Discrimination between enantiomers of structurally related molecules: Separation of benzodiazepines by molecularly imprinted polymers

Article abstract of DOI:10.1021/ja9926313

Molecular imprinting has been used to create synthetic receptor sites for a series of chiral benzodiazepines. A detailed HPLC analysis of binding properties using molecularly imprinted polymers (MIPs) as the stationary phase showed that binding, as measur

Full text of DOI:10.1021/ja9926313

460
J. Am. Chem. Soc. 2000, 122, 460-465
Discrimination between Enantiomers of Structurally Related
Molecules: Separation of Benzodiazepines by Molecularly Imprinted
Polymers
Bradley R. Hart, Daniel J. Rush, and Kenneth J. Shea*
Contribution from the Department of Chemistry, UniVersity of California, IrVine, California 92697-2025
ReceiVed July 26, 1999
Abstract: Molecular imprinting has been used to create synthetic receptor sites for a series of chiral
benzodiazepines. A detailed HPLC analysis of binding properties using molecularly imprinted polymers (MIPs)
as the stationary phase showed that binding, as measured by chromatographic retention, shows significant
dependence on the chiral match or mismatch. In addition, the shape and spatial orientation of functionality of
the imprinted binding site is also critical for recognition. Imprinted polymers, therefore, are not only able to
discriminate between enantiomers of the imprinted molecule, they also demonstrate an ability to discriminate
between a wide range of enantiomers of structurally related molecules that have not been imprinted. The
ability of MIPs to discriminate between enantiomers of molecules in favor of the imprinted absolute
configuration, even as the structural homology between the enantiomers and the original template decreases,
indicates that the synthetic benzodiazepine receptors may serve as crude mimics of the natural receptor.
The separation of enantiomers by molecularly imprinted
Scheme 1
polymers (MIPs) has proven to be one of the more challenging
and exciting applications of this growing technology.1 Addition-
ally, the use of MIPs for drug assays2,3 and as sensors3,4 is an
area of considerable research effort. We report the use of
molecular imprinting to create materials capable of shape
selective chiral recognition of benzodiazepine derivatives. Of
special significance is that, while these materials are able to
discriminate between enantiomers of the imprinted molecule,
they also demonstrate the ability to discriminate between a wide
range of enantiomers of structurally related molecules that have
not been imprinted. The ability to resolve enantiomers of a
family of molecules in favor of the imprinted absolute config-
uration follows trends that can be understood in terms of rational
concepts that control recognition in molecular imprinting.5 This
insight along with complimentarity of functionality can lead to
the creation of “designer” recognition sites that may ultimately
have the capability to sort molecules on the basis of their
physiological activity.2,3
.
5
from modulating the GABA system in the brain via specific
benzodiazepine receptors that form an integral part of the
GABAA receptor-chloride channel complex. The exact structure
of native GABAA receptor subtypes remains unknown, and there
has been no assignment of receptor subtypes to particular
behaviors or pathologies.6 Due to this lack of information about
the receptors involved in benzodiazepine action, we set out
preparing synthetic receptors for this class of molecule. To this
end we have synthesized a series of imprinted polymers using
a family of structurally related chiral benzodiazepines as
templates. These materials were then examined to determine
their ability to resolve enantiomers of benzodiazepines with
varying structures.
Benzodiazepines are a class of compounds that modify
affective responses to sensory perceptions. Their action results
(1) (a) Sellergren, B.; Shea, K. J. J. Chromatogr. A 1993, 635, 31-49.
(b) Mosbach, K. Trends Biol. Sci. 1994, 19, 9-14. (c) Kempe, M. Anal.
Chem. 1996, 68, 1948-1953. (d) Ramstrom, O.; Ansell, R. Chirality 1998,
10, 195-209. (e) Remcho, V. T.; Tan, Z. J. Anal. Chem. 1945, 17, 248A-
255A. (f) Molecular and Ionic Recognition with Imprinted Polymers;
Bartsch, A., Maeda, M., Eds.; ACS Symp. Ser. 703; American Chemical
Society: Washington, DC, 1998.
(2) (a) Vlatakis, G.; Andersson, L. I.; Muller, R.; Mosbach, K. Nature
1993, 361, 645-647. (b) Ramstrom, O.; Ye, L.; Krook, M.; Mosbach, K.
Anal. Commun. 1998, 35, 9-11. (c) Karlsson, P. K. O.; Lutz, E. S. M.;
Andersson, L. I Trends Anal. Chem. 1999, 18, 146-154.
Enantiomerically pure benzodiazepines 6-10 were synthe-
sized from L- and D-alanine, valine, phenylalanine, tyrosine, and
tryptophan by the procedure outlined in Scheme 1.7
(3) Haupt, K.; Mosbach, K. Trends Biol. Sci. 1998, 16, 468-475.
(4) (a) Kriz, D.; Ramstrom, O.; Mosbach, K. Anal. Chem. 1997, 69,
A345-A349. (b) Arnold, B. R.; Euler, A. C.; Jenkins, A. L.; Uy, O. M.;
Murray, G. M. Johns Hopkins APL Tech. Digest 1999, 20, 190-197. (c)
Yano, K.; Karube, I. Trends Anal. Chem. 1999, 18, 199-204.
