Impact of Azaproline on Amide Cis-Trans Isomerism:
Conformational Analyses and NMR Studies of Model Peptides
Including TRH Analogues
Wei-Jun Zhang,†Anders Berglund,†,|Jeff L.-F. Kao,‡Jean-Pierre Couty,§
Marvin C. Gershengorn,§and Garland R. Marshall*,†
Contribution from the Departments of Biochemistry and Molecular Biophysics and of Chemistry,
Washington UniVersity, St. Louis, Missouri 63110, and DiVision of Molecular Medicine,
Department of Medicine, Weil Medical College of Cornell UniVersity,
New York, New York 10021
Received July 19, 2002; E-mail: email@example.com
Abstract: The ?-turn is a well-studied motif in both proteins and peptides. Four residues, making almost
a complete 180°-turn in the direction of the peptide chain, define the ?-turn. Several types of the ?-turn are
defined according to Φ and Ψ torsional angles of the backbone for residues i + 1 and i + 2. One special
type of ?-turn, the type VI-turn, usually contains a proline with a cis-amide bond at residue i + 2. In an
aza-amino acid, the R-carbon of the amino acid is changed to nitrogen. Peptides containing azaproline
(azPro) have been shown to prefer the type VI ?-turn both in crystals and in organic solvents by NMR
studies. MC/MD simulations using the GB/SA solvation model for water explored the conformational
preferences of azPro-containing peptides in aqueous systems. An increase in the conformational preference
for the cis-amide conformer of azPro was clearly seen, but the increased stability was relatively minor with
respect to the trans-conformer as compared to previous suggestions. To test the validity of the calculations
in view of the experimental data from crystal structures and NMR in organic solvents, [azPro3]-TRH and
[Phe2, azPro3]-TRH were synthesized, and their conformational preferences were determined by NMR in
polar solvents as well as the impact of the azPro substitution on their biological activities.
The ?-turn is a well-studied motif in both proteins and cyclic
peptides (for reviews, see Rose et al.1or Richardson2). Four
sequential residues, making an almost complete 180°-turn in
the direction of the peptide chain, define a reverse turn, the most
rigorously defined subset of which is the ?-turn. There are
several types of ?-turns as described in the literature usually
requiring an internal hydrogen bond between residues one and
four.3,4The different types are defined according to the Φ- and
Ψ-backbone torsional angles for residues 2 and 3. One special
type of ?-turn is the type VI-turn defined by a proline with a
cis-amide bond located at residue 3. Richardson2divided this
class into two different subclasses, type VIa and VIb. Type VIa
usually has an internal hydrogen bond, while type VIb does
not usually make a hydrogen bond. The reported values for (Φ2,
Ψ2), (Φ3, Ψ3) torsions for the two types of VI-turns are as
The conformational space for a set of tripeptides containing
a cis-amide bond was studied by Nagarajaram et al.,5who
reported the minimum-energy conformation for a set of Xxx-
cis-Pro combinations including Ala-Pro and Pro-Pro. These
studies also ruled out the occurrence of a cis-amide bond for
the first proline in Pro-cis-Pro due to its high energy. Two
different types of hydrogen bond were found, type 4 f 1 and
type 1 f 2. There is a high degree of the cis-amide conformer
of proline observed in peptides and proteins; the intrinsic
probability for a cis-amide bond instead of the trans-conforma-
tion of the amide bond preceding a proline is 0.1-0.3 as
compared to less than 10-3for the rest of the amino acids.6
The energy barrier for cis-trans isomerization is also less for
proline as compared to those of the rest of the amino acids.
One reason of this is the greater length of the Xxx-Pro bond,
1.36 Å for proline as compared to 1.33 Å for the usual amide
bond. This comes from the loss of the amide hydrogen resulting
†Department of Biochemistry and Molecular Biophysics, Washington
‡Department of Chemistry, Washington University.
§Weil Medical College of Cornell University.
|Current address: Research Group for Chemometrics, Department of
Chemistry, Umeå University, S-901 87 Umeå, Sweden.
(1) Rose, G. D.; Gierasch, L. M.; Smith, J. A. AdV. Protein Chem. 1985, 37,
(2) Richardson, J. S. AdV. Protein Chem. 1981, 34, 167-339.
(3) Venkatachalam, C. M. Biopolymers 1968, 6, 1425-1436.
(4) Lewis, P. N.; Momany, F. A.; Scheraga, H. A. Biochim. Biophys. Acta
1973, 303, 211-229.
(5) Nagarajaram, H. A.; Paul, P. K. C.; Ramanarayanan, K.; Soman, K. V.;
Ramakrishnan, C. Int. J. Pept. Protein Res. 1992, 40, 383-394.
(6) Brandts, J. F.; R., H. H.; Brennan, M. Biochemistry 1975, 14, 4953-4963.
type VIa:(Φ2) -60°, Ψ2) 120°),
(Φ3) -90°, Ψ3) 0°)
type VIb:(Φ2) -120°, Ψ2) 120°),
(Φ3) -60°, Ψ3) 0°)
Published on Web 01/09/2003
10.1021/ja020994o CCC: $25.00 © 2003 American Chemical Society
J. AM. CHEM. SOC. 2003, 125, 1221-1235 9 1221
in a lack of resonance stabilizing and a redistribution of charge.
Jorgensen and Gao7and Ciepak and Kollman8have explored
the relative stabilities of cis-trans-amide conformers of N-
methylacetamide by ab initio calculations and molecular simula-
tions in the gas phase and in aqueous solution.
There are many examples of how the peptide backbone can
be modified to help stabilize a desired conformation. We have
previously studied the conformational space for the Pro-D-NMe-
amino acid sequence9and also the effect of N-methylation and
N-hydroxylation10on reverse-turn stabilization. Backbone con-
formations can be stabilized by incorporation of many different
modified amino acids and dipeptides (see Lubell11,12and
references therein). Another, less investigated, modification is
the aza-amino acid. In an aza-amino acid, the R-carbon is
changed to nitrogen; this definition precedes that of Mish et
al.,13who referred to the ∆2-pyrazoline-5-carboxylic acids
obtained by cycloaddition as azaproline. Azapeptides contain
aza-amino acids, and numerous aza-analogues of biologically
active peptides have been prepared, for example, angiotensin
II,14oxytocin,15eledoisin,16enkephalin,17and luliberin (LHRH)18,19
with one analogue [D-Ser(t-Bu),6azGly10]-LHRH, a commercial
product, Zoladex, ICI 118630, for the treatment of prostate
carcinoma. More recently, azaglycine has been studied as a
replacement for the central residue of the RGD recognition motif
of integrins.20,21The azapeptide linkage also appears to confer
resistant to degradation by many proteolytic enzymes as
originally discovered by Oehme et al.22and Dutta and Giles.23
The azapeptide linkage has been incorporated into inhibitors
of various enzymes, such as angiotensin converting enzyme,24
cysteine protease,25,26renin,27human leukocyte elastase,28and
human rhinovirus 3C protease.29A promising new HIV protease
inhibitor atazanavir (BMS-232632) contains a para-substituted
azaphenylalanine and is active against multiple-resistant strains.30
A special example of azapeptides is the azatide that is defined
as a “pure azapeptide”, where the R-carbon for each amino acid
is changed to nitrogen.31,32An inhibitor of renin prepared by
Gante et al.27was the first example of a biologically active
Lee et al.33studied by ab initio calculations the structural
perturbation introduced into formyl-amino acid-amides by
changing the R-carbon to nitrogen. The global minimum energy
conformation for these compounds (azGly, AzAla, AzLeu)
suggested a ?-turn motif with the aza residue at the i + 2nd
position. The peptide Boc-Phe-azLeu-Ala-OMe was prepared,
and its structural preference was determined in organic solvents.
The data supported a type II ?-turn as predicted. More relevant
to this study, azapeptides containing azaproline (azPro) have
been shown to prefer the ?-turn type VI in crystals34,35and also
by NMR studies36in organic solvents. The two adjacent
nitrogens in the ring show a clear pyramidal character in the
reported crystal structures. The amide bond is also longer, giving
a lower barrier for the cis-trans isomerization. This finding is
also supported by a theoretical study of diformylhydrazine by
Reynolds at al.37The calculations are focused on the Z,Z, Z,E,
and E,E conformations of diformylhydrazine as well as the
rotational barrier for the CO-N-N-CO torsional angle. In
these calculations, the two nitrogens had a pyramidal conforma-
tion for all of the conformations. The fact that diformylhydrazine
has a flat conformation in crystal structures is probably due to
crystal packing forces in that a planar conformation allows
stacked sheets of hydrogen-bonded networks. Aza-amino acids
have also been used in the backbone in a peptide/oligourea/
azapeptide hybrid for inducing a hairpin turn.38
The focus of this paper is the quantitative determination both
computationally and experimentally of the conformational
influence of azaproline (azPro) in stabilizing reverse-turn
conformations in peptides. The previous observations34-36that
the enhanced cis-amide conformation was induced by azPro
increased the probability of type VI reverse-turns. Analogues
of thyrotropin-releasing hormone (TRH) containing azPro were
included to further probe the receptor-bound conformation of
TRH. TRH is a natural tripeptide, L-pyroglutamyl-L-histidyl-
L-proline-amide (pGlu-His-Pro-NH2), a prominent neuromes-
senger released from the hypothalamus that controls the release
of thyroid stimulating hormone (TSH) from the pituitary. Several
papers have investigated the solution conformation of TRH39-42
in the hopes of gaining insight into the biologically relevant
conformation. One study has shown that the activity is correlated
(7) Jorgensen, W. L.; Gao, J. J. Am. Chem. Soc. 1988, 110, 4212-4216.
