On-bead combinatorial approach to the design of chiral stationary phases for HPLC

Article abstract of DOI:10.1021/ac981143v

A library of 36 L-amino acid anilides, which are potential selectors for chiral HPLC, was synthesized in solution and attached to functionalized macroporous polymer beads. The best selector from the library was identified by a deconvolution process using

Full text of DOI:10.1021/ac981143v

Anal. Chem. 1999, 71, 1278-1284
On -Be a d Co m b in a t o ria l Ap p ro a c h t o t h e De s ig n o f
Ch ira l S t a t io n a ry P h a s e s fo r HP LC
Pe te r Mure r, Ke vin Le w a ndow s ki, Fra ntis e k Sve c , a nd J e a n M. J . Fre´ c he t*
Department of Chemistry, University of California, Berkeley, California 94720-1460
cyclodextrins,5 macrocyclic antibiotics,6 and low-molecular-weight
synthetic selectors7 have been linked to, or adsorbed onto, solid
supports, typically macroporous silica beads, affording effective
CSPs. The small, synthetically accessible selectors used in Pirkle’s
remarkable “brush-type” CSPs offer some distinct advantages over
high-molecular-weight biopolymers. They are amenable to exten-
sive chemical manipulations, allowing a rational optimization of
both their structure and their immobilization mode. In this context,
we recently reported the design of brush-type CSPs with improved
performance based on our size monodisperse macroporous beads.
In particular, greatly enhanced enantioselectivities were observed
when compared to analogous silica-based CSPs as a result of
decreased nonspecific interactions.8 Similarly, a thorough study
of linkers led to the design of a binding chemistry that minimizes
its effects on enantiomer recognition.9
A library of 3 6 L-amino acid anilides, which are potential
selectors for chiral HP LC, was synthesized in solution and
attached to functionalized macroporous polymer beads.
The best selector from the library was identified by a
deconvolution process using the HP LC separation of
several racemic N-(3,5-dinitrobenzoyl)-r-amino acid alky-
lamides as a probe. In each deconvolution step, a series
of chiral stationary phases (CSP s) containing a subset of
the amino acid anilide selector library was screened for
enantioselectivity. After the best CSP was chosen, the
library was further deconvoluted until the single best
selector was found. The highest selectivity was obtained
with a -proline-1 -indananilide that exhibited r values up
L
to 2 3 under normal-phase HP LC conditions. In addition,
six CSP s were prepared using individual selectors from
the library, and screening results indicate that the decon-
volution process indeed led to the most selective receptor.
If a new racemic solute has to be resolved, commercially
available CSPs are first screened and separation conditions such
as mobile-phase composition, flow rate, and temperature opti-
mized. Since this whole process may not afford adequate enan-
tioselectivities, an alternative method is to develop a new CSP
with a tailored selector capable of efficient enantiomer recognition.
The synthesis and screening of combinatorial libraries10,11 is a
validated strategy for the identification and study of ligand/
The pharmaceutical field offers a rapidly expanding range of
applications for the resolution of racemates by chromatography.1
High-performance liquid chromatography (HPLC) with chiral
stationary phases (CSPs) for the determination of the composition
of enantiomeric mixtures in biological and pharmacological studies
is now a well-established analytical tool. Moreover, the application
of the method on a preparative scale for the production of
enantiomerically pure compounds in amounts suitable for biologi-
cal testing, toxicological studies, and even, at a later stage, clinical
testing is gaining increasingly wide acceptance. During the
preliminary test phase of new chiral drugs, chromatography allows
rapid access to both of the pure enantiomers and can replace
advantageously the often lengthy elaboration of an enantioselective
synthesis.
(3) Hesse, G.; Hagel, R. Chromatographia 1 9 7 3 , 6, 277. Okamoto, Y.; Ka-
washima, M.; Hatada, K. J. Chromatogr. 1 9 8 6 , 363, 173.
(4) Schwanghart, A.-D.; Backmann, W.; Blaschke, G. Chem. Ber. 1 9 7 7 , 110,
778.
(5) Haan, S. M.; Armstrong, D. W. HPLC Separation of Enantiomers and Other
Isomers with Cyclodextrin-Bonded Phases. In Chiral Separations by HPLC;
Krstulovic, A. M., Ed.; Ellis Horwood Ltd.: Chichester, 1989; Chapter 10,
pp 208-284.
(6) Armstrong, D. W.; Tang, Y.; Bagwill, C.; Chen, J. R. Anal. Chem. 1 9 9 4 , 66,
1473-1484. Chang, S. C.; Wang, L. R.; Armstrong, D. W. J. Liq. Chromatogr.
1 9 9 2 , 15, 1411.
(7) Pirkle, W. H.; Pochapsky, T. C. Chem. Rev. 1 9 8 9 , 89, 347. Welch, C. J.
Crawling out of the Chiral Pool: The Evolution of Pirkle-Type Chiral
Stationary Phases. In Advances in Chromatography; Brown, P. R., Grushka,
E., Eds.; Marcel Dekker: New York, 1995; Vol. 35, pp 172-197.
(8) Liu, Y.; Svec, F.; Fre´chet, J. M. J. Anal. Chem. 1 9 9 7 , 69, 61.