(5) (a) Spivak, D.; Gilmore, M. A.; Shea, K. J. J. Am. Chem. Soc. 1997,
119, 4388-4393. (b) Sellergren, B. Trends Anal. Chem. 1999, 3, 164-
174.
(6) (a) Hobbs, W.; Rall, T.; Verdoorn, T. In Goodman & Gillman’s The
Pharmaceutical Basis of Therapeutics, 9th ed.; Hardman, J., Limbird, L.,
Eds.; McGraw-Hill: New York, 1996, Chapter 17. (b) Braestrub, C.;
Nielsen, M.; Honore, T.; Jensen, L. H.; Petersen, E. N. Neuropharmacology
1983, 22, 1451-1457. (c) Takeuchi, T.; Tanaka, S.; Rechnitz, G. A. Anal.
Biochem. 1992, 203, 158-162.
(7) Sunjie, v.; Kajfez, F.; Stromar, I.; Blazevic, N.; Kolbah, D. J.
Hetercycl. Chem. 1973, 10, 591.
10.1021/ja9926313 CCC: $19.00 © 2000 American Chemical Society
Published on Web 12/31/1999

Separation of Benzodiazepines
J. Am. Chem. Soc., Vol. 122, No. 3, 2000 461
benzodiazepine (bold entries). We would anticipate this result
in that an imprinted void within the MIP should selectively bind
molecules that have the best size, shape, and functional group
complementarity. In the absence of competing effects, a
molecule that has a superior “fit” at the site should be retained
longer by the MIP stationary phase. For enantiomers, this
diastereomeric cavity/molecule interaction is the only discrimi-
nating factor. The ability of benzodiazepine MIPs to produce
high R values for a family of molecules in favor of the absolute
configuration of the template is an important result of this study.
In addition, the recognition sites created by imprinting reflect
not only the chirality of the microenvironment, but also the shape
and functional group complimentarity of the sites (Figure 2).
It was found that as structural similarity between the imprint
molecule and the analyate molecule diminishes, both retention
and selectivity decrease. In our study, this was explored by
modifying the side chains at the asymmetric center and
observing the resulting changes in the R values.
Figure 1. SEM image of 25-38 µm MIP particles. Surface area and
pore analysis indicate that these materials have an average surface area
of 256 m2/g, and an average pore diameter of 94 Å.
More specifically, the MIP made from the S-6 benzodiazepine
(methyl derivative) showed low R values for all nonmethyl
derivatives (row e). If one focuses on the steric environment at
the chiral center, the methyl group of S-6 will create an
imprinted “void” in the polymer cavity that should be too small
to accommodate any of the other side chains studied. We see
this expressed as moderate selectivity for enantiomers of
template 6 and negligible selectivity for the other benzodiazepine
derivatives.
This situation is not likely true for binding sites created by
benzodiazepines containing larger R groups at the asymmetric
center. The S-9 benzodiazepine (4-hydroxy benzyl derivative)
MIP is illustrative in this case because the imprinted cavity
should be large enough to accommodate most of the other side
chains. Discrimination between enantiomers of benzodiazepine
9 is high, R ) 3.8 (row a). Making a relatively small perturbation
to the analyate by removing the aromatic hydroxyl group results
in a lower, but still significant, separation factor of R ) 2.5.
The R values continue to erode for the remaining derivatives
in reasonable order of how well each of the S-enantiomers
should be accommodated at the binding site based on the relative
van der Waals volume of the side chains.9
This analysis can also be used to explain the other trends in
Table 1. The S-7 MIP might be expected to have a slightly
larger binding site based on the van der Waals volume of the
side chains than the S-6 MIP and, therefore, it can accommodate
S-7 and S-6, but not the larger residues (row d). In the S-10
benzodiazepine (indolyl derivative) imprinted polymers, the
erosion of R presumably stems from a non-optimal fit and
noncomplementarity of the slightly smaller aromatic residues
(row c). The lower R for analytes 6 and 7 stems perhaps from
these residues having too much freedom at the S-10 imprinted
binding site to produce the best chiral discrimination. This MIP
presents a case where the large binding site is shaped uniquely
enough that none of the other residues exhibit high selectivity
in binding, but because the smaller side chains can be accom-
modated, some chiral preference is observed, unlike with the
S-6 MIP. Similarly, the S-8 benzodiazepine (benzyl derivative)
MIP creates a cavity that roughly accommodates S-9, but does
not discriminate well among the other derivatives. The side chain
of S-10 is too big to be accommodated in the recognition site,
while the smaller sizes of the S-6 and S-7 side chains result in
lowered R values (row b).
Table 1. Benzodiazepine R Values for Retention on MIP
Stationary Phasesa,b
benzodiazepine derivative injected
4-hydroxy- benzyl indolyl isopropyl methyl
MIP template
benzyl (9) (8)
(10)
(7)
(6)
a. S-9 4-hydroxybenzyl
b. S-8 benzyl
c. S-10 indolyl
3.8
2.3
1.5
1.2
1.0
2.5
3.0
1.8
1.3
1.1
1.3
1.7
2.3
1.3
1.1
1.6
1.6
1.3
2.2
1.1
1.4
1.7
1.4
1.6
1.3
d. S-7 isopropyl
e. S-6 methylc
a Photochemically initiated polymer with acetonitrile used as the
porogen. b HPLC performed using acetonitrile as eluent. c A MIP
prepared using the opposite enantiomer, R-6, showed selectivity in favor
of the (R) enantiomers which was slightly higher than the data shown.
