Conformational manifold of α-aminoisobutyric acid (Aib) containing alanine-based tripeptides in aqueous solution explored by vibrational spectroscopy, electronic circular dichroism spectroscopy, and molecular dynamics simulations

Article abstract of DOI:10.1021/ja0738430

Replacement of the α-proton of an alanine residue to generate α-aminoisobutyric acid (Aib) in alanine-based oligopeptides favors the formation of a 3₁₀ helix when the length of the oligopeptide is about four to six residues. This research was aimed at experimentally identifying the structural impact of an individual Aib residue in an alanine context of short peptides in water and Aib's influence on the conformation of nearest-neighbor residues. The amide I band profile of the IR, isotropic and anisotropic Raman, and vibrational circular dichroism (VCD) spectra of Ac-Ala-Ala-Aib-OMe, Ac-Ala-Aib-Ala-OMe, and Ac-Aib-Ala-Ala-OMe were measured and analyzed in terms of different structural models by utilizing an algorithm that exploits the excitonic coupling between amide I′ modes. The conformational search was guided by the respective ¹H NMR and electronic circular dichroism spectra of the respective peptides, which were also recorded. From these analyses, all peptides adopted multiple conformations. Aib predominantly sampled the right-handed and left-handed 3₁₀-helix region and to a minor extent the bridge region between the polyproline (PPII) and the helical regions of the Ramachandran plot. Generally, alanine showed the anticipated PPII propensity, but its conformational equilibrium was shifted towards helical conformations in Ac-Aib-Ala-Ala-OMe, indicating that Aib can induce helical conformations of neighboring residues positioned towards the C-terminal direction of the peptide. An energy landscape exploration by molecular dynamics simulations corroborated the results of the spectroscopic studies. They also revealed the dynamics and pathways of potential conformational transitions of the corresponding Aib residues.

Full text of DOI:10.1021/ja0738430

Published on Web 10/05/2007
Conformational Manifold of r-Aminoisobutyric Acid (Aib)
Containing Alanine-Based Tripeptides in Aqueous Solution
Explored by Vibrational Spectroscopy, Electronic Circular
Dichroism Spectroscopy, and Molecular Dynamics
Simulations
Reinhard Schweitzer-Stenner,*,† Widalys Gonzales,† Gregory T. Bourne,‡
Jianwen A. Feng,‡ and Garland R. Marshall‡
Contribution from the Department of Chemistry, Drexel UniVersity, 3141 Chestnut Street,
Philadelphia, PennsylVania 19104 and Department of Biochemistry and Molecular Biophysics,
Washington UniVersity, St. Louis, Missouri 63110
Received May 28, 2007; E-mail: RSchweitzer-Stenner@drexel.edu
Abstract: Replacement of the R-proton of an alanine residue to generate R-aminoisobutyric acid (Aib) in
alanine-based oligopeptides favors the formation of a 310 helix when the length of the oligopeptide is about
four to six residues. This research was aimed at experimentally identifying the structural impact of an
individual Aib residue in an alanine context of short peptides in water and Aib’s influence on the conformation
of nearest-neighbor residues. The amide I band profile of the IR, isotropic and anisotropic Raman, and
vibrational circular dichroism (VCD) spectra of Ac-Ala-Ala-Aib-OMe, Ac-Ala-Aib-Ala-OMe, and Ac-Aib-Ala-
Ala-OMe were measured and analyzed in terms of different structural models by utilizing an algorithm that
exploits the excitonic coupling between amide I′ modes. The conformational search was guided by the
1
respective H NMR and electronic circular dichroism spectra of the respective peptides, which were also
recorded. From these analyses, all peptides adopted multiple conformations. Aib predominantly sampled
the right-handed and left-handed 310-helix region and to a minor extent the bridge region between the
polyproline (PPII) and the helical regions of the Ramachandran plot. Generally, alanine showed the
anticipated PPII propensity, but its conformational equilibrium was shifted towards helical conformations in
Ac-Aib-Ala-Ala-OMe, indicating that Aib can induce helical conformations of neighboring residues positioned
towards the C-terminal direction of the peptide. An energy landscape exploration by molecular dynamics
simulations corroborated the results of the spectroscopic studies. They also revealed the dynamics and
pathways of potential conformational transitions of the corresponding Aib residues.
dran plots reported by Marshall and Bosshard6 and by Burgess
and Leach7 indicate two troughs of the potential energy
Introduction
It is well-established that the natural, nonprotein amino acid
R-aminoisobutyric acid (Aib, R-methylalanine) has a very high
helix propensity and a high potential as a â-sheet (strand)
breaker.1-3 It is a common residue in peptides produced by
microbes.4,5 Examples are linear peptides of 15-20 residues,
such as alamethicin, antiamoebin, and emerimicin, which
produce voltage-gated channels in lipid membranes. Aib can
be considered as being produced by the replacement of the CR
proton of an alanine residue by a methyl group which
significantly limits the rotational degree of freedom with respect
to the dihedral backbone coordinates φ and ψ. The Ramachan-
landscape centered at (φ ) -57°, ψ ) -47°) and (φ ) 57°, ψ
) 47°), corresponding to right- and left-handed R-helices. Aib
is per se achiral, but differential sampling of the two enantio-
meric conformations corresponding to the above clusters in the
Ramachandran plot are induced by neighboring chiral amino
acid residues. The two clusters encompass both R-helices and
310-helices with a more frequent sampling of the latter for shorter
peptides.
The 310-helix represents an important secondary structure in
proteins and accounts for approximately 10% of helical residues
in crystal structures (PDB).8 Recent investigations of coil
libraries indicate that 310 rather than R is the preferred
conformation in the lower left-handed Ramachandran quadrant
of coil fragments of proteins.9-11 The 310-helical segments in
† Drexel University.
‡ Washington University.
(1) Karle, I. L.; Balaram, P. Biochemistry 1990, 29, 6747.
(2) Moretto, V.; Crisma, M.; Bonora, G. M.; Toniolo, C.; Balaram, H.; Balaram,
P. Macromolecules 1989, 22, 2939.
(6) Marshall, G. R.; Bosshard, H. E. Circ. Res. 1972, 30/31 (Suppl. II.), 143.
(7) Burgess, A. W.; Leach, S. J. Biopolymers 1973, 12, 2599.
(8) Barlow, D. J.; Thornston, J. M. J. Mol. Biol. 1988, 201, 601.
(9) Avbelj, F.; Baldwin, R. L. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 5742.
(10) Serrano, L. J. Mol. Biol. 1995, 254, 322-333.
(3) Kumita, J. R.; Weston, C. J.; Choo-Smith, L.-P.; Wooley, G. A.; Smart,
O. S. Biochemistry 2003, 42, 4492.
(4) Closse, A.; Huguenin, R. HelV. Chim. Acta 1974, 57, 533.
(5) Flippen, J. L.; Karle, I. L. Biopolymers 1976, 15, 1081.
9
10.1021/ja0738430 CCC: $37.00 © 2007 American Chemical Society
J. AM. CHEM. SOC. 2007, 129, 13095-13109
13095