(8) Cieplak, P.; Kollman, P. J. Comput. Chem. 1991, 12, 1222-1236.
(9) Chalmers, D. K.; Marshall, G. R. J. Am. Chem. Soc. 1995, 117, 5927-
(10) Takeuchi, Y.; Marshall, G. R. J. Am. Chem. Soc. 1998, 120, 5363-5372.
(11) Hanessian, S.; McNaughton-Smith, G.; Lombart, H.-G.; Lubell, W. D.
Tetrahedron 1997, 53, 12789-12854.
(12) Halab, L.; Lubell, W. D. J. Org. Chem. 1999, 64, 3312-3321.
(13) Mish, M. R.; Guerra, F. M.; Carreira, E. M. J. Am. Chem. Soc. 1997, 119,
(14) Hess, H.-J.; Moreland, W. T.; Laubach, G. D. J. Am. Chem. Soc. 1963, 85,
(15) Niedrich, H. J. Prakt. Chem. 1972, 314, 769.
(16) Oehme, P.; Bergmann, J.; Falck, M.; Reich, J. G.; Vogt, W. E.; Niedrich,
H.; Pirrwitz, J.; Berseck, C.; Jung, F. Acta Biol. Med. Germanica 1972,
(17) Han, H.; Yoon, J.; Janda, K. D. Bioorg. Med. Chem. Lett. 1998, 8, 117-
(18) Dutta, A. S.; Furr, B. J.; Giles, M. B.; Valcaccia, B. J. Med. Chem. 1978,
(19) Dutta, A. S.; Furr, B. J.; Giles, M. B.; Valcaccia, B.; Walpole, A. L.
Biochem. Biophys. Res. Commun. 1978, 81, 382-390.
(20) Gibson, C.; Goodman, S. L.; Hahn, D.; Holzemann, G.; Kessler, H. J. Org.
Chem. 1999, 64, 7388-7394.
(21) Sulyok, G. A.; Gibson, C.; Goodman, S. L.; Holzemann, G.; Wiesner, M.;
Kessler, H. J. Med. Chem. 2001, 44, 1938-1950.
(22) Oehme, P.; Katzwinkel, S.; Vogt, W. E.; Niedrich, E. Experientia 1973,
(23) Dutta, A. S.; Giles, M. B. J. Chem. Soc., Perkin Trans. 1 1976, 244-248.
(24) Greenlee, W. J.; Thorsett, E. D.; Springer, J. P.; Patchett, A. A. Biochem.
Biophys. Res. Commun. 1984, 122, 791-797.
(25) Magrath, J.; Abeles, R. H. J. Med. Chem. 1992, 35, 4279-4283.
(26) Xing, R.; Hanzlik, R. P. J. Med. Chem. 1998, 41, 1344-1351.
(27) Gante, J.; Krug, M.; Lauterbach, G.; Weitzel, R.; Hiller, W. J. Pept. Sci.
1995, 1, 201-206.
(28) Powers, J. C.; Boone, R.; Carroll, D. L.; Gupton, B. F.; Kam, C. M.;
Nishino, N.; Sakamoto, M.; Tuhy, P. M. J. Biol. Chem. 1984, 259, 4288-
(29) Venkatraman, S.; Kong, J.; Nimkar, S.; Wang, Q. M.; Aube, J.; Hanzlik,
R. P. Bioorg. Med. Chem. Lett. 1999, 9, 577-580.
(30) Witherell, G. Curr. Opin. InVest. Drugs 2001, 2, 340-347.
(31) Han, H.; Janda, K. D. J. Am. Chem. Soc. 1996, 118, 2539-2544.
(32) Han, H.; Yoon, J.; Janda, K. D. In Peptidomimetics Protocols; Kazmierski,
W. M., Ed.; Humana Press: Totowa, NJ, 1998; Vol. 23, Chapter 106, pp
(33) Lee, H.-J.; Ahn, I.-A.; Ro, S.; Choi, K.-H.; Choi, Y.-S.; Lee, K.-B. J. Pept.
Res. 2000, 56, 35-46.
(34) Lecoq, A.; Boussard, G.; Marraud, M.; Aubry, A. Biopolymers 1993, 33,
(35) Didierjean, C.; Del Duca, V.; Benedetti, E.; Aubry, A.; Zouikri, M.;
Marraud, M.; Boussard, G. J. Pept. Res. 1997, 50, 451-457.
(36) Zouikri, M.; Vicherat, A.; Aubry, A.; Marraud, M.; Boussard, G. J. Pept.
Res. 1998, 52, 19-26.
(37) Reynolds, C. H.; Horman, R. E. J. Am. Chem. Soc. 1996, 118, 9395-
(38) Soth, M. J.; Nowick, J. S. J. Org. Chem. 1999, 64, 276-281.
(39) Donzel, B.; Rivier, J.; Goodman, M. Biopolymers 1974, 13, 2631-2647.
(40) Montagut, M.; Lemanceau, B.; Belloco, A.-M. Biopolymers 1974, 13,
(41) Vicar, J.; Abillon, E.; Piriou, F.; Lintner, K.; Blaha, K.; Fromageot, P.;
Fermandjian, S. FEBS Lett. 1979, 97, 275-278.
(42) Flurry, R. L., Jr.; Abdulnur, S. F.; Bopp, J. M., Jr. Biopolymers 1978, 17,
A R T I C L E SZhang et al.
1222 J. AM. CHEM. SOC.9VOL. 125, NO. 5, 2003
to the stability of the cis-amide conformer for several TRH
analogues43and suggested that the cis-amide conformer repre-
sents the receptor-bound conformation. A variety of ap-
proaches12,44,45to stabilizing the cis-amide conformer of the
peptide bond have been proposed and reviewed.46A series of
both cis- and trans-amide conformationally constrained ana-
logues of TRH have been prepared,47and the consensus of the
data favors the trans-amide-bond conformation as biologically
Conformational searches and molecular dynamics were performed
with MacroModel50version 6.5. The MacroModel implementation of
the AMBER all-atom force field was used for all of the calculations.
The GB/SA continuum solvation model was used for the solution-phase
calculations.51The calculations were done on a SGI Power Challenge
L with 12 R10000 processors.
Conformational Searches. Conformational searches were performed
by the systematic Monte Carlo Method of Goodman and Still.52For
each search, 5000 starting structures were generated and minimized
until the gradient was less then 0.05 (kJ/mol)/A-1, using the truncated
Newton-Raphson method implemented in MacroModel.50Duplicate
conformations and those with energy greater than 50 kJ/mol above the
global minimum found were discarded.
Monte Carlo/Stochastic Dynamics. All simulations were performed
at 300 K with use of the Monte Carlo/Stochastic Dynamics (MC/SD)
hybrid simulation algorithm53with the AMBER* force field as
implemented in MacroModel 6.5.50A time step of 0.75 fs was used
for the stochastic dynamics (SD) part of the algorithm. The MC
simulation used random torsional rotations between (60° and (180°
that were applied to all rotatable bonds except for the C-N amide
bonds, of the azapeptide of interest, where the random rotation was
between (90° and (180°. No rotations were applied to the bonds in
the rings for proline and azaproline, as the transition barriers between
ring conformers are low enough to permit adequate sampling from the
SD part of the simulation. The total simulation was 2000 ps, and
samples were taken at 1 ps intervals, yielding 2000 conformations for
Chemical Syntheses of [azPro3]-TRH and [Phe2, azPro3]-TRH.
To test the validity of the calculations in view of the experimental data
from crystal structures and NMR in organic solvents suggesting a strong
stabilization of the cis-amide bond, [azPro3]-TRH and [Phe2, azPro3]-
TRH were synthesized by a novel synthetic route (Scheme 1) due to
the relative lack of reactivity of azaproline. The synthesis of azapeptides
was initially introduced by Goldschmidt and Wick54and actively
developed by Gante55,56and by Dutta and Morley.57More recently,
synthetic routes have been investigated by Gante,56Quibellet et al.,58,59
and Gibson et al.20utilizing solid-phase synthesis, and also by Han et
al.31using a liquid-phase approach. Incorporation of an aza-amino acid
residue into the peptide chain requires a combination of hydrazine and
peptide chemistry. The usual approach consists of adding a protected
hydrazine to an isocyanate derivative of the peptide N-terminal, but it
is not applicable when Pro occupies the N-terminal position. Another
method uses activated aryl esters of aza residues, but the reaction needs
high temperature and long reaction time, resulting in low yields with
numerous side products, oxadiazolones, and hydantoins. Andre et al.60
used triphosgene as the carbonylative reagent of a protected hydrazide.