Individual benzodiazepine derivatives S-6 through S-10 were
subsequently used to imprint ethylene glycol dimethacrylate
(EGDMA) polymers using methacrylic acid (MAA) as the
functional monomer. Solutions of template (1%), MAA (11%),
EGDMA (87%), and AIBN (1%) in acetonitrile were polym-
erized photochemically by UV irradiation for 12 h while
immersed in a circulating bath at 0 ( 2 °C. The resulting
polymers were crushed, extracted with methanol, ground, and
sieved to a 25-38 µm particle size (Figure 1). These particles
were slurry packed in 10 cm × 0.46 cm i.d. stainless steel
chromatography columns for evaluation by HPLC.5
Solutions of enantiomerically pure benzodiazepine derivatives
were injected onto HPLC columns containing MIP stationary
phases imprinted with the S-enantiomer of a benzodiazepine
derivative. Corrected retention volumes (capacity factors, k′)8
were calculated against a void volume marker (acetone) and
the ratios of the k′ values for each enantiomeric pair (R ) k′S/
k′R) are summarized in Table 1. The R values across each row
show the ability of a specific MIP to resolve a series of
benzodiazepine enantiomers. The R values in each column
indicate how a particular benzodiazepine is resolved on the series
of MIPs.
Looking across the rows in Table 1, one observes that each
S-MIP is most selective for the S-enantiomer over the R
regardless of the imprint molecule (R > 1). Additionally, the
highest R value in each row corresponds to that of the imprinted
(8) The capacity factor (k′) is defined as k′ ) [(V(t) - V(0))/V(0)], where
V(t) is the retention volume and V(0) is the dead volume or the retention
volume of a nonbinding substrate. The 95% confidence limits were
calculated for all capacity factors and separation factors. For the capacity
factors, all confidence limits were equal to or less than (0.06. For separation
factors, all confidence limits were equal to or less than (0.1.
We can understand the retention and selectivity data as the
result of both specific and nonspecific interactions of analyte
(9) Creighton, T. E. Proteins Structures and Molecular Properties, 2nd
ed.; W. H. Freeman & Co.: New York, 1993; Chapter 1.

462 J. Am. Chem. Soc., Vol. 122, No. 3, 2000
Hart et al.
Figure 2. (A) Receptor site simulated by the density surface of an X-ray crystal structure of S-(7). (B) An attempted “fit” of R-(7) in the simulated
receptor site. The orientation of R-(7) was chosen so as to maintain the position of the carbonyl oxygen (red circle) and the amide NH and imine
nitrogen (blue circles) in both enantiomers.
Table 2. Benzodiazepine Capacity Factors for Retention on MIP Stationary Phasesa,b
benzodiazepine derivative injected
4-hydroxybenzyl (9)
benzyl (8)
indolyl (10)
isopropyl (7)
methyl (6)
MIP template
k′R
k′S
k′R
k′S
k′R
k′S
k′R
k′S
k′R
k′S
a. S-9 4-hydroxybenzyl
b. S-8 benzyl
c. S-10 indolyl
d. S-7 isopropyl
e. S-6 methyl
0.33
0.44
0.51
0.23
0.36
1.25
0.99
0.79
0.26
0.38
0.27
0.38
0.43
0.22
0.29
0.68
1.15
0.76
0.29
0.32
0.48
0.47
0.57
0.27
0.40
0.62
0.82
1.31
0.34
0.41
0.27
0.32
0.42
0.19
0.30
0.43
0.52
0.54
0.41
0.31
0.38
0.44
0.52
0.23
0.41
0.53
0.75
0.70
0.36
0.54
a Photochemically initiated polymer with acetonitrile used as the porogen. b HPLC performed using acetonitrile as eluent.
with the polymer matrix. A close look at the corrected retention
volumes (k′) of the various derivatives indicates that the
R-enantiomers, with a chiral mismatch to the imprint cavity,
have narrowly spread corrected retention volumes (k′R) that are
mostly indicative of the relative polarities of the molecules
(Table 2). Their order of elution is not influenced by the choice
of the imprint molecule. However, the corrected retention
volumes of the matching S-enantiomers (k′S) are strongly
dependent on the imprint molecule and order themselves
according to a combination of specific and nonspecific matrix
effects. The R Values are largely determined by the changes in
k′S relatiVe to a nearly constant k′R. This trend is shown
graphically in Figure 3, which plots the selectivity factor and
the capacity factors for each benzodiazepine derivative analyzed
using the S-8 imprinted polymer. This plot is typical of the other
MIPs that were investigated. As the fit between the imprint
cavity and the analyte molecule improves, the selectivity and
therefore the separation factor increases. The repeated selectivity
of MIPs for the template molecule over so many structurally
similar species is striking.