A R T I C L E S
Schweitzer-Stenner et al.
proteins are generally short with an average length of four
residues.8,12 A 310-helix differs from the canonical R-helix not
only with respect to its dihedral coordinates but also by a (i, i
+ 3) hydrogen-bonding arrangement.13 Augspurger et al.
investigated several Aib-based, aliphatic peptides of intermediate
size in organic solvents and obtained a very thermostable 310-
structure.14 De Filippis et al. confirmed and explained the
thermostability of Aib with the reduction of entropy of this
amino acid residue in the unfolded state.15 Numerous studies
have been conducted to identify the conditions at which Aib
stabilizes a 310-helix in peptides. Basically, a peptide gains an
extra internal hydrogen bond in a 310-helix while steric
interactions between residues favor R-helical torsional angles.
At a critical length depending on the dielectric of the solvent,
the two interactions approximately balance, and populations of
both helices coexist. For this reason, short Aib-based peptides
generally prefer 310-helices, whereas R-helices are more likely
for longer peptides.1,16,17 Rather long peptides, however, e.g.,
pBrBz-Aib10-OtBu (pBrBz, para-bromobenzyl; OtBu, tert-
butoxy) have been found to adopt 310-helices in crystals.18
Balaram et al. used NMR and electronic circular dichroism
spectroscopy to determine that the rather hydrophobic Boc-Aib-
Val-Aib2-Val3-Aib-Val-Aib-OMe and Boc-Aib-Leu-Aib2-Leu3-
Aib-Leu-Aib-OMe prefer the 310 helix in CDCl3, whereas an
R-helix preference was found in (CD3)2SO.19 This indicated that
the solvent has a strong influence due to differences in dielectric
constant and relative stabilities of hydrogen bonds on structural
preference. Crystalline forces and the choice of the protection
group have been identified as affecting the overall propensity
of Aib-based peptides in crystals.20-24 DiBlasio et al. found that
crystallized (pBrBz-Aib-(Pro-Aib)n-OMe (n ) 3, 4) forms so-
called â-ribbons which exhibit the Aib residues in 310-like
conformations.25 A review of the (mostly crystal) structure of
a large number of Aib-containing peptides by Karle and Balaram
indicated that the 310-helix is the preferred structure provided
that the fraction of Aib residues exceeds 50%.1 A dilution of
the Aib content decreases the preference for 310 so that a
coexistence of both helix types is established. Smythe et al.
calculated the potential energy landscape (i.e., potentials of mean
force) and performed molecular dynamics calculations for (Ac-
Aib10-NHMe) in Vacuo, water, and CH3CN to obtain a higher
stabilization of the R-helix for all three cases.26 The stability of
RR was found to increase with the dielectric constant of the
solvent. The higher stability of RR was later confirmed by
another computational study by Smythe et al.27 However, MD
simulations with the Cedar force field by Zhang and Hermans
yielded nearly the opposite result in that these authors obtained
310 as the predominant structure of Aib in Vacuo and a 310/RR
mixture in water.28 A strong 310 propensity of Aib was recently
experimentally confirmed by Schievano et al.29 Bu¨rgi et al. used
the GROMOS96 package to perform MD simulations for the
octapeptide Z-Aib5-Leu-Aib2-OMe in DMSO and obtained a
sampling of right- and left-handed helical conformations with
comparable fractions of R-helical and 310-conformations.30
A
somewhat different result emerged from simulations of Ma-
hadevan et al. for Aib10 in DMSO and water, for which the
authors used the CHARMM22 force field.31 They observed a
mixture of R- and π-helix. Taken together, this review of the
literature leads to the conclusion that more experimental data
are required to obtain the RR- and 310-propensity of Aib in
peptides as well as the use of force fields that include multipole
electrostatics and polarizability in MD simulations.32
In spite of some earlier work on short peptides, the intrinsic
propensity of Aib as a guest in a natural host system has not
yet been thoroughly investigated experimentally. Winter, Jung
and colleagues reported the structure of crystallized Ac-Ala-
Aib-Ala-OMe, Boc-Ala-Aib-Ala-OMe, and Boc-Gly-Ala-Aib-
OMe.33-35 For the two Ala-Aib-Ala peptides the authors
identified a distorted â-turn. CD and NMR data indicate that
this structure was maintained in DMSO for Boc-Ala-Aib-Ala-
OMe. The crystal structure of Ac-Ala-Aib-Ala-OMe was of
interest because the reported dihedral angles suggested that the
central Aib adopts a 310-like conformation, whereas the two
alanines exhibited a polyproline II (PPII)-like structure.35 This
is in accordance with the recently established PPII propensity
of alanine in water.36-38 The third peptide, Boc-Gly-Ala-Aib-
OMe, was found to adopt a somewhat distorted 310 or type III
â-turn-like conformation.
Experiments reported in this paper were aimed at investigating
the structure of three blocked Aib-containing tripeptides, namely
Ac-Aib-Ala-Ala-OMe, Ac-Ala-Aib-Ala-OMe, and Ac-Ala-Ala-
Aib-OMe in aqueous solution. Different vibrational spectro-
scopic techniques (IR, polarized Raman, vibrational circular
1
dichroism (VCD)) were combined with ECD and H NMR
(11) Duan, Y.; Wu, C.; Chowdhury, S.; Lee, M. C.;, Xiong, G.; Zhang, W.;
Yang, R.; Cieplak, P.; Luo, R.; Lee, T.; Caldwell, J.; Wang, J.; Kollman,
P. J. Comput. Chem. 2003, 24, 1999-2012.
(12) Robbins, D. J.; Stout, C. D. Proteins: Struct., Funct., Genet. 1989, 5, 289.
(13) Prasad, B. V. V.; Balaram, P. CRC Crit. ReV. Biochem. 1984, 16, 307.
(14) Augspurger, J. D.; Bindra, V. A.; Scheraga, H. A.; Kuki, A. Biochemistry
1995, 34, 2566.
(15) De Fillipis, V.; De Antoni, F.; Frigo, M.; de Laureto, P. P.; Fontana, A.
Biochemistry 1998, 37, 1686.
(16) Bosch, R.; Jung, G.; Schmitt, H.; Winter, W. Biopolymers 1995, 24, 979.
(17) Tomicolo, C.; Bonora, G. M.; Bovaso, A.; Bendetti, E.; DiBlasio, B.;
Pavone, V.; Pedone, C. J. Biomol. Struct. 1985, 3, 585.
(18) Pavone, V.; DiBasio, B.; Santini, A.; Benedetti, E.; Pedone, C. J. Mol.
Biol. 1990, 214, 633.
(19) Balarram, H.; Sukumar, M.; Balaram, P. Biopolymers 1986, 25, 2209.
(20) Karle, I. L.; Flippen-Anderson, J. L.; Sakumer, M.; Balaram, P. Int. J.
Peptide Protein Res. 1988, 32, 536.
(21) Karle, I. L.; Flippen-Anderson, J. L.; Sukumar, M.; Balaram, P. J. Med.
Chem. 1992, 35, 3885.
(22) Marshall, G. R.; Hodgkin, E. E.; Langs, D. A.; Smith, G. D.; Zabrocki, J.;
Leplawy, M. T. Proc. Natl. Acad. Sci. U.S.A. 1990, 87, 487.
(23) Blank, J. T.; Guerin, D.; Miller, S. J. Org. Lett. 2000, 2, 1248.
(24) Karle, I. L.; Sukumar, M.; Balaram, P. Proc. Natl. Acad. Sci. U.S.A. 1989,
29, 6747.
(25) DiBasio, B.; Pavone, V.; Savino, M.; Lombardi, A.; Nastri, F.; Pedone,
C.; Benedetti, E.; Crisma, M.; Anzoli, M.; Toniolo, C. J. Am. Chem. Soc.
1992, 114, 6274.
spectroscopy as well as molecular dynamics simulations.
(26) Smythe, M. L.; Huston, S. E.; Marshall, G. R. J. Am. Chem. Soc. 1993,
115, 11594.
(27) Smythe, M. L.; Huston, S. E.; Marshall, G. R. J. Am. Chem. Soc. 1995,
117, 5445.
(28) Zhang, L.; Herman, J. J. Am. Chem. Soc. 1994, 116, 11915.
(29) Schievano, E.; Bisello, A.; Chorev, M.; Bisol, A.; Mammi, S.; Peggion, E.
J. Am. Chem. Soc. 2001, 123, 2743.
(30) Bu¨rgi, R.; Daura, X.; Mark, A.; Bellanda, M.; Mammi, S.; Peggion, E.;
van Gunsteren, W. J. Peptide Res. 2001, 57, 107.
(31) Mahadevan, J.; Lee, K.-H.; Kuczera, K. J. Phys. Chem. B 2001, 105, 1863.
(32) Ponder, J. W.; Case, D. A. Force fields for protein simulations. AdV. Protein
Chem. 2003, 66, 27-85.
(33) Bosch, R.; Jung, G.; Winter, W. Liebigs Ann. Chem. 1982, 1322.
(34) Jung, G.; Bru¨ckner, H.; Bosch, R.; Winter, W.; Schaal, H.; Stra¨hle, J. Liebigs
Ann. Chem. 1983, 1096.
(35) Bosch, R.; Jung, G.; Voges, K.-P.; Winter, W. Liebigs Ann. Chem 1984,
1117.
(36) Schweitzer-Stenner, R.; Eker, F.; Griebenow, K.; Cao, X.; Nafie, L. A. J.
Am. Chem. Soc. 2004, 126, 2768.
(37) Schweitzer-Stenner, R.; Measey, T.; Kakalis, L.; Jordan, F.; Pizanelli, S.;
Forte, C.; Greibenow, K. Biochemistry 2007, 46, 1587.
(38) Schweitzer-Stenner, R.; Measey, T. J. Proc. Natl. Acad. Sci U.S.A. 2007,
104, 6649-6654.
9
13096 J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007

