Imprinted polymers as protecting groups for regioselective modification of polyfunctional substrates

Article abstract of DOI:10.1021/ja982238h

Imprinted polymers were prepared using a functional monomer derived from boronophthalide and a number of steroid templates bearing spatially separated hydroxyl groups. The cooperative nature of the binding interaction was demonstrated in polymers imprinted with androst-5-ene-3β,17β-diol and its structural analogues. The stoichiometry and kinetics of binding were probed using IR spectroscopy, selective solvent extractions, and chemical modification experiments. The feasibility of using imprinted polymers as reusable protecting groups was established by the regioselective acylation of trihydroxysteroids bound to polymers imprinted with structurally related diols. In polymers prepared with tert-butyl ester templates "matched" to the substrate, regioselectivities as high as 23.1:1 (24-acetate:3-acetate) in the monoester products (65% of recovered material) were seen. In the "unmatched" case, the ratio fell to 5.4:1; however, in functionally identical control polymers, imprinted with ethylene glycol, the regioselectivity was completely reversed (<1:100), and only poor yields of monoesters (13%) were obtained.

Full text of DOI:10.1021/ja982238h

6640
J. Am. Chem. Soc. 1999, 121, 6640-6651
Imprinted Polymers as Protecting Groups for Regioselective
Modification of Polyfunctional Substrates
Cameron Alexander, Craig R. Smith, Michael J. Whitcombe,* and Evgeny N. Vulfson
Contribution from the Macromolecular Science Department, IFR, Reading Laboratory, Earley Gate,
Whiteknights Road, Reading RG6 6BZ, U.K.
ReceiVed June 26, 1998
Abstract: Imprinted polymers were prepared using a functional monomer derived from boronophthalide and
a number of steroid templates bearing spatially separated hydroxyl groups. The cooperative nature of the
binding interaction was demonstrated in polymers imprinted with androst-5-ene-3â,17â-diol and its structural
analogues. The stoichiometry and kinetics of binding were probed using IR spectroscopy, selective solvent
extractions, and chemical modification experiments. The feasibility of using imprinted polymers as reusable
protecting groups was established by the regioselective acylation of trihydroxysteroids bound to polymers
imprinted with structurally related diols. In polymers prepared with tert-butyl ester templates “matched” to
the substrate, regioselectivities as high as 23.1:1 (24-acetate:3-acetate) in the monoester products (65% of
recovered material) were seen. In the “unmatched” case, the ratio fell to 5.4:1; however, in functionally identical
control polymers, imprinted with ethylene glycol, the regioselectivity was completely reversed (<1:100), and
only poor yields of monoesters (13%) were obtained.
Introduction
drugs9 and pollutants.10 Most of the research has focused on
the synthesis of polymers capable of resolving racemic mix-
tures,11 but other applications, e.g., in solid-phase extraction,12
imprinted polymer-supported reagents,13 catalysis,14 sensors,15
removal of undesirable components from complex mixtures,16
Molecular imprinting is a methodology for the introduction
of selective recognition sites into highly cross-linked polymeric
matrixes via template-directed assembly of functionalized
monomers in a polymer-forming mixture. This enables the
positioning of complementary functionality in discrete cavities
in the polymer, with the precise spatial arrangement of functional
groups to provide specific interactions with the template on
rebinding.1 Over the past few years, imprinting has been
successfully applied to the preparation of polymers with
selectivity to a wide range of natural products such as amino
acids2 and short peptides,3 monosaccharides and their deriva-
tives,4 nucleotide bases,5,6 and steroids,7,8 as well as numerous
(5) (a) Shea, K. J.; Spivak, D. A. Macromolecules 1998, 31, 2160-
2165. (b) Shea, K. J.; Spivak, D. A.; Sellergren, B. J. Am. Chem. Soc. 1993,
115, 3368-3369.
(6) Mathew, J.; Buchardt, O. Bioconjugate Chem. 1995, 6, 524-528.