It was found that changing the composition of the mobile
phase from pure acetonitrile to binary mixtures of acetonitrile
and chloroform caused a bimodal change in the capacity factor
of the template molecule (Figure 4). This behavior is a general
observation that has been observed with a variety of solutes
Figure 3. Graphical representation of capacity factor and separation
factor data for benzodiazepine derivatives injected onto a HPLC column
packed with S-(8) (benzyl) benzodiazepine imprinted polymer: k′R (9),
k′s ([), R (O).
and chiral stationary phases.10,11 The phenomena has been
(10) Funk, M.; Frank, H.; Oesch, F.; Platt, K. L. J. Chromatogr. A 1994,
659, 57.
(11) (a) Pescher, P.; Caude, M.; Rosset, R. J. Chromatogr. 1986, 371,
159. (b) Siret, L.; Tambute´, A.; Caude, M.; Rosset, R. J. Chromatogr. 1990,
498, 67.

Separation of Benzodiazepines
J. Am. Chem. Soc., Vol. 122, No. 3, 2000 463
Scheme 2
possible. Also in this group is the racemic mixture of Lorazepam
that elutes as a somewhat broadened peak.
Figure 4. Effect of eluent composition on the capacity factor (b) of
S-(8) obtained from an HPLC column packed with S-(8) imprinted
polymer.
The third block of compounds includes racemic Lormetazepam,
which is structurally related to Lorazepam with the exception
of a methyl group on the amide nitrogen. Flunitrazepam, while
lacking a chiral center, also has the N-methyl group and different
substituents on the aromatic rings. The difference in retention
between the N-methyl containing compounds and the non-
methylated compounds deserves further comment. Having
already established the chiral center as a specific recognition
element, the amide nitrogen is implicated as a secondary site
for interaction with the matrix. If hydrogen bonding with the
carboxylic acid groups of the matrix contributes to the recogni-
tion, the methylation of the amide nitrogen would severely
disrupt the binding interaction. This seems to be the cause of
the lower k′ in Lormetazepam compared to that of Lorazepam,
since the rest of the molecule is identical. The fact that
Flunitrazepam, with no pendant chiral residue, has retention
similar to that of Lormetazepam suggests that the amide
hydrogen is a significant determinant of binding.
We currently have no pharmacological data on the amino
acid functionalized benzodiazepines. However, it is of interest
to note that while Lorazepam is an anxiolytic drug, Lormetazepam
and Flunitrazepam are classified as hypnotic drugs.13 All
benzodiazepines are CNS depressants and act by binding to a
receptor for γ aminobutyric acid (GABAA) in the brain and
allosterically modulating its activity.14 It is known that many
subtypes of GABAA receptors exist, but as yet no association
of subtypes with physiological effects has been made. It has
been hypothesized that the wide range of benzodiazepine
pharmacological effects is due to the complex interplay of
different drugs preferentially binding different GABAA receptor
subtypes.14
Table 3. Capacity Factors on an S-8 (Benzyl) Benzodiazepine
MIP Columna,b
molecule
(S)-8 benzyl
(S)-9 hydroxy benzyl
(S)-10 indolyl
k′
molecule
k′
1.15
0.99
0.82
0.75
0.52
0.51
0.47
(R)-9 hydroxy benzyl
(R)-6 methyl
(R)-8 benzyl
(R)-7 isopropyl
(()-Lormetazepam
benzylamine
0.44
0.44
0.38
0.32
0.29
0.27
0.19
(S)-6 methyl
(S)-7 iIsopropyl
(()-Lorazepam
(R)-10 indolyl
flunitrazepam
a Photochemically initiated polymer with acetonitrile used as the
porogen. b HPLC performed using acetonitrile as eluent.
attributed to the intermediate polarity of the stationary phase
with respect to the solvents being used.10 In acetonitrile, the
higher capacity factor is contributed to by nonpolar interactions
with the stationary phase. This interaction is reduced by addition
of a nonpolar solvent, chloroform. In pure chloroform, the
interaction with the stationary phase is polar in nature, and
therefore is reduced by addition of the polar solvent acetonitrile.
Table 3 shows the capacity factors (k′ ) for a variety of
substances obtained using a MIP imprinted with benzodiazepine
S-8. As would be expected, the template benzodiazepine had
the highest affinity for the stationary phase.
The S forms of other benzodiazepines show slightly lower
affinity for these stationary phases. As mentioned earlier, this
decrease in capacity factor most likely results from the mismatch
of the side chain of the amino acid portion of the molecules
with the cavity formed around the template molecule. The fact
that all of the S enantiomers are retained over the R enantiomers
indicates that the perturbation caused by the inversion of the
stereocenter has a more profound effect on the binding energy
than the structural changes associated with different side
chains.12
The R enantiomers of the benzodiazepines appear at signifi-
cantly lower capacity factors. This highlights the importance
of proper fit into the chiral cavity created by imprinting an S
enantiomer. It is important to note that selectivity between the
enantiomers of the isopropyl and methyl derivatives is achieved.
This demonstrates the wide range of structural variation that is
The fact that the hypnotic drugs elute differently than the
anxiolytic benzodiazepine suggests that MIPs might be able to
differentiate between molecules which bind to different subtypes
of receptors as well. Using a MIP model of a specific receptor
subtype could greatly increase the speed and ease with which
compounds could be developed having high binding specificity
for that individual receptor. In addition to obtaining new, more
specific drugs, we may also be able to learn more about the
likely composition and pharmacology of the GABAA receptor
subtypes.
