Chiral Recognition of Peptide Enantiomers by Cinchona Alkaloid Derived Chiral Selectors: Mechanistic Investigations by Liquid Chromatography, NMR Spectroscopy, and Molecular Modeling

Article abstract of DOI:10.1021/jo0346914

The chiral recognition mechanism of a cinchona alkaloid based chiral selector for N-protected peptide enantiomers was investigated. A chiral stationary phase derived from this selector was employed for liquid chromatographic enantiomer separations. It sho

Full text of DOI:10.1021/jo0346914

Ch ir a l Recogn ition of P ep tid e En a n tiom er s by Cin ch on a Alk a loid
Der ived Ch ir a l Selector s: Mech a n istic In vestiga tion s by Liqu id
Ch r om a togr a p h y, NMR Sp ectr oscop y, a n d Molecu la r Mod elin g
Christoph Czerwenka,† Mei Mei Zhang,‡ Hanspeter Ka¨hlig,§ Norbert M. Maier,†
Kenny B. Lipkowitz,‡ and Wolfgang Lindner*,†
Institute of Analytical Chemistry, University of Vienna, Wa¨hringerstrasse 38, 1090 Wien, Austria,
Department of Chemistry, North Dakota State University, Fargo, North Dakota 58105, and Institute of
Organic Chemistry, University of Vienna, Wa¨hringerstrasse 38, 1090 Wien, Austria
wolfgang.lindner@univie.ac.at.
Received May 22, 2003
The chiral recognition mechanism of a cinchona alkaloid based chiral selector for N-protected peptide
enantiomers was investigated. A chiral stationary phase derived from this selector was employed
for liquid chromatographic enantiomer separations. It showed exceptionally high enantiomer
discrimination for the (all-R)- and (all-S)-enantiomers of dialanine (R ) 20), while a pronounced
loss of chiral recognition occurred upon the insertion of an additional alanine residue into the peptide
backbone. This reduction of enantioselectivity was investigated in great detail by NMR spectroscopy
of complexes of the chiral selector and the analyte enantiomers accompanied by molecular modeling
studies. Investigation of intramolecular NOEs provided the conformational states of the free and
complexed forms of the selector. The analysis of complexation-induced shifts yielded information
on intermolecular interactions and allowed us to propose binding models, which were further
supported by the observation of intermolecular NOEs, indicating the relative arrangements of
selector and analytes. Stochastic molecular dynamics simulations were able to reproduce the
chromatographic retention orders and energy differences, as well as the intermolecular NOEs. The
computational data were used to evaluate the intermolecular forces responsible for analyte binding.
In addition, the relative contributions of the fragments of the chiral selector to the enantioselective
binding event were assessed. A spatial arrangement of the chiral selector and the analyte allowing
the primary ionic interaction as well as hydrogen bonding and π-π-stacking to take place
simultaneously was found to be essential to obtain very high enantioselectivities.
1. In tr od u ction
tailoring “receptor-like” selectors aimed at the specific
target of a single (group of) molecule(s). Further, a well-
understood chiral recognition mechanism can, in some
cases, allow the indirect assignment of the absolute
configuration of an analyte of unknown stereochemistry.
Unfortunately, the polymeric nature and supramolecu-
lar complexity of many of the widely used natural chiral
selectors (e.g. polysaccharides, antibiotics, or proteins)
make studies targeted at the elucidation of stereoselective
molecular recognition mechanisms difficult. In contrast,
low molecular weight synthetic selectors facilitate the
investigation of stereoselective binding phenomena.
Cinchona alkaloid derivatives, e.g., carbamoylated
quinine, represent one group of the large array of
synthetic selectors available nowadays. These chiral
selectors have been successfully employed for enantiomer
separations of acidic analytes with use of liquid chroma-
tography,2-5 capillary electrophoresis,6-8 and capillary
electrochromatography.9-11
The direct chromatographic separation of enantiomers
is of great importance in a variety of fields including the
pharmaceutical, agro, and fine chemicals industry, both
at analytical and preparative scale.1 For chromatographic
enantiomer separations a wide array of chiral selectors
are available; however, the choice of a selector suited for
a specific separation problem often relies on a trial-and-
error approach due to the lack of extensive understanding
of the underlying discrimination mechanisms. Therefore,
advancing the knowledge on the stereoselective recogni-
tion mechanism of chiral selectors is not only of interest
from an academic point of view but also of importance
for practical reasons. These include the choice of a
suitable selector for a given analyte (“selectand”) and the
development of new and refinement of existing selectors,
as well as for broadening their scope of application or
* Corresponding author. Phone: +43 1 4277 52300. Fax: +43 1 4277
9523.
(2) La¨mmerhofer, M.; Lindner, W. J . Chromatogr. A 1996, 741, 33-
48.
(3) Maier, N. M.; Nicoletti, L.; La¨mmerhofer, M.; Lindner, W.
Chirality 1999, 11, 522-528.
(4) Mandl, A.; Nicoletti, L.; La¨mmerhofer, M.; Lindner, W. J .
Chromatogr. A 1999, 858, 1-11.
† Institute of Analytical Chemistry, University of Vienna.
‡ North Dakota State University.
§ Institute of Organic Chemistry, University of Vienna.
(1) Maier, N. M.; Franco, P.; Lindner, W. J . Chromatogr. A 2001,
906, 3-33.
10.1021/jo0346914 CCC: $25.00 © 2003 American Chemical Society
Published on Web 10/02/2003
J . Org. Chem. 2003, 68, 8315-8327
8315

Czerwenka et al.
In a recent paper reporting high-performance liquid
chromatography (HPLC) enantiomer separation of N-
acylated oligopeptides on chiral stationary phases based
on these cinchona alkaloid derivatives,5 a particular
behavior of enantioselectivity in terms of chain length
was described. A very pronounced drop in enantioselec-
tivity was observed for the separation of 3,5-dinitroben-
zoylated peptide enantiomers between the dipeptide and
the tripeptide. As the elution order remained unchanged
and only a slight decrease in enantiomer discrimination
was noted when studying tetramers and further elon-
gated peptides, it can be assumed that the general chiral
recognition remained the same. However, a loss of one
(or more) stereoselective binding increment(s) was an-
ticipated to occur that effected the large loss of enantio-
meric discrimination. To investigate this phenomenon
and to enhance the knowledge on the underlying molec-
ular recognition mechanism a detailed NMR study was
carried out to elucidate the factors responsible for these
observations. In addition, molecular modeling was em-
ployed to study the intermolecular forces responsible for
chiral recognition from a theoretical point of view.
In this contribution we thus present the enhanced
insights into the chiral recognition mechanism of one
representative of the family of cinchona alkaloid derived
chiral selectors for peptide enantiomers based on the
results of our experimental and theoretical studies.
Specifically, the exceptional chiral recognition capabilities
of 6′-neopentoxy-9-O-tert-butylcarbamoylcinchonidine (Fig-
ure 1, SO1) for 3,5-dinitrobenzoyl dialanine (HPLC: R
≈ 20)5 versus those for the significantly less well resolved
3,5-dinitrobenzoyl trialanine (Figure 1, SA1 and SA2,
respectively) are investigated in detail.
F IGURE 1. Structures of chiral selector (SO1), the related
chiral stationary phase (CSP 1), and peptide analytes (selec-
tands SA1 and SA2; being of either (all-R) or (all-S) config-
uration) studied.
TABLE 1. HP LC En a n tiom er Sep a r a tion Resu lts on
6′-Neop en toxy-9-O-ter t-bu tylca r ba m oylcin ch on id in e-
Ba sed CSP a
sample
k1
R
RS
∆∆G (kJ ‚mol-1
)
elution orderb
2. Resu lts a n d Discu ssion
DNB-Ala 1.52 15.83 22.83
DNB-Ala2 1.14 19.96 20.68
DNB-Ala3 0.83 2.84 8.41
-6.85
-7.42
-2.59
(S)
(S,S)
(S,S,S)
2.1. Ch r om a togr a p h ic Deter m in a tion of En a n ti-
oselectivities a n d Differ en tia l F r ee Bin d in g En er -
gies. For HPLC enantiomer separations, an immobilized
version of the chiral selector was used. The chromato-
graphic enantioselectivities (R), which are defined as the
ratio of the retention factors of the two enantiomers kR
and kS, can be used to calculate the differential free
binding energies according to ∆S/R(∆G) ) -RT‚ln R (kS
> kR). If the nonselective retention increments are
negligible compared to the enantioselective ones and
immobilization of the chiral selector does not influence
its enantiodiscrimination capabilities, the chromato-
graphically determined ∆S/R(∆G) values represent a good
estimate for the intrinsic (thermodynamic) enantioselec-
tivities. In a recent study these conditions were found to
be fulfilled for the present selector and a selectand closely
a
Conditions: mobile phase, 80/20 methanol/0.5 M aqueous
ammonium acetate, adjusted to pHa 6.0 with acetic acid; flow rate,
1 mL‚min-1; 25 °C; UV detection at 254 nm. Indicating the
b
configuration of the longer retained enantiomer.
related to the investigated peptide analytes, thus en-
abling the above-mentioned deductions.12
To determine the R- and ∆S/R(∆G) values of the (all-
R)- and (all-S)-enantiomers of di- and trialanine peptides,
SO1 was immobilized to a mercaptopropyl modified silica
support yielding CSP 1 (Figure 1), which was then packed
into an HPLC column. Enantiomer separations of the
N-protected peptide enantiomers (for comparison reasons
the corresponding amino acid enantiomers, 3,5-dini-
trobenzoyl (R)- and (S)-alanine, were also included) were
performed with a buffered hydro-organic mobile phase.
The observed enantioselectivities and the differential free
binding energies calculated thereof are presented in
Table 1. Very high enantioselectivities were achieved for
the amino acid and dipeptide derivatives, whereas a
significant drop in enantioselectivity (from almost 20 to
below 3) was observed for the N-protected tripeptide. This
translates into a loss of differential free binding energy
(5) Czerwenka, C.; La¨mmerhofer, M.; Maier, N. M.; Rissanen, K.;
Lindner, W. Anal. Chem. 2002, 74, 5658-5666.
(6) Piette, V.; La¨mmerhofer, M.; Lindner, W.; Crommen, J . Chirality
1999, 11, 622-630.
(7) La¨mmerhofer, M.; Zarbl, E.; Lindner, W.; Simov, B. P.; Ham-
merschmidt, F. Electrophoresis 2001, 22, 1182-1187.
(8) Czerwenka, C.; La¨mmerhofer, M.; Lindner, W. Electrophoresis
2002, 23, 1887-1899.
(9) La¨mmerhofer, M.; Lindner, W. J . Chromatogr. A 1998, 829, 115-
125.
(10) Tobler, E.; La¨mmerhofer, M.; Lindner, W. J . Chromatogr. A
2000, 875, 341-352.
(11) La¨mmerhofer, M.; Peters, E. C.; Yu, C.; Svec, F.; Fre´chet, J .
M. J .; Lindner, W. Anal. Chem. 2000, 72, 4623-4628.
(12) Maier, N. M.; Schefzick, S.; Lombardo, G. M.; Feliz, M.;
Rissanen, K.; Lindner, W.; Lipkowitz, K. B. J . Am. Chem. Soc. 2002,
124, 8611-8629.
8316 J . Org. Chem., Vol. 68, No. 22, 2003