It is a mild, easy to handle, and efficient carbonylating agent for
azapeptide synthesis. Alhough this method works well, in general, the
activated species can only be prepared in situ (under N2at -10° C),
and formation of considerable amounts of side products, such as
Boc-hydrazine 1 was acylated by carbobenzoxychloride to afford
Boc-NH-NH-Z 2. Reaction with NaH in DMF and subsequent treatment
by 1,3-dibromopropane afforded Boc-azPro-OBzl 3. Removal of Z by
hydrogenation gave Boc-N,N′-propylhydrazine 4, which was activated
with triphosgene below -10° C, and then reacted with tritylamine to
give Boc-azPro-NH-Trt 5. Removal of Boc with 25% TFA for 25 min
gave azPro-NH-Trt 6, which was reacted with the protected dipeptides,
either pGlu-His(Trt)-OH or pGlu-Phe-OH after activation with TFFH/
HOAT/DIEA, to obtain the protected tripeptides 7. Next 95% TFA/
CH2Cl2was used to remove the trityl groups from both the carboxamide
and the imidazole groups, and the crude TRH analogues 8 were purified
by HPLC and characterized by ESI/MS (Scheme 1).
In this work, while making Boc-azPro-carbonyl chloride from 6 and
triphosgene with NMM, the reaction of phosgene with 1 was incom-
plete, and starting material remained. When Boc-azPro-amide was
prepared with anhydrous NH3in dioxane, side reactions occurred, and
(43) Liakopoulou-Kyriakides, M.; Galardy, R. E. J. Med. Chem. 1979, 22, 1952-
(44) Zabrocki, J.; Smith, G. D.; Dunbar, J. B., Jr.; Iijima, H.; Marshall, G. R.
J. Am. Chem. Soc. 1988, 110, 5875-5880.
(45) Paul, P. K. C.; Burney, P. A.; Campbell, M. M.; Osguthorpe, D. J. Bioorg.
Med. Chem. Lett. 1992, 2, 141-144.
(46) Etzkorn, F. A.; Travins, J. M.; Hart, S. A. In AdVances in Peptidomimetics;
Abell, A., Ed.; JAI Press Inc.: Greenwich, CT, 1999; Vol. 2, pp 1-34.
(47) Tong, Y. S.; Olczak, J.; Zabrocki, J.; Gershengorn, M. C.; Marshall, G.
R.; Moeller, K. D. Tetrahedron 2000, 56, 9791-9800.
(48) Laakkonen, L.; Li, W.; Perlman, J. H.; Guarnieri, F.; Osman, R.; Moeller,
K. D.; Gershengorn, M. C. Mol. Pharmacol. 1996, 49, 1092-1096.
(49) Chu, W.; Perlman, J. H.; Gershengorn, M. C.; Moeller, K. D. Bioorg. Med.
Chem. Lett. 1998, 8, 3093-3096.
(50) Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Liskamp, R.; Lipton, M.;
Caufield, C.; Chang, G.; Hendrickson, T.; Still, W. C. J. Comput. Chem.
1990, 11, 440-467.
(51) Still, W. C.; Tempczyk, A.; Hawley, R. C.; Hendrickson, T. J. Am. Chem.
Soc. 1990, 112, 6127-6129.
(52) Goodman, J. M.; Still, W. C. J. Comput. Chem. 1991, 12, 1110-1117.
(53) Guarnieri, F.; Still, W. C. J. Comput. Chem. 1994, 15, 1302-1310.
(54) Goldschmidt, S.; Wick, M. Liebigs Ann. Chem. 1952, 575, 217-231.
(55) Gante, J. Chem. Ber. 1965, 98, 540-547.
(56) Gante, J. Synthesis 1989, 405-413.
(57) Dutta, A. S.; Morley, J. S. J. Chem. Soc., Perkin Trans. 1 1975, 1712-
(58) Gray, C. J.; Quibell, M.; Baggett, N.; Hammerle, T. Int. J. Pept. Protein
Res. 1992, 40, 351-362.
(59) Quibell, M.; Turnell, W. G.; Johnson, T. J. Chem. Soc., Perkin Trans. 1
(60) Andre, F.; Marraud, M.; Tsouloufis, T.; Tzartos, S. J.; Boussard, G. J. Pept.
Sci. 1997, 3, 429-441.
Scheme 1. Synthesis of the Protected azPro-NH2Derivative and
Its Use in the Preparation of [AzPro3]-TRHa
aAn analogous procedure was used for the synthesis of [Phe2, azPro3]-
TRH by reacting pGlu-Phe-OH with compound 6.
Im pact of Azaproline on Am ide Cis−Trans Isom erism A R T I C L E S
J. AM. CHEM. SOC. 9 VOL. 125, NO. 5, 2003 1223
carbonyldi-N,N′(Boc)-propylhydrazine was formed. The probable reason
is that NH3is a weaker base then 1, so it did not compete with 1 to
react with 2. Instead of the desired reaction to produce the carboxamide,
1 left in solution due to the incomplete reaction with triphosphine
reacted with 2 to form the undesirable symmetrical carbonyldihydrazide
derivative 9. When tritylamine, a stronger base than 1 or 2, was used
instead of NH3, the tritylamine reacted smoothly with Boc-azPro-
carbonyl chloride to give the desired product 5.
Materials. BocNHNH2, bis(trichloromethyl)carbonate, and trity-
lamine were purchased from Aldrich. His(Trt)-OMe was purchased from
Bachem. Pyroglutamic acid (pGlu, pyrrolidone carboxylic acid) and
trifluoroacetic acid (TFA) were purchased from Advanced Chemtech.
HATu and HOAT were purchased from Richeliau Biotechnologies.
Solvents were all HPLC grade. Tetramethylfluoroformamidinium
hexafluorophosphate (TFFH)61was prepared in our laboratory. The
purity of the peptides was confirmed by analytical HPLC (SP8800
Spectraphysics) with a C18 column (4.3 × 250 mm) and mass
spectrometry. The mobile phase consisted of a gradient (B; 0.05% TFA
in H2O; B 0.038% TFA in 10% H2O, 90% acetonitrile). The peptides
were purified by HPLC chromatography using a Rainin Model HPX1,
equipped with a Ranin C18 column (5µ 10 × 250 mm). The mobile
phase consisted of the same two solvents as the analytical HPLC.
Procedure. 1. Boc-NHNH-Z. ZCl (0.1 M, 14.3 mL) was added
dropwise to Boc-NHNH2(0.1 M, 13.2 g) in anhydrous THF (75 mL)
containing NMM (0.1 M, 11 mL) at 0 °C. After being stirred for 12 h
at room temperature, NMM‚HCl salt was filtered off. THF was removed
under reduced pressure to get an oil that was dried in desiccator
overnight, when it became solid. The material was purified by flash
chromatography with the solvent system, 7:3 ) ethyl acetate:hexane,
and was recrystallized with THF and hexane. Yield 85%, mp 63-64
°C, Rf) 0.66 (TLC solvent system, EtAc:hexane, 7:3). Mass: M + 1
2. Boc-AzaPro Benzyl Ester. NaH (0.48 g, 0.02 M) in mineral oil
(60% dispersion) was suspended in DMF (20 mL) under nitrogen at
room temperature, and BocNHNHZ (0.01 M, 2.66 g) in DMF (10 mL)
was added dropwise before the addition of 1,3-dibromopropane (0.01
M, 1.05 mL) in DMF (5 mL). When 1,3-dibromopropane was added,
an ice-H2O bath was used for the initial exothermic reaction, and the
reaction solution was stirred under N2overnight. DMF was removed
under reduced pressure, and the crude sample was taken up by AcOEt
and washed successively with 0.1 M aqueous citric acid, 5% aqueous
NaHCO3, and brine. The crude sample obtained was purified by silica
gel chromatography using the solvent system AcOEt:hexane ) 7:3 as
eluent; 4.43 mmol of pure sample was obtained. Yield ) 44%, Rf)
0.56 (ETOAc:hexane ) 7:3). Mass: M + 1 ) 306.
3. Boc-azPro-tritylamide. Boc-azPro-benzylester (3 mmol, mw 305,
917 mg) was hydrogenated with Pd/C in methanol for 4 h. Pd/C was
separated by filtration, and the solution was evaporated to dryness. Ether
was added to the residue to give an oily solid (Boc-azPro-OH). NMM
(mw ) 101, d ) 0.92, 3 mmol, 0.33 mL in THF 0.4 mL) was added
dropwise under stirring to a cold solution (-10 °C) of triphosgene (mw
) 296.75, 1 mmol, 297 mg) and BocAzapro (3 mmol in THF 3 mL)
with stirring at -10 °C for 45 min. Tritylamine (mw 295.35, 3 mmol,
778.1 mg) and NMM (0.33 mL) were added progressively, and the
solution was stirred overnight, followed by filtration and solvent
evaporation. The residual oil was dissolved in EtOAc, and the solution
was washed with 1 N NaHCO3, 0.1 M citric acid, and brine. It was
dried with Mg2SO4and evaporated to dryness. The material was purified
with the solvent system EtOAc:hexane ) 4:6 on a silica gel column to
yield a liquidlike oil that crystallized with EtAc and petroleum ether.