(9) Lewandowski, K.; Murer, P.; Svec, F.; Fre´chet, J. M. J. Anal. Chem. 1 9 9 8 ,
70, 1629.
(10) For recent reviews, see: Gallop, M. A.; Barrett, R. W.; Dower, W. J.; Fodor,
S. P. A.; Gordon, E. M. J. Med. Chem. 1 9 9 4 , 37, 1233-1251, 1386-1401.
Terrett, N. K.; Gardner, M.; Gordon, D. W.; Kobylecki, R. J.; Steele, J.
Tetrahedron 1 9 9 5 , 51, 8135-8173. Thompson, L. A.; Ellman, J. A. Chem.
Rev. 1 9 9 6 , 96, 555-600. Gordon, E. M.; Gallop, M. A.; Patel, D. V. Acc.
Chem. Res. 1 9 9 6 , 29, 144-154. Balkenhohl, F.; von dem Bussche-
Hu¨ nnefeld, C.; Lansky, A.; Zechel, C. Angew. Chem., Int. Ed. Engl. 1 9 9 6 ,
35, 2288-2337. Lam, K. S.; Lebl, M.; Krchnak, V. Chem. Rev. 1 9 9 7 , 97,
411-448. Wilson, S. R., Czarnik, A. W., Eds. Combinatorial Chemistry; John
Wiley & Sons: New York, 1997.
A wide variety of CSPs for HPLC applications have been
developed, enabling enantiomer separation for a broad range of
compounds of different types with a variety of combinations of
functionalities. Proteins,2 polysaccharides,3 synthetic polymers,4
(1) Ahuja, S., Ed. Chiral Separations: Applications and Technology; American
Chemical Society: Washington, DC, 1996. Aboul-Enein, H. Y., Wainer, I.
W., Eds. The Impact of Stereochemistry on Drug Development and Use; Wiley-
VCH: New York, 1997. Jinno, K., Ed. Chromatographic Separations Based
on Molecular Recognition; Wiley-VCH: New York, 1997. Allenmark, S., Ed.
Chromatographic Enantioseparation: Methods and Applications; Ellis Hor-
wood: New York, 1991. Siret, L.; Bargmann-Leyder, N.; Tabute, A.; Caude,
M. Analusis 1 9 9 2 , 20, 427-435.
(2) Allenmark, S.; Bomgren, B. J. Chromatogr. 1 9 8 3 , 264, 63. Hermansson, J.
J. Chromatogr. 1 9 8 3 , 269, 71. Miwa, T.; Miyazawa, T.; Kayano, M.; Miyake,
Y. J. Chromatogr. 1 9 8 7 , 408, 316.
1278 Analytical Chemistry, Vol. 71, No. 7, April 1, 1999
10.1021/ac981143v CCC: $18.00 © 1999 American Chemical Society
Published on Web 02/25/1999

receptor interactions.12 It has been shown that the screening of
libraries of molecules tethered to solid supports in binding assays
with soluble molecules can greatly facilitate the identification of
key structural elements responsible for host/ guest interactions
and molecular recognition.10,13 To our knowledge, the only
reported attempts to use combinatorial techniques in chromatog-
raphy have only focused on affinity chromatography14 and capillary
electrophoresis.15 Our combinatorial methods are aimed at the
rapid preparation of tailor-made CSPs for HPLC designed for a
specific racemic solute. Recently, we described an application of
Pirkle’s principle of reciprocity16,17 in a combinatorial scheme that
led to the design of novel highly selective substituted dihydro-
pyrimidine-based CSPs.18 A single enantiomer of the target
racemate was immobilized on a macroporous polymeric support
and used for HPLC screening of a library of racemic compounds.
The best separated compound was prepared in enantiomerically
pure form and coupled to a support providing a CSP for the
efficient separation of the target racemate. Obviously, the CSP
prepared by this approach is theoretically optimized for the
resolution of one racemate only, although in practice, this CSP
may have a somewhat broader selectivity.
this deconvolution process to identify the most selective single
selector CSP is much smaller than the number of actual selectors
in the library.
EXPERIMENTAL SECTION
All reactions were carried out in standard oven-dried (120 °C)
glassware under an argon atmosphere blanket. Reagent-grade
chemicals were purchased from Aldrich or Sigma and used
without further purification. Tetrahydrofuran (THF) was freshly
distilled from sodium benzophenone ketyl radical under nitrogen.
Analytical thin-layer chromatography (TLC) was performed on
Merck silica gel 60 F254-coated plates. Compounds were visualized
by dipping the plates in a basic potassium permanganate solution
followed by heating. IR spectra were obtained on a Nicolet Mattson
Genesis II FT-IR spectrophotometer (KBr). 1H and 13C NMR
spectra were measured on Bruker AMX-300 or AMX-400 spec-
trometers in CDCl3.