Chiral Recognition of Peptide Enantiomers
F IGURE 2. Change of retention factors and enantioselectivity
in HPLC with increasing peptide length. Conditions: mobile
phase, 80/20 methanol/0.5 M aqueous ammonium acetate,
adjusted to pHa 6.0 with acetic acid; flow rate, 1 mL‚min-1
25 °C; UV detection at 254 nm.
;
of almost -4.9 kJ ‚mol-1. From Figure 2 it becomes clear
that this loss of enantioselectivity is mainly caused by a
strongly reduced retention of the more strongly bound,
i.e., the second eluted, enantiomer. This observation
suggests a loss or at least a strong reduction of complex-
stabilizing binding increments within the association
structure of the (all-S)-tripeptide enantiomer and the
chiral selector relative to that of the dipeptide.
2.2. NMR Sp ectr oscop y. Low substance require-
ments, the possibility for solution-phase studies, the
1
intrinsic sensitivity, etc. make H NMR spectroscopy an
ideal tool to elucidate these changes of binding incre-
ments. NMR spectroscopy has been shown to be a
powerful tool for studying enantioselective associations
between a chiral selector and the analyte enantiomers
in solution in various cases. For peptide enantiomer
analytes investigations employing modified cyclodextrins
as chiral selectors have been reported.13,14 Conversely,
the use of peptides as chiral selectors has also been
studied by NMR spectroscopy.15-17 In several studies the
insights into the chiral recognition mechanisms gained
by NMR spectroscopy have served to interpret observa-
tions made during enantiomer separation experi-
ments.12,14,16-20
To study the (noncovalent) binding properties of the
transient diastereomeric complexes of SO1 and the
enantiomers of SA1 and SA2, respectively, a combination
of several 1D and 2D NMR experiments were carried out.
First, the stoichiometry of the complexes was determined
by continuous variation-type titration experiments. In-
formation on the conformational preferences of the chiral
selector and their changes upon complexation was ex-
tracted by analysis of intramolecular NOEs found in
NOESY spectra. Further, complexation-induced shifts of
both the selector’s and the selectand’s protons upon
enantioselective binding were assessed to identify specific
intermolecular interactions. Finally, intermolecular NOEs
were studied to gain information on the time-averaged
geometry of the complexes. To generate medium condi-
tions similar to those used in the chromatographic
experiments all experiments were carried out in methanol-
F IGURE 3. Proton numbering scheme for the studied selector
and selectands.
d4. In the following discussions the individual protons of
SO1, SA1, and SA2 are referred to according to the
numbering scheme shown in Figure 3.
2.3. Com p lexa tion Stoich iom etr y. The knowledge
of the complexation stoichiometry is a prerequisite for
investigations on the molecular recognition process. Thus,
the stoichiometry of the complexation of SO1 with the
more strongly bound (all-S)-enantiomers of SA1 and SA2,
respectively, was determined by a continuous variation
NMR titration protocol.21 The chemical shifts of the
proton H9 of SO1 were measured for different molar
ratios of the chiral selector and the selectands. The thus
acquired J ob plots (available as Supporting Information)
showed a maximum at molfractions of 0.5 for both (all-
S)-SA1 and (all-S)-SA2, which indicates a 1:1 stoichi-
ometry of the complexes. The ion-pair-type binding of the
chiral selector and the selectands seems to occur only at
the stronger basic quinuclidine nitrogen of SO1.
In a previous study,12 in which the complexation of
SO1 with the enantiomers of the related N-3,5-dini-
trobenzoylleucine was studied, the same complexation
stoichiometry of 1:1 was established for both enanti-
omers. It was expected that the same stoichiometry be
present for both enantiomers of the two peptide analytes.
Therefore, only the complexation stoichiometries of the
stronger binding enantiomers of the two peptides were
probed.
(13) Kano, K.; Hasegawa, H.; Miyamura, M. Chirality 2001, 13,
474-482.
(14) Su¨â, F.; Kahle, C.; Holzgrabe, U.; Scriba, G. K. E. Electrophore-
sis 2002, 23, 1301-1307.
(15) McEwen, I. Biopolymers 1993, 33, 933-942.
(16) De Lorenzi, E.; Massolini, G.; Molinari, P.; Galbusera, C.;
Longhi, R.; Marinzi, C.; Consonni, R.; Chiari, M. Electrophoresis 2001,
22, 1373-1384.
(17) Marinzi, C.; Longhi, R.; Chiari, M.; Consonni, R. Electrophoresis
2001, 22, 3257-3262.
(18) Ta´rka´nyi, G. J . Chromatogr. A 2002, 961, 257-276.
(19) Yamamoto, C.; Yashima, E.; Okamoto, Y. J . Am. Chem. Soc.
2002, 124, 12583-12589.
(20) Zhou, L.; Thompson, R.; Reamer, R. A.; Miller, C.; Welch, C.;
Ellison, D. K.; Wyvratt, J . M. J . Chromatogr. A 2003, 987, 409-420.
(21) Homer, J .; Perry, M. C. J . Chem. Soc., Faraday Trans. 1 1986,
82, 533-543.
J . Org. Chem, Vol. 68, No. 22, 2003 8317