Conformational Manifold of
R
-Aminoisobutyric Acid
A R T I C L E S
with 0.05 nm resolution and a scan speed of 500 nm/min. A 0.2 mm
quartz cell was used for all experiments. We performed the measure-
ments in D2O rather than in H2O (concentration 3 mM) in order to
allow a direct comparison with structural data obtained by vibrational
spectroscopies. The samples were placed in a nitrogen-purged JASCO
CD module. The temperature at the cuvette was controlled by means
of a Peltier type heating system with an accuracy of (1 °C. For each
measurement, the sample was allowed to equilibrate for 5 min at the
adjusted temperature prior to acquisition. The spectra were obtained
by averaging 10 scans and were collected as ellipticity as a function of
wavelength and converted to difference molar absorptivities per residue
as earlier described.47
Analysis of Vibrational Spectra: The spectral analysis program
MULTIFIT was used to analyze all IR and Raman spectra.48 Calibration
of the Raman spectrum was checked using the NaClO4 peak at 934
cm-1. Reference spectra for both polarizations were measured which
were subtracted from the relevant peptide spectra to eliminate solvent
contributions. The isotropic and anisotropic Raman intensities were
calculated as
Analysis of the experimental results revealed the position-
dependent propensity of Aib within an alanine context in water.
Additionally, the influence of Aib on the conformational
manifold sampled by its nearest neighbors was obtained. This
is important for elucidating to what extent Aib affects the
propensity of its nearest neighbors already in the unfolded state
of peptides. The isolated-pair hypothesis often assumed in helix-
coil transition theory stipulates that in the unfolded state the
conformation of each amino acid residue is independent from
that of its neighbors.39 Pappu et al. have recently provided
computational evidence that this hypothesis does not apply when
a residue samples the right-handed helical region of the
Ramachandran plot.40 Since Aib doubtlessly prefers predomi-
nantly helical conformations, the set of peptides examined
provided a convenient way to check these predictions experi-
mentally.
Materials and Methods
Materials. Chlorotrityl resin (substitution ) 1 mmol/g) and all NR-
Fmoc-amino acids (peptide-synthesis grade) were purchased from
AdvanceChemtech (Louisville, KY) or Novabiochem (San Diego,
U.S.A.). Diisopropylethylamine and trifluoroacetic acid were purchased
from Aldrich (Milwaukee, USA). All other solvents including dichlo-
romethane, N,N-dimethylformamide, and HPLC-grade acetonitrile were
purchased from Fisher Scientific (USA).
Peptide Synthesis. All linear peptides were chemically synthesized
stepwise using Fmoc protecting groups and in situ HBTU activation
(or HATU for Aib) protocols as previously described.41-43 Coupling
efficiencies were determined by the quantitative ninhydrin test and
recoupled where necessary to obtain >99.5% efficiency.44
4
Iiso ) Ix - Iy
3
Ianiso ) Iy
(1)
where Ix and Iy denotes the Raman scattering polarized light parallel
and perpendicular to the polarization of the exciting laser.
Determination of 3JCRHNH Coupling Constants. To determine the
3JCRHNH coupling constants 1H NMR spectra of Ac-Ala-Ala-Aib-OMe,
Ac-Ala-Aib-Ala-OMe, and Ac-Aib-Ala-Ala-OMe were recorded at
room temperature on a VARIAN 500 INOVA instrument in the
Chemistry department of Drexel University. Pulse experiments used
3
Vibrational Spectroscopies. For IR, VCD, and Raman measure-
software supplied by VARIAN. The JCRHNH coupling constant of the
1
ments, the peptides were dissolved at a concentration of 0.07 M (20.0
Aib sequences were obtained from the H NMR spectrum. The Aib
-
mg/mL) in D2O with 0.025 M NaClO4 and the ClO4 Raman peak at
samples were dissolved in 700 µL of D2O at a concentration of 0.04
M for ABA and BA2 and 0.02 M for A2B.
934 cm-1 was used as an internal wavenumber standard.45 All
experiments were carried out at room temperature. The instrument for
the Raman experiments has been described in detail in previous papers.46
All spectra were recorded in the “continuous” mode and were measured
a total of 8 times for each of the polarization directions. The spectra
finally used for analysis were obtained by calculating the average of
the measured spectra to improve the signal-to-noise ratio. The reference
spectra of a 25 mM NaClO4 in D2O were subtracted from the respective
sample spectra for both polarization directions.
DFT Calculations. Some preliminary DFT calculations were carried
out for the investigated peptides in Vacuo. All calculations were carried
out with the TITAN program, a joint product from Wavefunction, Inc.
(Irvine, CA) and Schro¨dinger, Inc. (Portland, OR). The optimized
geometries of the deuterated peptides (NH substituted by ND) were
calculated at the B3LYP/6-31G** level for initial conditions specified
below. Normal mode calculations were subsequently carried out for
the thus obtained geometries.
The FTIR and VCD spectra were recorded with a Chiral IR Fourier
Transform VCD spectrometer from BioTools. The instrument has been
previously described.46 The peptide sample was placed into a cell with
a path length of 42 µm. The spectral resolution was 4 cm-1 for both
spectra. The VCD and IR spectra were both collected as one
measurement for a combined total time of 1440 min. To eliminate any
background and solvent contributions to the IR spectrum, the cell was
first filled with the reference solvent.
Molecular Dynamics Simulations
Initial Conformations. The Ac-Aib-Ala-Ala-OMe, Ac-Ala-
Ala-Aib-OMe, and Ac-Ala-Aib-Ala-OMe PDB files were cre-
ated using Sybyl 7.3.49 They were manually edited to satisfy
the atom-name requirements of the pdb2gmx file-conversion
program in GROMACS.50,51 GROMACS topology files were
manually appended to define the OMe residue and its torsion
parameters. Four initial conformers were generated for each
peptide sequence. The first conformer was a classical R-helix.
Rotating the Aib æ torsion in the first conformer from -58° to
+58° created the second conformer. The third and fourth
conformers were classical â-strands except the Aib æ torsion
was set to -58° or +58°, respectively. In total, 12 initial
conformers were created (3 peptides × 4 conformers/peptide).
Each peptide was solvated in a rhombic dodecahedron box of
ECD Spectroscopy. The UV ECD spectra in the wavelength range
of 180-240 nm of the peptides were measured as a function of
temperature with an earlier described JASCO J-810 spectrapolarimeter47
(39) Brant, D. A.; Flory, P. J. J. Am. Chem. Soc. 1995, 87, 2791.
(40) Pappu, R. V.; Srinivasan, R.; Rose, G. D. Proc. Natl. Acad. Sci. U.S.A.
2000, 97, 12565.
(41) Carpino, L. A.; El-Faham, A.; Minor, C. A.; Albericio, F. J. Chem. Soc.,
Chem. Commun. 1994, 1994, 201-203.
(42) Schno¨lzer, M.; Alewood, P. F.; Jones, A.; Alewood, D.; Kent, S. B. H.
Int. J. Pept. Protein Res. 1992, 40, 180-93.
(43) Alewood, P.; Alewood, D.; Miranda, L.; Love, S.; Meutermans, W.; Wilson,
D. Methods Enzymol. 1997, 289, 14-29.
(44) Sarin, V. K.; Kent, S. B. H.; Tam J. P.; Merrifield, R. B. Anal. Biochemistry
1981, 117, 147-157.
(48) Jentzen, W.; Unger, E.; Karvounis, G.; Shelnutt, J. A.; Dreybrodt, W.;
Schweitzer-Stenner, R. J. Phys. Chem. 1996, 100, 14184.
(49) Sybyl 7.3; Tripos: St. Louis, MO, 2007.
(45) Sieler, G.; Schweitzer-Stenner, R. J. Am. Chem. Soc. 1997, 119, 1720.
(46) Measey, T.; Schweitzer-Stenner, R. Chem. Phys. Lett. 2005, 408, 123.
(47) Hagarman, A.; Measey, T.; Doddasomayajula, R. S.; Dragomir, S.; Eker,
F.; Griebenow, K.; Schweitzer-Stenner, R. J. Phys. Chem. B 2006, 110.
(50) Lindahl, E.; Hess, B.; Van der Spoel, D. J. Mol. Model. 2001, 7 (8), 306-
317.
(51) Van Der Spoel, D. J. Comput. Chem. 2005, 26 (16), 1701-18.
9
J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007 13097

A R T I C L E S
Schweitzer-Stenner et al.
TIP4P water where the minimum distance between any peptide
atom and the edge of the box was 10 Å for â-strand conformers
and 13 Å for R-helices.52 This resulted in approximately the
same size box for both â-strand and R-helix conformations. The
system was energy minimized, heated up to 300 K in 50 K
increments where each increment was simulated for 20 ps at 1
fs timesteps. Production simulations (100 ns) were performed
at 300 K using GROMACS. The constant number of molecules,
pressure, and temperature (NPT) ensemble was used. Snapshots
were recorded every 10 ps. Force field parameters were defined
by the OPLS 2001 set.53 Dihedral angles were calculated using
the g-angle analysis program provided in the GROMACS
package. The simulations were performed on local computing
resources supplemented by the Teragrid supercomputer network.
Energy-Landscape Exploration. High-temperature (500 K)
30 ns simulations were performed to explore the energy
landscape of each peptide sequence. The initial conformations
of Ac-Aib-Ala-Ala-OMe, Ac-Ala-Ala-Aib-OMe, and Ac-Ala-
Aib-Ala-OMe were linear where the æ and ψ torsions were set
to +180°. The peptides were solvated, energy minimized, and
simulated at 500 K for 30 ns. The NPT ensemble was used.
Analysis Tools. Phi, psi, and omega torsions were calculated
for the 12 trajectories (3 peptide × 4 conformers/peptide). The
atoms specific for each dihedral were manually identified and
written to a file. Torsions from the first 3 ns of the simulations
were excluded from the calculations. The g-angle program
provided by GROMACS was used to calculate the angles. For
each trajectory, the torsions of each amino acid in a peptide
were binned into 90 × 90 squares, each covering an area of 4
× 4°.54 The percentages of torsions that fall within a square
were calculated. Squares with the highest percentages were
colored coded dark-red, and squares with the lowest percentages
were color-coded blue. White indicates that a square was not
populated.
All profiles were highly asymmetric. For Ac-Ala-Ala-Aib-OMe
and Ac-Ala-Aib-Ala-OMe (Figures 1 and 2), the first moment
of the isotropic Raman profile was clearly at higher wavenum-
bers than that of the respective IR band. This noncoincidence
is indicative of a predominant sampling of extended conforma-
tions in the upper left quadrant of the Ramachandran plot,58,59
As recently demonstrated by Torii, a large negative noncoin-
cidence (ΩIR-Ωiso < -10 cm-1, ΩIR: IR-peak position, Ωiso:
peak position of the isotropic Raman band) is indicative of a
peptide predominantly sampling the PPII trough in the Ram-
achandran plot.60 Ac-Aib-Ala-Ala-OMe, however, showed a
different behavior in that the noncoincidence between IR,
isotropic, and anisotropic Raman band peaks is substantially
smaller than observed for the other two peptides. The peak
position of the isotropic Raman band is downshifted and that
of the IR-band upshifted compared with the respective positions
in Ac-Ala-Aib-Ala-OMe and Ac-Ala-Ala-Aib-OMe. This could
be indicative of a substantial sampling of either helical or
turnlike structures58,59 and of a shift of intrinsic amide I
wavenumbers. The VCD spectra of Ac-Ala-Aib-Ala-OMe and
Ac-Aib-Ala-Ala-Ala-OMe (Figures 2 and 3) both showed a
purely negative signal, which were much less pronounced than
the negative VCD band earlier observed for Ala2-Lys-Ala and
Ala4.36,37 The spectra were qualitatively different from the
couplets observed for Ala3.61 The VCD spectrum of Ac-Ala-
Ala-Aib-OMe (Figure 1) did not show a clearly discernible
amide I signal.
Figure 4 exhibits the ECD spectra of the three peptides
measured at different temperatures between 10 and 80 °C. At
room temperature, their spectra all displayed a positive couplet,
but their amplitudes and wavelength positions were different.
We discuss these spectra in detail below.
Theoretical Concept for the Structure Analysis. The
excitonic coupling model for amide I has been described in detail
in several earlier papers.62,59 The model describes the amide I
band profiles of IR, isotropic Raman, anisotropic Raman, and
VCD in terms of nearest neighbor and non-nearest neighbor
excitonic coupling between the excited states of the local
oscillators63 and the dihedral angles φ and ψ of the residues
between the interacting amide I modes. One counts the first
central amide residue (starting at the N-terminal) as 1 and the
subsequent one as 2. It should be noted that the C-terminal
residue is not probed by the amide I′ band profile. Femtosecond
two-dimensional IR spectroscopy has shown for cationic tri-
alanine that vibrational coupling between the C-terminal car-
bonyl vibration and the adjacent amide I mode is negligibly
small.64 DFT calculations on Ac-Ala-Aib-Ala-OMe with deu-
terated amide groups in Vacuo reveal some small coupling
between the respective CO stretching vibration of the C-terminal
ester group and the adjacent amide I′. The corresponding
optimized geometry exhibits a PPII conformation for the first
alanine, a right-handed 310 for Aib, and an extended â-strand
for the C-terminal alanine. This mixing is likely to be even
weaker for the “real” peptide in aqueous solution since the
Results and Discussion
In this section of the paper, we first present the results of
our spectroscopic experiments on Ac-Ala-Ala-Aib-OMe, Ac-
Ala-Aib-Ala-OMe, and Ac-Aib-Ala-Ala-OMe and discuss them
in qualitative terms. The second part describes the simulation
of amide I′ band profiles in terms of different structural models.
The results of this analysis are then compared with the respective
ECD spectra. The last part describes the results of MD
simulations from which the energy landscapes of the peptides
were obtained.
Vibrational Spectra. Figures 1-3 show the amide I′ region
of the IR, isotropic Raman, anisotropic Raman, and VCD spectra
of Ac-Ala-Ala-Aib-OMe (Figure 1), Ac-Ala-Aib-Ala-OMe
(Figure 2), and Ac-Aib-Ala-Ala-OMe (Figure 3) in D2O. Using
D2O rather than H2O eliminated vibrational mixing between
solvent and amide I modes.45,55 D2O is known to slightly
destabilize hydrogen bonds compared with H2O,56 but this just
leads to a slight shift in equilibrium toward extended â-strands.57
(52) Jorgensen, W. L.; Madura, J. D. Mol. Phys. 1985, 56, 1381-1392.
(53) Kaminski, G. A. J. Phys. Chem. B 2001, 105 (28), 6474-6487.
(54) Hovmoller, S.; Zhou, T.; Ohlson, T. Acta Crystallogr., Sect. D 2002, 58,
768-76.
(55) Chen, X. G.; Schweitzer-Stenner, Asher, S. A.; Mirkin, N. G.; Krimm, S.
J. Phys. Chem. 1995, 99, 3074.
(56) Shi, Z.; Olson, C. A.; Kallenbach, N. R.; Sosnick, T. R. J. Am. Chem. Soc.
2002, 124 (13), 994-13 995.
(57) Eker, F.; Griebenow, K.; Schweitzer-Stenner, R. J. Am. Chem. Soc. 2003,
125, 8178-8185.
(58) Torii, H.; Tasumi, M. J. Raman Spectrosc. 1998, 29, 81-86.
(59) Schweitzer-Stenner, R. J. Phys. Chem. B 2004, 108, 16965.
(60) Torii, H. J. Phys. Chem. B 2007, 111, 5434.
(61) Eker, F.; Cao, X.; Nafie, L.; Schweitzer-Stenner, R. J. Am. Chem. Soc.
2002, 124, 14330.
(62) Schweitzer-Stenner, R. Biophys. J. 2002, 83, 523.
(63) Hamm, P.; Lim, M.; Hochstrasser, R. J. Phys. Chem. B 1998, 102, 6123.
(64) Woutersen, S.; Hamm, P. J. Phys. Chem. B 2000, 104, 11316-11320.
9
13098 J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007