(7) Ramstro¨m, O.; Ye, L.; Mosbach, K. Chem. Biol. 1996, 3, 471-477.
(8) (a) Cheong, S. H.; McNiven, S.; Rachkov, A.; Levi, R.; Yano, K.;
Karube, I. Macromolecules 1997, 30, 1317-1322. (b) McNiven, S.;
Yokobayashi, Y.; Cheong, S. H.; Karube, I. Chem. Lett. 1997, 1297-1298.
(9) (a) Fischer, L.; Mu¨ller, R.; Ekberg, B.; Mosbach, K. J. Am. Chem.
Soc. 1991, 113, 9358-9360. (b) Vlatakis, G.; Andersson, L. I.; Mu¨ller, R.;
Mosbach, K. Nature 1993, 361, 645-647. (c) Ramstro¨m, O.; Yu, C.;
Mosbach, K. J. Mol. Recognit. 1996, 9, 691-696. (d) Senholdt, M.;
Siemann, M.; Mosbach, K.; Andersson, L. I. Anal. Lett. 1997, 30, 1809-
1821. (e) Siemann, M.; Andersson, L. I.; Mosbach, K. J. Antibiot. 1997,
50, 89-91. (f) Tomioka, Y.; Kudo, Y.; Hayashi, T.; Nakamura, H.; Niizeki,
M.; Hishinuma, T.; Mizugaki, M. Biol. Pharm. Bull. 1997, 20, 397-400.
(g) Matsui, J.; Takeuchi, T. Anal. Commun. 1997, 34, 199-200.
(10) (a) Muldoon, M. T.; Stanker, L. H. J. Agric. Food Chem. 1995, 43,
1424-1427. (b) Haupt, K.; Dzgoev, A.; Mosbach, K. Anal. Chem. 1998,
70, 628-631. (c) Dickert, F. L.; Besenbo¨ck, H.; Tortschanoff, M. AdV.
Mater. 1998, 10, 149-151. (d) Lu¨bke, M.; Whitcombe, M. J.; Vulfson, E.
N. J. Am. Chem. Soc. 1998, 120, 13342-13348.
* To whom correspondence should be addressed. Tel: +44 118 9357000.
Fax: +44 118 9267917. E-mail: michael.whitcombe@bbsrc.ac.uk.
(1) For recent reviews, see: (a) Mosbach, K.; Ramstro¨m, O. Bio/
Technology 1996, 14, 163-170. (b) Shea, K. J. Trends Polym. Sci. 1994,
2, 166-173. (c) Steinke, J.; Sherrington, D. C.; Dunkin, I. R. AdV. Polym.
Sci. 1995, 123, 81-125. (d) Wulff, G. Angew. Chem., Int. Ed. Engl. 1995,
34, 1812-1832. (e) Vidyasankar, S.; Arnold, F. H. Curr. Opin. Biotechnol.
1995, 6, 218-224. (f) Mosbach, K. Trends Biochem. Sci. 1994, 19, 9-14.
(g) Alexander, C.; Whitcombe, M. J.; Vulfson, E. N. Chem. Br. 1997, 33,
23-26. (h) Haupt, K.; Mosbach, K. J. Mol. Recogn. 1998, 11, 62-68. (i)
Molecular and Ionic Recognition with Imprinted Polymers; Bartsch, R. A.,
Maeda, M., Eds.; ACS Symposium Series 703; American Chemical
Society: Washington, DC, 1998.
(2) (a) Andersson, L.; Sellergren, B.; Mosbach, K. Tetrahedron Lett.
1984, 25, 5211-5214. (b) Sellergren, B.; Lepisto¨, M.; Mosbach, K. J. Am.
Chem. Soc. 1988, 110, 5853-5860. (c) Andersson, L. I.; Mosbach, K. J.
Chromatogr. 1990, 516, 313-322.
(3) (a) Ramstro¨m, O.; Nicholls, I. A.; Mosbach, K. Tetrahedron:
Asymmetry 1994, 5, 649-656. (b) Andersson, L. I.; Mu¨ller, R.; Vlatakis,
G.; Mosbach, K. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 4788-4792. (c)
Yano, K.; Nakagiri, T.; Takeuchi, T.; Matsui, J.; Ikebukuro, K.; Karube, I.
Anal. Chim. Acta 1997, 357, 91-98.
(11) (a) Ramstro¨m, O.; Ansell, R. J. Chirality 1998, 10, 195-209. (b)
Sellergren, B. Trends Anal. Chem. 1997, 16, 310-320. (c) Kempe, M.;
Mosbach, K. J. Chromatogr. A 1995, 694, 3-13. (d) Kempe, M.; Mosbach,
K. J. Chromatogr. A 1994, 664, 276-279.
(12) (a) Sellergren, B. Anal. Chem. 1994, 66, 1578-1582. (b) Muldoon,
M. T.; Stanker, L. H. Anal. Chem. 1997, 69, 803-808. (c) Andersson, L.
I.; Paprica, A.; Arvidsson, T. Chromatographia 1997, 46, 57-62. (d) Rashid,
B. A.; Briggs, R. J.; Hay, J. N.; Stevenson, D. Anal. Commun. 1997, 34,
303-305. (e) Walshe, M.; Howarth, J.; Kelly, M. T.; O’Kennedy, R.; Smyth,
M. R. J. Pharm. Biomed. Anal. 1997, 16, 319-325. (f) Martin, P.; Wilson,
I. D.; Morgan, D. E.; Jones, G. R.; Jones, K. Anal. Commun. 1997, 34,
45-47. (g) Mullett, W. M.; Lai, E. P. C. Anal. Chem. 1998, 70, 3636-
3641. (h) Zander, A.; Findlay, P.; Renner, T.; Sellergren, B.; Swietlow, A.
Anal. Chem. 1998, 70, 3304-3314.
(4) (a) Wulff, G.; Poll, H. G. Makromol. Chem. 1987, 188, 741-748.
(b) Chen, G. H.; Guan, Z. B.; Chen, C. T.; Fu, L. T.; Sundaresan, V.; Arnold,
F. H. Nat. Biotechnol. 1997, 15, 354-357. (c) Mayes, A. G.; Andersson,
L. I.; Mosbach, K. Anal. Biochem. 1994, 222, 483-488.
(13) Bystro¨m, S. E.; Borje, A.; Akermark, B. J. Am. Chem. Soc. 1993,
115, 2081-2083.
10.1021/ja982238h CCC: $18.00 © 1999 American Chemical Society
Published on Web 06/25/1999