Czerwenka et al.
F IGURE 5. Intermolecular NOEs observed for the 1:1 com-
plex of 6′-neopentoxy-9-O-tert-butylcarbamoylcinchonidine (anti-
open conformation) with (all-S)-DNB-Ala2 in a tentative model
of the complex.
nated form, SO1‚HCl, NOEs characteristic for the anti-
open conformation were found, though only one very
weak cross-peak for the syn-closed state. The value of
the coupling constant also indicated the practically
exclusive presence of the anti-open conformation. These
results show that the protonation of SO1 significantly
changes its conformational status. The complexes of the
more strongly bound (all-S)-enantiomers of SA1 and SA2,
respectively, with SO1 showed a similar picture to the
one established for SO1‚HCl. Several strong NOEs
characteristic for the anti-open conformation were found,
the presence of which was supported by a vanishing
3J H8H9. However, one pronounced cross-peak character-
istic for the anti-closed conformation was also found.
These findings may be rationalized by a hybrid confor-
mation combining properties of both the anti-open and
anti-closed conformers. It seems that the complete as-
sociation of SO1 with the more strongly bound analyte
enantiomers, which includes not only a long-range ion-
pairing event but also short-range interactions (hydrogen-
bonding, π-π-stacking, and steric interactions5), induces
this conformation that does differ somewhat from the
solely protonated chiral selector. For the complexes of the
less strongly bound (all-R)-enantiomers of SA1 and SA2,
respectively, with SO1, NOE cross-peaks significant for
all three conformations were observed. The magnitude
F IGURE 4. Difference of complexation-induced 1H NMR
shifts for selected protons (for numbering see Figure 3) of the
1:1 complexes of 6′-neopentoxy-9-O-tert-butylcarbamoylcin-
chonidine with the (all-R)- and (all-S)-enantiomers of (a) DNB-
Ala2 and (b) DNB-Ala3 (the shifts are reported relative to the
free forms). Spectra acquired in methanol-d4.
2.4. Con for m a tion s of th e Selector in Its F r ee,
P r oton a ted , a n d Com p lexed F or m s. The conforma-
tional state of the chiral selector is of great importance
for its enantiodiscriminating properties, as the relative
arrangement of its structural features defines the dimen-
sion as well as functionalities of the enantioselective
binding site. Consequently, the conformational behavior
of the basic SO1 was investigated in detail, also consid-
ering potential conformational changes induced upon
complexation with the acidic analyte enantiomers.
Prior investigations have identified ionic attraction as
the primary interaction force during the complexation of
the selector and the selectands,2,3 which guides the two
components toward each other before other interactions
come into effect. Therefore, the conformations of SO1
were studied for its free form, the protonated state
(resembling an intermediate state of the complexation
process), and in the presence of the analytes. A detailed
description of the conformations of the free base (SO1)
and the protonated form (SO1‚HCl) has been published
recently.12 Three possible conformations of the chiral
selector were established: an anti-open one, in which the
lone pair of the quinuclidine nitrogen points away from
the quinoline ring (see also Figure 5), and two closed ones
(syn and anti, depending on the orientation of the
carbamoyl residue relative to the 6′-residue), in which
the lone pair is directed toward the aromatic ring. The
relative populations of these conformational states were
assessed by the presence/absence of inter-ring NOEs
between the quinoline and quinuclidine protons and the
3
of the J H8H9 coupling constant indicated open/closed
ratios of 79/21 and 90/10 for the SA1 and SA2 complexes,
respectively. Therefore, the conformational states of these
two complexes can be regarded as intermediates between
those of the free (nonprotonated) selector and its com-
plexes with the corresponding (all-S)-enantiomers.
Taking into account all results described above, the
chiral discrimination between the (all-R)- and (all-S)-
enantiomers of SA1 and SA2 is reflected in the confor-
mational status of the chiral selector. Moreover, the fact
that the conformations of all four complexes deviate from
that of SO1‚HCl shows that the conformational changes
upon complexation are not solely related to the protona-
tion of SO1 but must also include additional molecular
interaction increments, which obviously differ in mag-
nitude and/or direction (attraction or repulsion) for the
enantiomers. However, the conformational differences
between the (all-R)- and the (all-S)-complexes were
similar for both the di- and the tripeptide. This similarity
3
magnitude of the J H8H9 coupling constant.12
For SO1 all three conformations were found with an
approximate open/closed ratio of 63/37. For the proto-
8318 J . Org. Chem., Vol. 68, No. 22, 2003

Chiral Recognition of Peptide Enantiomers
TABLE 2. 1H NMR Ch em ica l Sh ifts (δ) of Selector , Selecta n d s, a n d Selector -Selecta n d Com p lexes a n d
Com p lexa tion -In d u ced Sh ifts (∆δ)
δ (ppm)a
∆δ (ppm)b
free
ionized
formse
(S,S)-
complex
(R,R)-
complex
(S,S,S)-
complex
(R,R,R)-
complex
(S,S)-
complex
(R,R)-
complex
(S,S,S)-
complex
(R,R,R)-
complex
protonc
H2′
H3′
H5′
H7′
H8′
H2a
H2b
H3
H4
H5a
H5b
H6a
H6b
H7a
H7b
H8
H9
H10
H11a
H11b
H12af
H12bf
H14
formsd
8.65
7.54
7.50
7.46
7.96
2.65
3.08
2.35
1.83
1.61
1.91
2.72
3.26
1.69
1.83
3.24
6.47
5.81
4.95
5.00
3.87
3.87
1.12
n.d.g
1.27
4.43
1.44
n.d.g
4.63
1.53
n.d.g
9.10
9.13
4.37
1.41
n.d.g
4.42
1.41
n.d.g
4.60
1.53
n.d.g
9.11
9.14
8.78
7.71
7.59
7.63
8.07
3.35
3.65
2.86
2.18
2.03
2.32
3.39
3.73
1.86
2.30
3.84
6.87
5.82
5.09
5.17
3.93
4.02
1.14
n.d.g
1.29
4.22
1.38
n.d.g
4.63
1.53
n.d.g
9.13
9.13
4.15
1.36
n.d.g
4.39
1.41
n.d.g
4.63
1.57
n.d.g
9.12
9.13
8.60
7.50
7.14
7.20
7.75
3.14
3.75
2.73
2.05
1.93
2.21
3.35
3.44
1.65
2.12
3.65
7.12
5.75
5.00
5.09
3.59
3.68
1.13
n.d.g
1.41
4.22
1.44
n.d.g
4.78
1.59
n.d.g
8.80
8.62
8.68
7.57
7.48
7.51
7.99
3.04
3.40
2.63
2.02
1.84
2.11
3.08
3.50
1.78
2.06
3.57
6.63
5.82
5.03
5.09
3.87
3.90
1.13
n.d.g
1.28
4.33
1.42
n.d.g
4.63
1.53
n.d.g
9.11
9.12
8.66
7.56
7.38
7.44
7.93
3.10
3.47
2.68
2.04
1.88
2.16
3.12
3.51
1.75
2.10
3.63
6.80
5.80
5.03
5.10
3.79
3.84
1.13
n.d.g
1.31
8.68
7.57
7.49
7.51
8.00
3.07
3.42
2.66
2.04
1.86
2.13
3.11
3.52
1.79
2.09
3.62
6.65
5.82
5.03
5.10
3.87
3.91
1.13
n.d.g
1.28
-0.05
-0.04
-0.36
-0.26
-0.22
0.49
0.67
0.38
0.22
0.33
0.30
0.64
0.18
-0.03
0.29
0.41
0.65
0.04
0.04
-0.02
0.05
0.03
0.39
0.32
0.28
0.19
0.23
0.20
0.36
0.25
0.09
0.23
0.33
0.16
0.01
0.08
0.09
0.00
0.03
0.01
n.d.g
0.01
-0.10
-0.02
n.d.g
-0.01
0.00
n.d.g
0.00
-0.01
0.02
0.02
-0.12
-0.03
-0.03
0.45
0.39
0.32
0.21
0.27
0.25
0.40
0.25
0.06
0.26
0.39
0.33
0.04
0.04
-0.01
0.05
0.03
0.42
0.35
0.30
0.21
0.26
0.22
0.39
0.27
0.10
0.25
0.38
0.17
0.01
0.09
0.10
0.00
0.04
0.01
n.d.g
0.01
-0.07
0.05
0.09
-0.01
0.08
0.10
-0.27
-0.19
0.01
-0.08
-0.03
0.01
H16
H18
n.d.g
0.13
n.d.g
0.04
H202
H203
H204
H206
H207
H208
H211+ H215
H213
H302
H303
H304
H306
H307
H308
H310
H311
H312
H315 + H319
H317
-0.21
0.00
n.d.g
0.14
0.06
n.d.g
-0.30
-0.51
4.28
1.39
n.d.g
4.41
1.40
n.d.g
4.63
1.54
n.d.g
9.05
9.05
4.28
1.39
n.d.g
4.41
1.41
n.d.g
4.61
1.55
n.d.g
9.10
9.12
-0.10
-0.02
n.d.g
-0.01
-0.01
n.d.g
-0.09
-0.02
n.d.g
-0.01
0.00
n.d.g
0.03
0.01
0.01
0.02
n.d.g
n.d.g
-0.06
-0.09
-0.01
-0.01
a
b
All ppm values are referenced to residual methanol (δ ) 3.31 ppm). Complexation-induced shifts (reported relative to the free forms);
negative values denote upfield shifts. c For numbering of protons see Figure 3. SO1/SA1, SA2. e SO1‚HCl/sodium salts of SA1, SA2.
d
f Arbitrary assignment of the shifts to the two protons. Not detected.
g
is remarkable, since the levels of enantiodiscrimination
are considerably different for these two analytes.
2.5. Com p lexa tion -In d u ced Sh ifts u p on En a n tio-
selective Bin d in g. Chemical shift changes occurring
upon complexation of the chiral selector and a selectand
enantiomer were studied to gain information on the
intermolecular interactions responsible for enantiomer
discrimination as well as on the relative orientation of
the association partners in the complexes. Because the
primary interaction involved in complexation is a (non-
enantioselective) ion-pair formation, the protonation and
deprotonation of the chiral selector and the selectand,
respectively, are expected to induce strong complexation-
induced shifts (CISs).
comparing the chemical shifts observed for the free base
SO1 and SO1‚HCl as well as for the free acids SA1 and
SA2 and their sodium salts (Table 2). Protonation of SO1
deshielded all protons, leading to pronounced downfield
shifts of the protons of the quinuclidine moiety and of
H9 (generally in the range of 0.4-0.7 ppm), while the
shifts of the quinoline protons were considerably smaller
(δ ≈ 0.1-0.2 ppm). These observations confirm that the
quinuclidine nitrogen is the sole site of protonation
(compare section 2.3). All other protons experienced only
small chemical shifts, with those of the tert-butyl groups
being practically unaffected. Deprotonation of SA1 and
SA2 had a significant impact only on the protons at-
tached to the C-terminal chiral center (H202/H302), which
shifted upfield by ≈0.2 ppm.
To deconvolute the effects of ionization from the other
intermolecular interactions present in the complexes, the
changes induced by the protonation of SO1 and the
deprotonation of SA1 and SA2 were investigated by
By using this information, the chemical shifts of the
protons in the complexes were evaluated with respect to
significant CISs. For the complexes of the SA1 enanti-
J . Org. Chem, Vol. 68, No. 22, 2003 8319