Conformational Manifold of
R
-Aminoisobutyric Acid
A R T I C L E S
Figure 2. Amide I′ profiles (dotted) of the isotropic Raman, anisotropic
Raman, IR, and VCD spectrum of Ac-Ala-Aib-Ala-OMe in D2O. The solid
line results from a simulation based on a multistate (per residue) model,
encompassing a PPII, â-strand, and right- and left-handed 310-helix and a
disordered type IV and IV′ â-turn.
Figure 1. Amide I′ profiles (dotted) of the isotropic Raman, anisotropic
Raman, IR, and VCD spectrum of Ac-AAB-OMe in D2O. The solid line
results from a simulation based on a multistate (per residue) model,
encompassing a PPII, â-strand, and right- and left-handed 310-helix.
In our analysis, we considered five representative conforma-
tions, i.e., PPII (j ) 1), â (j ) 2), right-handed helical (j ) 3),
left-handed helical (j ) 4), and type IV/IV′ â-turn (j ) 5,6).
The designation type IV normally encompasses a large variety
of turn structures, which do not fit within the type I, II, III, and
V category. In our case it solely represented conformation
sampling in the bridge region between PPII and right-handed
helical regions in the Ramachandran plot. Recent MD simula-
tions by Bu¨rgi et al.30 suggested that Aib is able to adopt such
conformations. The Gibbs energy of PPII was set to zero.
Initially, we based our simulation on (φ,ψ)1 ) (-68°, 150°)
and (φ,ψ)2 ) (-125°, 115°) for alanine, which represent
respective distributions in the coil library of Avbelj and Baldwin9
and correspond to the 3JCRHNH constants reported by Shi et al.65
Coil libraries as well as molecular dynamics simulations for
alanine dipeptides generally exhibit helical distributions with a
maximum at (φ,ψ)3 ) (-65°, -30°), which we assumed as
representative for right-handed helical conformations.10 These
coordinates are closer to a 310 than to a canonical R-helical
experimentally obtained difference between the wavenumbers
of the considered modes (72 cm-1) is much larger than the
theoretically predicted one (33 cm-1).
Since each configuration of a peptide gives rise to an
individual set of band profiles, the total intensity of a peptide
ensemble is written as37
nc1,nc2
nck
3
(Iij j (Ω) exp-( G )/RT)
∑ ∑
∑
jk
1 2
k)1
I(Ω) ) i)1 j ,j )1
(2)
1
2
Z
where Gj is the Gibbs energy of the k-th residue with the
k
conformation j. The subscript i labels the excitonic state of the
amide I′ oscillators. Iij j (Ω) is thus the intensity profile of the
1 2
i-th excitonic state associated with the configuration {j1,j2} of
the tetrapeptide. R is the gas constant, T is the absolute
temperature, Z is the partition sum of the ensemble, and nc
1
and nc are the number of conformations considered for the two
central residues.
2
(65) Shi, Z.; Olson, C. A.; Rose, G. A.; Baldwin, R. L.; Kallenbach, N. R. Proc.
Natl. Acad. Sci. U.S.A. 2005, 102, 17964.
9
J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007 13099

A R T I C L E S
Schweitzer-Stenner et al.
[(Jk - Jâ,k) - ø3,k(JRr - Jâ,k) - ø4,k(JRl - Jâ,k)]
Jp,k - Jâ,k
ø1,k
)
ø2,k ) 1 - ø1,k - ø3,k - ø4,k
(3)
where øj,k is the mole fraction of the k-th residue in the j-th
conformation, Jk is the measured coupling constant of the k-th
residue, and Jp,k and Jâ,k (4.8 and 9.8 Hz) are the coil-library
coupling constants of the k-th residue for PPII and â as reported
by Shi et al.65 JRr ) 4.1 Hz and JRl ) 7.0 Hz are the
representative coupling constant of the above right- and left-
handed helical conformation, respectively. The experimentally
determined 3JCRHNH coupling constants of the respective alanine
residues are listed in Table 1. The mole fraction ø3,k and ø4,k of
these conformations have been used as an adjustable parameter
in the simulation for alanine and Aib. For the latter, we
additionally used ø5,k and ø6,k as free parameters. The mole
fractions were used to calculate the relative Gibbs energies of
the residues:
øj,k
Gj ) RT ln
(4)
k
[
]
ø1,k
that were used in eq 2 to calculate the intensity profiles.
Structure Analysis of Ac-Ala-Ala-Aib-OMe. Since the
excitonic coupling depends solely on the dihedral angles of the
residues between the amide I′ modes of interacting peptide
bonds, the band profiles solely reflect the conformations of the
two alanine residues. First, we needed estimated values for the
intrinsic wavenumbers of the three coupled amide I′ vibrations
in order to carry out the simulation of the band profiles. To
this end a self-consistent spectral decomposition of Raman and
IR band profiles was performed61 that yielded the wavenumber
values of three subbands listed in Table 2. We also measured
the (IR) amide I band position for Ac-Ala-OMe in D2O and
obtained a value of 1629 cm-1, which suggests that the 1628
cm-1 subband of Ac-Ala2-Aib-OMe was assignable to the
N-terminal peptide group. A comparison with the subbands of
cationic and anionic tetraalanine36 suggests that the band at 1661
cm-1 and 1646 cm-1 were assignable to the central and
C-terminal peptide group, respectively. To calculate IR absorp-
tion and rotational strength (for VCD) in absolute units, the
transition dipole moments of dialanine were used for all peptide
groups.68 The C-terminal amide I mode of small peptides
generally exhibits some intrinsic rotational strength which is
not due to the coupling with another oscillator. It most likely
results from some electronic effect of the C-terminal carbonyl
group.36,61 In contrast to what was observed for Ac-Ala-OH,
however, the amide I′ VCD signal of Ac-Aib-OH was very weak
and barely discernible from noise (data not shown). The intrinsic
contribution of the C-terminal Aib of Ac-Ala-Ala-Aib-OMe to
the VCD band profile was, therefore, neglected. One would be
tempted to check whether the data could be reproduced solely
based on the crystal structure of Boc-Gly-Ala-Aib-OMe, a
somewhat distorted type III â-turn. Indeed, this yielded a
reasonable reproduction of Raman and IR band profiles but
overestimated the VCD signal (data not shown). Moreover, this
modeling was inconsistent with the 3JCRHNH coupling constants
Figure 3. Amide I′ profiles (dotted) of the isotropic Raman, anisotropic
Raman, IR, and VCD spectrum of Ac-Aib-Ala-Ala-OMe in D2O. The solid
line results from a simulation based on a multistate (per residue) model,
encompassing a PPII, â-strand, and right- and left-handed 310-helix and a
disordered type IV and IV′ â-turn.
conformation. In the following, we refer to this structural model
as the “coil-library model.” For Aib, we assumed the type III
and III′ â-turn coordinates representing right-handed ((φ,ψ)3 )
(-60°, -30°)) and left-handed ((φ,ψ)4 ) (60°, 30°)) 310-like
conformations. Alternatively, we considered the possibility that
Aib samples predominantly R-helix-like conformations, which
we represented by (φ,ψ)3 ) (-57°, -47°) (right-handed) and
(φ,ψ)4 ) (57°, 47°) (left-handed). The type IV â-turn manifold
was represented by (φ,ψ)5 ) (-60°, 10°), and the corresponding
IV′ enantiomers were represented by (φ,ψ)6 ) (60°, -10°). This
conformation could be described as a somewhat distorted C7
conformation, which short alanine-based peptides can adopt in
nonaqueous solutions.66
3
For the alanine residues, we used the JCRHNH constants of
the considered conformations obtained by employing a modified
Karplus equation67 with our experimentally determined coupling
constants and determined the respective mole fraction of these
conformations as follows:
(66) Jalkanen, K. J.; Suhai, S. Chem. Phys. 1996, 208, 81.
(67) Vuister, G. W.; Bax, A. J. Am. Chem. Soc. 1993, 115, 7772.
(68) Measey, T.; Hagarman, A.; Eker, F.; Griebenow, K.; Schweitzer-Stenner,
R. J. Phys. Chem. B 2005, 109, 8195.
9
13100 J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007