Thus 260 mg of crystals (mw 456) was obtained. Yield ) 60%, mp
132-133 °C. Mass: M + 1 ) 456.
pGlu-His(Trt)-COOMe. Pyroglutamic acid (mw 129.2, 4 mmol,
1.792 g), HATu (mw 380.2, 4 mmol, 1.52 g), and HOAT (mw 136.1,
4 mmol, 544.4 g) were dissolved in DMF. DIEA (8 mmol, 1.392 mL)
was added dropwise. His(Trt)-COOMe‚HCl (4 mmol, mw 411.5, 1.6
g) was then added after being neutralized with 0.696 mL of DIEA.
The pH was adjusted to 6.8, and the reaction was stirred overnight.
The solvent was evaporated, and the residue was washed with 5%
NaHCO3, brine, ice-cold 0.1 N citric acid, and brine again to give an
oily white solid. Mass: M + 1 ) 523. Next 0.79 mmol of pGlu-His-
(Trt)-COOMe (410 mg, mw 522.7) was dissolved in 0.79 mL of DMF.
In an ice bath, 0.79 mL (1 N NaOH) was added dropwise to the solution
with stirring. The extent of hydrolysis was checked by TLC (CHCl3:
CH3OH:HOAc ) 9:1:0.1) every 30 min, and the reaction was complete
by 1 h and 40 min. Evaporating the DMF gave a white solid (214 mg,
53% yield). mp 132 °C, Rf) 0.219 in solvent system 9:1:0.1 (CHCl3:
methanol:HOAc). Mass: M + 1 ) 509.
pGlu-His(Trt)-azPro-tritylamide. Boc-azPro tritylamide (182.4 mg,
mw 456, 0.4 mM) was deprotected with 25% TFA/CH2Cl2for 25 min
to give TFA‚azPro tritylamide. The mixture of TFFH (mw 266, 0.4
mM, 106.4 mg), HOAt (mw 136.1, 0.4 mM, 54.44 mg), and collidine
(0.0525 mL in DMF 0.5 mL) was stirred at room temperature for 3
min and was added at 0 °C to 2 mL of DMF containing pGlu-His-
(Trt)-COOH (203.48 mg, mw 508.7, 0.4 mM) and TFA. AzPro
tritylamide (after being neutralized with collidine, 0.0525 mL) and then
collidine 0.0571 mL were added; pH was 6. The trityl group was
removed with 95% TFA for 1 h, and then the TFA was evaporated.
After ethyl ether was added to the residue, a white precipitate developed
that was filtered and dried.The crude sample was purified by reverse
phase HPLC with the solvent system A: 0.05% TFA/H2O; B: 0.038%
TFA/90% acetonitrile and 10% H2O; gradient 2-100% B 30′. The
desired tripeptide, pGlu-His-azPro-NH2, came out at 7.8 min (mw 363,
48 mg). Total yield ) 35%, mp ) 121 °C, Rf) 0.836 in 9:1:0.1 )
CHCl3:MeOH:HOAc). Mass: M + 1 ) 364.
Biological Assay. The TRH analogues were tested for binding as
well as signaling at the TRH receptor type 1, TRH-R1. Until recently,
this was the only TRH receptor described, but a second receptor with
a distinct distribution in the CNS has been described. Little to no
differences were seen in binding or signaling between the two receptors
for a set of TRH analogues.62Receptor binding of the TRH analogues
was measured in TRH-R1 transfected COS-1 cells as previously
described.63TRH-R1-mediated phosphoinositol hydrolysis was mea-
sured in myo-[3H]inositol-labeled cells.64All analogues appear to be
full agonists. The experimental values in the two assays as compared
with control standards were as follows:
NMR Spectroscopy. NMR spectra were recorded with a Varian
Unity-600 (Varian Assoc., Palo Alto, CA) spectrometer, and the data
were processed off-line on a ULTRASPARC station with VNMR
software. Proton and carbon chemical shifts were measured in parts
per million (ppm) downfield from an external 3-(trimethylsilyl)propionic
acid (TSP) standard. Proton spectra were obtained in 6100-Hz spectral
width collected into 64K data points with 5.0 s pre-acquisition delay.
Carbon spectra were obtained with a 32 000-Hz spectral width collected
into 64k data points.
The total correlation (2D HOHAHA) spectra65were recorded using
an MELV-17 mixing sequence of 100 ms, flanked by two 2 ms trim
(61) Carpino, L. A.; El-Faham, A. J. Am. Chem. Soc. 1995, 117, 5401-5402.
(62) O’Dowd, B. F.; Lee, D. K.; Huang, W.; Nguyen, T.; Cheng, R.; Liu, Y.;
Wang, B.; Gershengorn, M. C.; George, S. R. Mol. Endocrinol. 2000, 14,
(63) Colson, A. O.; Perlman, J. H.; Smolyar, A.; Gershengorn, M. C.; Osman,
R. Biophys. J. 1998, 74, 1087-1100.
(64) Perlman, J. H.; Laakkonen, L. J.; Guarnieri, F.; Osman, R.; Gershengorn,
M. C. Biochemistry 1996, 35, 7643-7650.
(65) Bax, A.; Davis, D. G. J. Magn. Reson. 1985, 65, 355.
20 000 (7400-58 000)
A R T I C L E SZhang et al.
1224 J. AM. CHEM. SOC.9VOL. 125, NO. 5, 2003
pulses. Phase-sensitive 2D spectra were obtained by employing the
Hypercomplex method. A total of 2 × 256 × 2048 data matrix with
16 scans per t1 value were collected. Gaussian line broadening and a
sine-bell function were used in weighting the t2 and t1 dimensions,
respectively. After two-dimensional Fourier transform, the spectra
resulted in 2048 × 2048 data points that were then phase- and baseline-
corrected in both dimensions. A two-dimensional COSY2spectrum was
collected into a 512 × 2048 data matrix with 16 scans per t1 value.
The time-domain data were zero filled to yield a 2048 × 2048 data
matrix and were Fourier transformed using a sine-bell weighting
function in both t2 and t1 dimensions. The NOESY3spectrum resulted
form a 2 × 256 × 2049 data matrix with 32 scans per t1 value. Specta
were recorded with a 250 ms mixing time. The Hypercomplex method
was used to yield phase-sensitive spectra. The time-domain data were
zero filled to yield a 2k x 2k data matrix and were processed in a way
similar to that of the 2D HOHAHA spectrum described above.
The proton-detected heteronuclear multiple quantum coherence
(HMQC)4spectrum was recorded using a 0.35 s1H-13C nulling period,
and a 0.055 s delay was used in HMBC experiment. The 90°1H pulse
width was 11 µs, and the 90°13C pulse width was 14 µs. The proton
spectral width was set to 6100 Hz, and the carbon spectral width was
set to 31 000 Hz. Phase-sensitive 2D spectra were obtained by
employing the Hypercomplex method. A 2 × 256 × 2048 data matrix
with 64 scans per t1 value was collected. Gaussian line broadening
was used in weighting both the t2 and the t1 dimension. After two-
dimensional Fourier transform, the spectra resulted in 512 × 2048 data
points that were phase and baseline corrected in both dimensions.
Results and Discussion
NMR Studies. Proton and carbon chemical shifts of [azPro3]-
TRH and [Phe2, azPro3]-TRH (Scheme 1) were assigned by
analysis of COSY, TOCSY, NOESY, HMQC, and HMBC
spectra. Pyroglutamic acid (U) was identified first by the spin
propagation from the amide NH through γ, ?, and R protons
(Figure 1d), and the Uγproton was confirmed by the sequential
γ(i)-NH (i + 1) connections between the adjacent U-H (His)
or U-F (Phe) residues. Correlated proton resonances from NH
at 8.45 ppm in Figure 1d were unambiguously assigned to FR
and F?at 5.05 and 2.95 ppm, respectively. The assignments of
azPro (Z) Zδand Z?protons were made from the F (or H)-
CO-Zδand Z-CO-Z?HMBC correlations. Connections of F
(or H)-CO-Z?and Z-CO-Zδwere ruled out due to the weak
four-bond carbonyl carbon-proton couplings. F-CO resonance
at 177.33 ppm (Figure 1a) was assigned on the basis of the
unique F-CO to FR, F?, and Zδmultiple-bond carbon-proton
correlation. Likewise, the CO resonance at 174.52 ppm in Figure
2a was assigned to H-CO. Correlation from the carbonyl carbon
at 166.14 ppm (Figure 1a) was assigned to Z-CO due to the
connection to two Z?protons only. Complete assignment of the
azPro resonances in [azPro3]-TRH was obtained by tracing the
HMBC cross-peaks (Figures 1a and 2a) to cross-peaks of the
same proton chemical shift at the COSY and HMQC spectra.