Czerwenka et al.
omers these CISs are depicted in Figure 4a alongside the
respective ionization-induced CISs. The complex of SO1
with (all-S)-SA1 showed extensive downfield shifts of the
quinuclidine protons, indicating extensive protonation.
However, protons H2a, H6b, H7a, and H7b exhibited con-
siderably smaller shifts relative to the free form of SO1
than were found for SO1‚HCl (∆ > 0.2 ppm, Table 2).
This can be explained by SA1 being a weaker acid than
HCl and the different geometry of the SA1 anion com-
pared to chloride. Very pronounced upfield shifts were
noted for the quinoline’s H5′, H7′, and H8′ protons. The
same behavior was observed for the H12a,b protons located
in close proximity to H5′ and H7′. An analogous trend was
found for the aromatic protons of (all-S)-SA1, with H213
being shifted upfield 0.51 ppm. These findings provide
striking evidence for strong π-π-interactions between the
aromatic groups of SO1 and (all-S)-SA1, which effect
mutual shielding. Another strong CIS occurred for the
H9 signal. The downfield shift (0.65 ppm) can only partly
be explained by the deshielding effect of SO1 protonation
(see Figure 4a). The additional shift may be explained
by the change of conformation upon complexation (see
above) moving H9 into deshielding regions of the quino-
line moiety or the carbamate carbonyl group. Finally, an
interesting downfield shift of the H18 protons of the tert-
butyl carbamate was noted. As the carbamoyl group of a
related chiral selector was found to form a hydrogen bond
with the C-terminal amide of (all-S)-SA1,5 it seems very
likely that the carbamoyl group of SO1 engages in
hydrogen bonding with (all-S)-SA1, which then causes
the deshielding of the H18 protons. Contrary to these
findings, no specific CISs were found for the complex of
SO1 with (all-R)-SA1. The upfield shifts of the quinu-
clidine protons result from the ion-pair formation that
constitutes the primary interaction force for this complex
as well. For most protons these shifts are somewhat
smaller than those observed for the complex of SO1 with
(all-S)-SA1, which may reflect a lower proportion of the
selectand being present in the complexed form. This
relates to a smaller complexation constant for this
enantiomer, which as a consequence leads to lower
chromatographic retention (see above). The less pro-
nounced upfield shift of H9 compared to the (all-S)-SA1
complex (0.16 ppm versus 0.65 ppm) results from the
different conformation (see above). The signals of the
aromatic protons of both selector and analyte remained
almost unaffected by the complexation process, indicating
the absence of π-π-interactions, which thus seem to play
an important role for enantiomer discrimination.
F IGURE 6. NOESY spectra of the 1:1 complex of 6′-neopen-
toxy-9-O-tert-butylcarbamoylcinchonidine with the (all-S)-
enantiomer of DNB-Ala2 showing intermolecular interactions
of (a) the analyte’s DNB group (ortho protons H211 and H215
)
and (b) the selector’s neopentyl group (tert-butyl protons H14).
showed that some differences still exist (see Figure 4b);
however, their magnitude is much smaller than for the
complexes of SA1. Thus, the stereoselective π-π-interac-
tions, which are a major contributor to enantioselectivity,
are largely lost upon elongation of the peptide chain from
the dipeptide (SA1) to the tripeptide (SA2). Most prob-
ably, steric constraints do not allow ion-pairing and π-π-
stacking to take place simultaneously. The resonances
of the H18 protons are almost unchanged relative to the
free selector, suggesting the absence of hydrogen bonding
for the complexes of both SA2 enantiomers. Overall, the
differences between the diastereomeric complexes of the
two enantiomers of SA2 are much smaller than those
between the complexes of the SA1 enantiomers, which
is also reflected by the smaller CIS differences of H9 (0.16
ppm versus 0.49 ppm).
2.6. Sp a tia l Ar r a n gem en t of Selector a n d Selec-
ta n d s in Solu tion . To further elucidate the geometry
of the complexes, i.e., the relative arrangement of the
chiral selector and the analyte enantiomer, the NOESY
spectra (available as Supporting Information) were
scanned for the presence of resonances evoked by close
intermolecular contacts.22,23 For the complex of SO1 and
(all-S)-SA1 several intermolecular NOEs were detected
(Figures 5 and 6). One was found between an aromatic
ortho-proton of SA1 (H211/215) and H5′, further supporting
the proposed presence of face-to-face π-π-stacking. The
H211/215 protons of the 3,5-dinitrobenzoyl group also
showed intermolecular NOEs with the protons of the
selector’s neopentyl group (H12a,b and H14), suggesting a
The results obtained for the complexes of SO1 with
the SA2 enantiomers (see Figure 4b and Table 2) showed
a very different picture. The upfield shifts of the quinu-
clidine protons were practically identical for both com-
plexes, indicating a very similar degree of protonation.
Their extent was considerably smaller than that of the
shifts noted for the complex of SO1 with (all-S)-SA1. This
finding seems to point toward small complexation con-
stants for both SA2 enantiomers, their similar magnitude
resulting in only modest enantioselectivity. This deduc-
tion is supported by the HPLC results, where small
retention factors were found for both enantiomers of SA2.
The comparison of the CISs observed for the aromatic
protons of both SO1 and the (all-R)- and (all-S)-enanti-
omers of SA2 for the two diastereomeric complexes
(22) Pirkle, W. H.; Pochapsky, T. C. J . Am. Chem. Soc. 1986, 108,
5627-5628.
(23) Pirkle, W. H.; Pochapsky, T. C. J . Am. Chem. Soc. 1987, 109,
5975-5982.
8320 J . Org. Chem., Vol. 68, No. 22, 2003

