Conformational ensembles of flexible β-turn mimetics in DMSO-d 6

Article abstract of DOI:10.1039/b921794k

β-Turns play an important role in peptide and protein chemistry, biophysics, and bioinformatics. The aim of this research was to study short linear peptides that have a high propensity to form β-turn structures in solution. In particular, we examined conformational ensembles of β-turn forming peptides with a general sequence CBz-L-Ala-L-Xaa-Gly-L-Ala-OtBu. These tetrapeptides, APGA, A(4R)MePGA, and A(4S)MePGA, incorporate proline, (4R)-methylproline, and (4S)-methylproline, respectively, at the Xaa position. To determine the influence of the 4-methyl substituted prolines on the β-turn populations, the NAMFIS (NMR analysis of molecular flexibility in solution) deconvolution analysis for these three peptides were performed in DMSO-d₆ solution. The NBO (natural bond orbital) method was employed to gain further insight into the results obtained from the NAMFIS analysis. The emphasis in the NBO analysis was to characterize remote intramolecular interactions that could influence the backbone-backbone interactions contributing to β-turn stability. NAMFIS results indicate that the enantiospecific incorporation of the methyl substituent at the C ^γ (C4) position of the proline residue can be used to selectively control the pyrrolidine ring puckering propensities and, consequently, the preferred ,ψ angles associated with the proline residue in β-turn forming peptides. The NAMFIS analyses show that the presence of (4S)-methylproline in A(4S)MePGA considerably increased the type II β-turn population with respect to APGA and A(4R)MePGA. The NBO calculations suggest that this observation can be rationalized based on an n→π* interaction between the N-terminus alanine carbonyl oxygen and the proline carbonyl group. Several other interactions between remote orbitals in these peptides provide a more detailed explanation for the observed population distributions.

Full text of DOI:10.1039/b921794k

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/obc | Organic & Biomolecular Chemistry
Conformational ensembles of flexible b-turn mimetics in DMSO-d6†
Jari J. Koivisto, Esa T. T. Kumpulainen and Ari M. P. Koskinen*
Received 19th October 2009, Accepted 19th February 2010
First published as an Advance Article on the web 11th March 2010
DOI: 10.1039/b921794k
b-Turns play an important role in peptide and protein chemistry, biophysics, and bioinformatics. The
aim of this research was to study short linear peptides that have a high propensity to form b-turn
structures in solution. In particular, we examined conformational ensembles of b-turn forming peptides
with a general sequence CBz-L-Ala-L-Xaa-Gly-L-Ala-OtBu. These tetrapeptides, APGA,
A(4R)MePGA, and A(4S)MePGA, incorporate proline, (4R)-methylproline, and (4S)-methylproline,
respectively, at the Xaa position. To determine the influence of the 4-methyl substituted prolines on the
b-turn populations, the NAMFIS (NMR analysis of molecular flexibility in solution) deconvolution
analysis for these three peptides were performed in DMSO-d6 solution. The NBO (natural bond orbital)
method was employed to gain further insight into the results obtained from the NAMFIS analysis. The
emphasis in the NBO analysis was to characterize remote intramolecular interactions that could
influence the backbone-backbone interactions contributing to b-turn stability. NAMFIS results indicate
that the enantiospecific incorporation of the methyl substituent at the Cg (C4) position of the proline
residue can be used to selectively control the pyrrolidine ring puckering propensities and, consequently,
the preferred f,y angles associated with the proline residue in b-turn forming peptides. The NAMFIS
analyses show that the presence of (4S)-methylproline in A(4S)MePGA considerably increased the type
II b-turn population with respect to APGA and A(4R)MePGA. The NBO calculations suggest that this
observation can be rationalized based on an n → p* interaction between the N-terminus alanine
carbonyl oxygen and the proline carbonyl group. Several other interactions between remote orbitals in
these peptides provide a more detailed explanation for the observed population distributions.
the side chains of the central residues i + 1 and i + 2 in
such a manner that they can be recognized electronically and
sterically by the complementary receptor molecule. Therefore, b-
Introduction
Reverse turns are elements of secondary structure that have an
important role both in peptides and proteins.1 They are broadly
turns are widely implicated in hormone-receptor interactions as
defined as those regions of the polypeptide where the amino acid
well as recognition sites for immunological, metabolic, genomic
and endocrinological mechanisms.1 b-Turns are also involved in
chain turns back onto itself. The most common type of reverse
turn, the b-turn, represents about 25% of all residues in proteins.2
the biological activity of peptides, often postulated as the bioactive
b-Turns play a critical role in the topology of polypeptides by
orienting structural elements such as a-helices and b-strands.
It has also been proposed that they might be important in
the initiation of the protein folding process.3 Furthermore, b-
turns play many biological roles in peptides and proteins.1 They
are predominantly constituted of hydrophilic residues and thus
frequently occur on the surfaces of globular proteins.4 This and
the higher than random proportion of structural or functional
side chains such as proline, aspartic acid, serine, threonine, and
lysine makes them viable candidates for structural motifs for
molecular recognition.1 The b-turn structure is able to orient
conformation that interacts with another molecule. The above-
mentioned properties have led to great interest in mimicking b-
turns for the synthesis of medicines in the field of medicinal and
pharmacological chemistry.5
An understanding of the conformational behavior of the b-turn
forming peptides in solution has the potential to provide valuable
insights into the conformational requirements for biological
activity. In the present work, we have analyzed the conformational
profiles of short linear b-turn forming peptides in solution by using
nuclear magnetic resonance (NMR) spectroscopy. Generally, in
this method, interproton distances are derived from NOESY
(nuclear Overhauser spectroscopy) or ROESY (rotating frame
Overhauser spectroscopy) experiments and torsional angles are
calculated from J-coupling constants using empirical Karplus-
type equations.6 These parameters are used as geometrical re-
straints to define the conformation(s) of the molecule by the means
of various protocols including distance geometry and simulated
annealing. These methods assume the existence of a single, unique
conformation. However, most short peptides, even cyclic ones,
are present in solution as conformational ensembles containing
a large number of widely different conformers that are in rapid
interconversion on the NMR time scale and are characterized by
Aalto University School of Science and Technology, Faculty of Chemistry
and Materials Sciences, Department of Chemistry, P.O.Box 16100, FI-
00076 Aalto, Finland. E-mail: ari.koskinen@tkk.fi; Fax: +358 9 470 22538;
Tel: +358 9 470 22526
† Electronic supplementary information (ESI) available: Aggregation
studies. 1H and 13C NMR chemical shift assignments, 1H,1H coupling
constants, and off-resonance ROESY derived interproton distances for
APGA, A(4R)MePGA, and A(4S)MePGA. Classification of b-turns.
Detection of b-turns by 1H NMR. Synthetic procedures. 1H, 13C, and off-
resonance ROESY spectra. DFT energies and Cartesian coordinates for
(4R)MePG-I, (4R)MePG-II, (4S)MePG-I, and (4S)MePG-II. Additional
NAMFIS analyses. See DOI: 10.1039/b921794k
This journal is
The Royal Society of Chemistry 2010
Org. Biomol. Chem., 2010, 8, 2103–2116 | 2103
©

View Article Online
an averaged NMR spectrum.7–9 Consequently, NOE/ROE and J-
coupling constants correspond to population weighted averaged
values which, when used as restraints in the above-mentioned
calculations, will most probably lead to a virtual conformation that
combines multiconformer features in a single geometry. Therefore,
for flexible molecules, different types of procedures are needed to
identify and quantitate the ensemble of different conformations in
solution.
The puckering propensity can be modified by introduction of a
substituent at the Cg position of the pyrrolidine ring. It has been
shown that the (4R) epimers of hydroxy-, fluoro- and chloroproline
favor the exo puckering, while the (4S) epimers favor the endo
puckering.23–25 The relative stability of the exo puckering increases
with the electronegativity of the (4R) substituent, whereas the
endo puckering is favored by an electronegative (4S) substituent.
In the preferred puckering, these substituents adopt a pseudoaxial
orientation. However, in the case of (4R)- and (4S)-mercapto-
L-proline and related derivatives, the puckering propensities are
reversed: the (4R) epimers induce the endo puckering, while
the (4R) epimers generate the exo puckering, orienting the
substituents pseudoequatorially.26 In addition to conformational
control over the proline side chain, substitutions on Cg are
known to influence the equilibrium conformational populations
of the cis/trans prolyl bonds.27-29 Extensive computational and
NMR studies have suggested that the conformational effects
of the electronegative substituent is dictated by inductive and
stereoelectronic factors.23–30
In this research, we have employed the NAMFIS (NMR analysis
of molecular flexibility in solution)10–12 method to analyze the
conformational profiles of linear tetrapeptides in solution and
to determine the estimated populations of the b-turn containing
conformers among the structures in the conformational pool.
NAMFIS is an approach that deconvolutes the averaged NMR
data into a small family of conformations that optimally represents
the torsions derived from J-coupling constants and the distances
derived from NOEs/ROEs. In the NAMFIS analysis, explicit
force field energies are ignored during the deconvolution. There-
fore, artifacts emerging from energy misevaluation introduced
by long-range interactions among polar functional groups are
bypassed. NAMFIS has been used, for example, to analyze the
solution structures of short peptides,11,13–15 a tricyclic ketone,10
Taxol,16 discodermolide,17 several epothilones,18,19 laulimalide,12
and geldanamycin and radiacol.20 The research described in
this paper involves three tetrapeptides, with the general amino
acid sequence CBz-L-Ala-L-Xaa-Gly-L-Ala-OtBu, in which Xaa
is proline, or the nonproteinogenic amino acid (4R)- or (4S)-
methylproline. These peptides are illustrated in Scheme 1, along
with the shorthand designations used in the text.
Results and discussion
In this study, we have used the non-electronegative methyl
substituent at the Cg position to control the puckering propensity
of the pyrrolidine ring and, consequently, the preferred f,y angles
associated with the proline residue in b-turn forming peptides.
DFT calculations have previously indicated that the pyrrolidine
ring of (4R)-methyl-L-proline has a strong preference for the
Cg endo puckering, while that of (4S)-methyl-L-proline has a
strong preference for the Cg exo puckering.31 These ring puckering
preferences have also been observed in crystal structures of Ac-
(4R)-MePro-NHMe and Ac-(4S)-MePro-NHMe.32 In addition,
the predominant ring puckering observed in (4R)- and (4S)-tert-
butylprolines follow the same trend.33 In the preferred conforma-
tions, the methyl or tert-butyl group adopts a pseudoequatorial
orientation. The ring puckering preferences of these substituents
are strongly determined by steric effects.31
The puckering preference of the pyrrolidine ring could elicit
changes in the polypeptide chain f, y and w angles, leading
to a correlation of proline ring puckering with polypeptide
chain conformation: the Cg exo puckering is more common in
compact structures, while the Cg endo puckering is favored in
extended structures.22–24,28,34–37 These observations are related to
the fact that the Cg endo conformations are accompanied by
more negative f angles and larger y angles, whereas the opposite
is true for the Cg exo conformations.23,24 Therefore, selective
control of the pyrrolidine ring puckering provides the possibility
to influence the polypeptide chain conformation. 4-Substituted
proline derivatives have been incorporated into collagen25,26,31,38–42
and elastin-mimetic30,43 peptides, and into globular proteins,37,44,45
to selectively control the backbone conformations and stability of
these structures by defining the ring puckering of critical proline
residues using stereoelectronic effects.