General P rocedure for the P reparation of Amino Acid-
Based Libraries. The following amino acids and aromatic amines
were used to prepare libraries: N-t-Boc-L-valine (Val), N-t-Boc-L-
phenylalanine (Phe), N-t-Boc- -proline (Pro); 3,4,5-trimethoxya-
L
In this report, we demonstrate an alternative combinatorial
approach in which a CSP carrying a library of enantiomerically
pure potential selectors is used directly to screen for enantiose-
lectivity in the HPLC separation of target analytes. The best of
the bound selectors for the desired separation is then identified
in a few deconvolution steps. As a result of the “parallelism
advantage”, the number of columns that have to be screened in
niline, 3,5-dimethylaniline, 3-benyloxyaniline, 1-aminonaphthalene,
4-tritylaniline, 2-aminoanthracene, 5-aminoindane, 4-tert-butyla-
niline, 4-aminobiphenyl, 2-aminofluorene, 2-aminoanthraquinone,
and 3-amino-1-phenyl-2-pyrazolin-5-one.
The N-tert-butyloxycarbonyloxy-protected (Boc) amino acids
were activated for coupling with amines. First, 7.0 mmol of an
amino acid was dissolved in THF (35 mL), cooled to -15 °C, and
treated dropwise using a syringe with triethylamine (0.97 mL, 7.0
mmol) and ClCO2Et (0.67 mL, 7.0 mmol). After stirring at -15
°C for 1 h, a cold (-15 °C) mixture of aromatic amines composed
of an equimolar mixture of the desired amines (total amount of
amines 7.0 mmol) in THF (25-30 mL) was added dropwise.
Stirring was continued at -15 °C for 1 h and at room temperature
overnight. The resulting suspension was concentrated in vacuo,
diluted with ethyl acetate (200 mL), and extracted with 1 mol/ L
HCl (3 × 100 mL), saturated aqueous NaHCO3 (2 × 100 mL),
and saturated aqueous NaCl (2 × 100 mL). The organic phase
was separated, dried over MgSO4, and concentrated in vacuo to
afford quantitatively the product mixture as a colored solid or a
foam.
General P rocedure for the Deprotection of Libraries. The
product obtained from the procedure described above (7.0 mmol)
was dissolved in CH2Cl2 (10 mL), cooled to 0 °C, and treated with
trifluoroacetic acid/ acetic acid 1:1 (30 mL). After deprotection,
the resulting solution was stirred at room temperature overnight.
After TLC (hexane/ ethyl acetate 2:1) confirmed the disappearance
of starting materials, the reaction mixture was concentrated in
vacuo and diluted with H2O (30 mL) and CH2Cl2 (20 mL). A
solution of 2 mol/ L KOH was then added at 0 °C until the pH
was 9-10. The aqueous phase was further extracted with CH2Cl2
(3 × 100 mL), and the combined organic phases were washed
with H2O (100 mL) and saturated aqueous NaCl (100 mL), dried
over MgSO4, and concentrated in vacuo. Drying under high
vacuum provided a quantitative yield of the deprotected product
mixture as a colored solid. Integration of the individual 1H NMR
signals for the amide hydrogens of the compounds clearly showed
that all expected products were formed.
(11) For drug discovery, see examples in ref 6. Superconducting materials: Xiang,
X.-D.; Sun, X.; Briceno, G.; Lou, Y.; Wang, K. A.; Chang, H.; Wallace-
Freedman, W. G.; Chen, S.-W.; Schultz, P. G. Science 1 9 9 5 , 268, 1738.
Hsieh-Wilson, L. C.; Xiang, X.-D.; Schultz, P. G. Acc. Chem. Res. 1 9 9 6 , 29,
164. Magnetoresistant materials: Briceno, G.; Chang, H.; Sun, X.; Schultz,
P. G., Xiang, X.-D. Science 1 9 9 5 , 270, 273. Phosphors: Sun, X.-D.; Gao, C.;
Wang, J.; Xiang, X.-D. Appl. Phys. Lett. 1997, 70, 3353. Danielson, E.; Golden,
J. H.; McFarland, E. W.; Reaves, C. M.; Weinberg, W. H.; Wu, X. D. Nature
1 9 9 7 , 389, 944. Metal complexing ligands: Burger, M. T.; Still, W. C. J.
Org. Chem. 1 9 9 5 , 60, 7382. Francis, M. B.; Finney, N. S.; Jacobsen, E. N.
J. Am. Chem. Soc. 1996, 118, 8983-8984. Catalysts: Menger, F. M.; Eliseev,
A. V.; Migulin, V. A. J. Org. Chem. 1 9 9 5 , 60, 6666. Liu, G.; Ellman, J. A. J.
Org. Chem. 1 9 9 5 , 60, 7712. Gilbertson, S. R., Wang, X. J. Org. Chem. 1 9 9 6 ,
61, 434. Gilbertson, S. R.; Wang, X. Tetrahedron Lett. 1 9 9 6 , 37, 6475. Cole,
B. M.; Shimizu, K. D.; Krueger, C. A.; Harrity, J. P. A.; Snapper, M. L.;
Hoveyda, A. H. Angew. Chem., Int. Ed. Engl. 1 9 9 6 , 35, 1668. Taylor, S. J.;
Morken, J. P. Science 1 9 9 8 , 280, 267. Gennari, C.; Ceccarelli, S.; Piarulli,
U.; Monalbetti, C. H. G. N.; Jackson, R. F. W. J. Org. Chem. 1 9 9 8 , 63, 5312.
Polymers: Brocchini, S.; James, K.; Tangpasuthadol, V.; Kohn, J. J. Am.