Chiral Recognition of Peptide Enantiomers
TABLE 3. Aver a ged P oten tia l En er gies, Com p on en t En er gies, a n d Solva tion En er gies (k J ‚m ol-1) for th e In ter a ction s of
th e Com p lexes of SO1 w ith SA1 a n d SA2
SO1 +
(all-R)-SA1
SO1 +
(all-S)-SA1
SO1 +
(all-R)-SA2
SO1 +
(all-S)-SA2
∆Ea
∆Ea
total energy
stretch
bend
-651.3
156.9
205.7
120.0
28.7
-939.6
-223.0
-658.1
155.0
206.6
118.4
26.3
-956.4
-208.0
-6.8
-1.9
0.9
-1.6
-2.4
-16.8
15.0
-730.5
169.3
224.2
128.6
36.3
-1044.5
-244.3
-734.5
170.7
222.5
126.9
32.3
-1048.2
-238.5
-4.0
1.4
-1.7
-1.7
-4.1
-3.7
6.2
torsion
van der Waals
electrostatic
solvation
a
A negative value means that the (all-S)-complex is more stable.
well-defined arrangement of the quinoline ring plus its
6′-residue and the N-terminus of the 3,5-dinitrobenzoy-
lated peptide, as well as with the protons of the tert-
butylcarbamoyl group (H18). The observation of NOE
signals between the H211/215 protons and several protons
of SO1, which are located in different regions of the
selector molecule, may be explained by the simultaneous
presence of two or more different relative arrangements
of the selector and the analyte within the complex.
Finally, an intermolecular NOE was noted between the
tert-butyl protons of the neopentyl group of SO1 (H14) and
the proton attached to the chiral center of the C-terminal
amino acid of SA1 (H202). On the contrary, no inter-
molecular NOEs were found in the spectrum of the
complex of SO1 with (all-R)-SA1. This observation is
consistent with the weaker binding of the (all-R)-enan-
tiomer seen in the HPLC and CIS studies as well as with
the absence of π-π-stacking. For the SA2 complexes no
intermolecular NOEs were observed for both enanti-
omers, which is consonant with the considerably weaker
intermolecular interactions noted for these two complexes
above and the almost complete loss of π-π-stacking also
for the more strongly bound (all-S)-enantiomer.
organic solvent. Total energies and solvation energies for
these simulations are presented in Table 3. Because the
total interaction energies consist of energies from bond
stretching, angular deformation, and other contributing
force field terms, the time-averaged component energies
for all four complexes are also presented in Table 3. The
differences in energies between the (all-R)- and (all-S)-
complexes are given by ∆E, where a negative value
means the (all-S)-complex is more stable.
The results from these nanosecond simulations indi-
cate that the more retained enantiomer, corresponding
to the more stable complex, has (all-S)-configuration for
both SA1 and SA2. This is consonant with experiment.
The computed energy difference for the SA1 complexes
(-6.8 kJ ‚mol-1) underestimates slightly the experimental
value (-7.4 kJ ‚mol-1) while for the SA2 complexes the
energy difference (experimental ) -2.6 kJ ‚mol-1, com-
puted ) -4.0 kJ ‚mol-1) is overestimated. Hence, from
these simulations we find that the computed retention
order is correct for both selectands, while the differential
free energies of binding are reproduced well by the
AMBER* force field with use of a water continuum model
for SA1 but less well for SA2.
2.7. En er getics of Com p lexa tion w ith SO1. Com-
putational chemistry allows one to extract information
about intermolecular interactions not amenable to ex-
perimentation. However, before one can assess these
structural details, it must be ensured that essential
experimental findings, such as the HPLC elution order,
the differential free energies of the competing diastere-
omeric complexes, and the intra- and intermolecular
NOEs, are reproduced by the computations.
Anticipating that π-π-interactions may be involved in
the intermolecular association of the selector with the
selectands, the computations were performed with use
of a force field approach. This method is preferable over
density functional theory (DFT) for the following rea-
sons: With the latter method only picosecond simulation
times are possible compared to nanosecond results ob-
tainable with a force field. Second, as described in a
recent paper,24 DFT treats π-π-stacking poorly. For
example, DFT methods fail completely to describe the
attraction in the benzene dimer.24
We now consider the component energy terms in Table
3. For SA1, less stretching deformation is observed for
the more stable (all-S)-complex than for the (all-R)-
complex (∆E ≈ 2 kJ ‚mol-1). For SA2 this difference is
≈1 kJ ‚mol-1 and it favors the (all-R)-complex. The
bending deformation energies for the SA1 complexes are
similar in magnitude and favor the (all-R)-enantiomer,
while they are larger for the SA2 complexes and favor
the (all-S)-enantiomer. The torsional deformation ener-
gies are similar for the complexes of both enantiomers
of SA1 as well as SA2. For both selectands the more
stable complex contains less torsional deformations than
the less stable one. Even though these differences are
small in magnitude, it is evident that there are clear
differences in the energetics of the SA1 enantiomers
which are very well discriminated by SO1 versus the SA2
enantiomers which are relatively poorly discriminated.
The above-mentioned component energies are all con-
sidered to arise from “bonding” terms in the force field
because atoms involved in those terms are connected to
one another contiguously. Contrarily, the remaining
three component energies in Table 3 involve “nonbonded“
interactions, i.e., atom-atom interactions that are 1,3 or
greater in nature. In the SA1 simulations the van der
Waals (vdW) terms are within 2.4 kJ ‚mol-1 of one
another. For SA2 the energy difference is larger (4.1
kJ ‚mol-1). In both instances, however, there is less vdW
To account for the polar chromatographic conditions
used to evaluate the differential free binding energies we
implemented a polar solvent model in our simulations.
This solvent model is a continuum model of water and it
is expected that a pure water model may not reproduce
exactly the experimental condition of a mixed hydro-
(24) Czernek, J . J . Phys. Chem. A 2003, 107, 3952-3959.
J . Org. Chem, Vol. 68, No. 22, 2003 8321

Czerwenka et al.
TABLE 4. In ter m olecu la r En er gies (k J ‚m ol-1) of th e Com p lexes of SO1 w ith SA1 a n d SA2
SO1 +
(all-R)-SA1
SO1 +
(all-S)-SA1
SO1 +
(all-R)-SA2
SO1 +
(all-S)-SA2
∆Ea
∆Ea
total energy
van der Waals
electrostatic
-505.7
-22.8
-482.9
-529.5
-23.1
-506.4
-23.8
-0.3
-23.5
-480.0
-6.1
-473.9
-497.9
-14.3
-483.0
-17.8
-8.2
-9.1
a
A negative value means that the (all-S)-complex is more stable.
energy associated with the (all-S)-complexes, indicating
that there exists a better selector-analyte fit for the (all-
S)-complexes than for the (all-R)-complexes. The elec-
trostatic energies heavily favor the more stable (all-S)-
complex by 16.8 kJ ‚mol-1 for SA1. The genesis of this
stabilization is that the (all-S)-complex fits together
better than the (all-R)-complex. Because the electrostatic
attractions fall off with the square of the distance
between charges, even very small structural changes can
give rise to large stabilization energies. We point out here
that the energies described above refer to all atom-atom
interactions in the complex, i.e., the energies in Table 3
are a composite of both intramolecular and intermolecu-
lar energies. Below we will extract from these data only
the intermolecular energies and discuss them separately
with respect to chiral discrimination. For SA2, in con-
trast, we find that the electrostatics and the vdW energy
differences are comparable in magnitude and favor the
(all-S)-complex but to a lesser extent than for SA1.
2.8. In t er m olecu la r F or ces Lea d in g t o Ch ir a l
Discr im in a tion . To obtain a more detailed picture of
the enantiodiscriminating process, the intermolecular
forces responsible for chiral discrimination need to be
extracted from the simulation data. The intermolecular
terms that need to be considered are only the nonbonded
vdW energy and the electrostatic energies, hereafter
simply called “nonbonded“ energies. It is important to
recognize that the nonbonded energies in Table 3 are
composites of intramolecular and intermolecular ener-
gies. In molecular mechanics force fields the nonbonded
interactions are computed pairwise additive meaning
that all atoms experience all other nonbonded atoms
regardless of whether those other atoms are part of the
same molecule or of a different one. We have written a
program that extracts from these data only the intermo-
lecular components.25 Those values, averaged over the
simulation time period, are presented in Table 4.
favoring the (all-S)-enantiomers, the larger enantiodis-
criminating force is attributed to the long-range electro-
static effects rather than to the short-range dispersion
forces. We point out that this discriminating force does
not arise exclusively from the ion-pair formed between
the carboxylates of the selectands and the quinuclidine
nitrogen of the selector. Even though those regions of the
molecule are formally charged, all other atoms have
partial atomic charges and they too are contributing to
the electrostatic discriminating energies. In a forthcom-
ing section we will focus on which fragments of the
selector are actually most responsible for chiral discrimi-
nation. At this point we again highlight the difference
between the complexes of the well-separated SA1 and
the poorly resolved SA2 enantiomers. In the former
system the intermolecular chiral discrimination is almost
exclusively effected by electrostatic forces, while in the
latter the vdW and electrostatic forces are contributing
almost equally.
The intermolecular energies listed in Table 4 are
average values; they do not indicate how those energies
can fluctuate in time nor do they describe the distribution
of energies that occurred during the simulation time
period. During the simulation, as the molecules are free
to move about and collide with one another, both the
inter- and intramolecular forces have changed. There are
two ways of visualizing these changes, the first being
plots of how the intermolecular energies between SO1
and the analyte enantiomers change as a function of time
(trajectories). Huge variations in intermolecular energy
were observed as the system evolves in time with a
similar behavior for all four complexes. These trajectories
are available as Supporting Information. In particular,
due to thermal collisions the selectand moves and reori-
ents itself with respect to the selector, albeit while
remaining in a localized binding region around SO1.
During this reorientation both the electrostatic and vdW
energies are undergoing large fluctuations in magnitude
(the sharp, noise-like spikes in the figures).
Several points of significance are derived from Table
4. First, both the van der Waals and the electrostatic
interactions between the molecules are attractive (nega-
tive energies). Second, the dominant stabilizing force
holding each complex together is from the electrostatic
term. In this regard we see that for the SA1 complexes
95% of the total intermolecular energy is attributable to
electrostatic attractions, while 97-99% is due to electro-
statics for the SA2 complexes. Third, the most insightful
aspect of these results concerning chiral recognition is
the difference in energy between the diastereomeric
complexes, denoted as ∆E in Table 4. This difference is
a measure of chiral discrimination.
Another way of visualizing the intermolecular forces
is to make plots of the distribution of energies that were
observed during the total simulation time period. In these
figures the energies, binned in units of 2 kJ ‚mol-1, are
plotted as a function of the number of times a particular
energy was encountered during the simulation time
period. As the figures for the complexes of SA1 and SA2
showed similar patterns, only the plots for SA1 are
presented here (Figure 7), those for SA2 being available
as Supporting Information. In the plots of the vdW
energies (Figure 7e,f) note that some values are negative
(net attraction) while others are positive (net destabiliza-
tion). In contrast, the electrostatic energies (Figure 7c,d)
are always negative and attractive.
While we find both the vdW and electrostatic attrac-
tions between the molecules in the binary complexes
The distribution of electrostatic interactions between
selector and selectand is unimodal with a quasi-normal
(25) Peterson, M. A. Understanding Enantiodifferentiation Through
Molecular Simulations, Ph.D. Thesis, Purdue University, 1996.
8322 J . Org. Chem., Vol. 68, No. 22, 2003