Scheme 1 The structures of the peptides under study, along with the
shorthand designations used in the text.
A substituent at the Cg (C4) position of the pyrrolidine ring of
proline can provide a large degree of conformational control over
the side chain. The pyrrolidine ring has two predominant pucker-
ing modes: Cg exo (up) and endo (down) envelope conformers.21
In the case of the unsubstituted trans proline residue in peptides
and proteins, these two modes are almost equally probable.22
The simple model peptides used in this study are equally
accessible to both experimental and theoretical investigation. They
can be used to analyze and quantify the different interactions
contributing to b-turn stability and, therefore, b-turn formation in
peptides and proteins. The peptides examined consist of alanine,
glycine, and proline; thus side-chain interactions are minimal,
2104 | Org. Biomol. Chem., 2010, 8, 2103–2116
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
and the primary interactions contributing to b-turn stability
should result from the peptide backbone. In addition to the
above, these peptides provide a means to study the relationship
between the proline ring puckering and the peptide backbone
conformation. Our working hypothesis was that incorporation of
the 4-methylproline epimers with altered puckering propensities
into these linear tetrapeptides influences the bias in the populations
of the b-turn conformations in solution, thus allowing a means
to control the solution conformation of even a small peptide
in solution. In this investigation, we employed the NAMFIS10–12
method to analyze the conformational profiles of proline and 4-
methylproline containing tetrapeptides in DMSO-d6. By using
the NAMFIS, we determined the ring puckering preferences
of the proline and 4-methylproline residues incorporated into
these peptides, and analyzed the influence of the 4-methylproline
epimers on the estimated populations of the b-turn structures in
solution. In addition, we used the natural bond orbital (NBO)46
method to gain further insight into the results obtained from
the NAMFIS analysis. The emphasis in the NBO analysis was
to characterize remote intramolecular interactions that could
influence the peptide backbone conformation, thus providing
information on the underlying principles responsible for the
conformational preferences of b-turns.
region, which gave rise to negative exchange cross-peaks (same
sign as the diagonal) in the T-ROESY spectra and corresponded to
the slow cis/trans isomerization of the Ala1-Pro/4MePro peptide
bond. However, the integration of these signals in the 1D 1H
spectrum indicated that the population of the minor isomer was
less than 20% of the major one. Therefore, only the major isomers
were analyzed further. Based on the T-ROESY spectra, the proline
peptide bond conformation of the major species corresponded
to the trans isomer, as detected through the strong correlation
pro(R)/(S)
between Ala1-Ha and Pro-Hd
resonances for all three
1
peptides. H NMR chemical shifts and coupling constants for
APGA, A(4R)MePGA, and A(4S)MePGA are listed in Tables
S1 and S2 (Supporting Information). Please see Supporting
Information for the classification of b-turns and for the detailed
description of the detection of b-turns by 1H NMR.
Solution structures by NAMFIS analysis
In order to take conformational averaging into account, the
NAMFIS (NMR analysis of molecular flexibility in solution)10–12
procedure was utilized to evaluate the population distribution
of the conformational ensemble based on the distances de-
rived from NOEs/ROEs and torsion angles derived from the
J-coupling constants. The NAMFIS method deconvolutes an
averaged NMR spectrum into a small family of conformers,
identifies individual conformational minima, and assigns weights
to the conformers that optimally represent the experimentally
obtained NOEs/ROEs and J-couplings. Consequently, the burden
of accommodating the complete set of averaged NMR data is
shared among about 5 to 20 conformers. Interproton distances
for the NAMFIS analysis were obtained from the adiabatic
version of off-resonance ROESY experiments.47 The backbone
3J(NH,Ha) and the pyrrolidine ring 3J(H,H) coupling constants
1H NMR spectroscopy
Complete 1H chemical shift assignments for APGA,
A(4R)MePGA, and A(4S)MePGA were made using
a
combination of standard TOCSY and T-ROESY experiments.
Interproton distances were obtained from analysis of off-
resonance ROESY experiments.47 The 1H NMR parameters
were analyzed using the PERCH NMR software.48,49 Because the
peptides were protected as a benzyl carbamate and a tert-butyl
ester, they were insoluble in water. Therefore, the NMR solvent
in this study was chosen to be DMSO-d6, because it has similar
polarity and dielectric constant as water. Aggregation studies
over a concentration range of 1.6 to 100 mM revealed that the
chemical shifts were essentially insensitive to concentration (see
Supporting Information for details). This observation indicates
that the intermolecular associations were negligible over this
concentration range. Therefore, the samples were prepared at
relatively high concentrations (ca. 80 mM). Accordingly, the
signal-to-noise level of the spectra was high, enabling accurate
determination of NMR parameters needed for conformational
analysis. One might question the relevance of conformations
identified in DMSO in the biological context. However, based
on the observation that the liquid phases in the cytoplasm,
intersynaptic space, and receptor-binding sites are characterized
by a higher viscosity than water, it has been suggested that
DMSO, or another high viscosity solvent, might be more realistic
than water in modeling the above-mentioned environments.50 In
addition, it has been reported that DMSO does not necessarily
alter either biological activity or conformational preferences of
peptides and, therefore, is a pragmatic choice.51–54 Furthermore,
a preliminary study by O’Connell et al.55 on unprotected APGA
peptide in aqueous solution showed that this peptide behaves in
the same way both in aqueous solution and in DMSO.
1
were extracted from the 1D H NMR spectra using the PERCH
NMR software.48,49 The latter coupling constants were included
into the NAMFIS deconvolution in order to analyze the puckering
propensity of the pyrrolidine ring in solution. The 3J(H,H)
coupling constants and the ROE-derived distances used in the
NAMFIS analysis are provided in Supporting Information in
Tables S2 and S3, respectively. Extensive Monte Carlo confor-
mational searches for all three peptides were carried out using
OPLS_2005, AMBER*, and MMFFs force fields in MacroModel
9.556 combined with the GB/SA H2O continuum solvent model.57
The fully optimized structures were relieved from duplicate confor-
mations and loaded into a common structural database, yielding
2259, 1996, and 2485 conformers for APGA, A(4R)MePGA, and
A(4S)MePGA, respectively. These conformers were then subjected
to the NAMFIS deconvolution procedure using interproton dis-
tances and torsion angles obtained from the averaged NMR data.
This analysis, which couples calculated structures with structures
determined by NMR, resulted in an ensemble of conformations
with varying populations for the peptides under study. NAMFIS
populations and geometric parameters for APGA, A(4R)MePGA,
and A(4S)MePGA are provided in Tables 1–3, respectively.
The obtained NAMFIS conformations were grouped into 7
structural motifs (Table 4): ideal and distorted type I b-turn,
ideal and distorted type II b-turn, type I bend, type II bend, and
extended structures. In general, a structure was classified as a b-
1
The 1D H NMR spectrum of APGA, A(4R)MePGA, and
a
a
˚
A(4S)MePGA displayed two sets of signals in the amide proton
turn if the distance between C (i)–C (i + 3) was <7 A. More
This journal is
The Royal Society of Chemistry 2010
Org. Biomol. Chem., 2010, 8, 2103–2116 | 2105
©

View Article Online
Table 1 NAMFIS populations and geometric parameters for APGA
conformer pop (%) f(i + 1) (deg) y(i + 1) (deg) f(i + 2) (deg) y(i + 2) (deg) d(Ala -C –Ala -C ) (A) Pro conf. H-bond Ala1-CO–Ala2-NH
1
a
2
a
˚
1
2
3
4
5
6
7
8
9
10
11
12
13
28.7
12.3
11.5
10.7
9.3
7.4
5.5
4.8
3.9
2.2
1.5
-73.5
-68.0
-74.2
-73.7
-47.2
-65.5
-77.7
-67.3
-69.1
-72.9
-63.7
-76.6
-66.4
129.6
-26.5
108.5
130.9
109.0
121.4
-176.0
123.4
-25.6
110.3
-16.2
161.0
120.7
171.8
-94.1
87.9
88.8
143.9
93.8
-103.0
88.0
-88.9
86.8
-141.6
5.1
15.6
3.4
-43.8
-0.6
5.2
10.9
-86.4
22.1
24.0
21.9
8.3
9.0
5.2
4.9
5.8
5.8
5.4
8.6
5.3
5.4
4.9
4.8
7.0
5.2
endo
exo
endo
endo
exo
exo
endo
exo
exo
endo
exo
endo
exo
-
+
+
+
+
+
-
+
+
+
+
-
-122.9
84.6
88.7
1.2
0.6
+
Table 2 NAMFIS populations and geometric parameters for A(4R)MePGA
conformer pop (%) f(i + 1) (deg) y(i + 1) (deg) f(i + 2) (deg) y(i + 2) (deg) d(Ala -C –Ala -C ) (A) Pro conf. H-bond Ala1-CO–Ala2-NH
1
a
2
a
˚
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
16.4
15.3
11.3
9.7
8.6
7.0
6.3
5.5
4.7
3.9
3.4
3.3
1.5
1.5
1.0
0.7
-72.0
-73.9
-76.6
-76.7
-59.7
-50.8
-68.2
-81.0
-90.4
-72.1
-70.2
-49.3
-70.6
-79.2
-57.8
-73.2
122.7
131.8
-28.1
167.2
125.0
109.8
153.7
99.1
95.7
88.2
4.9
3.8
5.2
5.8
6.6
9.8
8.4
5.8
9.4
8.1
7.7
5.1
5.4
5.6
5.5
8.5
8.2
5.4
endo
endo
endo
endo
exo
+
+
-
-123.5
-88.1
165.8
144.9
169.2
173.2
179.0
-116.0
96.2
144.3
100.3
-102.9
162.1
-87.0
-144.4
157.4
-140.1
-43.5
-127.0
-147.7
163.5
-60.7
1.1
-
-
exo
+
-
endo
endo
endo
endo
endo
exo
endo
endo
endo
endo
-
-21.4
-2.4
-
+
+
+
+
-
127.4
106.4
122.6
-173.2
133.5
-15.9
-48.7
-1.5
5.6
-120.2
-89.6
-
+
Table 3 NAMFIS populations and geometric parameters for A(4S)MePGA
conformer pop (%) f(i + 1) (deg) y(i + 1) (deg) f(i + 2) (deg) y(i + 2) (deg) d(Ala -C –Ala -C ) (A) Pro conf. H-bond Ala1-CO–Ala2-NH
1
a
2
a
˚
1
2
3
4
5
6
7
8
9
10
11
12
13
21.0
15.7
14.2
13.2
5.9
5.1
4.8
4.7
4.2
3.5
3.4
-48.9
-65.8
-67.1
-57.9
-69.8
-71.4
-66.4
-66.3
-74.2
-70.6
-44.3
-63.3
-75.3
109.2
121.9
123.6
130.9
167.8
-32.5
-14.9
123.0
-176.9
-34.6
116.8
132.6
173.5
143.8
89.3
87.9
-42.7
7.4
10.9
-143.6
-124.8
-162.2
41.6
5.1
52.7
29.5
5.8
5.2
5.3
9.0
10.2
7.2
4.7
5.4
8.2
5.8
5.6
5.7
9.1
exo
exo
exo
exo
exo
exo
exo
exo
endo
exo
exo
exo
endo
+
+
+
-
171.9
-91.1
-93.4
-129.9
89.4
-97.2
-171.2
117.4
103.9
-112.5
-
-
+
+
-
-
-18.7
-23.6
12.2
+
+
-
3.3
1.1
Table 4 Total populations of the NAMFIS motifs for APGA, A(4R)MePGA, and A(4S)MePGAa
APGA
pop, %
A(4R)MePGA
A(4S)MePGA
motif
members
pop, %
members
pop, %
members
ideal bII
distorted bII
ideal bI
37.3
9.3
13.9
3.9
3, 4, 6, 8, 10, 13
5
2, 11
9
1
36.6
10.3
1, 2, 11, 13
6, 12
41.3
21.0
2, 3, 8, 11, 12
1
distorted bI
15.9
15.1
4.7
3, 10, 16
5, 8, 15
9
8.3
13.2
5.1
7, 10
4
6
bend II
bend I
28.8
extended
6.7
7, 12
17.5
4, 7, 14
11.2
5, 9, 13
a Populations are the normalized sum of the individual conformer populations.
