,
W94−98.
(16) Van Der Spoel, D.; Lindahl, E.; Hess, B.; Groenhof, G.; Mark, A.
E.; Berendsen, H. J. J. Comput. Chem. 2005, 26, 1701−18.
(17) Jorgensen, W. L.; Tirado-Rives, J. J. Am. Chem. Soc. 1988, 110,
1657−1666.
(18) Berendsen, H. J. C.; Postma, J. P. M; DiNola, A.; Haak, J. R. J.
Chem. Phy.s 1984, 81, 3684−3690.
(19) Hess, B.; Bekker, H.; Berendsen, H. J. C.; Fraaije, J. G. E. M. J.
Comput. Chem. 1997, 18, 6933−6939.
(20) Pettersen, E. F.; Goddard, T. D.; Huang, C. C.; Couch, G. S.;
Greenblatt, D. M. J. Comput. Chem. 2004, 24, 1605−1612.
(21) Perutz, M. F. Philos. Trans. R. Soc. London A 1993, 345, 105−
112.
ASSOCIATED CONTENT
■
S
* Supporting Information
Derivation of eq 1 and four tables (Tables S1−S4). This
material is available free of charge via the Internet at http://
pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
gautamda@gmail.com or gautam@boseinst.ernet.in
Notes
The authors declare no competing financial interest.
(22) Umezawa, Y.; Tsuboyama, S.; Takahashi, H.; Uzawa, J.; Nishio,
M. Bioorg. Med. Chem. 1999, 7, 2021−2026.
ACKNOWLEDGMENTS
■
G.B. acknowledges financial support from Department of
Science and Technology, Government of India (SR/SO/BB/-
11/2008).
(23) Harigai, M.; Kataoka, M.; Imamoto, Y. J. Am. Chem. Soc. 2006,
128, 10646−10647.
(24) Meng, H. Y.; Thomas, K. M.; Lee, A. E.; Zondlo, N. J.
Biopolymers 2006, 84, 192−204.
(25) Tatko, C. D.; Waters, M. L. J. Am. Chem. Soc. 2004, 126, 2028−
2034.
REFERENCES
■
(1) Nishio, M.; Hirota, M.; Umezawa, Y. The CH/π Interaction.
Evidence, Nature and Consequences, Wiley-VCH: New York, 1988.
(2) Brandl, M.; Weiss, M. S.; Jabs, A.; Suhnel, J.; Hilgenfeld, R. J. Mol.
Biol. 2001, 307, 357−377.
(3) Plevin, M. J.; Bryce, D. L.; Boisbouvier, J. Nat. Chem. 2010, 2,
466−471.
(4) Dasgupta, B.; Chakrabarti, P.; Basu, G. FEBS Lett. 2007, 581,
4529−32.
(5) (a) Stewart, D. E.; Sarkar, A.; Wampler, J. E. J. Mol. Biol. 1990,
214, 253−260. (b) MacArthur, M. W.; Thornton, J. M. J. Mol. Biol.
1991, 218, 397−412. (c) Weiss, M. S.; Jabs, A.; Hilgenfeld, R. Nat.
Struct. Biol. 1998, 5, 676. (d) Jabs, A.; Weiss, M. S.; Hilgenfeld, R. J.
Mol. Biol. 1999, 286, 291−304. (e) Pal, D.; Chakrabarti, P. J. Mol. Biol.
1999, 294, 271−288.
(6) Maigret, B.; Perahia, D.; Pullman, B. J. Theor. Biol. 1970, 29,
275−291.
(7) Reimer, U.; Scherer, G.; Drewello, M.; Kruber, S.; Schutkowski,
M.; Fischer, G. J. Mol. Biol. 1998, 279, 449−460.
(26) Kessler, H.; Bats, J. W.; Griesinger, C.; Koll, S.; Will, M.;
Wagner, K. J. Am. Chem. Soc. 1988, 110, 1033−1049.
(27) (a) Rai, R.; Aravinda, S.; Kanagarajadurai, K.; Raghothama, S.;
Shamala, N.; Balaram, P. J. Am. Chem. Soc. 2006, 128, 7916−28.
(b) Chatterjee, B.; Saha, I.; Raghothama, S.; Aravinda, S.; Rai, R.;
Shamala, N.; Balaram, P. Chem.Eur. J. 2008, 14, 6192−204.
(28) Ferreon, J. C.; Hilser, V. J. Protein Sci. 2003, 12, 447−57.
(29) Pachler, K. G. R. Spect. Chim. Acta 1963, 19, 2085−2092.
(30) Altschul, S. F.; Gish, W.; Miller, W.; Myers, E. W.; Lipman, D. J.
J. Mol. Biol. 1990, 215, 403−410.
(31) Crooks, G. E.; Hon, G.; Chandonia, J. M.; Brenner, S. E.
Genome Res. 2004, 14, 1188−1190.
(8) (a) Lorenzen, S.; Peters, B.; Goede, A.; Preissner, R.; Frommel,
C. Proteins 2005, 58, 589−595. (b) Exarchos, K. P.; Exarchos, T. P.;
Papaloukas, C.; Troganis, A. N.; Fotiadis, D. I. BMC Bioinformatics
2009, 10, 113. (c) Wathen, B.; Jia, Z. J. Proteome Res. 2008, 7, 145−
153.
(9) (a) Tweedy, N. B.; Nair, S. K.; Paterno, S. A.; Fierke, C. A.;
Christianson, D. W. Biochemistry 1993, 32, 10944−10949. (b) Mayr,
L. M.; Willbold, D.; Rosch, P.; Schmid, F. X. J. Mol. Biol. 1994, 240,
288−293. (c) Dodge, R. W.; Scheraga, H. A. Biochemistry 1996, 35,
1548−1559. (d) Jin, L.; Stec, B.; Kantrowitz, E. R. Biochemistry 2000,
39, 8058−8066. (e) Wu, Y.; Matthews, C. R. J. Mol. Biol. 2002, 322,
7−13. (f) Guan, R. J.; Xiang, Y.; He, X. L.; Wang, C. G.; Wang, M.;
Zhang, Y.; Sundberg, E. J.; Wang, D. C. J. Mol. Biol. 2004, 341, 1189−
1204. (g) Birolo, L.; Malashkevich, V. N.; Capitani, G.; De Luca, F.;
Moretta, A.; Jansonius, J. N.; Marino, G. Biochemistry 1999, 38, 905−
913. (h) Forstner, M.; Muller, A.; Rognan, D.; Kriechbaum, M.;
Wallimann, T. Protein Eng. 1998, 11, 563−568.
(10) Kemmink, J.; Creighton, T. E. J. Mol. Biol. 1993, 234, 861−878.
(11) (a) Grathwohl, C.; Wuthrich, K. Biopolymers 1976, 15, 2025−
2041. (b) Yao, J.; Feher, V. A.; Espejo, B. F.; Reymond, M. T.; Wright,
P. E.; Dyson, H. J. J. Mol. Biol. 1994, 243, 736−753. (c) Wu, W. J.;
Raleigh, D. P. Biopolymers 1998, 45, 381−394.
(12) (a) Kemmink, J.; Creighton, T. E. J. Mol. Biol. 1995, 245, 251−
260. (b) Nardi, F.; Kemmink, J.; Sattler, M.; Wade, R. C. J. Biomol.
NMR 2000, 17, 63−77.
4669
dx.doi.org/10.1021/ja209334v | J. Am. Chem. Soc. 2012, 134, 4661−4669