Conformational Manifold of
R
-Aminoisobutyric Acid
A R T I C L E S
Figure 4. Temperature-dependent ECD spectra of Sc-Ala-Ala-Aib-OMe, Ac-Ala-Aib-Ala-OMe, and Ac-Aib-Ala-Ala-OMe measured from 10 to 80 °C in
increments of 10 °C. The arrow indicates the change with increasing temperature. (Inset) Difference ECD spectrum obtained by subtracting the 10 °C from
the 80 °C spectrum.
3
Table 1. JCRNH Constant of Alanine Residues
Table 2. Wavenumbers of Three Subbands Obtained from a
Self-Consistent Spectral Decomposition of the Amide I′ Band
Profile
residue 1
residue 2
Ac-Aib-Ala-Ala-OMe
Ac-Ala-Aib-Ala-OMe
Ac-Ala-Ala-Aib-OMe
5.1 Hz
Ac-Aib-Ala-
Ala-OMe
Ac-Ala-Aib-
Ala-OMe
Ac-Ala-Ala-
Aib-Ome
6.2 Hz
5.5 Hz
5.5 Hz
Ω1 [cm-1
1659Ω2 [cm-1
Ω3 [cm-1
]
1631
1649
1667
1626
1637
1659
1628
1646
1661
]
]
and also with the ECD spectrum in Figure 4, which was clearly
indicative of a substantial PPII fraction at room temperature.
In a second attempt a three-state model for the two alanine
residues was invoked, which was recently used to reproduce
the amide I′ band profiles of short alanine-based peptides. It
considers the coexistence of the aforementioned PPII (j ) 1),
â-strand (j ) 2), and right-handed helix (j ) 3).37 It should be
noted that only the molar fraction of the helical conformation
was used as a free parameter. Thus, the simulated band profiles
were close to the experimental ones, but we did not judge the
agreement as satisfactory, particularly for VCD and anisotropic
Raman spectra. This prompted us to allow for slight differences
between the PPII coordinates of the alanine residues, as done
earlier for tetraalanine.36 The band profiles shown as solid lines
in Figure 1 were produced with the PPII coordinates (φ,ψ)1(j
) 1) ) (-63°, 150°) and (φ,ψ)2(j ) 1) ) (-70°, 140°) that
9
J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007 13101

A R T I C L E S
Schweitzer-Stenner et al.
increased population of â-strand-like conformations. At 80 °C,
the ECD spectrum shows two negative CD bands with a
minimum above 220 nm. The high-temperature spectra clearly
resemble what Drake et al. obtained when they measured the
CD spectrum of poly-L-lysine in aqueous solution at neutral
pH and high temperatures.72 They interpreted their spectra as
reflecting a mixture of PPII and helical conformations. Thus, it
seems that contrary to expectation the alanine residues sample
helical conformations at higher temperatures. The temperature
dependence of ∆ꢀ at 215 nm is nearly linear, which rules out
any involvement of cooperativity (data not shown).
Structure Analysis of Ac-Ala-Aib-Ala-OMe. The band
profiles of this peptide are shown in Figure 2. The decomposi-
tion into three subbands assignable to individual peptide groups
yielded the wavenumbers listed in Table 2. The band at 1626
cm-1 was again assigned to the N-terminal A. The highest
wavenumber band results from the central peptide, and the
central band was assignable to the C-terminal peptide. The
3JCRHNH coupling constant of the N-terminal alanine was 6.1
Hz (Table 1), slightly higher than the corresponding value for
Ac-A2B-OMe, indicating a reduced PPII content. We used the
recently observed intrinsic magnetic transition moments of
amide I′ of neutral Ac-Ala for the C-terminal amide I′ mode.37
As a first step, we assumed three conformations (PPII, â, right-
handed helix) for alanine and a mixture of right-handed and
left-handed helices for Aib. We performed simulations for both
R- and 310-helices for different sets of molar fractions for the
considered conformations but did not obtain a satisfactory
agreement with all experimental profiles. All simulated spectra
underestimated the intensity of the isotropic Raman band profile
at low wavenumbers and the IR absorption at high wavenum-
bers. This led us to the conclusion that a conformation had to
be considered for which the coupling constant between adjacent
amide I modes was negative. The contour plot reported by Torii
and Tasumi58 suggests that this applies solely to conformations
assignable to the bridge region connecting the PPII and helical
troughs in the Ramachandran plot.58 MD simulations of Aib-
based peptides in DMSO by Bu¨rgi et al. indeed suggested that
Aib can sample this region and to a lesser extent also the
corresponding enantiomers in the lower right quadrant of the
Ramachandran plot.30 This lead us to consider the aforemen-
tioned type IV/IV′ (j ) 5,6) for the Aib-residue. The solid line
in Figure 2 represents a simulation performed for the following
molar fractions: ø1 ) 0.41, ø2 ) 0.49, and ø3 ) 0.1 for the
N-terminal alanine; ø3 ) 0.55, ø4 ) 0.21, ø5 ) 0.12, and ø6 )
0.11 for Aib. A slightly modified 310-like conformation ((φ,ψ)1,2-
(j ) 3) ) (-55°, -30°), (φ,ψ)1,2(j ) 4) ) (55°, 30°) was
assumed for conformations 3 and 4. The agreement with the
experimental data was satisfactory. Replacing the two 310
conformations by corresponding R-helical conformations led to
deterioration of the quality of the IR and Raman profiles. It
produces a W-shaped VCD signal with two negative and one
positive band, not in agreement with the experimental data. We
therefore conclude that 310 was the preferred conformation for
Aib in Ac-Ala-Aib-Ala-OMe. Figure 6 displays the structure
of one of the dominant peptide sequences with PPII(Ala)-right-
handed 310(Aib)-PPII.
Figure 5. Structure of Ac-Ala-Ala-Aib-OMe with (φ,ψ)1 ) (φ,ψ)
(-60°, 150°) (PPII) and (φ,ψ)3 ) (-60°, -30°) (right-handed 310).
)
2
were only slightly different from canonical PPII-values.69 We
like to reiterate in this context that these conformations were
representatives for distributions in the Ramachandran plot. We
found that an addition of even small fractions of the helical
conformation led to deterioration rather than improvement in
the simulation spectra. The same conformational distribution
was obtained for both residues, i.e., molar fractions of 0.78 for
PPII and 0.22 for â-strand. This was indicative of a very
pronounced PPII propensity of alanine and resembled the results
obtained for cationic tetraalanine in D2O.36,38 Hence, we can
conclude that the C-terminal Aib does not have a significant
influence on the propensity of the two alanine residues. Figure
5 shows the most probable conformation with the sequence PPII-
PPII-310. We assumed a right-handed 310 conformation for Aib,
but of course nothing can be said about which enantiomer was
predominant. The weak VCD signal (compared with that of
tetraalanine) can be explained by the substantial alkylation-
induced downshift of the N-terminal amide I wavenumber (from
ca. 1675 cm-1 in trialanine to 1628 cm-1 in Ac-Ala-Ala-Aib-
OMe).
The ECD spectrum of the peptide was measured as a function
of temperature as shown in Figure 4. The spectra taken at room
temperature display a negatively biased couplet that is normally
indicative of PPII.70,71 However, both the negative and the
positive signals were weaker than what was observed for
tetraalanine, for which we obtained similar PPII fractions.36,37
This is most likely due to the contribution of the C-terminal
Aib, which can be expected to adopt a mixture of right- and
left-handed helical conformations. As shown by Hagarman at
al., the C-terminal residue of small peptides significantly
contributes to the ECD signal, which most likely reflects
coupling between the π f π* transitions of the peptide groups
and the C-terminal ester group.47 Interestingly, the difference
of the spectra taken at 80 °C and 10 °C shown in the inset of
Figure 4 shows the typical minimum at 215 nm but does not
display the strong positive signal at 195 nm observed for tri-
and tetraalanine,36,61 which together are indicative of an
Figure 4 exhibits the ECD spectrum of Ac-Ala-Aib-Ala-OMe
measured between 10 and 80 °C. At 10 °C the couplet is rather
(69) Cowan, P. M.; McGavin, S. Nature 1955, 176, 501.
(70) Tiffany, M. L.; Krimm, S. Biopolymers 1968, 6, 137.
(71) Sreerama, N.; Woody, R. W. Biochemistry 1994, 33, 10022.
(72) Drake, A. F.; Siligardi, G.; Gibbons, W. A. Biophys. Chem 1988, 31, 143.
9
13102 J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007