2106 | Org. Biomol. Chem., 2010, 8, 2103–2116
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
precisely, a structure was considered as an ideal type I or type
II b-turn if three of the f,y angles of the central residues i + 1
and i + 2 deviated less than 30◦ from the idealized values for
type I or II b-turn, and the fourth angle deviated less than 45◦.
Idealized f,y values, as determined by Lewis et al.,58 for type I
and type II b-turns are as follows: f(i + 1) = -60◦, y(i + 1) =
-30◦, f(i + 2) = -90◦, y(i + 2) = 0◦ for type I b-turn, and f(i +
1) = -60◦, y(i + 1) = 120◦, f(i + 2) = 80◦, y(i + 2) = 0◦ for type
II b-turn. In a distorted b-turn structure, f(i + 2) and/or y(i + 2)
angles were allowed to deviate more than 45◦ from the idealized
values. In addition, a structure was considered as distorted if both
of these angles deviated more than 30◦, but less than 45◦ from
the idealized values. A structure was classified as type I bend, type
II bend or extended, if the distance between Ca(i)–Ca(i + 3) was
Fig. 2 The structures of the most populated members of each motif for
A(4R)MePGA. (A) Ideal type II b-turn, (B) distorted type II b-turn, (C)
distorted type I b-turn, (D) type II bend, (E) type I bend, and (F) extended
motif. Protecting groups have been omitted for clarity.
˚
≥7 A. In a type I or type II bend, the f(i + 1) and y(i + 1) angles
were allowed to deviate less than 30◦ from the idealized values
for type I or II b-turn. In these structures, the f(i + 2) and y(i +
2) angles were allowed to take values greater than 45◦ from the
idealized b-turn values. In extended structures, all f and y angles
were allowed to adopt values greater than 45◦ from the idealized
values. Hydrogen bonds in the NAMFIS structures were defined
by relations between four atoms: the donor hydrogen atom (H),
the acceptor atom (A), the donor atom (D) bonded to H, and
another neighbor atom (B) bonded to A. The following values
were used determine a valid hyd◦rogen bond: H ◊ ◊ ◊ A distance <
◦
˚
2.50 A, D–H ◊ ◊ ◊ A angle > 120 , and H ◊ ◊ ◊ A–B angle > 90 .
The structures of the most populated members of each motif for
APGA, A(4R)MePGA, and A(4S)MePGA are given in Fig. 1–3,
respectively.
Fig. 3 The structures of the most populated members of each motif for
A(4S)MePGA. (A) Ideal type II b-turn, (B) distorted type II b-turn, (C)
distorted type I b-turn, (D) type II bend, (E) type I bend, and (F) extended
motif. Protecting groups have been omitted for clarity.
methylproline epimers with altered puckering propensities into
linear tetrapeptides introduces a bias in the populations of the
b-turn conformations in solution.
NAMFIS analysis of APGA
The NAMFIS analysis of APGA in DMSO-d6, using 19 ROE
distances, 4 3J(NH,Ha) backbone couplings, and 10 3J(H,H)
pyrrolidine ring couplings, resulted in 13 conformers above a
molar fraction of 0.01 with estimated populations ranging from 1
to 29% (Table 1). The most populated conformer (29%) represents
the type II bend motif, while the second (12%) and third most
populated (12%) conformers represent the ideal type I b-turn and
the ideal type II b-turn motifs, respectively. The most plentiful
motif among the NAMFIS conformers is the ideal type II b-
turn (37%), although its top contributor is only the third most
populated structure. This motif contains 6 out of 13 NAMFIS
conformers. The second most populated motif is the type II
bend (29%) containing only the most populated conformer. The
remaining 6 conformers are divided between 4 motifs: ideal type
I b-turn (14%), distorted type II b-turn (9%), extended (7%), and
distorted type I b-turn (4%). None of the NAMFIS conformers
belong to the type I bend motif. The estimated combined
population of the ideal type II b-turn and the distorted type II
b-turn motifs is 46%, and that of the ideal type I b-turn and the
Fig. 1 The structures of the most populated members of each motif for
APGA. (A) Ideal type II b-turn, (B) distorted type II b-turn, (C) ideal type
I b-turn, (D) distorted type I b-turn, (E) type II bend, and (F) extended
motif. Protecting groups have been omitted for clarity.
In addition to the backbone conformation, the proline ring
puckering (Cg endo/exo) of the NAMFIS structures was deter-
mined, and the influence of the proline ring conformation on
the estimated populations of the b-turn structures among the
other structures in solution was analyzed. As discussed above,
the pyrrolidine ring of the (4R)- and (4S)-methylproline residues
have alternative ring puckering preferences: the (4R) epimer of
methylproline favors endo puckering, while the (4S) epimer favors
exo puckering. The puckering preference of the pyrrolidine ring,
in turn, has an effect on the polypeptide chain f,y angles,
leading to a correlation of proline ring puckering with polypeptide
chain conformation. As discussed below, incorporation of 4-
This journal is
The Royal Society of Chemistry 2010
Org. Biomol. Chem., 2010, 8, 2103–2116 | 2107
©

View Article Online
distorted type I b-turn motifs is 18%. Thus, the combined b-turn
population of APGA in DMSO-d6 solution is 64%. As expected,
the type II b-turn conformations are much more populated than
the type I b-turn conformations. The bend structures represent
29% and the extended structures 7% of the total population of the
NAMFIS conformers. Hydrogen bonding between Ala1-CO and
Ala2-NH is present in all NAMFIS structures of APGA belonging
to the ideal or distorted type I/II b-turn motif. The conformation
of the pyrrolidine ring of the unsubstituted proline residue in
APGA is expected to be a time average of a series of different
puckered conformers. Among these conformers, the energetically
equivalent Cg endo and Cg exo puckers are preferred. These
two puckerings are essentially equivalently populated among the
NAMFIS conformers of APGA: in 7 out of 13 structures, the
pyrrolidine ring is in the Cg exo puckering, while in the remaining
6 structures it is in the Cg endo puckering. The Cg exo puckering is
more common in b-turn structures, evidenced by the observation
that the pyrrolidine ring is in the exo conformation in 7 out of
10 b-turn conformers. On the contrary, in the bend and extended
structures, the Cg endo puckering prevails. This is consistent with
the observation that the Cg exo puckering is more common in
compact structures, while the Cg endo puckering is favored in
extended structures.
NAMFIS analysis of A(4S)MePGA
The NAMFIS deconvolution of A(4S)MePGA in DMSO-d6,
3
using 25 ROE distances, 4 J(NH,Ha) backbone couplings, and
3
6 J(H,H) pyrrolidine ring couplings, yielded 13 conformations
above a molar fraction of 0.01 with estimated populations ranging
from 1 to 21% (Table 3). The most populated conformer (21%) rep-
resents the distorted type II b-turn motif, while the second (16%)
and third most populated (14%) conformers represent the ideal
type II b-turn motif. Similarly with APGA and A(4R)MePGA, the
most populated motif is the ideal type II b-turn (41%), containing
5 out of 13 NAMFIS conformers. The second most populated
motif is the distorted type II b-turn (21%), containing only one
conformer. The remaining 7 conformers are divided between 4
motifs: type II bend (13%), extended (11%), distorted type I b-turn
(8%), and type I bend (5%). None of the NAMFIS conformers
represent the ideal type I b-turn motif. The estimated combined
population of the ideal type II b-turn and the distorted type II
b-turn motifs is 62%, and that of the ideal type I b-turn and the
distorted type I b-turn motifs is 8%. Therefore, the combined
b-turn population of A(4S)MePGA in DMSO-d6 solution is
70%. As expected, the type II b-turn conformations are very
strongly favored over the type I b-turn conformations. The bend
structures represent 18% and the extended structures 11% of the
total population of the NAMFIS conformers. The pyrrolidine
ring of the proline residue in A(4S)MePGA strongly favors
Cg exo puckering: in 11 out of 13 NAMFIS-derived structures
the pyrrolidine ring exhibits the Cg exo conformation. In the
preferred puckering mode, the (4S) methyl substituent adopts a
pseudoequatorial orientation. Hydrogen bonding between Ala1-
CO and Ala2-NH is present in all except one NAMFIS conformer
of A(4S)MePGA belonging to the ideal or distorted type I/II
b-turn motif.