Combined use of the adjacent H1-H1 connectivity in COSY
and direct C13-H1 connectivity in HMQC allows for the
Figure 1. Expansions of the 600 MHz spectra of [Phe2, azPro3]-TRH in D2O at 25 °C. (a) The HMBC spectrum in which the correlations of F-CO-
FR-F?-Zδand Z-CO-Z?are connected by lines. (b) The COSY spectrum shows the adjacent H1-H1connectivity. (c) The HMQC spectrum shows the
H1-C13connectivity. (d) The TOCSY spectrum shows the U and F amide NH to R, ?, and γ correlations and (e) the NOE cross-peaks of Zδto FRin trans
F-Z amide conformation. The NOESY cross-peaks were distinguished from diagonal peaks by their negative character.
Im pact of Azaproline on Am ide Cis−Trans Isom erism A R T I C L E S
J. AM. CHEM. SOC. 9 VOL. 125, NO. 5, 2003 1225
removal of the ambiguity of assigning the geminal protons of
the azaproline. For example, Zδresonance at 3.79 ppm (Figure
1b) shows two cross-peaks Zδ-δand Zδ-γat 3.20 and 1.85 ppm.
However, the HMQC reveals that both proton resonances at
3.79 and 3.20 ppm connect to the same carbon at 46.4 ppm
(Z-Cδ) (Figure 1c), indicating that these are the Zδgeminal
protons. Resonance at 1.85 ppm (Figure 1b) was assigned to
the Zγproton because it connects to a different carbon at 26.43
ppm (Z-Cγ) (Figure 1c). The same scheme was used for the
assignment of the proton and carbon chemical shifts in [Phe2,
azPro3]-TRH (Figure 2b and c).
The presence of azPro in the peptide produces cis- and trans-
isomers of the F-Z and H-Z amide bond. This is reflected by
the observation of two sets of cross-peaks for each individual
H1 and C13 resonance, such as HR, FR, and Uγin Figures 1c and
2c. To further differentiate the azaproline resonances between
cis- and trans-isomers, use was made of the TOCSY spectra that
correlate the spin propagation Zδ-Zγ-Z?to the same isomer.
Correlated peaks from resonances at 4.17 ppm (Z?) and 3.98
ppm (Zδ) showing in Figure 2d all belong to the same isomer,
while spin propagation from 4.05 and 4.02 ppm in Figure 2d
belongs to the other isomer. Furthermore, strong NOE cross-
peaks between HR(at 5.06 ppm) and Zδ(at 3.98 and 3.23 ppm)
(Figure 2e) were observed, indicating that the His-azPro amide
bond appeared to be in the trans-conformation. Meanwhile, a
weak correlation between HRat 5.22 ppm and Zδat 4.02 ppm
(Figure 2e) suggests these were the resonances of the cis-
conformation. The same strategy was used to assign the cis-
and trans-conformers of [Phe2, azPro3]-TRH (Figure 1e), and
complete assignments of H1and C13resonance for each residue
of [azPro3]-TRH and [Phe2, azPro3]-TRH are listed in Table 1.
Interestingly, it was found that the chemical shift differences
for the two azPro ? protons (Z?t) are quite large for both of the
trans-amide conformers of [Phe2, azPro3]-TRH (∆δ ) 1.75
ppm) and [azPro3]-TRH (∆δ ) 1.58 ppm). The large chemical-
shift difference is rationalized by the more rigid azaproline ring
that places one of the Z?proton directly over the deshielding
zone of the Z-CO carbonyl group and leads to the strong
nonequivalence of the two ? protons. The observed Uγ-FNH
(or HNH) and FR (or HR)-Zδ NOEs allowed one to identify
[azPro3]-TRH and [Phe2, azPro3]-TRH with the all trans-amide
backbone conformation. However, the NOESY spectra of
[azPro3]-TRH (Figure 2e) showed that the HRnot only has NOE
to Zδbut also to Z?protons, indicating that the H-Z amide
bond deviates from a complete trans-amide conformation.
Populations of the cis-trans-conformer estimate from the
volume integral in HMQC and COSY spectra were ap-
proximately 40/60 for [azPro3]-TRH and 20/80 for [Phe2,
Side-chain rotamer populations5of the Phe and His residues
of trans-[azPro3]-TRH and trans-[Phe2, azPro3]-TRH were
calculated using the best fit for the values of the coupling
constant JRH-?H. The g- and t rotamer constituted the most
dominate population (Table 1). The g+ rotamer is sterically
Figure 2. Expansions of the 600 MHz spectra of [azPro3]-TRH in CD3OD at -10 °C. (a) The HMBC spectrum in which the correlation of H-CO-HR-
H?-Zδis connected by a line. (b) The COSY spectrum shows the adjacent H1-H1connectivity. (c) The HMQC spectrum shows the H1-C13connectivity.
(d) The TOCSY spectrum differentiates the Zδ-Zγ-Z?spin propagation between cis- and trans-isomers of [azPro3]-TRH, and (e) the NOE cross-peaks of
Zδto HRindicate the trans H-Z amide conformation.
A R T I C L E SZhang et al.
1226 J. AM. CHEM. SOC.9VOL. 125, NO. 5, 2003
unfavored due to interaction between the F, H aromatic groups
and the azaproline ring. The t rotamer may constitute up to 50%
of the population in [azPro3]-TRH that places the His aromatic
ring in the region near the Zδprotons. Ring-current effects from
the His imidazole may contribute more significantly to the
nonequivalence of two Zδprotons in [azPro3]-TRH (∆δ ) 0.75
ppm) than does Zδ in [Phe2, azPro3]-TRH (∆δ ) 0.59 ppm).
Calculations. In earlier versions of MacroModel, no param-
eters were available for aza-amino acids, so the proper
parameters for the AMBER* force field were estimated with
ab initio calculations. With the release of MacroModel 6.5, this
effort became redundant because this version included param-
eters for aza-amino acids in agreement with those estimated.
By focusing on azaproline, we minimized the impact of the
parameters for the central N-N bond because the torsional
rotation of this bond is relatively fixed for the azPro ring.
Conformational Searches. Monte Carlo searches were
performed on the blocked aza-containing tetrapeptides Ac-Ala-
Xxx-Yyy-Ala-NHMe, the blocked aza-tripeptides Boc-Ala-Xxx-
Ala-NHiPr, and on the TRH tripeptides. The calculations
included the GB/SA solvation model and the AMBER* force
field parameters as implemented in MacroModel version 6.5.
As reported by Boussard et al.,34,35the nitrogens in azaproline,
especially the R-nitrogen, are not planar due their sp3character.
The reported out-of-plane distance for the nitrogen is between
0.18 and 0.40 Å. This perturbation results in both nitrogens being
prochiral; they can either be S,S or R,R. Both diastereomers
have been studied to compare the influence of the conformation
on the reverse-turn propensity. The energy differences between
the cis- and trans-conformations are calculated between the
conformations with the lowest energy independent of whether
it is (R,R) or (S,S). Hence, we assumed that the atoms attached
to the nitrogens in the azPro ring can invert from R to S and
TRH. The results from the conformational searches (see
Table 2) show that the trans-conformation is more stable than
the cis-conformation for TRH. The same is seen for TRH with
azaproline in the C-terminal position; the trans-conformation
is more stable than the cis-conformation. Comparing the cis-
and trans-conformations reveals that the trans-conformation may
be stabilized in these two compounds by an internal hydrogen
bond. Changing the imidazole ([azPro3]-TRH) ring to a phenyl
ring ([Phe2, azPro3]-TRH) removed the possibility of an internal
hydrogen bond. This makes the cis-conformation more stable,
as can be seen in the Table 2.
The conformational preference of these TRH analogues was
examined by NMR in water and in methanol. In methanol at
-10°, the signals from the cis- and trans-conformers of [azPro3]-
TRH were resolved and could be clearly assigned. The ratio of
cis-trans-conformers estimated from the volume integrals in
the HMQC and COSY spectra is approximately 40/60 for
[azPro3]-TRH and 20/80 for [Phe2, azPro3]-TRH. Comparing
the cis- and trans-conformations reveals that the trans-
conformation of [azPro3]-TRH is stabilized by an internal
Table 1. Proton and Carbon Chemical Shifts (ppm) of [Phe2, azPro3]-TRH in D2O at 25 °C and [azPro3]-TRH in CD3OD at -10 °Ca
g-, t ≈ 48%
Z 2.03, 1.85 3.79, 3.20
2.282.39, 1.97 4.18
JR?) 5.8, 8.9c
g- ≈ 58%, t ≈ 29%
aChemical shifts of cis-amide conformers are shown in parentheses.bIn 90% H2O at 25 °C.cJ coupling constant in Hz with rotamer population g- and
t in percentage.