Chiral Recognition of Peptide Enantiomers
F IGURE 7. Distributions of the total intermolecular energy (a, b), the intermolecular electrostatic energy (c, d), and the
intermolecular van der Waals energy (e, f) for the (all-S)-enantiomer (a, c, e) and the (all-R)-enantiomer (b, d, f) of SA1 interacting
with SO1.
Waals and electrostatic contributions. Note that the sums
of energies in Table 5 are slightly smaller than that in
Table 4 because C9 was not included in the calculation
of the fragment energies.
From this analysis we find that the fragment most
responsible for holding the complexes together is the
quinuclidine group (see Table 5). The second most
stabilizing interaction comes from the quinoline ring.
This ring is seen in molecular graphics movies to form
π-π-stacking associations with the 3,5-dinitrobenzoyl
group of the analyte (see below). The carbamate frag-
ment, being positive in energy, is actually repulsive. This
is a consequence of the molecules being held together
tightly by electrostatic forces and some parts of the
F IGURE 8. Partitioning of SO1 into four molecular frag-
ments.
complexes are experiencing steric repulsions due to
congestion.
(Gaussian) distribution for all complexes. The centers of
the distributions of the more tightly bound (all-S)-
The second major point concerning these results in-
volves the difference each fragment experiences when
associating with each selectand enantiomer. This differ-
ence is a measure of chiral discrimination. For the SA1
complexes the fragment showing the greatest difference
between the enantiomers is the carbamate moiety. In
contrast, the enantiodiscrimination arising from the
carbamate when SA2 binds to SO1 is negligible; in this
system the most enantiodiscrimination is associated with
the quinoline moiety. In an earlier paper where simula-
tions were carried out for SO1 and a 3,5-dinitrobenzoyl
leucine analyte we found that the most discriminating
fragment was the carbamate.12 In that study the chro-
matographic enantioselectivity was extremely large (R )
32.6) as for the SA1 system described here. Hence, a
pattern is emerging from the simulation results: effective
chiral recognition by the cinchona alkaloid carbamate
receptors involves large enantiodiscrimination by the
carbamate moiety. It is clear here that the SA2 enanti-
omers, which are resolved with considerably less selectiv-
ity, do not experience this type of discrimination. This is
in accordance with the CISs results (see above). Finally,
enantiomers are shifted to more negative (attractive)
energies than for the corresponding (all-R)-antipodes as
expected. The plots further illustrate that the shapes of
the distributions for the (all-R)- and the (all-S)-enanti-
omers interacting with SO1 are not much different from
one another, indicating that only very subtle differences
in intermolecular interactions are involved in the dis-
crimination process.
2.9. F r a gm en t In ter a ction s. To better understand
how SO1 binds and discriminates between the analyte
enantiomers we consider now the interactions of molec-
ular fragments that constitute the chiral selector. We
emphasize here that fragment energies are accessible by
computations only and not by experiment. Partitioning
the selector into fragments is a subjective and arbitrary
decision. However, because there exist several charac-
teristic, identifiable groups comprising this SO we have
divided the molecule into the four segments that are
linked to C9 as illustrated in Figure 8.
In Table 5 we compile the average intermolecular
energies associated with each of these fragments inter-
acting with the analyte molecule as well as the van der
J . Org. Chem, Vol. 68, No. 22, 2003 8323

Czerwenka et al.
TABLE 5. In ter m olecu la r En er gies (k J ‚m ol-1) for F r a gm en ts of th e Com p lexes of SO1 w ith SA1 a n d SA2
SO1 +
SO1 +
SO1 +
SO1 +
fragment
quinoline
intermolecular energy
(all-R)-SA1
(all-S)-SA1
∆Ea
(all-R)-SA2
(all-S)-SA2
∆Ea
total
-84.6
-45.5
-39.2
-410.4
38.5
-448.8
68.3
-14.3
82.6
-100.5
-47.0
-53.5
-400.2
39.4
-439.6
48.0
-14.2
62.2
-15.9
-1.5
-14.3
10.2
0.9
9.2
-20.3
0.1
-75.8
-33.6
-42.2
-416.7
39.9
-456.5
67.7
-10.8
78.5
-92.1
-35.2
-56.9
-414.9
38.8
-453.6
67.0
-16.0
83.0
-16.3
-1.6
-14.7
1.8
-1.1
2.9
-0.7
-5.2
4.5
-0.6
0.0
van der Waals
electrostatic
total
van der Waals
electrostatic
total
van der Waals
electrostatic
total
van der Waals
electrostatic
quinuclidine
carbamate
-20.4
1.0
0.4
hydrogen H9
-15.6
0.5
-16.2
-14.6
0.9
-15.5
-15.6
0.2
-15.6
-16.2
0.2
-16.2
0.7
-0.6
a
A negative value means that the (all-S)-complex is more stable.
we see from Table 5 that for fragments that are most
discriminating, the differential interaction energy (∆E)
is larger for electrostatic contributions than for vdW
contributions. Hence, these fragments discriminate pri-
marily by long-range electrical effects rather than by
short-range dispersion forces.
Another way to discuss these fragmentation energies
is to plot the distribution of intermolecular energies for
each fragment. For each fragment the electrostatic ener-
gies, the vdW energies, and the total intermolecular
energies of all four complexes are available as Supporting
Information. Regardless of which enantiomer binds to
SO1 we find that most of the plots are similar in shape,
varying only slightly in their skewness. In some in-
stances, however, there are distinct differences, e.g. for
the electrostatic energy distributions of the quinoline
fragment for the complex of SO1 with SA1, where the
distribution for the (all-S)-enantiomer has bimodal char-
acter while that for the (all-R)-enantiomer is unimodal.
The shapes of these distributions may serve as key
signatures associated with chiral recognition, a topic that
we are pursuing.
2.10. Str u ctu r al Featu r essCom par ison s with NMR
Da t a . One experimental key finding is that an inter-
molecular NOE exists between an aromatic ortho-proton
(H211/215) of (all-S)-SA1 and H5′ of SO1 in the complex.
The propinquity of the π-acidic and π-basic aromatic rings
(visualized from MD trajectories as π-π-stacking) ex-
tracted from the molecular simulation data is consistent
with this finding. An illustration of the π-π-stacking
distance is shown in Figure 9. It is to be noted in this
figure that both enantiomers form π-π-stacking but the
more stable (all-S)-enantiomer can form better π-π-
stacking, which shows up as a greater number of short-
range distances in the histogram.
The H211/215 protons also sense the hydrogens of the
selector’s neopentyl group. The average distance between
the H211/215 protons and the terminal methyl protons of
the neopentyl group (H14) was computed to be 5.85 Å.
This distance is larger than one anticipates for two
reasons. First, we are using a pure water solvent to carry
out the simulations compared to a methanol solution used
experimentally. The more polar aqueous phase is over-
solvating the ionic complex making it less tight than it
would be in a less polar solvent. Second, and more
important, we are reporting an average value here. This
average includes many configurations where part of the
system has dissociated and then reassociated during the
F IGURE 9. Distribution of intermolecular distances between
the aromatic ring centroids of SO1 and the (all-S)-enantiomer
(a) and the (all-R)-enantiomer (b) of SA1. The distance
between the centroids is related to π-π-stacking in the
complexes.
simulation. What is important is that there exist many
short-range interactions which relate to the NOE (in
addition to the non-NOE-inducing long-range contacts).
A relevant paper describing short-range and long-range
proton-proton distances obtained from MD simulations
and their relationship to NOE observations has been
published by Feller et al.26
Another NOE of the aromatic ortho-protons also indi-
cates the proximity of the carbamate’s tert-butyl group.
The computed average distance between these hydrogens
is 4.87 Å, a value that is again somewhat overestimated
for the same reasons as above. Finally, the proton
attached to the stereogenic center of the C-terminal
amino acid of SA1 (H202) shows an interaction with the
tert-butyl protons of the neopentyl group of SO1. The
computed average distance between these protons is 5.03
Å. Hence, most intermolecular NOEs found experimen-
tally could be computationally reproduced in terms of
appropriate distances between the respective protons.
3. Con clu sion s
An advanced understanding of the chiral recognition
mechanism governing the enantioselective binding of
N-3,5-dinitrobenzoyl derivatives of dialanine and triala-
(26) Feller, S. E.; Huster, D.; Gawrisch, K. J . Am. Chem. Soc. 1999,
121, 8963-8964.
8324 J . Org. Chem., Vol. 68, No. 22, 2003