NAMFIS analysis of A(4R)MePGA
A deconvolution of the average structure of A(4R)MePGA in
DMSO-d6 with the NAMFIS procedure, using 34 ROE distances,
4 3J(NH,Ha) backbone couplings, and 6 3J(H,H) pyrrolidine ring
couplings, delivered 16 conformations above a molar fraction of
0.01 that accommodate the NMR data with estimated populations
ranging from 1 to 16% (Table 2). The most (16%) and the second
most populated (15%) conformers represent the ideal type II
b-turn motif, while the third most populated conformer (11%)
represents the distorted type I motif. The most populated motif
is the ideal type II b-turn (37%), although it contains only 4 out
of 16 NAMFIS conformers. However, as mentioned above, the
two most populated conformers belong to this motif. The second
most populated motif is the extended motif (18%), containing 3
conformers. The rest of the 9 conformers are divided between 4
motifs: the distorted type I b-turn (16%), type II bend (15%),
distorted type II b-turn (10%), and type I bend (5%). The ideal
type I motif contains no NAMFIS conformers. The combined
population of the ideal type II b-turn and the distorted type
II b-turn motifs is 47%, while the combined population of the
ideal type I b-turn and the distorted type I b-turn motifs is 16%.
Consequently, the combined b-turn population of A(4R)MePGA
in DMSO-d6 solution is approximately 63%. As in the case of
APGA, the type II b-turn conformations are strongly favored over
the type I b-turn conformations. The bend structures represent
20% and the extended structures 18% of the total population of
the NAMFIS conformers. The conformation of the pyrrolidine
ring in A(4R)MePGA strongly favors Cg endo puckering: in 13 out
of 16 NAMFIS structures, the pyrrolidine ring is in the Cg endo
puckering mode. Consequently, the (4R) methyl substituent adopts
a pseudoequatorial orientation in these structures. All except one
NAMFIS structure of A(4R)MePGA belonging to the ideal or
distorted type I/II b-turn motif exhibit a hydrogen bond between
Ala1-CO and Ala2-NH.
Effect of n → p* interaction on proline w angles
The most populated structural motifs for all peptides under study
are those in which the y(i + 1) dihedral angle is close to values
compatible with the type II b-turn conformation, i.e. in which the
y(i + 1) deviates less than 30◦ from the idealized value for type II
b-turn (120◦). We propose that this observation can be rationalized
based on an energetic stabilization associated with a non-bonded n
→ p* interaction between the lone pair of the Ala1 carbonyl oxygen
and the antibonding orbital of the proline carbonyl group. The Cg
exo ring puckering preorganizes the main chain f,y angles of the
=
proline residue for this favorable O(i) ◊ ◊ ◊ C¢ O(i + 1) interaction,
which is strongest when O(i) is positioned proximally and along
59-62
=
the Bu¨rgi-Dunitz trajectory to C¢ O(i + 1).
This type of n
→ p* interaction has been proposed to have an important role
in stabilizing trans peptide bond.23 In addition, the n → p*
interaction has been suggested to stabilize polypeptide structures
such as a-helix and polyproline type II (PPII) helix.63
A significant n → p* interaction can be defined as one in
˚
which the O(i) ◊ ◊ ◊ C¢(i + 1) distance is dBD < 3.00 A and the
O(i) ◊ ◊ ◊ C¢ O(i + 1) angle is tBD = 107 10◦. These parameters are
=
listed for APGA, A(4R)MePGA, and A(4S)MePGA in Table 5.
The NAMFIS structures of A(4S)MePGA belonging to the ideal
and distorted type II b-turn motif and those belonging to the type
˚
II bend motif have values of dBD ≤ 3.00 A and tBD within the range
2108 | Org. Biomol. Chem., 2010, 8, 2103–2116
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
=
Table 5 The O(i) ◊ ◊ ◊ C¢(i + 1) distance (dBD) and the O(i) ◊ ◊ ◊ C¢ O(i + 1) angle (tBD) for APGA, A(4R)MePGA, and A(4S)MePGA
APGA
A(4R)MePGA
A(4S)MePGA
˚
˚
˚
conformer
dBD (A)
tBD (deg)
conformer
dBD (A)
tBD (deg)
conformer
dBD (A)
tBD (deg)
1
2
3
4
5
6
7
8
9
10
11
12
13
2.99
3.05
3.11
3.12
2.84
2.97
3.32
3.01
3.11
3.10
2.94
3.23
3.00
108.8
120.4
126.1
112.0
113.6
113.8
79.2
113.6
120.4
124.4
125.6
94.2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
3.08
3.13
3.30
3.23
2.89
2.86
3.06
3.10
3.29
3.23
3.03
2.82
3.03
3.34
2.93
3.18
116.5
111.6
113.0
90.5
107.9
114.8
92.9
137.9
117.7
133.7
111.0
114.9
115.4
78.8
1
2
3
4
5
6
7
8
9
10
11
12
13
2.85
2.99
3.00
2.83
3.12
3.14
2.99
2.99
3.27
3.09
2.92
2.93
3.27
114.2
113.3
113.4
100.5
86.5
116.2
125.0
113.4
78.0
114.9
104.0
104.2
83.7
114.4
103.0
125.4
of 100–113◦, thus producing a favorable n → p* interaction. In
all these structures, the proline ring is in the Cg exo puckering.
The structures belonging to the distorted type I b-turn motif,
type I bend motif, or extended motif exhibit O(i) ◊ ◊ ◊ C¢(i + 1)
wave functions are interpreted in terms of a set of occupied Lewis-
type natural bond orbitals and a set of non Lewis-type localized
orbitals (either two-center antibonding or one-center Rydberg).
Delocalization effects can be identified from the presence of off
diagonal elements of the Fock matrix in the NBO basis. Ap-
proximate stabilization energies, Ei,j*(2), from these delocalization
interactions can be determined for specific bond and antibond
or Rydberg orbital pairs using second-order perturbation theory.
Another NBO method, natural steric analysis (NSA),64,65 was also
˚
distances longer than 3.00 A and/or significant deviation of the
◦
=
O(i) ◊ ◊ ◊ C¢ O(i + 1) angle from the ideal value of 107 . Therefore,
no significant energetic stabilization from the n → p* interaction
could be reasonably expected for the latter structures.
In A(4R)MePGA, in which the pyrrolidine ring strongly prefers
the Cg endo puckering, only 4 NAMFIS structures have dBD
≤
applied to study certain repulsive interactions in these peptides.
(st)
˚
3.00 A. All three structures in which the pyrrolidine ring has
Cg exo puckering belong to this group. The structures in which
the y(i + 1) dihedral angle is close to values compatible with
the type II b-turn conformation exhibits the shortest O(i) ◊ ◊ ◊ C¢(i
In this method, based on NBOs, steric interaction energies, Ei,j ,
can be calculated by identifying interatomic exchange repulsion
energy components associated with wave function antisymmetry.
The tetrapeptide unit was truncated to a capped moiety
corresponding to the fragment MeCO-Xaa-Gly-NHMe (Xaa =
(4R)- or (4S)-MePro), and the resulting dipeptides were confor-
mationally manipulated to produce the type I and type II b-
turn structures. The puckering of the pyrrolidine ring was set
to match Cg endo in the (4R)-MePro peptides and Cg exo in
the (4S)-MePro peptides. Consequently, the methyl substituent
is oriented pseudoequatorially in the corresponding four peptides.
These structures were minimized without any constraints using the
OPLS_2005 force field combined with the GB/SA H2O continuum
solvent model.57 Each b-turn structure was further optimized,
first at B3LYP/6-31+G* level of theory in gas phase, and second
at B3LYP/6-311++G** level of theory using the PBF solvation
model with DMSO as solvent.66,67 Finally, NBO calculations were
carried out for all four structures at the B3LYP/6-311++G** level
in DMSO using the PBF solvation model and geometries obtained
at the same level of theory. For ease of discussion, the (4R)-MePro
and (4S)-MePro peptides have been designated (4R)MePG-I,
(4R)MePG-II, (4S)MePG-I and (4S)MePG-II, in which the I and
II implies the type I or type II b-turn conformation, respectively.
All four MeCO-Xaa-Gly-NHMe units exhibit f,y angles for
proline and glycine residues that are within 30◦ of the idealized
f,y values corresponding to the (i +1) and (i + 2) residues of
the respective turn type. In addition, on the grounds of geometric
parameters listed in the section “Solution structures by NAMFIS
analysis”, all four peptides exhibit hydrogen bonding between O(i)
˚
+ 1) distances (dBD ≤ 3.13 A). Excluding one structure, these
=
conformations also exhibit the O(i) ◊ ◊ ◊ C¢ O(i + 1) angle within
the range of 103–116◦. The structures belonging to the distorted
type I b-turn motif, type I bend motif, or extended motif exhibit
˚
longer O(i) ◊ ◊ ◊ C¢(i + 1) distances (dBD ≥ 3.18 A), thus lacking
the ability to form a meaningful n → p* interaction. In only one
extended structure is the O(i) ◊ ◊ ◊ C¢(i + 1) distance is shorter (dBD
=
˚
3.06 A). In all these latter structures, the pyrrolidine ring is in
the Cg endo puckering. The trend observed for A(4S)MePGA and
A(4R)MePGA is also evident for APGA. The NAMFIS structures
in which the pyrrolidine ring is in the Cg exo puckering and in which
the y(i + 1) dihedral is close to values compatible with the type
II b-turn conformation typically exhibit dBD and tBD values that
can provide a significant energetic stabilization from the n → p*
interaction. On the contrary, with one exception, the structures
in which the pyrrolidine ring is in the Cg endo puckering exhibit
dBD and tBD values that cannot provide a meaningful n → p*
interaction.
NBO analysis of Pro-Gly turns
The selected stabilizing hyperconjugative interactions in
A(4R)MePGA and A(4S)MePGA were analyzed with density
functional theory (DFT) calculations using the natural bond
orbital (NBO) paradigm.46 In the NBO analysis, the electronic
This journal is
The Royal Society of Chemistry 2010
Org. Biomol. Chem., 2010, 8, 2103–2116 | 2109
©

View Article Online
Table 6 Relative energies and geometric parameters for (4R)MePG-I, (4R)MePG-II, (4S)MePG-I and (4S)MePG-IIa
DE (kcal mol-1)
f(i + 1) (deg)
y(i + 1) (deg)
f(i + 2) (deg)
y(i + 2) (deg)
dBD (A)
tBD (deg)
˚
peptide
(4R)MePG-I
(4R)MePG-II
(4S)MePG-I
(4S)MePG-II
0.9
0.5
1.1
0.0
-75.7
-59.0
-65.7
-50.8
-1.8
130.0
-15.1
128.0
-74.6
84.7
-71.2
85.6
-11.2
-5.3
-12.2
-6.3
3.147
2.868
3.009
2.805
132.4
104.8
125.8
101.8
a Calculated at the B3LYP/6-311++G** level of theory using the PBF solvation model with DMSO as solvent.
and the NHMe group. The relative energies and the geometric
parameters for (4R)MePG-I, (4R)MePG-II, (4S)MePG-I, and
(4S)MePG-II are listed in Table 6. The B3LYP/6-311++G**
optimized geometries of these dipeptides are shown in Fig. 4.
therefore, not optimal for the nucleophilic attack. The distance
BD between the nucleophile and the electrophilic carbonyl carbon
d
plays an important role in the stabilization that arises from the n →
p* interaction. This distance increases in the order of (4S)MePG-
II < (4R)MePG-II < (4S)MePG-I < (4R)MePG-I. (Table 6).