Table 2. Results from the Conformational Searches To Determine
the Relative Stability of cis-trans-Conformersa
aThe minimum found for each set of four diastereomeric (RR, SS) azPro
cis-trans-conformers is indicated in bold. In the peptides not containing
azPro, the minimum found for the cis-trans-conformer pairs is indicated
in bold. The ∆E column denotes the difference between the lowest energy
conformer (bold) and the lower of the other cis-trans-amide column for
Im pact of Azaproline on Am ide Cis−Trans Isom erismA R T I C L E S
J. AM. CHEM. SOC. 9 VOL. 125, NO. 5, 2003 1227
hydrogen bond between its His2imidazole ring and a backbone
amide bond. Changing the imidazole (His2) ring to a phenyl
ring (Phe2) removes the possibility of this internal hydrogen
bond. Thus, while the exchange of the carbon-R by nitrogen
enhances the propensity for the cis-conformer, the conforma-
tional preference of azPro is obviously environmentally sensitive
as seen both experimentally and computationally. The strength
of this constraint is, therefore, insufficiently strong to hold the
amide predominately in the cis-conformer regardless of the
environment. Other constraints such as the 1,5-tetrazole amide-
bond surrogate66or other bicyclic analogues47may be more
appropriate to probe recognition of the cis-amide bond con-
former by the receptor.
The affinity of [azPro3]-TRH for the TRH receptor from the
anterior pituitary was reduced 40-fold as compared to that for
TRH itself, consistent with significantly reduced activity seen
in general for analogues with cis-amide constraints.47[Phe2]-
TRH has a lower affinity for the TRH receptor than TRH itself,
and the substitution of azPro for Pro in the [Phe2, azPro3]-TRH
analogue causes an approximately 200-fold further loss in
Tetrapeptides. The results from the conformational search
are shown in Table 2. It can be seen that the S,S-conformer is
the most stable for the cis-conformation, in agreement with
Boussard et al.34The same thing is not observed for the trans-
conformations; the (S,S) does not seem to be favored over the
(R,R) diastereomers. For the Ac-Ala-Pro-azPro(S,S)-Ala-NHMe
peptide, the cis-conformation is the most stable one, 4.6 kJ/
mol more stable than the trans-conformation. The minimum
energy conformation of Ac-Ala-Pro-azPro(S,S)-Ala-NHMe is
shown in Figure 3a. It is clear from the figure that the structure
forms a type VIa ?-turn with a cis-conformation for the
preceding azPro residue (ω2) 19.0°) and with Φ2) -37.4°,
Ψ2) 122.6°, Φ3) -92.5°, and Ψ3) 14.8°. The conformation
in Figure 3a is stabilized with an internal hydrogen bond
between the carbonyl oxygen at residue one and the amide
hydrogen of residue four. There is also a second hydrogen bond
between the two end groups.
Moving the azaproline to the second residue only gave a small
difference between the lowest energy conformations with a cis-
versus a trans-amide bond; the difference was only 0.5 kJ/mol.
This is in accordance with what Boussard et al.34found, that
the amide bond preceding the azPro is favored, and not the one
after the azPro.
Changing the second residue to an alanine, Ac-Ala-Ala-azPro-
Ala-NHMe, which is more flexible than proline, does not change
the energy difference between cis- and trans-conformers much,
but the minimum is a type VIb ?-turn instead of a type VIa
?-turn. The lowest energy conformation for Ac-Ala-Ala-azPro-
(S,S)-Ala-NHMe is shown in Figure 3b with the following
torsional angles: ω2) 11.7°, Φ2) -159.3°, Ψ2) 138.4°,
Φ3) -91.0°, and Ψ3) 13.9°. The reverse turn is not stabilized
with any internal hydrogen bonds, also in good agreement with
what has been found in other examples of type VIa ?-turns.1
Also, the differences between the S,S- and the R,R-configurations
are less, indicating that these peptides are more flexible than
the one with two pralines, as expected.
For comparison, we also included two tripeptides appearing
in the paper by Boussard et al.67studied both in the crystalline
state and, more recently, in chloroform solution by NMR. The
two tripeptides are Boc-Ala-azPro-Ala-NHiPr and Boc-Ala-
azPro-Ala-NHiPr, both reported to induce a ?-turn in the
sequence. Both show similar behavior according to the calcula-
tions; both favor the S,S-diasteriomer and the cis-conformation.
This was also seen in the published NMR studies; however,
there was no evidence for cis-trans equilibrium for the ω1-
amide bond. It is important to remember that the NMR studies
were not done in water, but in CHCl3. The corresponding results
with the CHCl3GB/SA solvation model are also shown in Table
2. The most stable conformation by simulation under similar
conditions is also the S,S-diasteriomer with the cis-amide
Monte Carlo/Stochastic Dynamics. The results from the
MC/SD runs are less conclusive than the results from the
conformational searches. The differences between the R,R- and
the S,S-diastereomers are quite large for [azPro3]-TRH and,
especially, for [Phe2, azPro3]-TRH.
The same set of geometrical parameters for characterizing
the reverse turn was used as those previously by Takeuchi et
al.10in a similar study on reverse-turn induction in tetrapeptides.
We also included the number of conformations from the MD/SD
simulation that have a cis-conformation for the central amide
bond, because this is required for a type VI ?-turn. The virtual
torsion, ?, defined by CR1, CR2, CR3, and N4,68,69gives a
measurement on the tightness of the turn. A torsion angle less
than (30° indicates a turn. The third parameter is that the CR1-
CR4 distance must be less than 7 Å. The fourth and fifth para-
meters are the distance between the carbonyl at position one and
the amide hydrogen of residue four with the cutoff distances set
at 4 and 2.5 Å, respectively; within 4 Å indicates an interaction
between these groups,70and within 2.5 Å indicates a hydrogen
bond.9A slightly modified definition was used for the tripeptides
because these only have three residues, but the results are still
comparable. Also, in these calculations, both of the diastereomers
were considered to account for the inversion of the nitrogens
in the azPro ring that occurs, but is hard to simulate quantita-
(66) Zabrocki, J.; Marshall, G. R. In Peptidomimetics Protocols; Kazmierski,
W. M., Ed.; Humana Press: Totowa, NJ, 1998; Vol. 23, Chapter 424, pp
(67) Boussard, G.; Marraud, M. J. Am. Chem. Soc. 1985, 107, 1825-1828.
(68) Ball, J. B.; Andrews, P. R.; Alewood, P. F.; Hughes, R. A. FEBS Lett.
1990, 273, 15-18.
(69) Ball, J. B.; Hughes, R. A.; Alewood, P. F.; Andrews, P. R. Tetrahedron
1993, 49, 3467-3478.
(70) Constantine, K. L.; Mueller, L.; Andersen, N. H.; Tong, H.; Wandier, C.
F.; Friedrichs, M. S.; Bruccoleri, R. E. J. Am. Chem. Soc. 1995, 117,
Figure 3. Structures of the global energy minimum found by a Monte
Carlo using the AMBER* force field and the GB/SA solvation model. (a)
Ac-Ala-Ala-AzPro(S,S)-Ala-NHMe. (b) Ac-Ala-Pro-AzPro(S,S)-Ala-NHMe.
A R T I C L E SZhang et al.
1228 J. AM. CHEM. SOC.9VOL. 125, NO. 5, 2003
tively without extensive parametrization. For TRH and [azPro3]-
TRH, only the percentage of cis-amide bond was used as a
descriptor, because other parameters were not easily measured.
A summary of the MC/SD runs is shown in Table 3. To
compare the conformational space for the different peptides, a
plot of the backbone torsional angle Φ against angle Ψ for
residue two and three was also used with other similar plots.
TRH and [azPro3]-TRH. The results from the MC/SD
simulation on both TRH and [azPro3]-TRH are given in Table
2. Both of the tripeptides have a similar number of conforma-
tions with a cis-amide bond, 35.7% for TRH, 34.7% for
[azPro3]-TRH (S,S), and a slightly higher 51.1% for [azPro3]-
TRH (R,R). The Φ and Ψ plots for the His2and Pro3residues
are shown in Figure 4a,b for TRH, and in Figure 4c,d for
[azPro3]-TRH. The conformational spaces are similar for the
two tripeptides; the His2residue is restricted for both TRH and
[azPro3]-TRH. A similar trend is seen for the Pro3and azPro3
residues. Changing the His2to a Phe increases the number of
conformations with a cis-amide bond in the MD/SD simulation,
as seen in Table 2. Only the [azPro3]-TRH (R,R) diastereomer
shows a lower degree of cis-amide bond. The conformational
space is still restricted, and the Φ and Ψ angles for pGlu-Phe-
Pro(S,S)NH2 are shown in Figure 4 e,f; there are no large
differences as compared to the Φ and Ψ plot for [azPro3]-TRH.
Ac-Ala-Pro-Pro-Ala-NHMe. As a “reference” peptide, the
Ac-Ala-Pro-Pro-Ala-NHMe tetrapeptide was included to enable
comparison of this unmodified peptide with other analogues to
determine the impact of modifications. Table 2 shows that 28.9%
of the conformations from the MD/SD simulation have a cis-
amide bond between the two central prolines. This indicates
that the energy difference between the cis- and the trans-
conformation of the intervening amide bond is not large. Also
notable, only 3.5% of the conformations form a strong hydrogen
bond between the carbonyl oxygen at residue one and the amide
hydrogen at residue four, and 28.8% have a CR1-CR4 distance
of less than 7 Å. The plot of torsional angles Φ2versus Ψ2and
Φ3versus Ψ3is shown in Figure 5a,b. The torsional space for
the second residue is restricted, as seen in Figure 5a. The
conformation is also independent of the conformation for the
amide bond because both the squares and the diamonds are
mixed in the plot. The third residue is less constrained as the
Φ3versus Ψ3plot shows in Figure 5b. The angle Ψ3is less
well defined, ranging from about -70° to 180°. The cis-
conformation for the central amide bond is a requirement for a
hydrogen bond between the carbonyl oxygen and the amide
hydrogen, and this can be seen in Figure 5c, the distance
between the carbonyl oxygen and the amide hydrogen versus
the amide torsion (ω2). The Pro-Pro peptide forms a type VI
?-turn, but it is not the dominant conformation in the MD/SD
simulation, as can be seen in Table 3.