In summary, molecularly imprinted polymers were prepared
that can be used to separate enantiomers of a series of
(12) At this time we have no information about the solution conforma-
tions of the benzodiazepines being studied and what effect, if any, that
may have on recognition.
(13) Budavari, S., Ed. The Merck Index, 11th ed.; Merck & Co., Inc.:
Rahway, NJ, 1989; pp 5456 and 5458.
(14) Sieghart, W. J. Psychiatr. Neurosci. 1994, 19, 24-29.

464 J. Am. Chem. Soc., Vol. 122, No. 3, 2000
Hart et al.
General Procedure for the Preparation of 1-5. To a stirred
solution of 2-amino-5-chlorobenzophenone (0.838 g, 3.60 mmol) and
the particular N-Boc-amino acid (3.3 mmol) in either methylene chloride
(5 mL, N-Boc-alanine and N-Boc-valine) or tetrahydrofuran (5 mL,
N-Boc-tryptophan, N-Boc-phenylalanine, N-Boc-tyrosine) was added
dicyclohexylcarbodiimide (DCC) (0.746 g, 3.6 mmol) in methylene
chloride (5 mL) dropwise, over 30 min at 0 °C. The reaction mixture
was stirred an additional 8 h at room temperature. The dicyclohexyl
urea formed was filtered off and the filtrate concentrated. The crude
products were purified as described below.
General Procedure for the Preparation of 6-10. To a stirred
solution of the appropriate benzophenone (1-5, 2.0 mmol) in chloro-
form (50 mL) at room temperature was bubbled hydrogen chloride gas
slowly. After 20 min, the bubbling was stopped and the solution was
stirred overnight at room temperature. The reaction mixture was washed
with saturated sodium bicarbonate solution (2×, 50 mL) and water (2×,
50 mL). The organic layer was dried (MgSO4) and concentrated. The
residual oil was dissolved in methanol-water (1:1, 30 mL) and the
pH was adjusted to 8.5 by the addition of sodium hydroxide (1 N).
The reaction mixture was stirred for 10 h at room temperature. The
solution was concentrated and water (50 mL) was added. The solution
was extracted with CH2Cl2 (3 × 50 mL) and the organic layer was
dried (MgSO4) and concentrated. The crude products were purified as
described below.
benzodiazepine molecules. A detailed HPLC analysis of binding
properties using MIP stationary phases showed significant
dependence on the chiral match or mismatch with the shape of
the imprinted binding site. These results demonstrate the ability
of MIPs to discriminate between enantiomers of molecules in
favor of the imprinted absolute configuration, even as the
structural homology between the enantiomers and the original
template decreases.
The reported results indicate that these synthetic benzodiaze-
pine receptors may serve as crude mimics of the natural receptor.
While MIPs used as synthetic drug receptors may not mimic
protein microenvironments, they may nevertheless prove useful
for screening large libraries of compounds for molecules with
similar or related affinities. As molecular structure is key to
reactivity, it is possible that MIPs may be able to sort molecules
based on their pharmacological properties. In this specific case,
the response of various drugs to a set of imprinted polymers
may provide clues to their binding of GABAA receptors,
providing a new technique for use in developing better, more
selective pharmaceuticals.
Experimental Section
2-N-(N′-Boc-alanyl)amino-5-chlorobenzophenone (R-1 and S-1).
Recrystallization of the crude product from cyclohexane afforded 1
(1.088 g, 82%). Mp 150-152 °C; 1H NMR (CDCl3) δ 11.19 (s, 1 H),
8.62 (d, J ) 8.6 Hz, 1 H), 7.68 (d, J ) 7.90 Hz, 2 H), 7.61 (t, J ) 6.96
Hz, 1 H), 7.52-7.48 (m, 4 H), 5.10 (br s, 1 H), 4.31 (br s, 1 H), 1.46
(d, J ) 7.1 Hz, 3 H), 1.41 (s, 9 H); 13C NMR (CDCl3) δ 197.8, 172.0,
155.2, 138.4, 137.7, 133.7, 132.8, 132.5, 129.8, 128.4, 127.4, 124.9,
122.9, 80.2, 51.6, 28.2, 18.4 ppm; IR (KBr, cm-1) 3321, 2979, 2933,
1647, 1597; HRMS calcd for C21H23N2ClO4 402.1346, found 402.1334;
[R]26D(R-1) ) +49.9 (c 1.15, CHCl3), [R]26D(S-1) ) -49.3 (c 0.965,
CHCl3).
Instrumentation. Proton nuclear magnetic resonance spectra (1H
NMR) were recorded at 500 MHz on Bruker spectrometers. Carbon
nuclear magnetic resonance spectra (13C NMR) were recorded on Bruker
spectrometers at 125.8 MHz. Infrared spectra (IR) were measured on
an Analect RFX-40 FTIR spectrometer. HPLC analyses were performed
with a Shimadzu LC-10AS dual pump gradient solvent delivery system
equipped with an SPD-10AV UV/Vis detector and a Hewlett-Packard
3396A integrator. A Thomas-Hoover capillary melting point apparatus
was used to observe uncorrected melting points. High-resolution mass
spectra (EI, 70 eV or CI, isobutane or ammonia) were obtained on a
VG 7070E high-resolution mass spectrometer or Fisons Autospec mass
spectrometer. Optical rotations were obtained using a JasCo DIP-360
digital polarimeter. UV irradiation experiments utilized a Hanovia
medium-pressure mercury arc lamp.