Thus, taking the dBD and tBD values into consideration, the n →
p* stabilization is expected to be strongest for (4S)MePG-II and
weakest for (4R)MePG-I. Second-order perturbation estimates
from NBO do indeed show increasing n → p* stabilization
energies, Ei,j*(2), with decreasing distance between O(i) and C¢(i +
1) (Tables 6 and 7). The type II b-turn peptides, (4S)MePG-II and
(4R)MePG-II, show the highest stabilization energies (1.94 and
1.39 kcal mol-1, respectively), while the type I b-turn peptides,
(4S)MePG-I and (4R)MePG-I, show clearly reduced n → p*
stabilization (0.59 and0.21kcalmol-1, respectively). Consequently,
=
the stabilizing n → p* interaction between O(i) and C¢ O(i + 1)
is more effective in the type II b-turn peptides than in the type
I b-turn peptides. Furthermore, the (4S)MePro peptides exhibit
stronger stabilization than the corresponding (4R)MePro peptides.
A representation of the n → p* interaction between O(i) and
=
C¢ O(i + 1) is given in Fig. 5.
Fig.
4
The B3LYP/6-311++G** optimized geometries of (A)
(4R)MePG-I, (B) (4R)MePG-II, (C) (4S)MePG-I, and (D) (4S)MePG-II.
The strongest donor character is shown by the lone pair of
nitrogen atoms and the lone pair of oxygen atoms. The n → s*
and n → p* interactions between the lone pair of nitrogen atoms
=
and the antibonding orbitals of vicinal C O bonds correspond
to the Ei,j* value as high as about 84 kcal mol-1. The n → s*
(2)
interactions between the lone pair of oxygen atoms and the vicinal
(2)
C–C and C–N bonds reach the Ei,j* value of approximately
24 kcal mol-1. The interactions between the above-mentioned
=
lone pair orbitals and their vicinal C–C, C–N, or C O filled
orbitals exhibit the largest steric energies, Ei,j(st). Most stabilizing
and destabilizing interactions in these peptides occur between
vicinal NBOs. In addition, several long-range interactions between
remote orbitals are also present. This analysis is focused on the
latter interactions owing to the influence they can play on the
peptide backbone conformation. Selected hyperconjugative and
steric repulsive interactions between remote orbitals are given in
Table 7.
Fig. 5 3D representation of the n → p* interaction between O(i) and
=
C¢ O(i + 1) in (4S)MePG-II.
In addition to the n → p* interaction between O(i) and
=
First, the stabilizing n → p* interaction between O(i) and
C¢ O(i + 1), other stabilizing interactions between remote filled
=
C¢ O(i + 1) is considered. In both (4R)MePG-II and (4S)MePG-
and unfilled orbitals are also present in MeCO-Xaa-Gly-NHMe
(Xaa = (4R)- or (4S)-MePro) peptides (Table 7). The strongest
remote interaction in all four peptides involves the O(i) lone
pairs as donor and the N–H antibond orbital, s*(N–H), of
the NHMe group as acceptor, with n(Lp1O) being a stronger
donor than n(Lp2O). The stabilization energies, Ei,j*(2), associated
II, the tBD angle from the amide oxygen O(i) to the proline
=
carbonyl group C¢ O(i + 1) is similar and very close to the
optimal Bu¨rgi-Dunitz trajectory (tBD = 107◦) for the attack of a
nucleophile on a carbonyl group. On the contrary, in (4R)MePG-
I and (4S)MePG-I, the tBD angle is much larger (>125◦) and,
2110 | Org. Biomol. Chem., 2010, 8, 2103–2116
This journal is The Royal Society of Chemistry 2010
©

View Article Online
Table 7 Selected hyperconjugative and steric repulsive interactions between remote orbitals for (4R)MePG-I, (4R)MePG-II, (4S)MePG-I, and
(4S)MePG-IIa
(2)
(st)
peptide
stabilizing interaction
Ei,j* (kcal mol-1)
repulsive interaction
Ei,j (kcal mol-1)
(4R)MePG-I
n(Lp2O2) → p*(C6O7)
n(Lp1O2) → s*(N24H23)
n(Lp2O2) → s*(N24H23)
n(Lp2O7) → s*(N4C5)
n(Lp2O20) → s*(N17C18)
n(Lp2O2) → s*(C5C8)
n(LpN17) → p*(C19O20)
p(C1O2) → s*(N24H23)
n(Lp2O2) → p*(C6O7)
n(Lp1O2) → s*(N24H23)
n(Lp2O2) → s*(N24H23)
n(Lp2O20) → s*(N17C18)
n(Lp2O2) → s*(C5C8)
n(LpN4) → p*(C6O7)
n(LpN17) → p*(C19O20)
p(C1O2) → s*(N24H23)
n(Lp2O2) → p*(C6O7)
n(Lp1O2) → s*(N19H18)
n(Lp2O2) → s*(N19H18)
n(Lp2O7) → s*(N4C5)
n(Lp2O15) → s*(N12C13)
n(Lp2O2) → s*(C5C8)
n(LpN12) → p*(C14O15)
p(C1O2) → s*(N19H18)
n(Lp2O2) → p*(C6O7)
n(Lp1O2) → s*(N19H18)
n(Lp2O2) → s*(N19H18)
n(Lp2O15) → s*(N12C13)
n(Lp2O2) → s*(C5C8)
n(LpN4) → p*(C6O7)
n(LpN12) → p*(C14O15)
p(C1O2) → s*(N19H18)
0.21
2.76
1.71
1.08
1.08
0.68
0.62
1.10
1.39
4.94
2.14
1.13
0.63
0.69
0.54
0.86
0.59
3.44
1.34
1.06
1.09
0.69
0.60
1.29
1.94
5.55
2.15
1.13
0.50
0.58
0.52
0.55
n(Lp1O2) ↔ s(N24H23)
n(Lp2O2) ↔ s(N24H23)
n(Lp2O7) ↔ s(N4C5)
n(Lp2O20) ↔ s(N17C18)
2.26
1.04
0.39
0.28
p(C1O2) ↔ s(N24H23)
n(Lp2O2) ↔ p(C6O7)
n(Lp1O2) ↔ s(N24H23)
n(Lp2O2) ↔ s(N24H23)
n(Lp2O20) ↔ s(N17C18)
0.84
0.87
4.19
1.17
0.31
(4R)MePG-II
(4S)MePG-I
(4S)MePG-II
n(LpN4) ↔ p(C6O7)
0.51
p(C1O2) ↔ s(N24H23)
n(Lp2O2) ↔ p(C6O7)
n(Lp1O2) ↔ s(N19H18)
n(Lp2O2) ↔ s(N19H18)
n(Lp2O7) ↔ s(N4C5)
n(Lp2O15) ↔ s(N12C13)
0.89
0.27
2.82
0.83
0.34
0.29
p(C1O2) ↔ s(N19H18)
n(Lp2O2) ↔ p(C6O7)
n(Lp1O2) ↔ s(N19H18)
n(Lp2O2) ↔ s(N19H18)
n(Lp2O15) ↔ s(N12C13)
0.99
1.14
4.75
1.14
0.32
n(LpN4) ↔ p(C6O7)
0.40
p(C1O2) ↔ s(N19H18)
0.63
a The numbering of the atoms is shown in Fig. 4. Blank cells correspond to a value lower than the threshold printing value 0.1 kcal mol-1.
with these hyperconjugative interactions are obtained as 7.70,
7.08, 4.78, and 4.74 kcal mol-1 for (4S)MePG-II, (4R)MePG-
II, (4S)MePG-I and (4R)MePG-I, respectively, which quantify
the extent of hydrogen bonding between O(i) and the NHMe
group in these peptides. Another remote interaction present in
all four peptides is the n → s* interaction between the lone
pair of O(i + 2) and the glycine N–Ca antibond orbital. This
interaction is strongest when the N–Ca-C¢-O dihedral angle of
the glycine residue is close to 180◦ or, in other terms, when the
y(i + 2) angle approaches 0◦ (Fig. 6). Consequently, this type
of hyperconjugative interaction contributes to the stability of the
peptide when the y(i + 2) angle approaches a value compatible
with the idealized type I and type II b-turn conformation. The
Ei,j* value associated with the n(Lp2O) → s*(N-Ca) interaction
(2)
is about 1.1 kcal mol-1 for all four peptides (Table 7). This type of n
→ s* interaction is also present in (4R)MePG-I and (4S)MePG-I
between the lone pair of O(i + 1) and the proline N–Ca antibond
orbital. In analogy to the above, this interaction is strongest
when the N–Ca–C¢–O dihedral angle of the proline residue is
close to 180◦. Consequently, the n → s* interaction gives the
strongest stabilization when the y(i + 1) angle is near 0◦, which
is about 30◦ away from the idealized value for the type I b-turn
(2)
conformation (Fig. 7). As above, the stabilization energy, Ei,j*
,
associated with the n(Lp2O) → s*(N-Ca) interaction is about 1.1
kcal mol-1 for both (4R)MePG-I and (4S)MePG-I, while the y(i
+ 1) angle is -1.8◦ in (4R)MePG-I and -15.1◦ in (4S)MePG-I. As
can be seen from Table 7, steric repulsive interactions between the
above-mentioned remote orbitals are outweighed by the stabilizing
hyperconjugative interactions. Thus, the net result from these
interactions is a stabilizing contribution to the b-turn formation.
Besides the interactions discussed above, Table 7 also includes the
energies of other selected remote interactions. However, these are
not discussed further here.
A particular repulsive n ↔ s interaction, not included in
Table 7, deserves some comment. This interaction is apparent in
(4R)MePG-II, in which the pyrrolidine ring of the proline residue
Fig. 6 3D representation of the n → s* interaction between O(i + 2) and
the glycine N–Ca antibond orbital in (4S)MePG-II.