Chiral Recognition of Peptide Enantiomers
nine peptides (“selectands” SA1 and SA2, respectively)
and 6′-neopentoxy-9-O-tert-butylcarbamoylcinchonidine
(SO1) has been achieved employing a combination of
chromatographic and NMR spectrocopic experiments as
well as molecular modeling.
nances with intermolecular NOEs involving the aromatic
ortho-protons of the 3,5-dinitrobenzoyl group of SA1, once
again indicating the presence of π-π-stacking. No inter-
molecular NOEs were observed for the two complexes of
the (all-R)-configured analytes nor for the complex of (all-
S)-SA2, which is in agreement with a less tight binding
of the selectands to the selector in these complexes.
Besides the NMR investigations, nanosecond stochastic
dynamics simulations on the complexes were carried out
by using a polar continuum model for solvent. The
simulation results agree with all experimental facts
including the following: (1) The retention orders were
correctly computed with the (all-S)-enantiomers being
longer retained on the column. (2) The computed energy
difference for the well-separated SA1 was reproduced
very well, while that for SA2 is slightly overestimated.
(3) Most intermolecular NOEs depicted in Figure 5 are
accounted for by close contacts between the respective
atom pairs during the simulation. On the basis of the
good agreement between experiment and theory, infor-
mation extracted from the simulations was deemed
reliable.
Force field component energies for the complexes were
analyzed and differences between the complexes were
explained. More important, however, was the extraction
of the intermolecular energies giving rise to complex
formation as well as to stereodiscrimination. In all
instances the intermolecular force most responsible for
formation of the complexes is electrostatic in nature. For
SA1, which is highly discriminated by SO1, we find that
chiral recognition is almost exclusively electrostatic,
while the poorly resolved SA2 has an equal mixture of
van der Waals and electrostatic forces contributing to the
discrimination.
To visualize the numerical data derived from the
simulations different forms of presentation were provided
for the purpose of better illuminating the differences
between the various complexes as well as to illustrate
their similarities. Histogram plots of energy distributions
were provided in addition to typical MD trajectories. Plots
of intermolecular van der Waals energies, intermolecular
electrostatic energies, and their sums (total intermolecu-
lar energies) revealed similar unimodal energy distribu-
tions with the center of those distributions shifted to
lower energies for the more stable complexes. The widths
and shapes of the distributions, albeit different for the
(all-S)- versus the (all-R)-complexes, are surprisingly
similar, thus highlighting the fact that the intermolecular
forces giving rise to chiral discrimination differ only in
very subtle ways.
It was assessed which fragments of the chiral selector
are most responsible for analyte binding as well as for
enantiomer discrimination. The roles of these fragments
differ for the two peptide analytes (SA1 versus SA2).
Intermolecular energies were presented as histogram
distribution plots to illustrate the similarities and dif-
ferences between the fragment energies of the various
complexes. It was found that effective chiral recognition
by the cinchona carbamate receptors involves large
stereoselective discrimination by the carbamate moiety.
The significantly diminished enantioselectivity ob-
served for tripeptide SA2 as compared to dipeptide SA1
is a consequence of the spatially more extended and
conformationally flexible backbone. The (all-S)-enanti-
Liquid chromatographic experiments performed with
a chiral stationary phase comprising an immobilized
version of SO1 showed that the (all-S)-enantiomers of
SA1 and SA2 were bound more strongly than the
corresponding (all-R)-enantiomers, indicating closely re-
lated enantioselective binding mechanisms. The level of
enantioselectivity, however, was found to depend on the
number of interlinked amino acids. While a very high
enantioselectivity factor of R ) 20.0 was observed for
SA1, the enantiomers of SA2 were significantly less well
separated (R ) 2.8). Thus, in terms of differential free
energy, the insertion of an additional amino acid into the
dipeptide backbone induced a dramatic loss of 4.9
kJ ‚mol-1. This observation suggested an unfavorable
change of stereoselective binding increments. Thus, the
relative contributions of the intermolecular interaction
increments being responsible for this phenomenon were
assessed by an in-depth NMR study of the four complexes
between SO1 and the (all-R)- and (all-S)-enantiomers of
SA1 and SA2, respectively.
Continuous variation-type titration experiments es-
tablished 1:1 stoichiometry for all complexes, indicating
that only the (more basic) quinuclidine nitrogen of SO1
is involved in the binding of the acidic analytes. The
analysis of inter-ring NOEs between the quinoline and
quinuclidine protons of SO1 provided evidence for the
existence of three specific conformations of the selector,
the relative populations of which depended on its proto-
nation status and the nature and stereochemistry of
analyte associated with SO1.
While a mixture of open and closed conformers was
detected for the free SO1, only the anti-open conforma-
tion was observed for the selector upon protonation or
complexation with the more strongly bound (all-S)-
enantiomers of both selectands. Contrarily, all three
conformations were found for the complexes of the (all-
R)-enantiomers, although a higher proportion of the anti-
open state than for the free SO1 was noted. The
conformational differences between the (all-R)- and the
(all-S)-complexes were similar for both the di- and the
tripeptide.
Complexation-induced shifts (CISs) were studied to
acquire information on the nature and strength of the
intermolecular interactions active in the complexes.
Thereby, the large effects of unspecific ionization of the
binding partners were taken into account separately to
allow an unhindered assessment of the effects of the
enantioselective interactions. The chemical shifts ob-
served for the two (all-R)-complexes of SA1 and SA2,
respectively, were practically identical in all cases and
no specific CISs were observed. The two (all-S)-complexes
showed several CISs in the same directions but their
extent was much smaller for the SA2 complex, indicating
a loss of hydrogen bonding and a strong reduction of π-π-
interactions.
Finally, the relative arrangements of the selector and
the selectands were investigated based on the observed
intermolecular NOEs. The NOESY spectrum of the
complex of SO1 with (all-S)-SA1 showed several reso-
J . Org. Chem, Vol. 68, No. 22, 2003 8325