General Procedures. All solvents were distilled from drying agents
(CaH2 or Na/benzophenone) just before use. Ethylene glycol dimethacry-
late (EGDMA, Aldrich) was distilled under reduced pressure (10
mmHg, 60 °C). Methacrylic acid (MAA, Aldrich) was distilled over
CaH2 (10 mmHg, 80 °C). AIBN (Aldrich) was recrystallized from
methanol. Glassware was cleaned by soaking in an alcoholic KOH
solution overnight and rinsing with water; glassware was oven dried
overnight. All reactions were run under a N2 atmosphere. Volatile
solvents were removed under reduced pressure using a Bu¨chi rotary
evaporator. Thin-layer chromatography was run on precoated plates
of silica gel with a 0.25 mm thickness containing 60F-254 indicator
(Merck). Column chromatography was run using 230-400 mesh silica
gel (Merck).
General Procedure for Molecularly Imprinted Polymer Synthe-
sis. A solution of ethylene glycol dimethacrylate (3.47 g, 17.5 mmol),
methacrylic acid (0.203 g, 2.40 mmol), and AIBN (0.033 g, 0.20 mmol)
in acetonitrile or chloroform (4.5 mL) was deoxygenated using nitrogen
gas for 15 min. Additional solvent was added to bring the volume of
the solution to its initial level. The appropriate enantiomer of benzo-
diazepine derivative (6-10, 0.20 mmol) was added and the solution
was divided between two 20 mL flame-dried scintillation vials. After
further deoxygenating (5 min) the vials were sealed. The polymeriza-
tions were initiated photochemical by a mercury arc UV light source
at 0 ( 2 °C and allowed to proceed for 12 h under constant irradiation.
After crushing, the polymers were extracted with methanol for 36 h,
then dried under vacuum. The polymers were further crushed using a
mortar and pestle and sieved to isolate the 25-38 µm particles for use
in chromatographic experiments.
2-N-(N′-Boc-valyl)amino-5-chlorobenzophenone (R-2 and S-2).
Flash column chromatography (CH2Cl2) afforded a colorless oil. Flash
column chromatography of this oil using 10:5:1 petroleum ether/CH2-
Cl2/Et2O provided 2 (0.925 g, 65%). Mp 106-108 °C; 1H NMR
(CDCl3) δ 11.21 (s, 1 H), 8.66 (d, J ) 9.7 Hz, 1 H), 7.69 (d, J ) 8.3
Hz, 2 H), 7.63 (t, J ) 7.5 Hz, 1 H), 7.53-7.50 (m, 4 H), 5.10 (br d,
J ) 7.6 Hz, 1 H), 4.19 (br s, 1 H), 2.35 (m, 1 H), 1.42 (s, 9 H), 1.03
(d, J ) 6.9 Hz, 3 H), 0.94 (d, J ) 6.9 Hz, 3 H); 13C NMR (CDCl3) δ
198.0, 171.1, 156.8, 138.5, 137.8, 133.9, 132.9, 132.7, 129.9, 128.5,
127.5, 124.8, 122.9, 80.2, 31.0, 30.8, 28.3, 19.4, 17.4; IR (KBr, cm-1
)
3320, 2969, 2931, 1695, 1506; HRMS calcd for C23H27N2ClO4
430.1659, found 430.1662; [R]26D(R-2) ) +47.2 (c 0.93, CHCl3), [R]26
(S-2) ) -46.4 (c 0.85, CHCl3).
-
D
2-N-(N′-Boc-phenylalanyl)amino-5-chlorobenzophenone (R-3 and
S-3). Flash column chromatography (10:1 CH2Cl2/Et2O) afforded 3
1
(0.978 g, 62%). Mp 136-138 °C; H NMR (CDCl3) δ 10.92 (br s,
1H), 8.62 (d, J ) 8.9 Hz, 1 H), 7.64 (m, 3 H), 7.51 (m, 4 H), 7.20 (m,
4 H), 7.07 (m, 1 H), 5.04 (br s, 1 H), 4.52 (br s, 1 H), 3.16 (d, J ) 6.2
Hz, 2 H), 1.38 (s, 9 H) ppm; 13C NMR (CDCl3) δ 197.4, 170.7, 155.2,
138.0, 137.6, 136.1, 133.7, 132.8, 132.4, 129.9, 129.1, 128.7, 128.4,
127.6, 127.0, 125.0, 122.9, 80.4, 57.0, 38.3, 28.2 ppm; IR (KBr, cm-1
)
2940, 1672, 1508, 1243, 1128; HRMS calcd for C27H27N2ClO4
478.1659, found 478.1658; [R]26D(R-3) ) +52.1 (c 1.27, CHCl3), [R]26
(S-3) ) -52.8 (c 1.30, CHCl3).