This journal is
The Royal Society of Chemistry 2010
Org. Biomol. Chem., 2010, 8, 2103–2116 | 2111
©

View Article Online
the b-turn stabilizing hydrogen bond forming between Ala1-CO
and Ala2-NH more probable. Remote n → s* interaction between
the lone pair of O(i + 2) and the glycine N–Ca antibond orbital
may help to increase the probability of hydrogen bonding by
stabilizing conformations in which the glycine y angle approaches
values around 0◦. As a consequence of the above, the peptide
spends more time in conformations related to type II b-turn,
increasing the type II turn population. The Cg endo ring puckering
in A(4R)MePGA, on the contrary, favors proline f,y angles that
=
result in less effective n → p* interaction between O(i) and C¢ O(i
+ 1). Therefore, in A(4R)MePGA, conformations in which the
proline y dihedral is close to values compatible with the type
II b-turn conformation are less stabilized in comparison with
A(4S)MePGA, in which the Cg exo ring puckering of the proline
residue prevails. Consequently, although the above-mentioned n
→ s* interaction is effective also in A(4R)MePGA, there may
be a reduced probability that the glycine residue samples f,y
angles compatible with the type II b-turn conformation. This, in
turn, makes the b-turn stabilizing hydrogen bond forming between
Ala1-CO and Ala2-NH less probable, decreasing the type II b-
turn population of the peptide. The magnitude of the stabilizing
n → p* interaction between the lone pair of the Ala1 carbonyl
oxygen and the antibonding orbital of the proline carbonyl group
is significantly reduced in structures in which the y(i + 1) dihedral
is close to values compatible with type I b-turn conformation, i.e.
in type I b-turns and bends. The same applies to the extended
conformations. Therefore, in these structures, stabilization from
Fig. 7 3D representation of the n → s* interaction between O(i + 1) and
the proline N–Ca antibond orbital in (4S)MePG-I.
is in the Cg endo conformation, thus bringing the lone pair of
O(i + 1) and the proline Cg –H filled orbital in close contact to
each other. The distance between the oxygen lone pair and the
Cg –H bond, and consequently, the steric energy, Ei,j(st), associated
with this interaction is dependent on the proline f angle. The
more negative values of proline f angle bring the oxygen lone
pair and the Cg –H bond closer to each other, thus increasing the
repulsion. The proline f angle in the DFT optimized structure
=
the hyperconjugative interaction between O(i) and C¢ O(i + 1)
of (4R)MePG-II is -59.0◦, which results in an Ei,j value of
(st)
is very small. However, the type I b-turn conformations may be
somewhat stabilized by the remote n → s* interaction between
the lone pair of O(i + 1) and the proline N–Ca antibond orbital.
This interaction gives the strongest stabilization when the y(i + 1)
angle is near 0◦, which is about 30◦ away from the idealized value
for the type I b-turn conformation.
0.61 kcal mol-1. If the f angle alone is decreased to -65◦, steric
repulsion between the lone pair and the Cg –H bond increases to
1.21 kcal mol-1. This repulsion can be relieved in three ways (if the
pyrrolidine ring stays in the endo conformation): 1) the proline y
angle approaches larger values (positive or negative) compatible
with the extended structures, 2) the proline y angle approaches
small negative values compatible with type I b-turn conformations,
and 3) the pyrrolidine ring adopts a more flattened conformation.
The repulsive n ↔ s interaction between the lone pair of O(i + 1)
and the proline Cg –H filled orbital may provide one explanation
why the combined population of type I b-turns, type I bends,
and extended structures is considerably higher for A(4R)MePGA
(38%) than that for APGA and A(4S)MePGA (25% for both
peptides).
Conclusion
In this study, we have determined the ensembles of confor-
mations for three linear tetrapeptides (APGA, A(4R)MePGA,
and A(4S)MePGA) in DMSO-d6 solution using the NAMFIS
methodology. It was shown that enantiospecific incorporation of
the methyl substituent at the Cg (C4) position of the proline residue
can be used to selectively control the pyrrolidine ring puckering
propensities and, consequently, the preferred f,y angles associated
with the proline residue in b-turn forming peptides. Thus, the
populations of the b-turn conformations of the peptide in solution
can be controlled by a minor chemical change on the natural
amino acid. In addition, we have employed the NBO method to
gain further insight into the results obtained from the NAMFIS
analysis. The outcome from the NBO calculations shed light
on the underlying principles responsible for the conformational
preferences of b-turns.
The NAMFIS selected conformations could be reduced to 7
structural families (motifs) to reproduce the interproton distances
and the torsion angles obtained from the averaged NMR data
(Table 4). As expected, the NAMFIS analysis demonstrated that
the pyrrolidine ring of (4R)-methylproline in A(4R)MePGA has
a strong preference for the Cg endo puckering, whereas (4S)-
methylproline in A(4S)MePGA has a strong preference for the Cg
Effect of remote interactions on peptide conformation
The results from the NBO analysis suggest that the non-bonding
interactions between remote filled and unfilled orbitals have
important implications in determining solution conformations of
the peptides under study. Particularly interesting is the n → p*
interaction between the lone pair of the Ala1 carbonyl oxygen and
the antibonding orbital of the proline carbonyl group. The Cg exo
ring puckering in A(4S)MePGA preorganizes the main chain f,y
=
angles of the proline residue for this favorable O(i) ◊ ◊ ◊ C¢ O(i + 1)
interaction, which stabilizes conformations in which the proline
y dihedral angle is close to values compatible with the type II
b-turn conformation. This, in turn, may increase the probability
that the f,y angles of the glycine residue approach values also
compatible with the type II b-turn conformation, thus making
2112 | Org. Biomol. Chem., 2010, 8, 2103–2116
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
exo puckering. In the preferred conformations, the 4-methyl sub-
stituent adopts a pseudoequatorial orientation. In APGA, the two
puckerings are essentially equally populated among the NAMFIS
conformers. According to the NAMFIS analysis, the combined
type II b-turn population of A(4S)MePGA is considerably higher
than that of APGA and A(4R)MePGA. The results from the
NBO analysis suggest that this observation can be rationalized
by an energetic stabilization associated with a non-bonded n →
p* interaction between the lone pair of the Ala1 carbonyl oxygen
and the antibonding orbital of the proline carbonyl group. The
dominant Cg exo ring puckering in A(4S)MePGA preorganizes
the main chain f,y angles of the proline residue for this remote
Experimental section
Tetrapeptide synthesis
Detailed procedures for the peptide synthesis are provided in the
Supporting Information. 4-Methylprolines suitable for solid phase
synthesis were prepared according to Scheme S1 (Supporting
Information). The solid phase synthesis of APGA, A(4R)MePGA,
and A(4S)MePGA is described in Scheme S2 (Supporting Infor-
mation).
NMR spectroscopy of tetrapeptides
=
O(i) ◊ ◊ ◊ C¢ O(i + 1) interaction, whereas in A(4R)MePGA the
NMR studies were carried out on an Bruker Avance DPX400
spectrometer operating at 400.13 and 100.61 MHz for H and
prevailing Cg endo ring puckering favors proline f,y angles that
1
results in less effective n → p* interaction between O(i) and C¢ O(i
13C, respectively. All experiments were performed in DMSO-d6
solution at 299 K. The sample concentration was 80 mM for
APGA, 77 mM for A(4R)MePGA, and 74 mM for A(4S)MePGA.
Chemical shifts are reported in ppm relative to TMS (d 0.00)
for 1H NMR. For 13C spectra, the residual DMSO (d 39.51)
was used as internal standard. All 1D and 2D spectra were
recorded using standard pulse sequences from the Bruker pulse
program library in TopSpin 1.3. Data processing was made using
=
+ 1). In APGA, the more flexible pyrrolidine ring of the proline
residue results in a combined type II b-turn population that is
comparable to that of A(4R)MePGA.
Several other remote interactions provide stabilization for
both the type I and II b-turn conformations. The strongest of
these occur between the O(i) lone pairs as donor and the N–H
antibonding orbital, s*(N–H), of the NHMe group as acceptor,
quantifying the extent of hydrogen bonding between O(i) and
the NHMe group. The n → s* interaction between the lone
pair of O(i + 2) and the glycine N–Ca antibonding orbital, can
potentially help the formation of the above-mentioned hydrogen
bond by stabilizing conformations in which the glycine y angle
approaches values around 0◦. Type I b-turn conformations may
gain some stabilization from the n → s* interaction between the
lone pair of O(i + 1) and the proline N–Ca antibonding orbital.
This interaction gives the strongest stabilization when the y(i + 1)
angle approaches 0◦, which is about 30◦ away from the idealized
value for the type I b-turn conformation. Finally, the repulsive n
↔ s interaction between the lone pair of O(i + 1) and the proline
Cg –H filled orbital may provide an explanation for the observation
that the combined population of the type I b-turns, type I bends,
and extended structures is considerably higher for A(4R)MePGA
than that for APGA and A(4S)MePGA.
1
TopSpin 2.0 software. The H chemical shift assignments were
made using a combination of TOCSY and T-ROESY experiments.
Interproton distances were obtained from the adiabatic version of
off-resonance ROESY experiments.47 This experiment was chosen
because it has been shown to provide both efficient suppression
1
of HOHAHA transfers and reduction of offset effects. The H
NMR spectral analyses were performed with the PERCHit iterator
under PERCH software.48,49 The Total-Line-Shape-Fitting option
of the PERCHit iterator was used for the final refinement of the
1
result. Prior to the spin system analysis, the resolution of the H
NMR spectra was enhanced using the combination of exponential
and trapezoidal windowing before Fourier transformation.68 The
13C chemical shifts were assigned by a combination of HSQC
and CIGAR-HMBC experiments. These assignments are given in
Table S4 (Supporting Information).
1
One-dimensional (1D) H spectra were acquired using a 1 s
There is a great interest in mimicking b-turns for the synthesis of
medicines in the field of medical and pharmacological chemistry.
Although the peptides studied are not biologically important
per se, they provided a simple model system to investigate
different backbone-backbone interactions that contribute to b-
turn stability and, therefore, b-turn formation in peptides and
proteins. In addition, they lend themselves to examination of the
factors behind the relationship between the proline ring puckering
and the peptide backbone conformation. Further, peptides such
as A(4R)MePGA and A(4S)MePGA potentially offer a probe
for molecular recognition studies since the 4-methyl substituent
of the pyrrolidine ring strongly favors pseudoequatorial orien-
tation, thus providing the possibility to deduce directionality
requirements in the binding of b-turns and of polypeptides in
general. We anticipate that this investigation has the potential
to improve our understanding of structural parameters that
influence conformational behavior of flexible b-turn forming
peptides in liquid phase. The findings from this study could be
useful in the de novo design of peptides, peptidomimetic com-
pounds, and proteins, and in the development of peptide-based
drugs.
relaxation delay, 64 transients, a spectral width of 10 ppm, and
64 K data points zero-filled to 128 K data points before Fourier
transformation and baseline correction. 1D 13C spectra were
recorded using 2 s relaxation delay, a spectral width of 180 or
190 ppm, 1 K transients, and 64 K data points. The phase-
sensitive TOCSY spectra were obtained with a mixing time of
80 ms using the MLEV17 spin lock sequence. A typical experiment
was recorded using a 2 s relaxation delay, a spectral width of
8.6 ppm in both dimensions, 2 K data points, 512 t1 increments,
and 8 transients. The data was zero-filled to a 2 K ¥ 2 K matrix
before Fourier transformation. A sine-bell weighting function was
used in both dimensions. The phase-sensitive T-ROESY spectra
were obtained with mixing times of 100, 200, and 300 ms. Typical
spectra were recorded using 2 s relaxation delay, a spectral width
of 8.6 ppm in both dimensions, 2 K data points, 512 t1 increments,
and 8 transients. The data was zero-filled to a 2 K ¥ 2 K matrix
before Fourier transformation. A sine-bell weighting function
was used in both dimensions. The phase-sensitive off-resonance
ROESY spectra were obtained with mixing times of 70, 140,
210, 280, 350, and 420 ms to check the linearity of the cross-
relaxation buildup. The mixing time of 180 ms was chosen for
This journal is
The Royal Society of Chemistry 2010
Org. Biomol. Chem., 2010, 8, 2103–2116 | 2113
©

View Article Online
the quantitative off-resonance ROESY experiments. The mixing
sequence was achieved by adiabatic rotation. The spin-lock field
strength gB1 was 5107 Hz and the H angle was equal to 45◦.