Chem. Soc. 1 9 9 7 , 119, 4553-4554. Sensors: Dickinson, T. A.; Walt, D. R.,
White, J.; Kauer, J. S. Anal. Chem. 1 9 9 7 , 69, 3413-3418.
(12) Gennari, C.; Nestler, H. P.; Piarulli, U.; Salom, B. Liebigs Ann./ Recl. 1 9 9 7 ,
637-647.
(13) Weingartner, M. D.; Sekanina, K.; Still, W. C. J. Am. Chem. Soc. 1 9 9 8 ,
120, 9112-9113. Gavin, J. A.; Deng, N.; Alcala, M.; Mallouk, T. E. Chem.
Mater. 1 9 9 8 , 10, 1937-1944.
(14) Huang, P. Y.; Carbonell, R. G. Biotechnol. Bioeng. 1 9 9 5 , 47, 288. Huang,
P. Y.; Baumbach, G. A.; Dadd, C. A.; Buettner, J. A.; Musecar, B. L.; Hentsch,
M.; Hammond, D. J.; Carbonell, R. G. Bioorg. Med. Chem. 1 9 9 6 , 4, 699.
(15) Jung, G.; Hofstetter, H.; Feiertag, S.; Stoll, D.; Hofstetter, O.; Wiesmo`ller,
K.-H.; Schurig, V. Angew. Chem., Int. Ed. Engl. 1 9 9 6 , 35, 2148. Chiari, M.;
Desperati, V.; Manera, E.; Longhi, R. Anal. Chem. 1 9 9 8 , 70, 4967.
(16) Pirkle, W. H.; House, D. W.; Finn, J. M. J. Chromatogr. 1 9 8 0 , 192, 143.
Pirkle, W. H.; Welch, C. J.; Lamm, B. J. Org. Chem. 1 9 9 2 , 57, 3854. Pirkle,
W. H.; Liu, Y. J. Org. Chem. 1 9 9 4 , 59, 6911.
(17) Pirkle, W. H.; Hyun, M. H.; Bank, B. J. Chromatogr. 1 9 8 4 , 316, 585. Pirkle,
W. H.; Welch, C. J. J. Chromatogr. 1 9 9 2 , 589, 45.
(18) Lewandowski, K.; Murer, P.; Svec, F.; Fre´chet, J. M. J. Chem. Commun.
1 9 9 8 , 2237. Lewandowski, K.; Murer, P.; Svec, F.; Fre´chet, J. M. J. J. Comb.
Chem. 1 9 9 9 , 1, 105-112.
Analytical Chemistry, Vol. 71, No. 7, April 1, 1999 1279

General P rocedure for the P reparation of Single Selectors.
N-t-Boc- -proline (1.5 g, 7.0 mmol) was dissolved in CH2Cl2 (30
L
mL), cooled to 0 °C, and treated with 2-ethoxy-1-(ethoxycarbonyl)-
1,2-dihydroquinoline (EEDQ; 2.0 g, 8.0 mmol) in one portion. After
dropwise addition of an aromatic amine (7.0 mmol), the resulting
reaction mixture was stirred at 0 °C for 2 h and at room
temperature overnight. The reaction mixture was then diluted with
CH2Cl2 (100 mL) and washed successively with 1 mol/ L HCl (3
× 50 mL), saturated aqueous NaHCO3 (2 × 50 mL), H2O (50 mL),
and saturated aqueous NaCl (2 × 50 mL). The organic phase was
dried (MgSO4) and concentrated in vacuo to afford the product
quantitatively as a colored foam or solid. The following amines
were used in this procedure: 5-aminoindane (4 ) to afford 1 3 ,
4-tert-butylaniline (5 ) (f1 4 ), 4-aminobiphenyl (6 ) (f1 5 ), 3,5-
dimethylaniline (2 ) (f1 6 ) 1-aminonaphthalene (7 ) (f1 7 ), and
4-tritylaniline (8 ) (f1 8 ).
Cleavage of the t-Boc groups was performed with trifluoroacetic
acid/ acetic acid as described above, giving 80-90% of pure
1
product according to H NMR.
General P rocedure for the P reparation of Chiral Station-
ary P hases. To a slurry of 4-nitrophenyl carbonate-activated poly-
(2-hydroxyethyl methacrylate-co-ethylene dimethacrylate) beads
prepared according to the method described previously8,19 (1.6
g) in THF (15 mL) was added triethylamine (2.09 mL, 15 mmol)
at 0 °C. The suspension was then treated dropwise with a solution
of the product mixture obtained from the deprotection step
(maximum 7.0 mmol), in THF (20 mL) and stirring was continued
at room temperature for 3 h and at 60 ˚C overnight, providing a
brownish reaction mixture. The modified beads were then filtered
and washed repeatedly with THF, H2O, 1 mol/ L KOH, CH3OH,
acetone, and Et2O and then dried under high vacuum. The selector
content was determined by elemental analysis for nitrogen.
Chromatography. The chiral stationary phases were slurry
packed at a constant pressure of 150 bar into 150 × 4.6 mm i.d.
stainless steel columns. A Waters HPLC system consisting of two
510 HPLC pumps, a 717 plus autosampler, and a 486 UV detector,
and controlled by Millennium 2010 software, was used for all of
the chromatography.