-
D
2-N-(N′-Boc-tyrosyl)amino-5-chlorobenzophenone (R-4 and S-4).
Flash column chromatography (10:1 CH2Cl2/Et2O) afforded 4 (1.109
g, 68%). Mp 156-158 °C; 1H NMR (CDCl3) δ 10.93 (s, 1 H), 8.59 (d,
J ) 9.0 Hz, 1 H), 7.65 (d, J ) 7.3 Hz, 1 H), 7.62 (t, J ) 7.5 Hz, 1 H),
7.50 (m, 4 H), 7.02 (d, J ) 8.3 Hz, 2 H), 6.66 (d, J ) 8.5 Hz, 2 H),
5.08 (br s, 1 H), 4.45 (br s, 1 H), 3.08 (br s, 2 H), 1.38 (s, 9 H) ppm;
13C NMR (CDCl3) δ 197.6, 170.9, 155.3, 154.9, 138.0, 137.7, 133.8,
132.9, 132.5, 130.3, 129.9, 128.5, 127.7, 125.1, 123.0, 115.7, 80.6,
77.5, 57.1, 37.5, 28.2 ppm; IR (KBr, cm-1) 3318, 2932, 1686, 1514,
1250, 1160; HRMS calcd for C27H27N2ClO4 (M + Na) 517.1506, found
517.1502; [R]26D(R-4) ) +58.2 (c 0.972 CHCl3), [R]26D(S-4) ) -59.0
(c 0.965, CHCl3).
A slurry was made of the 25-38 µm particles in acetonitrile. This
slurry was forced into 4.6 mm i.d. × 10 cm stainless steel HPLC
columns using a stainless steel slurry reservoir and a Shimadzu LC-
10AS solvent delivery system.

Separation of Benzodiazepines
J. Am. Chem. Soc., Vol. 122, No. 3, 2000 465
2-N-(N′-Boc-tryptophanyl)amino-5-chlorobenzophenone (R-5 and
S-5). Flash column chromatography (10:1 CH2Cl2/Et2O) afforded 5
(1.197 g, 70%). Mp 150-152 °C; 1H NMR (CDCl3) δ 10.93 (s, 1 H),
8.63 (d, J ) 8.9 Hz, 1 H), 7.96 (br s, 1 H), 7.61-7.57 (m, 4 H), 7.51-
7.46 (m, 3 H), 7.43 (s, 1 H), 7.19 (d, J ) 8.1 Hz, 1 H), 7.11 (t, J ) 7.0
Hz, 1 H), 7.04 (m, 2H), 5.17 (br s, 1 H), 5.60 (br s, 1 H), 3.45 (m, 1
H), 3.34 (dd, J ) 14.6, 6.1 Hz, 1 H), 1.40 (s, 9 H) ppm; 13C NMR
(CDCl3) δ 197.3, 171.4, 155.4, 138.2, 137.6, 136.1, 133.7, 132.8, 132.4,
129.9, 128.3, 127.4, 124.9, 123.0, 122.9, 122.2, 119.6, 118.7, 111.1,
110.1, 80.3, 56.4, 29.7, 28.2 ppm. 24 signals were detected due to
coincident signals in the aromatic region; IR (KBr, cm-1) 3336, 2978,
2929, 1698, 1506; HRMS calcd for C29H28N3ClO4 517.1768, found
517.1756; [R]26D(R-5) ) +81.6 (c 1.01, CHCl3), [R]26D(S-5) ) -82.0
(c 0.91, CHCl3).
7-Chloro-1,3-dihydro-3-methyl-5-phenyl-2H-1,4-benzodiazepine-
2-one (R-6 and S-6). Recrystallization from acetone/water (1:1)
provided 6 (0.460 g, 81%). Mp 168-170 °C; 1H NMR (CDCl3) δ 8.62
(s, 1 H), 7.53 (d, J ) 7.2 Hz, 2 H), 7.51-7.44 (m, 2 H), 7.39 (m, 2 H),
7.34 (d, J ) 2.4 Hz, 1 H), 7.10 (d, J ) 8.6 Hz, 1 H), 3.75 (q, J ) 6.4
Hz, 1 H), 1.75 (d, J ) 6.5 Hz, 3 H) ppm; 13C NMR (CDCl3) δ 172.8,
167.8, 138.6, 137.1, 131.7, 130.5, 130.4, 129.7, 128.9, 128.6, 128.3,
122.6, 58.8, 17.0 ppm; IR (KBr, cm-1) 3226, 3135, 2929, 1685, 1608;
HRMS calcd for C16H13N2ClO (M - H) 283.0638, found 283.0639;
[R]26D(S-6) ) +124.4 (c 1.05, CHCl3), [R]26D(R-6) ) -123.6 (c 0.99,
CHCl3).