Czerwenka et al.
of methanolic sodium methoxide, evaporating the solvent, and
drying the resulting products in vacuo.
omer of SA1 appears to provide ideal structural proper-
ties in terms of size as well as spatial disctance between
the carboxylic group and the 3,5-dinitrobenzoyl function
to enable simultaneous ion-pairing, hydrogen bonding,
and π-π-interaction with SO1. Evidently, the insertion
of an additional alanine unit compromises this ideal
situation, weakening or even disrupting these selector-
selectand interactions for (all-S)-SA2. Chromatographi-
cally, this effect should lead to a diminished overall
binding energy for (all-S)-SA2 relative to (all-S)-SA1,
which is fully consistent with the observed retention
behavior.
All compounds (free and ionized forms of the selector and
the selectands) were dissolved in methanol-d4 to give 20 mM
solutions. For the complexes of SO1 and the individual
enantiomers of SA1 and SA2, respectively, the single com-
pounds were dissolved in methanol-d4 and mixed in various
ratios for the measurements of the complexation stoichiometry
(J ob plots) and in 1:1 ratio for the determination of the
complexation-induced shifts and the intermolecular NOEs.
1
Signal assignment of the H NMR spectra of SO1, SO1‚HCl,
SA1, SA2, and the four complexes was achieved by the help
of 1H-1H correlation (COSY), 1H-13C heteronuclear single
quantum coherence (HSQC), and nuclear Overhauser enhance-
ment (NOESY) spectroscopy.
The NMR spectra for the J ob plot measurements were
recorded on a 400-MHz spectrometer, while all other spectra
were recorded on a 600-MHz spectrometer equipped with a
5-mm triple probe (1H, 13C, broadband) and x,y,z-gradients, at
4. Exp er im en ta l Section
4.1. Syn th esis of 6′-Neop en toxy-9-O-ter t-bu tylca r ba m -
oylcin ch on id in e (SO1) a n d th e Cor r esp on d in g Ch ir a l
Sta tion a r y P h a se (CSP 1). 6′-Neopentoxy-9-O-tert-butylcar-
bamoylcinchonidine was prepared from tert-butylisocyanate
and 6′-neopentoxycinchonidine, which in turn was synthesized
from cupreine and neopentylbromide, as described in a recent
publication.12 CSP 1 was prepared by coupling SO1 to mer-
captopropyl-modified silica as decribed elsewhere.3 The selector
loading on the chiral stationary phase was calculated by using
its nitrogen content determined by CHN analysis and found
1
frequencies of 600.13 MHz for H and 150.90 MHz for 13C. The
following experiments were performed at a temperature of 300
K: 1H NMR (16 scans, sweep width 6100 Hz); 13C NMR with
the attached proton test sequence (5000 scans on average,
sweep width 33 000 Hz, power-gated proton decoupling during
acquisition with WALTZ-16); double quantum filtered correla-
tion spectroscopy (DQF-COSY) with pulsed field gradient
coherence selection according to the scheme of Davis et al.27
(2048 data points in f2, 256 data points in f1, 16 scans, sweep
width 6100 Hz, 1 ms sine-shaped gradient pulses with 20%
maximum amplitude, absorption mode in f1 using time pro-
portional phase increments (TPPI)); NOESY (2048 data points
in f2, 256 data points in f1, 64 scans, sweep width 6100 Hz,
800 ms mixing time, absorption mode in f1 using TPPI);
to be 0.24 mmol‚g-1
.
4.2. Syn th esis of th e (a ll-R)- a n d (a ll-S)-En a n tiom er s
of DNB-Ala 2 (SA1) a n d DNB-Ala 3 (SA2). All peptide enan-
tiomers ((all-R)- and (all-S)-Ala-Ala and (all-R)- and (all-S)-
Ala-Ala-Ala) were purchased from Bachem (Bubendorf, Swit-
zerland). The peptides and a 2-fold molar excess of sodium
hydrogencarbonate were dissolved in water and a 1.2-fold
molar excess of 3,5-dinitrobenzoyloxysuccinimide (prepared
from 3,5-dinitrobenzoyl chloride and hydroxysuccinimide by
Hu¨nig base coupling) was added. The resulting suspension was
stirred at ambient temperature until a clear yellow solution
was obtained (minimum 48 h). To remove the 3,5-dinitroben-
zoic acid byproduct the reaction solutions were purified by
preparative HPLC, using a tert-butylcarbamoylquinidine based
chiral stationary phase (Chiral-AX QD-1, Bischoff Chroma-
tography, Leonberg, Germany) and an 80/20 mixture of
methanol/1 M aqueous ammonium acetate (0.5 M for (all-R)-
Ala3) adjusted to an apparent pH (pHa) of 6.0 with acetic acid
as the mobile phase. The fractions containing the products
were pooled, and the solutions were concentrated on a ro-
tavapor, acidified with hydrochloric acid (pH < 2) and ex-
tracted three times with 5 mL ethyl acetate. The combined
organic layers were dried with MgSO4 and the solvents were
then removed with a stream of nitrogen, yielding yellow oils.
Recrystallization with chloroform gave off-white powders.
Purity was checked by HPLC and found to be >99% and 100%
ee for all four products.
1
sensitivity enhanced H-13C HSQC with echo/anti-echo selec-
tion28,29 (1024 data points in f2, 256 data points in f1, 16 scans,
sweep width in f2 6100 Hz, sweep width in f1 25 600 Hz, 1 ms
sine-shaped gradient pulses with 80% maximum amplitude,
adiabatic 13C decoupling during acquisition using a 1.5 ms
CHIRP30 pulse).
Processing was carried out as follows: All two-dimensional
spectra were zero filled, doubling the data points in the direct
dimension, and in the indirect dimension data points were
extended two times by linear prediction forward using 64
coefficients. In both dimensions, the data were multiplied with
a 90° shifted square sine window function and the spectra were
phase-corrected to absorption mode.
4.5. Com p u ta tion a l Meth od s. Conformational analyses,
molecular mechanics geometry optimization, and molecular
dynamics simulations were done with MacroModel 7.1.31 The
GB/SA continuum model32 for solvent was used throughout.
Conformation searching was done by using the grid search
method.33 Energy minimization was done by using the AM-
BER* force field with no cutoff of any kind invoked and
implementing a conjugate gradient minimizer, using Macro-
Model’s default convergence criteria. The molecular simula-
tions were done by using the stochastic dynamics method to
simulate the random collisions with solvent as well as solvent
4.3. HP LC En a n tiom er Sep a r a tion s. The chromato-
graphic enantioselectivities R for the (all-R)/(all-S) enanti-
omers of N-3,5-dinitrobenzoylated alanine and di- and triala-
nine peptides (N-protection was carried out according to ref
5) were measured by HPLC, using a 150 × 4 mm i.d. column
packed with the CSP 1. The mobile phase consisted of a
mixture of 80% methanol and 20% 0.5 M aqueous ammonium
acetate, which was then adjusted to pHa 6.0 with acetic acid.
A flow rate of 1 mL‚min-1 was employed and the column was
(27) Davis, A. L.; Laue, E. D.; Keeler, J .; Moskau, D.; Lohman, J . J .
Magn. Reson. 1991, 94, 637-644.
(28) Kay, E. L.; Keifer, P.; Saarinen, T. J . Am. Chem. Soc. 1992,
114, 10663-10665.
(29) Schleucher, J .; Schwendinger, M.; Sattler, M.; Schmidt, P.;
Schedletzky, O.; Glaser, S. J .; Sørensen, O. W.; Griesinger, C. J .
Biomol. NMR 1994, 4, 301-306.
thermostated at 25 °C. Sample solutions (50 µL of 1 mg‚mL-1
were injected. UV detection was performed at 254 nm.
)
(30) Bo¨hlen, J . M.; Bodenhausen, G. J . Magn. Reson. A 1993, 102,
293-301.
4.4. NMR Exp er im en ts. The monohydrochloride of SO1
was prepared by first adding an excess of methanolic hydrogen
chloride to SO1 and subsequently evaporating the solvent to
yield SO1‚2HCl. To this dihydrochloride an equivalent amount
of SO1 was added, the solvent was evaporated, and the
obtained SO1‚HCl was dried in vacuo. The sodium salts of
SA1 and SA2 were prepared by adding equivalent amounts
(31) 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.
(32) Still, W. C.; Tempczyk, A.; Hawley, R. C.; Hendrickson, T. J .
Am. Chem. Soc. 1990, 112, 6127-6129.
(33) Leach, A. In Reviews in Computational Chemistry; Lipkowitz,
K. B., Boyd, D. B., Eds.; VCH: Weinheim, Germany, 1991; Vol. 2, pp
1-47.
8326 J . Org. Chem., Vol. 68, No. 22, 2003

Chiral Recognition of Peptide Enantiomers
friction forces.34 The time step used in the numerical integra-
tion of Newton’s equations was 1 fs. Initial geometries for the
binary complexes were based on a crystal structure of SO1
with 3,5-dinitrobenzoyl-(S)-leucyl-(S)-leucine.35 By using those
atomic coordinates, the structure of (all-S)-SA1 was generated
by modifying the leucine groups to alanine groups. The
complex of (all-R)-SA1 was then obtained from that structure
by inverting the stereochemistries at the analyte’s stereogenic
centers. This was done by swapping the hydrogen and the
methyl group on each stereogenic center so as to preserve the
salt bridge and the π-π-stacking. These structures were then
geometry optimized before the simulation warm-up and equili-
bration protocol. For SA2 we inserted, de novo, an additional
alanine group into the structure of (all-S)-SA1. The stereo-
centers of that structure were inverted as above to generate
(all-R)-SA2. These structures were also geometry optimized
prior to the simulation warm-up and equilibration. A warm-
up protocol beginning from 0 to 298 K was done over a time of
5 ps. The system was then equilibrated for an additional 100
ps and then a production run of 1000 ps was carried out. This
heating-equilibration-simulation protocol was used for all
complexes described in this paper. Details of the methodology
for simulating selector-selectand interactions in chromatog-
raphy can be found in a previous paper.36 Structures were
sampled uniformly during 1-ns simulations and saved to disk
for postprocessing (10 000 total structures for each diastere-
omeric complex). Post-simulation analysis of the SD trajecto-
ries was performed with an in-house program anout, which
computes, among other properties, intermolecular energies and
the center-of-mass (COM) positions of one molecule relative
to another.25
Ack n ow led gm en t . This work was carried out by
grants from the Austrian Science Fund (project no.
P14179-CHE, N.M., W.L.), the National Science Foun-
dation (CHE-9982888), and the Petroleum Research
Fund (35172-AC4 and 39454-AC1) administered by the
American Chemical Society in addition to the North
Dakota Biomedical Research Infrastructure Network
(K.B.L.). C.C. thanks the Austrian Academy of Sciences
for supporting his work by a Ph.D. grant (DOC-Stipen-
dium).
Su p p or tin g In for m a tion Ava ila ble: Figures showing the
J ob plots of selector-selectand complexes, the NOESY spectra
of all four complexes, the trajectories for all four complexes,
the distributions of the intermolecular energies for the SA2
complexes, and the distributions of the intermolecular frag-
ment energies for all four complexes. This material is available
free of charge via the Internet at http://pubs.acs.org.
J O0346914
(34) Gunsteren, W. F. v.; Berendsen, H. J . C. Mol. Simul. 1988, 1,
173-185.
(35) Unpublished data.
(36) Lipkowitz, K. B. Acc. Chem. Res. 2000, 33, 555-562.
J . Org. Chem, Vol. 68, No. 22, 2003 8327