Conformational Manifold of
R
-Aminoisobutyric Acid
A R T I C L E S
dues.72 The contribution from the latter is certainly less
pronounced than that in the spectrum of Ac-Ala-Ala-Aib-OMe.
The temperature dependence of ∆ꢀ is less linear than that
obtained for Ala-Ala-Aib but still not indicative of cooperativity
(data not shown).
Structure Analysis of Ac-Aib-Ala-Ala-OMe. The band
profiles of this peptide are shown in Figure 3. The decomposi-
tion into three subbands assignable to individual peptide groups
yielded the wavenumbers listed in Table 2. The low wavenum-
ber band at 1631 cm-1 was assigned to the N-terminal Aib, as
suggested by the corresponding amide I′ wavenumber of Ac-
Aib, i.e., 1629 cm-1. The highest wavenumber band resulted
again from the central peptide,36 and the central band was
assignable to the C-terminal amide. Again, we used the magnetic
transition moment obtained for the amide I′ mode of Ac-Ala
for the C-terminal amide I′ mode.37 As argued below, however,
these numbers did not reflect the complex composition of the
Figure 6. Structures of Ac-Ala-Aib-Ala-OMe with (φ,ψ)1 ) (-60°, 150°)
(PPII), (φ,ψ)2 ) (-60°, -30°) (right handed 310), and (φ,ψ)3 ) (-60°,
150°) (PPII).
3
band profile. The JCRHNH coupling constant of the central
alanine residue was 5.0 Hz (Table 1). This low value suggested
either a substantial fraction of PPII or a significant population
of a helical conformation.
broad and symmetric thus deviating from the negatively biased
canonical PPII signal.70,71 Again, only the maximum decreased
with increasing temperature, which gives rise to the broad
minimum at 215 nm in the difference spectrum ∆ꢀ(80 °C) -
∆ꢀ(10 °C) (inset of Figure 4). Generally, the observed spectrum
is indicative of some type of left-handed turn. That would be
in perfect agreement with the crystal structure of Ac-Ala-Aib-
Ala-OMe34 that exhibited both alanines in a PPII and the central
Aib-residue in a left-handed 310 conformation. Analysis of amide
I′ profiles, however, yielded a surplus of right-handed 310 for
Aib. We performed simulations for different ratios of the two
Aib enantiomeric conformers and found that even a modestly
higher fraction of left-handed 310 reduces the VCD signal
significantly. If the left-handed conformation was predominant,
the VCD spectrum would exhibit a positive Cotton effect. It
should be noted in this context that a specific interpretation of
the ECD spectrum is difficult in view of the fact that the
C-terminal alanine, likely to adopt a mixture of PPII and
â-strand, also contributes to the ECD signal.47 Generally, ECD
spectra are less of a local probe than VCD spectra. For short
peptides they reflect electronic interactions between all π-elec-
tron moieties including the C-terminal carbonyl. Moreover,
mixing with high energy σ f σ* of side chains has been
proposed to contribute to the spectrum in the 180-230 nm
region.73,74 In the case of Ac-Ala-Aib-Ala-OMe, the peptide
adopted multiple conformations. If one considers the representa-
tive type IV turn also as left-handed, one obtains a fraction of
nearly 0.2 for those peptides with all their residues adopting a
left-handed conformation. Moreover, sequences with mixtures
of residues adopting PPII, â, and right-handed 310, which
account for a fraction of 39%, overall exhibit a left-handed
helicity. The C-terminal Aib certainly samples the right- as well
as the left-handed 310-trough of the Ramachandran plot. Hence,
we concluded that the ECD spectrum did not contradict the
result of our overall structure analysis.
The symmetry of the band profiles and the alignment of its
peaks indicated that the peptide sampled predominantly helical
conformations.58,59 As a consequence, a simulation based on
the assumption that the central alanine samples predominantly
PPII with some admixture of â-strand and helical conforma-
tion37,38 yielded band profiles quite distinct from the experi-
mentally observed ones in Figure 3. None of the experimental
band shapes and positions observed were reproduced. The band
profiles in Figure 3 were then simulated by allowing various
fractions of right-handed and left-handed helical conformations
for the central alanine residue. However, none of these attempts
were successful either, in that they all aligned the bands at
significantly higher wavelengths than indicated by the experi-
mental data. This is understandable since helical conformations
are known to have their amide I intensities predominantly in
the band associated with the in-phase combination of coupled
oscillators.58,59 In order to account for this discrepancy, we
assumed that the intrinsic wavenumbers of the individual amide
I′ modes shifted down if both residues Aib and Ala adopted
either a right-handed or left-handed 310 helical structure. These
peptide conformations would be stabilized by hydrogen bonding
between the N-terminal carbonyl and C-terminal amide group
(cf. Figure 7 for the right handed conformation). A very
satisfactory simulation shown in Figure 3 was eventually
achieved by this procedure. The necessary wavenumber down-
shifts for peptides with both Aib and Ala in a right-handed 310
conformation, were 20 cm-1 and 33 cm-1 for the terminal amide
I′ modes and the central amide I′, respectively. Such a downshift
of the intrinsic amide I wavenumber upon a PPII f 310 transition
can indeed be inferred from theoretical calculations of Wout-
ersen et al.75 For Aib, we obtained the molar fractions ø3 )
0.45, ø4 ) 0.35, and ø5 ) 0.20, whereas ø1 ) 0.14, ø2 ) 0.06,
ø3 ) 0.6, and ø4 ) 0.2 were obtained for the central alanine
residue. The molar fractions of the helical fractions of Aib were
somewhat uncertain in that the inverse ratio still yields satisfac-
tory simulations. In the framework of this model, however, it
is clear that both enantiomers of Aib were substantially
The minimum at 190 nm maintained its strength and red-
shifted with rising temperature. The positive CD band at 216
nm was concomitantly reduced. A comparison with the poly-
L-lysine spectra measured by Drake et al. suggests a mixture of
PPII, â-strand, and helical conformation for the alanine resi-
(73) Woody, R. W. J. Chem. Phys. 1968, 49, 4797.
(74) Woody, R. W.; Seerama, N. J. Chem. Phys. 1999, 111, 2844.
(75) Woutersen, S.; Pfister, R.; Hamm, P.; Mu, Y.; Kosov, D. S.; Stock, G. J.
Chem. Phys. 2002, 117, 6833.
9
J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007 13103

A R T I C L E S
Schweitzer-Stenner et al.
in the Ramachandran plot to ensure that a comparison of the
amide I wavenumbers is at least of qualitative relevance. The
vibrational analysis yielded substantial vibrational coupling
between the amide I′ modes for both peptides investigated.
Compared with Ac-Ala-Aib-Ala-OMe, the wavenumbers of the
amide I′ vibrations mostly associated with the central and the
C-terminal residues of Ac-Aib-Ala-Ala-OMe are reduced by 10
cm-1, whereas the amide I′ of the N-terminal residue shifts up
by 19 cm-1. The latter certainly result from intrapeptide
hydrogen bonding between the CO of residue 1 and the
C-terminal amide group. In aqueous solution the situation would
be different owing to additional hydrogen bonding to H2O,
which generally causes a substantial amide I downshift.76
Altogether, these results qualitatively support the assumption
we made regarding the intrinsic amide I′ wavenumbers of Ac-
Aib-Ala-Ala-OMe.
An initial inspection of the room-temperature ECD spectrum
in Figure 4 reveals a weak PPII signal. However its minimum
appears at 200 nm that would be characteristic of a left-handed
turn. Moreover, this minimum did not shift or decrease with
increasing temperature, which would not be expected if it
reflected a PPII conversion to â-strand. The difference spectrum
∆ꢀ(80 °C) - ∆ꢀ(10 °C) (inset of Figure 4) shows a rather large
minimum at 216 nm (-0.8 M-1 cm-1 per residue), which is
again not accompanied by a maximum below 200 nm. It could
be indicative of a right-handed 310-helix, but the (room
temperature) maximum at 215 nm clearly reflects some left-
handed turn or PPII conformation. It decreased with increasing
temperature, and a comparison with the poly-L-lysine spectra
of Drake et al. suggests again a mixture of PPII, â-strand, and
helical conformation.72 It is arguable that a mixture of right-
handed, left-handed, and mixed structures could give rise to
the observed ECD spectrum at room temperature. In this context,
we have to recall again the contribution of the C-terminal alanine
whose structural behavior could not be explored. The temper-
ature dependence of ∆ꢀ departs from linearity above 50 °C (data
not shown), but the statistical errors of the data points are too
large to allow any conclusions concerning cooperativity.
Energy-Landscape Exploration. High-temperature simula-
tions can cross energy barriers and provide accurate estimates
of the energy landscape of peptide sequences.77,78 Simulations
of Ac-Aib-Ala-Ala-OMe, Ac-Ala-Aib-Ala-OMe, and Ac-Ala-
Ala-Aib-OMe at 500 K showed that Aib samples both the right-
and left-handed R-helix areas of the Ramachandran plot (Figure
8). The alanine residues sample a much larger region compared
to Aib. While the æ angles of the alanine residues are mostly
limited to the -180° to -60° region, the ψ angles sample the
entire region between -180° and 180°. Some alanine conforma-
tions also sampled the 60° phi region.
Figure 7. Structures of Ac-Aib-Ala-Aib-OMe with (a) (φ,ψ)1 ) (-60°,
-30°), (φ,ψ)3 ) (-60°, -30°) (both right-handed 310), and (φ,ψ)3 ) (-60°,
150°) (PPII).
populated. We tried the same simulation based on the assump-
tion that the residues adopt RR rather than 310, but this did not
reproduce the VCD signal and the agreement with IR and
Raman profiles was less satisfactory.
Being concerned by the large number of parameter changes
we had to invoke for the satisfactory simulations shown in
Figure 3, we tried a more simplistic model by assuming that
even in the absence of an overall helical structure the wave-
number of the amide I′ mode of the central peptide group was
downshifted due to the fact that Aib adopted a helical
conformation. Thus, we obtained a satisfactory reproduction of
the Raman and IR band profiles for a -24 cm-1 shift and slight
adjustments for the wavenumbers of the terminal amide I′
modes. The VCD signal, however, was less well reproduced.
The corresponding molar fractions for alanine were ø1 ) 0.62,
ø2 ) 0.13, ø3 ) 0.2, and ø4 ) 0.05. We would not totally dismiss
this model just based on the somewhat less satisfactory
agreement with the experimental data because of the smaller
number of adjustable parameters. However, if we did not invoke
an overall helical conformation of the peptide it is difficult to
understand why Aib would per se influence the amide I′ mode
only in Ac-Aib-Ala-Ala-OMe, but not in Ac-Ala-Aib-Ala-OMe,
for which our data (i.e., the measured band profiles) rule out
this possibility.
In order to search for further evidence in support of the
assumed influence of the residue conformation on the intrinsic
amide I wavenumber we carried out normal mode calculations
for Ac-Ala-Aib-Ala-OMe and Ac-Aib-Ala-Ala-OMe in Vacuo
based on a DFT calculation using the B3LYP /6-31G** level.
The initial dihedral angles chosen for the geometry optimization
exhibited (φ,ψ)N ) (-60°, 150°), (φ,ψ)Ce ) (-60°, -30°), and
(φ,ψ)C ) (-60°, 150°) for Ac-Ala-Aib-Ala-OMe and (φ,ψ)N
) (-60°, -30°), (φ,ψ)Ce ) (-60°, -30°), and (φ,ψ)C ) (-60°,
150°) for Ac-Aib-Ala-Ala-OMe (N: N-terminal, Ce: central
residue, C: C-terminal) The optimization process yielded
backbone conformations with (φ,ψ)N ) (-81°, 90°), (φ,ψ)Ce
) (-59°, -34°), and (φ,ψ)C ) (-151°, 178°) for Ac-Ala-Aib-
Ala-OMe and (φ,ψ)N ) (-62°, -33°), (φ,ψ)Ce ) (-85°, 0°),
and (φ,ψ)C ) (-98°, 80°). Since we did not consider solvation,
these are not exactly the conformations we derived from our
experimental data, but the conformation of the central alanine
in Ac-Aib-Ala-Ala-OMe is close enough to the helical trough
To determine if simulations at 300 K could sufficiently sample
the estimated landscape at 500 K, we performed five 30 ns
simulations of Ac-Aib-Ala-Ala-OMe where the initial conform-
ers were 310-helix, R-helix, â-strand, linear strand, and a
structure with random torsion angles. These 300 K simulations
showed that the energy barrier for Aib between the right-handed
helix (æ ) -65°, ψ ) -45°) and the left-handed helix (æ )
65°, ψ ) 45°) regions was too high to be sampled within the
(76) Williams, S.; Causgrove, T. P., Gilmanshin, R.; Fang, K. S.; Callender, R.
H.; Woodruff, W. H.; Dyer, R. B. Biochemistry 1996, 35, 691.
(77) Day, R. J. Mol. Biol. 2002, 322 (1), 189-203.
(78) Li, A.; Daggett, V. Proc. Natl. Acad. Sci. U.S.A. 1994, 91 (22), 100-4.
9
13104 J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007