7-Chloro-1,3-dihydro-3-isopropyl-5-phenyl-2H-1,4-benzodiazepine-
2-one (R-7 and S-7). Recrystallization from petroleum ether/CH2Cl2
(40:1) provided 7 (0.537 g, 86%). Mp 189-191 °C; 1H NMR (CDCl3)
δ 9.29 (s, 1 H), 7.53-7.31 (m, 7 H), 7.14 (2, J ) 8.7 Hz, 1 H), 3.11
(d, J ) 9.3 Hz, 1 H), 2.78-2.72 (m, 1 H), 1.21 (d, J ) 6.7 Hz, 3 H),
1.10 (d, J ) 6.5 Hz, 3 H) ppm; 13C NMR (CDCl3) δ 170.8, 167.5,
138.8, 137.1, 131.6, 130.5, 130.4, 129.7, 128.8, 128.6, 128.3, 122.6,
69.4, 29.0, 20.3, 19.0 ppm; IR (NaCl, cm-1) 3201, 3104, 2958, 1683,
1606; calcd for C18H16N2ClO (M - H) 311.0951, found 311.0951;
[R]24D(S-7) ) +115.4 (c 1.05, CHCl3), [R]24D(R-7) ) -113.0 (c 0.97,
CHCl3).
raphy (7:3 Hex/EtOAc) afforded 8 (0.582 g, 80%). Mp 108-110 °C;
1H NMR (CDCl3) δ 9.19 (br s, 1 H), 7.45 (m, 4 H), 7.38 (m, 4 H),
7.32 (t, J ) 7.6 Hz, 2 H), 7.23 (m, 2 H), 7.12 (br s, 1 H), 3.79 (t, J )
6.8 Hz, 1 H), 3.62 (br s, 2 H) ppm; 13C NMR (CDCl3) δ 171.9, 168.1,
138.9, 138.6, 137.0, 131.7, 130.5, 130.4, 129.8, 129.7, 128.7, 128.6,
129.3, 128.2, 126.2, 122.8, 64.9, 37.5 ppm; IR (KBr, cm-1) 2940, 2362,
1685, 1605, 1478, 1323; HRMS calcd for C22H17N2ClO 360.1029, found
360.1017; [R]24D(S-8) ) +34.2 (c 1.100, CHCl3), [R]24D(R-8) ) -33.8
(c 1.06, CHCl3).
7-Chloro-1,3-dihydro-3-p′-hydroxybenzyl-5-phenyl-2H-1,4-ben-
zodiazepine-2-one (R-9 and S-9). Flash column chromatography (8:2
Hex/EtOAc) followed by recrystallization (Et2O/cyclohexane) provided
1
9 (0.603 g, 80%). Mp 139-142 °C; H NMR (CDCl3) δ 9.47 (br s, 1
H), 7.44 (m, 4 H), 7.36 (t, J ) 7.5 Hz, 2 H), 7.22 (d, J ) 2.4 Hz, 1 H),
7.20 (d, J ) 8.2 Hz, 2 H), 7.12 (d, J ) 8.6 Hz, 1 H), 6.76 (d, J ) 8.4
Hz, 2 H), 3.72 (br t, J ) 6.5 Hz, 1 H), 3.55-3.44 (m, 2 H) ppm; 13C
NMR (CDCl3) δ 171.3, 168.4, 154.3, 136.9, 132.8, 131.5, 132.0, 130.9,
130.6, 130.5, 129.9, 129.7, 128.8, 128.4, 122.8, 115.2, 65.0, 26.9 ppm;
IR (NaCl, cm-1) 3230, 2932, 1685, 1608; 1515, 1230; HRMS calcd
for C22H18N2ClO2 (M + H) 377.1057, found 377.1057; [R]24D(S-9) )
+27.9 (c 1.05, CHCl3), [R]24D(R-9) ) -26.2 (c 0.97, CHCl3).
7-Chloro-1,3-dihydro-3-(3′-indolyl)methyl-5-phenyl-2H-1,4-ben-
zodiazepine-2-one (R-10 and S-10). Water was added to the reaction
mixture to facilitate precipitation. Filtration of the precipitate followed
by recrystallization from acetone/water (1:1) provided 10 (0.630 g,
1
79%). Mp 149-151 °C; H NMR (CDCl3) δ 9.70 (s, 1 H), 8.11 (s, 1
H), 7.70 (d, J ) 7.8 Hz, 1 H), 7.49-7.33 (m, 7 H), 7.23-7.06 (m, 5
H), 3.86-3.81 (m, 2 H), 3.73-3.66 (m, 1H) ppm; 13C NMR (CDCl3)
δ 171.9, 168.0, 138.7, 136.9, 136.0, 131.8, 130.5, 129.7, 128.7, 128.6,
128.3, 127.7, 123.3, 122.7, 121.8, 119.1, 119.0, 112.8, 111.1, 64.2,
26.9 ppm. 21 signals were detected due to coincident signals in the
aromatic region; IR (KBr, cm-1) 3411, 3328, 3058, 2927, 1684, 1606;
HRMS calcd for C24H18N3ClO 399.1138, found 399.1129; [R]24D(S-
10) ) +27.4 (c 0.85, CHCl3), [R]24D(R-10) ) -25.2 (c 1.04, CHCl3).
Acknowledgment. We are grateful to the NIH and NSF for
financial support of this work.
7-Chloro-1,3-dihydro-3-benzyl-5-phenyl-2H-1,4-benzodiazepine-
2-one (R-8 and S-8). The residual oil was disolved in methanol and
water was added to facilitate precipitation. Flash column chromatog-
JA9926313