Ac-Ala-Pro-azPro(S,S)-Ala-NHMe and Ac-Ala-Pro-azPro-
(S,S)-Ala-NHMe. As seen in Table 2, 54.7% of the samples in
the simulations Ac-Ala-Pro-azPro(S,S)-Ala-NHMe have a cis-
conformation for the central amide bond; for example, it likely
forms a type VI ?-turn. The same value for the R,R diastereomer
is 61.9%. The Ψ2versus Φ2and Ψ3versus Φ3plot for Ac-
Ala-Pro-azPro(S,S)-Ala-NHMe shows that the ΦΨ space is well
defined for both residues two and three, Figure 6a,b. The torsion
angles seem to be independent of the ω2 torsion because the
diamonds (cis) and squares (trans) are mixed in both of the plots.
For the cis-conformations, all conformations have the type VIa
?-turn with mean torsion angles, for all conformations with a
cis-bond, of ω2) 16.5°, Φ2) -53.2°, Ψ2) 135.2°, Φ3)
-94.6°, and Ψ3) 11.5°. The population in the simulations that
had a virtual torsion of |?| < 30° was 26.2% for the S,S
conformation; Figure 6c shows the distance between the
carbonyl oxygen of residue one and the amide hydrogen of
residue four. It is clear that those conformations with a central
cis-amide bond (diamonds) have a virtual torsion angle around
30° (mean virtual torsion angle for the cis-conformations is
30.8°). 39.4% of the conformations in the simulations have a
strong hydrogen bond with a length less than 2.5 Å, a
significantly high value. This indicates that only conformations
with a cis-amide bond are able to form a reverse turn for this
peptide. Figure 6d, the hydrogen bond distance versus ω2,
clearly indicates that this is the case; the only allowed ?-turn is
the type VIa ?-turn. The number of structures from the MD/
SD simulation with a CR1-CR4 distance less than 4 Å is 53.4%.
The results for the R,R diastereomer differ in several ways.
As seen in Figure 6e, the hydrogen bonds are only possible for
Table 3. Results from the 1000 ps, 300 K MC/SD Run Using the GB/SA Solvation Model and the AMBER* Force Field in MacroModel 6.5
%d<7Å <4Å <2.5Å
Im pact of Azaproline on Am ide Cis−Trans Isom erismA R T I C L E S
J. AM. CHEM. SOC. 9 VOL. 125, NO. 5, 2003 1229
Figure 4. Results from the MC/SD simulations. ] - Structures with a cis-amide bond; [ - structures with a trans-amide bond. (a) Ψ2vs Φ2for TRH.
(b) Ψ3vs Φ3for TRH. (c) Ψ2vs Φ2for [azPro3]-TRH. (d) Ψ3vs Φ3for [azPro3]-TRH. (e) Ψ2vs Φ2pGlu-Phe-azPro(S,S)NH2. (f) Ψ3vs Φ3for pGlu-
A R T I C L E SZhang et al.
1230 J. AM. CHEM. SOC.9VOL. 125, NO. 5, 2003
the trans-conformations; 19.5% of the population has a hydrogen
bond length less than 2.5 Å. The reverse turn is also tighter as
can be seen from Figure 6f, comparing the hydrogen bond
distance versus the virtual torsion ?. The mean value for the
virtual torsion angle is 6.5° for the conformation with a trans-
amide bond (squares). The mean Φ and Ψ torsion angles for
the second and third residue are Φ2) -58.2°, Ψ2) 131.2°,
Φ3) 83.5°, and Ψ3) -8.9°. Both of the two diastereomers
show a high degree of reverse turn, even if the R,R diastereomer
requires a trans-amide bond, while the S,S diastereomer requires
a cis-amide bond, that is, a type VIa ?-turn.
Ac-Ala-azPro(S,S)-Pro-Ala-NHMe and Ac-Ala-azPro-
(R,R)-Pro-Ala-NHMe. Few of the sampled conformations of
these peptides form a tight turn; only 7.5% and 10.5% for the
R,R and S,S, respectively, have a virtual torsion less than 30°.
Table 2 shows that few of the sampled conformations form a
strong hydrogen bond between the carbonyl oxygen at residue
one and the amide hydrogen at residue four. The R,R diaste-
reomer has some interactions between the carbonyl oxygen and
the amide hydrogen: 19.3% of the samples have a distance less
than 4 Å, but only 0.2% less than 2.5 Å, indicating few strong
hydrogen bonds. The degree of reverse turn is even less than
that for the tetrapeptide with two ordinary residues at residue
two and three. An azPro (SS or RR) at residue two decreases
the likelihood for a cis-amide bond at its C-terminal, and,
therefore, it is also a type VI ?-turn for the tetrapeptide.
Ac-Ala-Ala-azPro(R,R)-Ala-NHMe and Ac-Ala-Ala-az-
Pro(S,S)-Ala-NHMe. The Ac-Ala-Ala-azPro-Ala-NHMe tet-
rapeptides are less restricted as compared to the tetrapeptides
with two central proline residues. As seen in Table 3, a higher
degree of the conformations in the MD/SD simulation has the
cis-conformation; for the S,S diastereomer, the value is 77.8%.
The S,S diastereomer also shows a high degree of ?-turn
according to the virtual torsion angle and the CR1-CR4
distance, 67.3 and 66.8%, respectively. The Φ and Ψ plots for
residues two and three are shown in Figure 7a,b. There is no
difference in Φ and Ψ space between conformations with a
cis-amide bond (diamonds) or conformations with a trans-amide
bond (squares). The mean Φ2 angle for conformations with a
cis-amide bond is shifted -100° degrees as compared to that
of the Ac-Ala-Pro-azPro(S,S)-Ala-NHMe peptide. The mean
value for the torsional angles are ω2) 11.0°, Φ2) -149.0°,
Ψ2) 138.7°, Φ3) -91.6°, and Ψ3) 8.43°. This is closer to
the type VIb ?-turn instead of the VIa ?-turn for the Ac-Ala-
Pro-azPro(S,S)-Ala-NHMe peptide. From Figure 7c, ω2versus
hydrogen bond distance, it is clear that the cis-conformation
gives the shortest distances between the carbonyl oxygen and
the amide hydrogen. The distance is still too long for making
good hydrogen bonds. We can also see from Figure 7d, virtual
torsion angle versus hydrogen bond distance, that all of the
conformations with a cis-amide bond (diamonds) have a virtual
torsion angle around zero (mean virtual torsion for the cis-
conformations is 7.0°). This indicates that the peptide makes a
tight reverse turn, but there is no internal hydrogen bonding
involved. The R,R diastereomer showed less propensity for
reverse-turn formation as seen in Table 1. All of the conforma-
tions from the MD/SD simulation that showed a reverse turn
had a trans-conformation for the middle amide bond, similar
for the R,R diasteriomer of the Ac-Ala-Pro-azPro-Ala-NHMe
Figure 5. Results from the MC/SD simulations for Ac-Ala-Pro-Pro-Ala-
NHMe. ] - Structures with a cis-amide bond; [ - structures with a trans-
amide bond. (a) Ψ2vs Φ2for Ac-Ala-Pro-Pro-Ala-NHMe. (b) Ψ3vs Φ3
for Ac-Ala-Pro-Pro-Ala-NHMe. (c) Hydrogen bond distance vs ω2for Ac-
Im pact of Azaproline on Am ide Cis−Trans Isom erism A R T I C L E S
J. AM. CHEM. SOC. 9 VOL. 125, NO. 5, 2003 1231
Figure 6. Results from the MC/SD simulations for Ac-Ala-Pro-azPro-Ala-NHMe. ] - Structures with a cis-amide bond; [ - structures with a trans-
amide bond. (a) Ψ2vs Φ2for Ac-Ala-Pro-azPro(S,S)-Ala-NHMe. (b) Ψ3vs Φ3for Ac-Ala-Pro-azPro(S,S)-Ala-NHMe. (c) Hydrogen bond distance vs the
virtual torsion ? for Ac-Ala-Pro-azPro(S,S)-Ala-NHMe. (d) Hydrogen bond distance vs ω2for Ac-Ala-Pro-azPro(S,S)-Ala-NHMe. (e) Hydrogen bond distance
vs ω2for Ac-Ala-Pro-azPro(R,R)-Ala-NHMe. (f) Hydrogen bond distance vs the virtual torsion ? for Ac-Ala-Pro-azPro(R,R)-Ala-NHMe.
A R T I C L E SZhang et al.