Chiral separations were carried out using a 20% hexane/
dichloromethane mixture as a mobile phase. The separation
factors R (selectivity) were calculated using the equation R ) k′2/
k′1 where k′1 and k′2 are the retention factors of the enantiomers
defined as k′i ) (tR - t0)/ t0. tR and t0 represent the retention times
of the compound and 1,3,5-tri-tert-butylbenzene (void volume
marker), respectively. The racemic analytes N-(3,5-dinitrobenzoyl)-
R-amino acid alkyl amides were prepared by methods similar to
those reported previously.20
Fig u re 1 . Illustration of the concept applied in an on-bead combi-
natorial deconvolution scheme to identify an effective selector H* from
a multicomponent chiral stationary phase (selector library A*-S*).
selectivity. Deconvolution consist in the stepwise preparation of
a series of “sublibrary columns” of lower diversity, each of which
contains a CSP with a reduced number of library components (see,
for example, three columns with subgroups of selectors A*-F*,
G*-M*, and N*-S* in Figure 1).
The feasibility of this approach is demonstrated with a model
library of 36 compounds constructed from a combination of three
L-amino acids (valine, phenylalanine, proline) and 12 aromatic
amines (3,4,5-trimethoxyaniline (1 ), 3,5-dimethylaniline (2 ), 3-be-
nyloxyaniline (3 ), 5-aminoindane (4 ), 4-tert-butylaniline (5 ), 4-bi-
phenylamine (6 ), 1-aminonaphthalene (7 ), 4-tritylaniline (8 ),
2-aminoanthracene (9), 2-aminofluorene (10), 2-aminoanthraquino-
ne (1 1 ), and 3-amino-1-phenyl-2-pyrazolin-5-one (1 2 )) (Figure 2).
The libraries were prepared by a two-step procedure that includes
the coupling of amino acids with a mixture of amines followed by
deprotection of the resulting Boc-protected amides (Scheme 1).
Both steps were performed in solution to allow for fast analysis
of the homogeneity of the mixtures. The amine mixtures used in
the synthesis contained equimolar quantities of building blocks
1 -1 2 and were selected for their similar reactivities toward the
activated amino acids.
RESULTS AND DISCUSSION
P reparation of Library of Chiral Selectors. As outlined
schematically in Figure 1, our strategy consists of the following
steps. A mixture of chiral compounds (from A* to S*) is
immobilized on a solid support and packed to afford a “library
column” which is tested in the resolution of targeted racemic
compounds. If some separation is achieved, the column may be
“deconvoluted” to identify the selector possessing the highest
Activation of the N-Boc-protected R-amino acids and acylation
of the various amines was accomplished by conversion of the
carboxylic acids to mixed anhydrides with NEt3/ ClCO2Et followed
(19) Lewandowski, K.; Murer, P.; Svec, F.; Fre´chet, J. M. J. Anal. Chem. 1 9 9 8 ,
70, 1629-1638.
(20) Pirkle, W. H.; Pochapsky, T. C. J. Chromatogr. 1 9 8 6 , 369, 175.
1280 Analytical Chemistry, Vol. 71, No. 7, April 1, 1999

S c h e m e 1
in solution were then immobilized onto a polymeric solid support
in order to test them for the resolution of racemates. Coupling of
the amide mixtures with 5-µm macroporous 4-nitrophenyl carbon-
ate activated poly(hydroxyethyl methacrylate-co-ethylene dimethacry-
late) beads (HEMA-EDMA)24 afforded the chiral stationary
phases with a multiplicity of selectors (Scheme 1). The individual
library members are assumed to show similar reactivities toward
the carbonate groups of the activated support beads since they
all contain a primary amine reactive group derived from the amino
acid component. Activation of size monodisperse hydroxyl-
functionalized HEMA-EDMA beads was accomplished by reac-
tion with 4-nitrophenyl chloroformate as published elsewhere8 and
yielded 0.94 mmol/ g carbonate functionalities. A slurry of these
beads in THF was treated with an excess of the respective amide
mixtures. In this way, the heterogeneous population of all selectors
present in the mixture should be attached to each bead. The use
of columns with mixed dissimilar selectors has not been recom-
mended for actual enantioseparation25 since they may exhibit
opposite signs of enantioselectivity, interact with each other, and/
or interact simultaneously with an analyte molecule and thus
negatively affect the overall selectivity. However, the majority of
mixed selectors in our combinatorial approach is not completely
dissimilar. They are based on identical chiral core and differ only
in the structure of the auxiliary substituents. Systematic studies
of the effects of substituents attached to chiral selectors on their
preferences to interact stronger with the specific enantiomer that
would support this claim are scarce. For example, Pirkle’s
separation of 54 5-arylhydantoins has demonstrated systematically
higher retention of (+)-enantiomers.26 Our study of the separation
of over 150 compounds with identical dihydropyrimidine scaffold
and various substituents also did not indicate any changes in
retention order.18 This suggests that the use of an identical scaffold
decreases the probability of inversion in selectivity. Therefore,
the mixed selector method appears to constitute a worthwhile
approach to optimized CSPs.
Fig u re 2 . Building blocks used for the preparation of π-basic-
substituted amino acid-based selector mixtures.
by reaction with the amine mixture to afford a mixture of chiral
amides. In all cases, the yields of the last step were essentially
quantitative with no side products detected by NMR analysis.