Conformational Manifold of
R
-Aminoisobutyric Acid
A R T I C L E S
Figure 8. Ramachandran plots of Ac-Ala-Ala-Aib-OMe, Ac-Ala-Aib-Ala-OMe, and Ac-Aib-Ala-Ala-OMe snapshots from 30 ns, 500 K simulations. The
initial structures were linear peptides with æ, ψ, and ω torsions set to 180°. Red indicates the highest percentage of the population, and white indicates that
a region was not populated.
simulation time scale (data not shown). The alanine residues
were able to sample the same regions as those found at 500 K.
The barrier for Aib between helical regions was much higher
than that of alanine due to Aib’s extra methyl group on the
R-carbon. Because of this limitation, it was necessary to seed
the Aib æ torsion of the initial conformers to 58° or -58° in
separate simulations. Seeding the torsions forced the Aib residue
to sample either the right-handed or left-handed helical region,
while the alanine residues sampled the allowed regions of the
Ramachandran plot.
Transition Pathway between Right- and Left-Handed
Helical Regions. During the 500 K simulations, Aib underwent
transitions between left-handed and the right-handed helical
regions. The æ torsion of Aib was mostly limited to the (60°
region, but it also explored the (180° region. The ψ torsion,
on the other hand, rotated more freely in all three peptide
sequences (Figure 8). The reaction pathways for transitions from
the left-handed helical region to the right-handed helical region
were mapped for the Ac-ABA-OMe sequence. Figure 9 shows
the æ torsion of Aib as it moves from the lower energy 60°
state to the higher energy 180° intermediate state. Snapshots
were recorded every 50 fs to capture the short-lived transitional
conformers. Four snapshots spanning 200 fs were observed
during the transition. From the 180° intermediate state, Aib can
fall into either the left- or right-handed helical conformations.
Two transition pathways were layered on top of a potential
energy contour plot of acetyl-aminoisobutyric N-methylamide
(Ac-Aib-NMA) (Figure 10). The potential energy contour plot
was generated by systematically scanning the æ and ψ torsions
of Aib in Ac-B-NMA followed by energy minimization in an
implicit solvent (GB/SA). Transition A in Figure 10 shows a
pathway taken by Aib as its æ transitions from 60° to (180° to
-60°. It starts at an energy minimum in the left-handed helix
regions (60°, 30°) and climbs over the lowest-energy barrier
(approximately 3.5 kcal/mol) (120°, -20°) to reach the inter-
mediate states at æ ) (180°. From the intermediate state, Aib
9
J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007 13105

A R T I C L E S
Schweitzer-Stenner et al.
Figure 9. Change of Aib Ψ torsion angle in Ac-Ala-Aib-Ala-OMe during a concerted transition between right- and left-handed helical conformers of Aib
residue at time 9.384 ns during a 30 ns 500 K Ac-Ala-Aib-Ala-OMe simulation. The æ angle moved from the left-handed helix torsion of 60 to the higher
energy torsion at (180.
Figure 10. Pathways in Aib Φ,Ψ-space for transitions between right- and left-handed helical conformers of the Aib residue in Ac-Ala-Aib-Ala-OMe.
Pathways A (Φ ) 60 to ( 180 to -60) and B (Φ ) -60 to ( 180 to 60) are layered on top of a potential energy contour plot of Ac-Aib-NMA. Each b
represent an Aib conformation as it undergoes a transition between two states.
climbs over an energy barrier that is approximately 4.0 kcal/
mol (0.5 kcal/mol higher than the lowest-energy barrier) to reach
the right-handed helix region (-60°, 30°). Two transitions from
180° to -60° show Aib climbing over the lowest-energy barrier
(approximately 3.5 kcal/mol) to reach the (-60°, 15°) region
(data not shown). Transition B in Figure 10 shows a pathway
taken by Aib as its æ moves from -60° to (180° to 60°. It
starts at the right-handed helical region (-60°, -30°) and climbs
over the lowest-energy barrier (3.5 kcal/mol) (-120°, 20°) to
reach the intermediate states at ((180°, 30°). From the
intermediate state, Aib climbs over an energy barrier at (120°,
-20°) to reach the left-handed helix region (60°, 30°). The ψ
angles range mostly from -60° to 60° in the transition pathways
shown in Figure 10. The ψ angles in other pathways stay
between 120° and 180° as the æ angle rotates from 60° to (180°
to -60° (data not shown). The potential energy plot of Aib in
Figure 10 shows an accessible pathway for ψ to rotate between
(60° to (180°. When the ψ angle is between 120° and 180°,
the energy barrier for æ to rotate from 60° to 180° is higher
than 5.0 kcal/mol.
All observed æ angle transitions between 60° and -60° pass
through an intermediate state for æ as æ increased towards
(180°. A æ value of zero was not observed in any of the three
500 K peptide simulations. The potential energy profile of Ac-
Aib-NMA indeed showed a lowest-energy path between the two
R-helical regions through the (180 region of æ (Figure 10).
Similar to our 500 K simulation, the potential energy surface
surrounding the æ ) 0 region was too high to be plausibly
explored.
300 K Simulations. The most populated regions of the
Ramachandran plot of each residue in the three sequences are
listed in Table 3. Their respective Ramachandran plots are
shown in Figures S1 to S3 in the Supporting Information. The
results for all three peptides were independent of whether the
initial conformation was an R-helix or a â-strand. The æ angle
of Aib impacted the neighboring amino acids’ conformational
preference for all three sequences with Ac-Aib-Ala-Ala-OMe
showing the largest differences.
Table 3 lists the center of the most populated regions of the
Ramachandran plot for the three peptides. Listed in column A
9
13106 J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007

Conformational Manifold of
R
-Aminoisobutyric Acid
A R T I C L E S
Table 3. Phi and Psi Torsions of Individual Residues in the Three Peptide Sequencesa
A:
R
-helix ΦB ) −58
B:
â
-strand ΦB ) −58
C:
R
-helix ΦB ) +58
D:
â
-strand ΦB ) +58
res 2
res 1
res 2
res 3
res 1
res 2
res 3
res 1
res 2
res 3
res 1
res 3
Ala-Ala-Aib -80, 150 -80, 150
-65, -45 -80, 150 -80, 145
-65, -45 -80, 150 -80, 145
65, 45
-80, 150 -80, 145
65, 45
-80, -25 -80, -30 -65, 155 -80, -25 -80, -30 -65, 155 -80, -25 -75, -25 65, -165 -80, -25 -80, -25 65, -165
Ala-Aib-Ala -80, 140 -65, -30 -80, -50 -80, 140 -65, -30 -80, -50 -75, 130 65, 25
-75, -15 -80, 120 -75, -15 -80, 120 -80, -20
-80, 130 -75, 130 65, 25
-80, -45 -80, -20
-80, 130
-80, -45
Aib-Ala-Ala -65, -25 -80, -20 -80, 130 -65, -25 -80, -20 -80, 130 65, 30
-80, 150
-80, -25 -80, -50
-150, 160
-80, 130 65,30
-80, 150
-80, 130
-80, -25 -80, -50
-150, 160
-140, 20
-80, 150
-140, 155
-80, -50
-140, 60
-140, 20
-80, 150
-140, 155
-80, -50
-140, 60
a The conformations of the initial structures are indicated by the top row. ΦB ) -58 indicates that the phi torsion of Aib was set to -58 in the initial
conformation. The backbone torsions are listed in (æ,ψ) pairs with the most populated region placed on top of each box.
are the æ-ψ torsions of the three peptides where the initial
conformer for the simulations were R-helices. Listed in column
B are the torsions where the initial conformers were â-strands
except that the æ torsion of Aib was set to -58°. Listed in
columns C and D are torsions where the æ torsion of Aib was
set to +58° and the rest of the residues were either R-helices
(C) or â-strands (D). When the initial æ torsion of Aib was set
to -58°, results for the three peptides where the initial conformer
was an R-helix (A) were nearly identical to those where the
initial conformer was a â-strand (B). Similarly, when the initial
æ torsion of Aib was set to +58°, results for the three peptides
were nearly identical regardless of the conformations of the
initial structures. This suggests that the simulations had con-
verged. While the simulation results were nearly identical when
the æ torsion of Aib was set to either -58° or +58°, the alanine
conformations depended on the initial handedness of the Aib
residue. The degree of dependency was greatest for Ac-Aib-
Ala-Ala-OMe.
The best simulation of the amide I′ profiles were observed by
confining ourselves to PII and â-strand, for which we obtained
molar fractions of 0.78 and 0.22, respectively. The results of
the MD simulations, however, indicate a significant sampling
of (right-handed) helical conformations. We performed simula-
tions for different mole fractions of a representative right-handed
310-helix and found that the simulations deviate substantially
from the experimental profile for helical fractions above 0.1. It
should be mentioned in this context that MD simulations based
on a variety of force fields generally overestimate the sampling
of helical conformations for alanine.80-82
Ac-Ala-Aib-Ala-OMe. When the Aib2 torsions were centered
at (65°, 25°), the majority of Ala1 conformations were centered
at (-75°, 130°) (Figure S2). A small percentage of conforma-
tions occupied the (-80°, -20°) helical region. When Aib2
torsions were centered at (-65°, -30°), Ala1 populated the
(-80°, 140°) and (-75°, -15°) regions of the Ramachandran
plot. Aib induces a higher percentage of Ala1 helical conforma-
tions when its æ and ψ were centered at the (-65°, -30°)
region. Ala1 conformation depended on the handedness of Aib2.
Ala3 conformations populated the (-80°, 120°) and (-80°,
-50°) regions and were less dependent on Aib handedness. The
population shifted toward the (-80°, -50°) region when the
Aib2 torsions were centered at (-65°, -30°). Similar to Ac-
Ala-Ala-Aib-OMe, the most populated regions explored by the
500 °K simulation were also explored by the four independent
300 °K simulations.
Ac-Ala-Ala-Aib-OMe. Ala1 mainly populated the two re-
gions at (-80°, 150°) and (-80°, -25°), but some conforma-
tions also explored the (-150°, 160°) region (Figure S1). Ala1
conformation was not influenced by the initial handedness of
Aib3. The conformation of Ala2, however, was slightly depend-
ent on the initial æ torsion of Aib3. When the æ torsion of Aib3
was centered at +65°, the ψ torsion of Ala2 shifted toward the
region at 145°, increasing its polyproline II population. Like
Ala1, Ala2 also explored the (-150°, 160°) region of the
Ramachandran plot. There was also a small population centered
at (-140°, 30°). Aib3 mainly populated the (-65°, -45°) and
(65°, 45°) region, corresponding to the right-handed and left-
handed helix, respectively. Two smaller regions at (-65°, 165°)
and (65°, -165°) were also populated. The æ torsion was more
flexible because Aib3 was at the C-terminus, and there was no
additional amino acid side chain to restrict its rotation. The most
populated regions explored by the 500 K simulation were also
explored by the four 300 K simulations indicating that all low
energy conformations were well sampled in the four 300 K
simulations.
The results of these simulations are in excellent agreement
with the structural models used for the simulation of the
corresponding amide I′ profiles. For the central Aib, sampling
of both helical enantiomers is suggested by theory and experi-
ment. For the N-terminal alanine, PPII is clearly the dominant
conformation with some sampling of helical and â-strand
conformations. Apparently, the extent to which these conforma-
tions are sampled depends on the initial conditions of the
simulation (Figure S2). In accordance with Bu¨rgi et al.30 we
obtained some minor sampling of the (type IV) bridge region
slightly above ψ ) 0°, which again agrees at least qualitatively
with what we inferred from the spectroscopic data.
The results of the above simulations are apparently in good
agreement with the peptide’s structural manifold inferred from
the amide I′ band profile of Ac-Ala-Ala-Aib-OMe in that both
indicate that PPII is the predominant conformation sampled by
the two alanine residues. This agrees well with multiple
experiments and with coil libraries which all suggest a high
PPII propensity of alanine.36,37,79 A discrepancy exists concern-
ing the occupancy of the less populated â-strand and 310 helix.
(79) Shi, Z.; Cheng, K.; Liu, Z.; Ng, A.; Bracken, W. C.; Kallenbach, N. R.
Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 17964.
(80) Duan, Y.; Wu, C.; Chowdhury, S.; Lee, M. C.; Xiong, G.; Zhang, W.;
Yang, R.; Cieplak, P.; Luo, R.; Lee, T.; Caldwell, J.; Wang, J.; Kollman,
P. J. Comput. Chem. 1999, 24, 1999.
(81) Gnanakaran, S.; Garcia, A. J. Phys. Chem. B 2003, 107, 12555.
(82) Zagrovic, B.; Lipfert, J.; Sorin, E. J.; Millett, I. S.; Van Gunsteren, W. F.;
Doniach, S.; Pande, V. S. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 11698.
9
J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007 13107