1232 J. AM. CHEM. SOC.9VOL. 125, NO. 5, 2003
Boc-Ala-azPro(R,R)-Ala-NHiPr and Boc-Ala-azPro(S,S)-
Ala-NHiPr. Because these are tripeptides, the measurements
had to be modified. For the virtual torsion angle, the ester
oxygen -O- was used instead of the CR1; the same thing was
done for the CR1-CR4 distance where the ester oxygen was
used instead of CR4. For internal hydrogen bonding, the
carbonyl oxygen on the Boc-group was used instead of the
carbonyl oxygen for the first residue. The same tendency is seen
for this peptide as the ones mentioned above; the S,S conforma-
tion is the diastereomer that forms a reverse turn more easily.
72.7% of the S,S conformations from the MD/SD run have a
cis-conformation for the central amide bond. The conformations
are similar to those in the MD/SC run for the tetrapeptide with
alanines, Ac-Ala-Ala-azPro-Ala-NHMe. The Φ and Ψ plots
(Figure 8a,b) show that there is no difference in Φ and Ψ space
between the cis- and trans-conformations. It can also be seen
that most of the conformations have the type VIb ?-turn, but
there is also a small cluster of conformations that have a type
VIa ?-turn. Figure 8c, hydrogen bond distance versus ω2, shows
that a cis-conformations are required to give a hydrogen bond.
The figure also reveals that few of the conformations have a
strong hydrogen bond; only 3.5% of the conformations have a
distance of less than 2.5 Å between the carbonyl oxygen and
the amide hydrogen. The R,R diastereomer showed the same
behavior as that of the previous R,R diastereomer.
Boc-Ala-azPro(R,R)-D-Ala-NHiPr and Boc-Ala-azPro-
(S,S)-D-Ala-NHiPr. Table 3 shows that the Boc-Ala-azPro(S,S)-
D-Ala-NHiPr has the highest degree of cis-amide bond for all
of the peptides in this study, 99%. Also in the virtual torsion
angle and CR1-CR4 parameters, it has high values, 71.5 and
77.1%, respectively. The Φ and Ψ plots (Figure 9a,b) show
that the spread in Φ2and Ψ2is greater than that for the other
Figure 7. Results from the MC/SD simulations for Ac-Ala-Ala-azPro-Ala-NHMe. ] - Structures with a cis-amide bond; [ - structures with a trans-
amide bond. (a) Ψ2vs Φ2for Ac-Ala-Ala-azPro(S,S)-Ala-NHMe. (b) Ψ3vs Φ3for Ac-Ala-Ala-azPro(S,S)-Ala-NHMe. (c) Hydrogen bond distance vs ω2
for Ac-Ala-Ala-azPro(S,S)-Ala-NHMe. (d) Hydrogen bond distance vs the virtual torsion ? for Ac-Ala-Ala-azPro(S,S)-Ala-NHMe.
Im pact of Azaproline on Am ide Cis−Trans Isom erismA R T I C L E S
J. AM. CHEM. SOC. 9 VOL. 125, NO. 5, 2003 1233
Figure 8. Results from the MC/SD simulations for Boc-Ala-azPro-Ala-
NHiPr tripeptide. ] - Structures with a cis-amide bond; [ - structures
with a trans-amide bond. (a) Ψ2vs Φ2for Boc-Ala-azPro(S,S)-Ala-NHiPr.
(b) Ψ3 vs Φ3 for Boc-Ala-azPro(S,S)-Ala-NHiPr. (c) Hydrogen bond
distance vs ω2for Boc-Ala-azPro(S,S)-Ala-NHiPr.
Figure 9. Results from the MC/SD simulations for Boc-Ala-azPro-D-Ala-
NhiPr tripeptide. ] - Structures with a cis-amide bond; [ - structures
with a trans-amide bond. (a) Ψ2vs Φ2for Boc-Ala-azPro(S,S)-D-Ala-NHiPr.
(b) Ψ3 vs Φ3 for Boc-Ala-azPro(S,S)-D-Ala-NHiPr. (c) Hydrogen bond
distance vs the virtual torsion ? for Boc-Ala-azPro(S,S)-D-Ala-NHiPr.
A R T I C L E S Zhang et al.
1234 J. AM. CHEM. SOC.9VOL. 125, NO. 5, 2003
peptides. The Φ3 versus Ψ3 plot, Figure 9a, shows two Download full-text
distinctive clusters, one around Φ3) -90°, Ψ3) 0° and the
other cluster around Φ3) -90° and Ψ3) -180°. 20.8% of
the conformations have a hydrogen bond distance less than 4
Å, while only 3.5% are within 2.5 Å. This indicates that the
structures are not stabilized with an internal hydrogen bond.
Figure 9c shows that the cis-conformations can form both a
reversed turn with the virtual torsion around zero and also
extended conformations with a virtual torsion larger than 90°.
This is a bit unexpected because it is usually the trans-
conformation that has the less tight turn, as can be seen for the
peptides discussed above. The cis-conformation seems to be
the far more stable conformation independent of the other torsion
angles whether it forms a reverse turn or not. These findings
are also true for the R,R diastereomer, even though the degree
of reverse turns in the conformations from the MD/SD simula-
tions is less, as can be seen from Table 3.
In the search for a new type VI ?-turn-inducer, azaproline
was of significant interest on the basis of initial crystal structures
and NMR studies in organic solvent. As compared to many other
types of type VI ?-turn inducers, azPro makes only a small
change to the overall geometry and conformation of proline and
does not introduce additional steric bulk that could compromise
receptor interactions. Conformational searches and MD/SD
simulations using the GB/SA solvation model indicated that an
azaproline increases the propensity for a cis-conformation in
the amide bond preceding the azPro and, by formation of that
cis-amide, the likelihood for a type VI ?-turn. The results from
MD/SD simulation for the Ac-Ala-Pro-azPro-Ala-NHMe tet-
rapeptide showed an increased degree of type VIa ?-turn, with
more than 50% of the conformations having a cis-amide bond
and also a high degree of internal hydrogen bonding. Changing
the second residue to an alanine resulted in a slightly different
type of ?-turn. The results for Ac-Ala-Ala-azPro-Ala-NHMe
also show the same trend in the degree of reverse turn, but it is
closer to a type VIb ?-turn with no stabilizing internal hydrogen
bond. The conformational searches revealed that the cis-
conformation is the most stable for all of the tetrapeptides with
an azaproline, ranging from 0.5 to 4.6 kJ/mol more stable than
the trans-conformation. Also, the tripeptides studied earlier by
Boussard et al.67showed the same behavior, a high degree of
The calculations for TRH and [azPro3]-TRH did not generate
the same results as the calculations for the tetrapeptides, or the
experimental studies, supporting enhanced cis-conformational
stability. In conformational searches, the trans-conformation was
the most stable conformation for both TRH and [azPro3]-TRH.
The higher stability of the trans-conformation is likely to come
from the interaction between the carboxamide and the His2-
imidazole ring. This hydrogen-bonding interaction, only possible
with a trans-amide bond, favors the trans-amide bond over the
cis-amide bond. Changing the imidazole group to a phenyl ring
removed the possibilities of this interaction, with results on
relative stabilities of cis-trans-amide conformers similar to the
results from the calculations for [Phe2, azPro3]-TRH as com-
pared with [azPro3]-TRH.
The biological activity of the two analogues of TRH
containing azPro are perfectly consistent when compared with
TRH and [Phe2]-TRH. The enhanced relative activity of TRH
may be due, in part, to the conformation seen when the
imidazole ring is hydrogen bonded to the backbone.71This was
supported by studies of 6-(1-methylhistidine) and 6-(3-meth-
ylhistidine) analogues of TRH by Rivier et al.,72where
[3-MeHis2]-TRH had 8-fold enhanced activity as compared to
that of TRH, and [1-MeHis2]-TRH was essentially inactive.
One problem with azPro is that the two hydrazine nitrogens
have a prochiral motif; they are either R,R or S,S. Both
diastereomers were explicitly used in these studies because the
nitrogens can readily invert at room temperature between these
two diastereomers. For most of the examples, the two diaster-
eomers give similar results in the MD/SD simulations, but, for
example, in [azPro3]-TRH, the differences are larger.
Acknowledgment. We acknowledge the National Institutes
of Health (GM 53630) for partial support of this research as
well as a postdoctoral fellowship (A.B.) from the Knut and Alice
Wallenberg Foundation. The Washington University Mass
Spectroscopy Resource Center partially supported by NIH
(RR00954) and the Washington University High Resolution
NMR Resource Center (NIH RR02004, RR05018 and RR07155)
were utilized to characterize the peptide analogues synthesized
as part of this study.
(71) Grant, G.; Ling, N.; Rivier, J.; Vale, W. Biochemistry 1972, 11, 3070-
(72) Rivier, J.; Vale, W.; Monahan, M.; Ling, N.; Burgus, R. J. Med. Chem.
1972, 15, 479-482.
Im pact of Azaproline on Am ide Cis−Trans Isom erism A R T I C L E S
J. AM. CHEM. SOC. 9 VOL. 125, NO. 5, 2003 1235