Other activation methods such as 1-[3-(dimethylamino)propyl]-
3-ethylcarbodiimide hydrochloride (EDC)/ 1-hydroxy-1H-benzot-
riazole (HOBt),21 N-hydroxysuccinimide (HOSu)/ dicyclohexyl-
carbodiimid (DCC),22 or benzotriazol-1-yloxytripyrrolidinophos-
23
phonium hexafluorophosphate (PyBOP)
gave less reactive
activated forms of the acid that did not fully acylate the aromatic
amines. The hydrophobic nature of the libraries produced with
the protected building blocks allows their separation from traces
of unreacted amines or amino acids by simple extraction. After
the coupling step was carried out, the Boc-protecting group was
readily removed by treatment with a trifluoroacetic acid/ acetic
acid/ CH2Cl2 (3:3:2) mixture. Surprisingly, only partial cleavage
of the Boc group was obtained even after extended reaction times
if the cleavage was carried out with a standard 5% solution of
trifluoroacetic acid in CH2Cl2. The resulting product mixtures were
obtained as colored foams or powders. Characterization by NMR,
MS, and IR analyses indicated that all members of the library were
present in the mixtures.
P reparation of Chiral Stationary P hases with Mixed
Selectors. The homogeneous mixtures of amino acid derivatives
Screening for Enantioselectivity. As expected from the
design of the experiment, the HPLC column packed with CSP 1
containing all 36 members of the library with π-basic substituents
(21) Seebach, D.; Overhand, M.; Ku¨ hnle, F. N. M.; Martinoni, B.; Oberer, L.;
Hommel, U.; Widmer, H. Helv. Chim. Acta 1 9 9 6 , 79, 913.
(22) Weygand, F.; Hoffmann, D.; Wu¨ nsch, E. Z. Naturforsch. Teil B 1 9 6 6 , 21,
426. Wu¨ nsch, E.; Drees, F. Chem. Ber. 1 9 6 6 , 99, 110.
(23) Coste, J.; LeNguyen, D.; Castro, B. Tetrahedron Lett. 1 9 9 0 , 31, 205.
(24) Lewandowski, K.; Svec, F.; Fre´chet, J. M. J. Chem. Mater. 1 9 9 8 , 10, 385.
(25) Pirkle, W. H.; Welch, C. J. J. Chromatogr., A 1 9 9 6 , 731, 322.
(26) Pirkle, W. H.; Gan, K. Z. J. Chromatogr., A 1 9 9 7 , 790, 65.
Analytical Chemistry, Vol. 71, No. 7, April 1, 1999 1281

Ta b le 1 . S e p a ra t io n o f De riva t ize d Am in o Ac id s o n CS P 2 -CS P 1 4 a
selectivity factor R
b
f
series
amino acid
amine
loading (mmol/ g)
Ic
IId
IIIe
IV
1
CSP 2
CSP 3
CSP 4
CSP 5
CSP 6
CSP 7
CSP 8
CSP 9
CSP 1 0
CSP 1 1
CSP 1 2
CSP 1 3
CSP 1 4
Val
Phe
Pro
Pro
Pro
Pro
Pro
Pro
Pro
Pro
Pro
Pro
Pro
1-12
1-12
1-12
1-6
7-12
1-3
4-6
4
5
6
1
2
0.68
0.60
0.74
0.78
0.77
0.75
0.74
0.75
0.74
0.74
0.63
0.74
0.70
4.9
5.8
13.7
13.6
7.3
17.4
14.9
23.1
11.5
12.4
24.7
2.5
5.1
6.3
14.4
19.8
7.7
19.8
15.3
22.1
10.7
13.5
25.2
2.9
4.6
4.5
6.6
10.5
5.4
9.0
12.2
13.2
10.6
8.1
13.5
1.8
6.7
4.5
8.5
9.7
5.8
8.6
9.0
12.8
9.6
7.7
12.8
1.9
2.7
2
3
4
control
3
3.6
4.6
3.5
a
Conditions: column 150 × 4.6 mm i.d.; mobile phase, 20% hexane in dichloromethane; flow rate, 1 mL/ min; UV detection at 254 nm. b For
structures of amines, see Figure 2. c (3,5-Dinitrobenzoyl)leucine diallylamide. d (3,5-Dinitrobenzoyl)leucine diethylamide. e (3,5-Dinitrobenzoyl)alanine
diallylamide. f (3,5-Dinitrobenzoyl)alanine diethylamide.
separates π-acid-substituted amino acid amides. Although encour-
aging since it suggests the presence of at least one useful selector,
this result does not reveal which of the numerous selectors on
CSP 1 is the most powerful. Therefore, a deconvolution process
that involves the preparation of series of beads with smaller
numbers of attached selectors is used. Thus, in the next step,
each single amino acid is coupled separately with the set of 12
amines resulting in 3 new mixed-ligand CSPs (CSP 2 -CSP 4 ).
Once packed into HPLC columns these CSPs were evaluated.
Table 1 summarizes the separation results for the enantiomers
of various N-(3,5-dinitrobenzoyl)-R-amino acid derivatives I-IV.