A typical experiment was recorded using 2 s relaxation delay, a
spectral width of 8.6 ppm in both dimensions, 2 K data points, 512
t1 increments, and 16 transients. The data was zero-filled to a 2 K ¥
2 K matrix before Fourier transformation. A sine-bell weighting
function was used in both dimensions. Typical phase-sensitive
HSQC spectra were acquired with spectral widths of 8.6 ppm for
1H and 190 ppm for 13C using 2 s relaxation delay, 2 K data points,
512 t1 increments, and 4 transients. The data was zero-filled to a
2 K ¥ 2 K matrix before Fourier transformation. A sine squared
window multiplication was used in both dimensions. The CIGAR-
HMBC spectra were obtained in the magnitude mode typically
NMR parameters. SSD values for APGA, A(4R)MePGA, and
A(4S)MePGA were 8.7, 7.9, and 9.1, respectively. Interproton
distances for the NAMFIS analysis were obtained from the off-
resonance ROESY experiment with a mixing time of 180 ms using
the initial rate approximation.72 Average cross-peak volumes were
determined via integration of rectangular peak areas from both
sides of the diagonal using TopSpin 2.0. In the case of APGA,
the obtained distances were calibrated with respect to the distance
2
2
˚
between Ala -Ha and Ala -CH3 (2.45 A). For A(4R)MePGA and
pro(R)
pro(S)
A(4S)MePGA, the distance between proline Hd
(1.78 A) was used. Error bars used by NAMFIS for off-resonance
and Hd
˚
˚
ROESY derived distances (dNOE, in A) were as follows: dNOE
<
2.5 ( 0.1), 2.5 ≤ dNOE < 3.0 ( 0.2), 3.0 ≤ dNOE < 3.5 ( 0.3),
and 3.5 ≤ dNOE < 6.0 ( 0.4). An extended Karplus equation
developed by Haasnoot et al.73 (parameter set E) was used to
1
with spectral widths of 8.6 ppm for H and 190 ppm for 13C, 6–
3
10 Hz optimization range for long-range heteronuclear couplings,
140–180 Hz range for suppression of one-bond coupling responses,
and the J-scaling factor of 0. These experiments were recorded
using 2 s relaxation delay, 2 K data points, 512 t1 increments, and
16 transients. The data was zero-filled to a 2 K ¥ 2 K matrix before
Fourier transformation. A sine squared window multiplication
was used in the F2 dimension and a sine-bell weighting function
in the F1 dimension. A polynomial baseline correction was applied
for all 2D experiments mentioned above.
calculate the pyrrolidine ring J(H,H) coupling constants. For
3
the backbone J(NH,Ha) couplings, the Karplus-type equation
parameterized by Pardi et al.74 was employed. Both equations
were associated with an error of 1.5 Hz. NAMFIS deconvolution
was performed on all possible permutations of methylene protons.
Permutations were handled by an external program permute.py.
Specific permutation with lowest SSD corresponded to the correct
assignment. Initial conformer populations were randomized using
the random option implemented in NAMFIS program in order
to avoid “local minima” as well as to ensure that the solution
was unique. Several additional NAMFIS analyses were also
performed in order to check the robustness of the NAMFIS-
selected peptide conformations and to explore the sensitivity of
NAMFIS populations to variations of the experimental datasets.
These data are provided as Supporting Information.
NAMFIS analysis
The extended structures of APGA, A(4R)MePGA, and
A(4S)MePGA were used as starting points for all conforma-
tional searches. For each peptide, three different force fields
(OPLS_2005, AMBER*, and MMFFs) were used to generate
conformations in MacroModel 9.556 combined with the GB/SA
H2O continuum solvent model.57 This solvent model is well-
parameterized for a range of organic structures and was chosen
to mimic polar solvent to dampen unrealistic gas-phase dipole–
dipole interactions. NAMFIS analyses, in which conformational
searches have been performed with the GB/SA H2O model and
the NMR parameters have been derived in DMSO-d6, have
been published earlier.17,20,69,70 Each force field was applied to
two separate random-seeded searches of 90000 MCMM (Monte
Carlo Multiple Minimum)71 steps using the PRGC algorithm. The
minimized conformers from the duplicate runs were combined and
optimized to convergence with their respective force fields using
the PRGC algorithm and a 6.0 kcal mol-1 (25 kJ mol-1) energy
cutoff. Duplicate conformers were eliminated at this point. The
global energy minimum was found at least 179 times for each force
field and peptide. Finally, all structures from the multiple force
field optimizations were combined using a redundant conformer
elimination to form a single conformation pool for each peptide.
This resulted in 2259, 1996, and 2485 conformers for APGA,
A(4R)MePGA, and A(4S)MePGA, respectively. These conform-
ers were subjected to the NAMFIS deconvolution procedure
using interproton distances and torsion angles obtained from the
averaged NMR data. NAMFIS performs a least-squares fit of
the above-mentioned experimental data against the same data ex-
tracted from the conformer database. Goodness of fit is expressed
as the sum of square differences (SSD) between experimental and
calculated variables. The result is a population of conformers
(set of mole fractions) that optimally represents the experimental
Density functional calculations
All DFT calculations were performed using the Jaguar 7.0
program package.75 Natural bond orbital (NBO) analysis was
performed using the NBO 5.0 program76 interfaced into the Jaguar
7.0 software. Orbital graphics were produced by the NBOView
1.1 program.77 For visualization purposes, 3D orbital diagrams
display the pre-orthogonal NBO (PNBO) associated with each
NBO. PNBOs are a set of localized Lewis NBO precursors that
lack the final interatomic orthogonalization step. They provide
a convenient way to visualize interactions between orbitals in
different bond regions, because their overlap is proportional to
their interaction energy.78 Starting geometries for the calculations
were build using the Maestro 8.0 program.79 The tetrapeptide unit
A(4R)MePGA and A(4S)MePGA was truncated to a capped
moiety corresponding to the fragment MeCO-Xaa-Gly-NHMe
(Xaa = (4R)- or (4S)-MePro), and the resulting dipeptides were
conformationally manipulated to produce the type I and type
II b-turn structures. The puckering of the pyrrolidine ring was
set to match Cg endo in the (4R)-MePro peptides and Cg exo in
the (4S)-MePro peptides. Consequently, the methyl substituent
is oriented pseudoequatorially in corresponding four peptides.
These structures were minimized without any constraints using
the OPLS_2005 force field combined with the GB/SA H2O
continuum solvent model.57 Each b-turn structure was further
optimized, first at B3LYP/6-31+G* level of theory in gas phase,
and second at B3LYP/6-311++G** level of theory using the
PBF solvation model66,67 with DMSO as solvent. Frequency
2114 | Org. Biomol. Chem., 2010, 8, 2103–2116
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
calculations at the B3LYP/6-311++G** level of theory (in gas
phase) were used to confirm that the optimized structures were
minima, as characterized by positive vibrational frequencies.
Finally, NBO calculations were carried out for all four structures
at the B3LYP/6-311++G** level in DMSO using the PBF solvent
model and geometries obtained at the same level of theory.
The stabilizing and destabilizing NBO interaction energies were
considered down to 0.1 kcal mol-1.
25 M. D. Shoulders, I. A. Guzei and R. T. Raines, Biopolymers, 2008, 89,
443–454.
26 S. A. Cadamuro, R. Reichold, U. Kusebauch, H.-J. Musiol, C. Renner,
P. Tavan and L. Moroder, Angew. Chem., Int. Ed., 2008, 47, 2143–2146.
27 E. S. Eberhardt, N. Panasik and R. T. Raines, J. Am. Chem. Soc., 1996,
118, 12261–12266.
28 L. E. Bretscher, C. L. Jenkins, K. M. Taylor, M. L. DeRider and R. T.
Raines, J. Am. Chem. Soc., 2001, 123, 777–778.
29 C. Renner, S. Alefelder, J. H. Bae, N. Budisa, R. Huber and L. Moroder,
Angew. Chem., Int. Ed., 2001, 40, 923–925.
30 W. Kim, A. McMillan, J. P. Snyder and V. P. Conticello, J. Am. Chem.
Soc., 2005, 127, 18121–18132.
31 M. D. Shoulders, J. A. Hodges and R. T. Raines, J. Am. Chem. Soc.,
2006, 128, 8112–8113.
Acknowledgements
32 J. L. Flippen-Anderson, R. Gilardi, I. L. Karle, M. H. Frey, S. J. Opella,
L. M. Gierasch, M. Goodman, V. Madison and N. G. Delaney, J. Am.
Chem. Soc., 1983, 105, 6609–6614.
33 A. M. P. Koskinen, J. Helaja, E. T. T. Kumpulainen, J. Koivisto, H.
Mansikkama¨ki and K. Rissanen, J. Org. Chem., 2005, 70, 6447–6453.
34 P. Chakrabarti and D. Pal, Prog. Biophys. Mol. Biol., 2001, 76, 1–102.
35 B. K. Ho, E. A. Coutsias, C. Seok and K. A. Dill, Protein Sci., 2005,
14, 1011–1018.
ETTK wishes to thank The Graduate School of Organic Chem-
istry and Chemical Biology, Orion Farmos Research Foundation
and Tekniikan Edista¨missa¨a¨tio¨ (TES) for support. We are greatly
indebted to Dr James P. Snyder (Emory University, Atlanta,
GA, USA) for providing the NAMFIS program. We also thank
Matthew Geballe (Emory University, Atlanta, GA, USA) for
teaching the use of the NAMFIS software.
36 J.-C. Horng and R. T. Raines, Protein Sci., 2006, 15, 74–83.
37 D. Naduthambi and N. J. Zondlo, J. Am. Chem. Soc., 2006, 128, 12430–
12431.