A R T I C L E S
Schweitzer-Stenner et al.
Ac-Aib-Ala-Ala-OMe. Among the three peptide sequences
examined, Ac-Aib-Ala-Ala-OMe showed the strongest depen-
dency on Aib handedness. Thus, the MD simulations strongly
corroborate the results from our amide I′ analysis. Mainly, Ala2
conformations changed dramatically when Aib1’s conformation
moved from the (-65°, -25°) region to the (65°, 30°) region
(Figure S3). On the other hand, Ala3 conformations were much
less sensitive to Aib1 handedness and mainly populated the
(-80°, 130°) and (-80°, -50°) regions of the Ramachandran
plot. When Aib1 torsions were centered at (-65°, -25°), Ala2
populated four regions located at (-80°, -20°), (-140°, 20°),
(-80°, 150°), and (-140°, 155°). When Aib1 was in the (65°,
30°) region, A2 populated mainly three regions located at (-80°,
150°), (-80°, -25°), and (-150°, 160°). The major differences
between these two sets of Ala2 conformations were the
disappearance of the (-140°, 20°) region and the significant
increase in population at (-80°, 150°) when A2 changed from
right-handed helix (-65°, -25°) to left-handed helix (65°, 30°).
Similar to the cases for Ac-Ala-Ala-Aib-OMe and Ac-Ala-Aib-
Ala-OMe, the most populated regions explored by the 500 K
simulation were also explored by the four independent 300 K
simulations.
The consistent results obtained from spectroscopy and MD
simulations for Ac-Aib-Ala-Ala-OMe provide conclusive evi-
dence for the notion that the N-terminal Aib, for which we
obtained a mixture of right-handed and left-handed 310-helices,
imposes a helical conformation on its nearest neighbor. This is
indicative of a break down of the isolated pair hypothesis, which
solely considers the two peptide neighbors of a given residue
as relevant for local interactions. Pappu et al. performed
simulations for an alanine polypeptide and found that non-
nearest neighbor interactions have to be taken into account if a
residue samples the helical region.40 That might apply even more
to Aib since it has been shown to substantially increase the
thermostability of peptides and proteins. This has been attributed
to the fact that its conformational space is not much increased
in the unfolded state, so that a substitution of, e.g., alanine with
Aib reduces the entropy difference between folded and unfolded
states.15 Our results suggest that this entropy reduction extends
to the next neighbor toward the C-terminus.
The number of studies on peptides with Aib residues in
aqueous solution is rather limited. Since Aib is often incorpo-
rated into peptides with a high hydrophobic content, the
crystalline state or organic solvents are generally observed
experimentally.14,15,19,21,83,84 One of the few comprehensive
experimental studies using water as solvent was carried out by
Schievano et al.29 who studied a series of Ac(Asn-Aib-Aib-
Lys)nAla-OH (n ) 1, 2, 3) peptides in water in SDS-containing
micelles. Their NMR data led them to conclude that a 310-helix
was adopted for dimers (n ) 2) and trimers (n ) 3). NMR and
ECD data suggested an increase of 310-helix formation upon
chain elongation. For n ) 1 their ECD data suggest a PPII-like
conformation, but that is certainly not the case for Aib-residues.
Thus, their data were not indicative of the capability of short
Aib-containing peptides to form a helix-like turn in solu-
tion.
Reaction pathways that allow Aib in Ac-Ala-Aib-Ala-OMe
to undergo transitions between right-handed to left-handed
helical conformers were explored with higher-temperature
simulations. This was an expedient approach to observing the
infrequent conformational transitions between helical states;
while the free energy landscape will differ somewhat at the
higher temperature, these preliminary studies facilitate future
detailed studies of the transitions using umbrella sampling and/
or replica exchange. All observed pathways went through a
higher-energy state where the æ angle approaches 180°. The
500 K simulations provide potential reaction pathways for Aib
to undergo transitions between the left-handed and right-handed
helix. Multiple pathways were accessible in the simulation, and
the most likely and lowest-energy pathways are shown in
Figure 10.
Conclusions and comparison with Literature. The crystal
structure of Ac-Ala-Aib-Ala-OMe and Boc-Gly-Ala-Aib-OMe
have been determined by Jung and associates.33-35 For the
former the authors derived a new type of â-turn stabilized by a
very weak 4 f 1 hydrogen bond.34 While the two alanine
residues adopt a clear PPII conformation, the Aib residue
exhibited a slightly distorted left-handed 310 or type III â-turn
conformation. Our results indicate that the peptide adopts
multiple conformations in solution. For the N-terminal alanine
residue, PPII, â-strand and to a minor extent right-handed 310
conformations coexist. The PPII propensity was slightly less
than that observed for the central residue of trialanine.61 For
Aib the experimental data indicate some preference for a right-
handed 310-like helical structure and additional sampling of the
bridge region connecting the PPII and helix region on the
right- and left-hand side of the Ramachandran plot. This
was in full agreement with results from MD simulations of
an Aib-rich peptide in DMSO.30 Our MD simulations suggest
that both right- and left-handed helical conformations are
sampled.
The crystal structure of Ac-Gly-Ala-Aib-OMe, which one
might regard as a good model for Ac-Ala2-Aib-OMe, showed
a somewhat deformed type III â-turn conformation with the
glycine and alanine residues in right-handed 310 conformations.33
This is definitely not the solution structure. We found that in
Ac-Ala2-Aib-OMe both alanine residues behave just as they do
in (cationic) tetraalanine.36,37,60 They exhibit a rather high PPII
propensity with some admixture of â-strand. This shows that
the C-terminal Aib does not have a detectable influence on
residues on its N-terminal side. Altogether, the comparison of
crystal and solution structure revealed that the former might be
of limited use for understanding the structural properties in
solution, since lattice forces apparently select and stabilize
particular peptide conformations in the crystal. Bosch et al. made
an opposite claim for Boc-Ala-Aib-Ala-OMe in methanol, but
we doubt that their results from 1H NMR and ECD were
sufficient for assessing the solution structure.33
Acknowledgment. We are indebted to Dr. Robert Woody,
who provided us with some important suggestions concerning
the interpretation of the ECD spectra at an early stage of the
project. The Ac-Aib-OMe potential energy plot was generously
provided by Daniel J. Kuster of the Marshall Lab. This research
was supported in part by an NIH research grant (GM 08460) to
G.R.M. J.F. also acknowledges graduate support from the
(83) Benedetti, E.; Bavoso, A.; DiBasio, B.; Pavone, V.; Pedone, C.; Crisma,
M.; Bonora, G. M.; Toniolo, C. J. Am. Chem. Soc. 1982, 104, 2437.
(84) Pispisa, B.; Palleschi, A.; Amato, M. E.; Segre, A. L.; Vernanzi, M.
Macromolecules 1997, 30, 4905.
9
13108 J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007

Conformational Manifold of
R
-Aminoisobutyric Acid
A R T I C L E S
Division of Biology and Biomedical Science of Washington
University in St. Louis, the Computational Biology Training
Grant (GM 008802), and the Kauffman Foundation. Computa-
tion resources were provided in part by grants from Teragrid.
The research in the laboratory of R.S.S. was partially supported
by a grant from the National Science Foundation (MCB
0318749).
Supporting Information Available: The description of pep-
tide synthesis and three figures exhibiting the results of MD
simulations. This material is available free of charge via the
Internet at http://pubs.acs.org.
JA0738430
9
J. AM. CHEM. SOC. VOL. 129, NO. 43, 2007 13109