For all these CSPs, the S-enantiomers are always retained longer
than are the R-enantiomers. The use of a hexane/ dichloromethane
mobile phase affords separations of these derivatives as exempli-
fied in the collection of chromatograms obtained on various CSPs
for analyte I (Figure 3). Although the peaks for some of the
columns are somewhat broader due to less perfect packing and
lower efficiency, the separation factors R can safely be calculated
to select the best performing column in each series.
The highest selectivity of 13.7 in the first series of columns,
CSPs 2 -4 , was found for the proline-based CSP 4 while the R
Fig u re 3 . Overlay of chromatograms obtained with CSPs 2-11
prepared in four deconvolution series. Conditions: analyte, (3,5-
dinitrobenzoyl)leucine diallylamide; column, 150 × 4.6 mm i.d.; mobile
phase, 20% hexane in dichloromethane; flow rate, 1 mL/min, UV
detection 254 nm.
values of the two other columns are close to 5 (see analyte I,
Table 1). The increased selectivity for proline-based selectors
attached to silica beads has also been observed by others.27 To
further determine which of the 12 compounds in the proline-based
column affords the highest selectivity, a third set of columns is
prepared (CSP 5, CSP 6) by splitting the 12 members of the amine
building blocks 1 -1 2 (Figure 2) in two subgroups. Thus, the
first proline-based sublibrary column (CSP 5 ) is prepared using
six amines with smaller aromatic substituents (amines 1 -6 ), and
the second column (CSP 6 ) with amines characterized with larger
substituents (7 -1 2 ). CSP 5 and CSP 6 exhibit selectivities of
13.6 and 7.3, respectively. In the next step, the six amine group
present in the more selective column CSP 5 is divided again into
two groups (1 -3 , 4 -6 ), containing three amines with mainly
meta-substituted aromatic amines and another three with para-
substituted amines. Quite unexpectedly, segregation of selectors
with meta- and para-substituted amines in two CSPs increases
selectivities of both new columns indicating that synergetic effects
25
occur in the mixtures of selectors (vide infra). The respective
columns CSP 7 and CSP 8 exhibit rather high R values of 17.4
and 14.9 for analyte I and indicate that both groups involve at
least one selector with a very high selectivity. This is not surprising
for CSP 7 , since a selector substituted with 2 that is also present
in this column was previously shown to be quite successful.8 Since
the performance of these two columns was similar (CSP 7
separates better analytes I and II while CSP 8 performs better
with analytes III and IV), we decided to further deconvolute CSP
8 , which involves entirely new selectors. Three columns, CSPs
(27) Pirkle, W. H.; Murray, P. G. J. Chromatogr. 1 9 9 3 , 641, 11. Ihara, T.;
Sugimoto, Y.; Asada, M.; Nakagama, T.; Hobo, T. J. Chromatogr., A 1 9 9 5 ,
694, 49.
1282 Analytical Chemistry, Vol. 71, No. 7, April 1, 1999

the use of all 20 natural amino acids with 12 amines would lead
to a library of 240 selectors. While the preparation and testing of
240 columns would be time-consuming, the mixture of these
selectors could be deconvoluted using our approach in 15 columns
representing only 1/ 16 of the total number of columns that would
be otherwise required. The parallelism advantage of the “library-
on-bead” approach with mixed-selector column would be even
more impressive with much larger libraries of selectors for which
the deconvolution by splitting the library in each step to two or
three sublibraries would rapidly lead to the most selective CSP.
Obviously, this approach can dramatically decrease the time
required for the development of a novel CSP.
Although the power of the combinatorial approach is clearly
demonstrated, this method has also some limitations. For example,
in a hypothetical situation in which only a single selector is active
and all members of a much larger library are attached to the beads
in equal amounts, the percentage of the active selector in the
mixture is low, and despite its possibly high specific selectivity
(selectivity per unit of loading), the actual selectivity of a mixed-
selector CSP may be rather small because the loading of the
specific selector is very low. Accordingly, the peaks for both
enantiomers may elute close to each other and the actual
separation may become impossible to discern within the limits of
experimental errors. Thus the sensitivity of the chromatographic
screening may somewhat limit this approach. However, the
number of selectors that may be screened in a single column is
still impressive.
Fig u re 4 . Retention factors k′ determined for (3,5-dinitrobenzoyl)-
leucine diallylamide on CSPs 2-11. For conditions, see Figure 3.
9 -1 1 , packed with beads containing only individual selectors
were prepared (CSPs 9 -1 1 with amines 4 -6 , respectively).
Although all these columns exhibit rather high selectivities, an R
value of 23.1 is achieved with CSP 9 featuring 4 as a part of the
proline selector. Figure 4 shows the changes in separation factors
k′ determined for (3,5-dinitrobenzoyl)leucine diallylamide on CSPs
2 -1 1 .
Since our method of screening initially operates by selecting
groups of molecules rather than individual compounds and since
the difference between both CSP 7 and CSP 8 was small, it is
possible that our best CSP 9 is not the most efficient selector of
the original mixture. To confirm this as well as to satisfy our
curiosity to uncover which other selector is powerful, we prepared
three additional columns CSPs 1 2 -1 4 containing single proline-
based selectors with amines 1 -3 as a control. As expected from
our previous research,8 CSP 1 2 prepared with amine 2 also
exhibits very high selectivity (R ) 24.7 for analyte I) similar to
that of CSP 9 . Surprisingly, CSP 1 3 and CSP 1 4 , prepared with
amines 7 and 8 , respectively, afford only modest R values of less
than 4.