38 N. B. Malkar, J. L. Lauer-Fields, J. A. Borgia and G. B. Fields,
Biochemistry, 2002, 41, 6054–6064.
39 M. Doi, Y. Nishi, S. Uchiyama, Y. Nishiuchi, T. Nakazawa, T. Ohkubo
and Y. Kobayashi, J. Am. Chem. Soc., 2003, 125, 9922–9923.
40 A. V. Persikov, J. A. M. Ramshaw, A. Kirkpatrick and B. Brodsky,
J. Am. Chem. Soc., 2003, 125, 11500–11501.
Notes and references
1 G. D. Rose, L. M. Gierasch and J. A. Smith, Adv. Protein Chem., 1985,
37, 1–109.
2 W. Kabsch and C. Sander, Biopolymers, 1983, 22, 2577–2637.
3 A. M. C. Marcelino and L. M. Gierasch, Biopolymers, 2008, 89, 380–
391.
41 D. Barth, A. G. Milbradt, C. Renner and L. Moroder, ChemBioChem,
2004, 5, 79–86.
4 E. G. Hutchinson and J. M. Thornton, Protein Sci., 1994, 3, 2207–2216.
5 K. S. Kee and S. D. S. Jois, Curr. Pharm. Des., 2003, 9, 1209–1224.
6 K. Wu¨thrich, NMR of Proteins and Nucleic Acids, Wiley, New York,
1986.
42 F. W. Kotch, I. A. Guzei and R. T. Raines, J. Am. Chem. Soc., 2008,
130, 2952–2953.
43 W. Kim, A. George, M. Evans and V. P. Conticello, ChemBioChem,
2004, 5, 928–936.
7 O. Jardetzky, Biochem. Biophys. Acta, 1980, 621, 227–232.
8 H. Kessler, C. Griesinger, J. Lautz, A. Mu¨ller, W. F. van Gunsteren and
H. J. C. Berendsen, J. Am. Chem. Soc., 1988, 110, 3393–3396.
9 X. Daura, I. Antes, W. F. van Gunsteren, W. Thiel and A. E. Mark,
Proteins: Struct., Funct., Genet., 1999, 36, 542–555.
10 N. Nevins, D. Cicero and J. P. Snyder, J. Org. Chem., 1999, 64, 3979–
3986.
44 S. C. R. Lummis, D. L. Beene, L. W. Lee, H. A. Lester, R. W. Broadhurst
and D. A. Dougherty, Nature, 2005, 438, 248–252.
45 C. Boule`gue, A. G. Milbradt, C. Renner and L. Moroder, J. Mol. Biol.,
2006, 358, 846–856.
46 A. E. Reed, L. A. Curtiss and F. Weinhold, Chem. Rev., 1988, 88,
899–926.
47 H. Desvaux, P. Berthault, N. Birlirakis, M. Goldman and M. Piotto,
J. Magn. Reson., Ser. A, 1995, 113, 47–52.
48 R. Laatikainen, M. Niemitz, U. Weber, J.-P. Sundelin, T. Hassinen and
J. Vepsa¨la¨inen, J. Magn. Reson., Ser. A, 1996, 120, 1–10.
49 PERCH NMR Software, version 1/2005, PERCH Solutions Ltd.,
Kuopio, Finland, 2005.
11 D. Cicero, G. Barbato and R. Bazzo, J. Am. Chem. Soc., 1995, 117,
1027–1033.
12 P. Thepchatri, D. O. Cicero, E. Monteagudo, A. K. Ghosh, B. Cornett,
E. R. Weeks and J. P. Snyder, J. Am. Chem. Soc., 2005, 127, 12838–
12846.
13 J. P. Snyder, A. S. Lakdawala and M. J. Kelso, J. Am. Chem. Soc., 2003,
50 P. Amodeo, A. Motta, D. Picone, G. Saviano, T. Tancredi and P. A.
Temussi, J. Magn. Reson., 1991, 95, 201–207.
125, 632–633.
14 G. Barbato, E. Bianchi, P. Ingallinella, W. H. Hurni, M. D. Miller, G.
Ciliberto, R. Cortese, R. Bazzo, J. W. Shiver and A. Pessi, J. Mol. Biol.,
2003, 330, 1101–1115.
15 M. J. Kelso, R. L. Beyer, H. N. Hoang, A. S. Lakdawala, J. P. Snyder,
W. V. Oliver, T. A. Robertson, T. G. Appleton and D. P. Fairlie, J. Am.
Chem. Soc., 2004, 126, 4828–4842.
51 K. Sorimachi and D. J. Craik, Eur. J. Biochem., 1994, 219, 237–251.
52 O. M. A. El-Agnaf, G. B. Irvine, G. Fitzpatrick, W. K. Glass and D. J. S.
Guthrie, Biochem. J., 1998, 336, 419–427.
53 P. Desai, M. Prachand, E. Coutinho, A. Saran, J. Bodi and H. Su¨li-
Vargha, Int. J. Biol. Macromol., 2002, 30, 187–195.
54 G. Shanmugam and P. L. Polavarapu, Biophys. J., 2004, 87, 622–630.
55 J. F. O’Connell, A. M. P. Koskinen, M. Krische and H. Rapoport,
unpublished work.
16 J. P. Snyder, N. Nevins, D. O. Cicero and J. Jansen, J. Am. Chem. Soc.,
2000, 122, 724–725.
17 E. Monteagudo, D. O. Cicero, B. Cornett, D. C. Myles and J. P. Snyder,
J. Am. Chem. Soc., 2001, 123, 6929–6930.
18 J. H. Nettles, H. Li, B. Cornett, J. M. Krahn, J. P. Snyder and K. H.
Downing, Science, 2004, 305, 866–869.
19 M. Erde´lyi, B. Pfeiffer, K. Hauenstein, J. Fohrer, J. Gertsch, K.-H.
Altmann and T. Carlomagno, J. Med. Chem., 2008, 51, 1469–1473.
20 P. Thepchatri, T. Eliseo, D. O. Cicero, D. Myles and J. P. Snyder, J. Am.
Chem. Soc., 2007, 129, 3127–3134.
21 D. F. DeTar and N. P. Luthra, J. Am. Chem. Soc., 1977, 99, 1232–1244.
22 L. Vitagliano, R. Berisio, A. Mastrangelo, L. Mazzarella and A. Zagari,
Protein Sci., 2001, 10, 2627–2632.
56 MacroModel, version 9.5, Schro¨dinger, LLC, New York, NY, 2007.
57 W. C. Still, A. Tempczyk, R. C. Hawley and T. Hendrickson, J. Am.
Chem. Soc., 1990, 112, 6127–6129.
58 P. N Lewis, F. A. Momany and H. A. Scheraga, Biochem. Biophys. Acta,
1973, 303, 211–229.
59 H. B. Bu¨rgi, J. D. Dunitz and E. Shefter, J. Am. Chem. Soc., 1973, 95,
5065–5067.
60 H. B. Bu¨rgi, J. M. Lehn and G. Wipff, J. Am. Chem. Soc., 1974, 96,
1956–1957.
61 H. B. Bu¨rgi, J. D. Dunitz, J. M. Lehn and G. Wipff, Tetrahedron, 1974,
30, 1563–1572.
23 M. L. DeRider, S. J. Wilkens, M. J. Waddell, L. E. Bretscher, F.
Weinhold, R. T. Raines and J. L. Markley, J. Am. Chem. Soc., 2002,
124, 2497–2505.
24 R. Improta, C. Benzi and V. Barone, J. Am. Chem. Soc., 2001, 123,
12568–12577.
62 H. B. Bu¨rgi, J. D. Dunitz and E. Shefter, Acta Crystallogr., Sect. B:
Struct. Crystallogr. Cryst. Chem., 1974, 30, 1517–1527.
63 J. A. Hodges and R. T. Raines, Org. Lett., 2006, 8, 4695–4697.
64 J. K. Badenhoop and F. Weinhold, J. Chem. Phys., 1997, 107, 5406–
5421.
This journal is
The Royal Society of Chemistry 2010
Org. Biomol. Chem., 2010, 8, 2103–2116 | 2115
©

View Article Online
65 J. K. Badenhoop and F. Weinhold, Int. J. Quantum Chem., 1999, 72,
269–280.
66 D. J. Tannor, B. Marten, R. Murphy, R. A. Friesner, D. Sitkoff, A.
Nicholls, M. Ringnalda, W. A. Goddard, III and B. Honig, J. Am.
Chem. Soc., 1994, 116, 11875–11882.
72 D. Neuhaus and M. P. Williamson, The Nuclear Overhauser Effect in
Structural and Conformational Analysis, 2nd edn, Wiley-VHC, New
York, 2000.
73 C. A. G. Haasnoot, F. A. A. M. de Leeuw and C. Altona, Tetrahedron,
1980, 36, 2783–2792.
67 B. Marten, K. Kim, C. Cortis, R. A. Friesner, R. B. Murphy, M. N.
Ringnalda, D. Sitkoff and B. Honig, J. Phys. Chem., 1996, 100, 11775–
11788.
74 A. Pardi, M. Billeter and K. Wu¨thrich, J. Mol. Biol., 1984, 180, 741–
751.
75 Jaguar, version 7.0, Schro¨dinger, LLC, New York, NY, 2007.
76 E. D. Glendening, J. K. Badenhoop, A. E. Reed, J. E. Carpenter, J. A.
Bohmann, C. M. Morales and F. Weinhold, NBO 5.0, Theoretical
Chemistry Institute, University of Wisconsin, Madison, WI, 2001.
77 M. Wendt and F. Weinhold, NBOView 1.1, NBO Orbital Graphics,
Department of Chemistry, University of Wisconsin, Madison, WI,
2001.
68 R. Laatikainen, K. Tuppurainen, Y. Hiltunen and S. Lo¨tjo¨nen, Magn.
Reson. Chem., 1990, 28, 939–946.
69 L. Barboni, C. Lambertucci, G. Appendino, D. G. Vander Velde, R. H.
Himes, E. Bombardelli, M. Wang and J. P. Snyder, J. Med. Chem., 2001,
44, 1576–1587.
70 L. Barboni, A. Datta, D. Dutta, G. I. Georg, D. G. Vander Velde, R. H.
Himes, M. Wang and J. P. Snyder, J. Org. Chem., 2001, 66, 3321–3329.
71 G. Chang, W. C. Guida and W. C. Still, J. Am. Chem. Soc., 1989, 111,
4379–4386.
78 S. J. Wilkens, W. M. Westler, J. L. Markley and F. Weihold, J. Am.
Chem. Soc., 2001, 123, 12026–12036.
79 Maestro, version 8.0, Schro¨dinger, LLC, New York, NY, 2007.
2116 | Org. Biomol. Chem., 2010, 8, 2103–2116
This journal is
The Royal Society of Chemistry 2010
©