Once again, the S-enantiomers are retained longer than the
R-enantiomers for all CSPs 9 -1 4 prepared from individual
selectors. This indicates that there is no inversion of selectivity
in this group as a result of the use of different substituents.
The rapid increase in the separation factors observed for the
individual series of columns reflects not only the improvement in
the intrinsic selectivities of the individual selectors but also the
effect of increased loading with the most potent selector. Although
the overall loading determined from nitrogen content remains
virtually constant at about 0.7 mmol/ g for all CSPs 1 -1 2 , the
fractional loading of each selector increases as the number of
selectors in the mixture decreases, provided the reactivity of all
selector molecules in the immobilization step is essentially equal.
Thus, the whole method of building block selection and sublibrary
synthesis can be also viewed as an amplification process.
In the classical one-column/ one-selector approach, the number
of columns that have to be tested equals the number of selectors.
Using the chemistry described above, this would require the
preparation, packing, and testing of 36 CSPs. In contrast, our
combinatorial scheme allows us to obtain a very highly selective
CSP from the same group of 36 selectors using only 11 columns
or less than one-third. A simple theoretical calculation reveals that
Another limitation of our approach is common to many
combinatorial techniques. The rapid screening of mixtures of
compounds can fail to discover some hits. This penalty is
counterbalanced by the advantage of considerably increased speed
of screening and large number of tested selectors. However, the
danger of missing good selectors should not be overestimated.
In a hypothetical case, simple averaging would imply that an
equimolar mixture that contains one very potent selector (R )
15) together with two completely inactive compounds would
exhibit an overall selectivity factor of 5 while another mixture of
three selectors each with R ) 6 affords a separation medium with
selectivity factor of 6. Following the procedure outlined above,
the latter mixture would be further deconvoluted to find a
mediocre selector with R ) 6 and miss the best one. However,
this may not be the typical case because this assumes (i) a linear
proportionality of selectivity to the amount of attached selector,
(ii) the perfect additivity of selectivities, and (iii) the lack of
synergistic effects of selectors within the beads. A comparison of
the data shown in Table 1 for mixed selectors CSP 7 and CSP 8
with averaged selectivities for individual CSPs 9-11 and CSPs
12-14, respectively, clearly documents that the mixed-selector
column has a selectivity for the critical analyte higher than the
average value. For example, mixed CSP 7 is characterized by an
R value of 19.8 for (3,5-dinitrobenzoyl)leucine diethylamide while
the average selectivity of CSPs 9-11 is only 15.4. Obviously, the
most potent selector prevails in the mixture, thus decreasing the
danger of missing a real hit. These data also indicate again that
the mixtures are unlikely to contain selectors with opposite signs
of enantioselectivity.
In contrast, there are less limitations from the chemical point
of view. The preparation of large, well-defined, libraries that involve
Analytical Chemistry, Vol. 71, No. 7, April 1, 1999 1283

amino acid building blocks has been demonstrated many times.10
Carefully optimized reaction conditions for the preparation of other
mixed libraries can also ensure that each desired compound is
present in sufficient amount. However, the reaction rates of some
individual selectors with the activated solid support may be lower
than others and those that react more readily would then occupy
a majority of the sites within the beads. Since the most reactive
selectors may not be the most selective, testing of a slightly larger
number of specifically designed CSPs may be required to avoid
false negative results.
Although high levels of enantioselectivity in the range of R
values well over 20 may be unnecessary or even undesired for
the analytical separations of enantiomers since they lead to rapid
increase in the overall analysis time, they are desirable for
preparative applications in columns operating under overload
conditions.
general method are set by the sensitivity of the detection that is
required for a system characterized by low loading of the best
selector at the early stage of deconvolution and by the remote
possibility of simultaneous attachment of selectors with opposite
signs of enantioselectivity. Although we have chosen simple
π-basic-substituted amino acids for our model library of selectors
used in the separation of π-acidic analytes to demonstrate the
concept, many existing libraries of organic compounds could also
be attached to reactive beads. These media would then allow the
screening for chiral recognition and accelerate the design of novel
chiral stationary phases with high selectivity for the desired target
racemate. As an additional benefit, such a study carried out with
structurally related families of selectors can further improve the
general understanding of chiral recognition.
ACKNOWLEDGMENT
Support of this research by a grant of the National Institute of
General Medical Sciences, National Institutes of Health (GM-
44885) is gratefully acknowledged. P.M. thanks the Swiss National
Science Foundation for a postdoctoral fellowship.
CONCLUSION
Combinatorial chemistry, a powerful tool in many areas such
as drug discovery, materials research, and catalysis, can also be
used in the area of molecular recognition to discover new selectors
for chiral HPLC. The strategy involving simultaneous attachment
of a library of selector candidates on each bead and packing a
single column followed by screening of the separation ability for
various racemates has now been validated. The limits of this
Received for review October 20, 1998. Accepted January
15, 1999.
AC981143V
1284 Analytical Chemistry, Vol. 71, No. 7, April 1, 1999