Imprinted Polymers as Protecting Groups
J. Am. Chem. Soc., Vol. 121, No. 28, 1999 6641
controlling crystallization,17 and separation and concentration
of proteins18 and microorganisms,19 have also been reported.
The formation of a template-monomer complex in the
polymerization mixture can be accomplished either by covalent
attachment of monomers to the template using appropriate
chemistry or by noncovalent complexation in situ. However,
regardless of the methodology employed, the selectivity of the
resulting materials is dependent on a number of contributing
factors, i.e., interactions between template/ligand and polymer-
forming components,20 the density and rigidity of the polymer
matrix as determined by the degree and chemical nature of cross-
linkers,21 and perhaps most importantly, the interaction between
functional groups in the polymer’s recognition site with those
of the ligand. Indeed, in most cases it is charge interactions,
strong hydrogen bonding and coordination with metals, and the
formation of a labile covalent bond that are the main contributors
to the overall energy of binding. Consequently, the polymer’s
ability to discriminate between the target and its structural
analogues is largely determined by the cumulative effect of
individual interactions in terms of free energies or, in other
words, by the cooperativity between functional groups in the
recognition site.
detail. Furthermore, Shea and Sasaki25 have established, in a
comprehensive FT-IR and solid-state NMR study, that rebinding
of bifunctional ligands leads to the formation of two covalent
bonds in a high proportion of bifunctional sites, as might be
expected considering the chemistry initially employed to prepare
the imprinted polymers. Although some unresolved issues with
regard to the kinetics of binding, e.g., the relative rates of
reaction between the first and the second pairs of functional
groups in the imprinted sites, still remain, it is generally accepted
that the binding of ligands to imprinted polymers is driven by
the reconstitution of the same multiple interactions in the sites
which were engaged in the assembly of template-polymer
complexes. However, if this is the case, then it should be feasible
to produce imprinted polymers capable of “holding” the ligand
in place by interacting with some of the functional groups while
leaving other groups on the ligand accessible for selective
chemical modification. This approach, where imprinted poly-
mers are effectively used as protecting groups for regioselective
synthesis, is rather different from other “synthetic” applications
reported for imprinted polymers, for example their use as
catalysts,14 supported reagents,13 and “microreactors” for chiral
synthesis.26,27
Cooperative interactions, especially in covalently imprinted
polymers, have been extensively investigated, notably by the
groups of Shea and Wulff and co-workers (see reviews1b,d). It
has been conclusively shown that the specificity of imprinted
sites is increased dramatically when two functional groups rather
than one are introduced into the polymer’s recognition site, in
terms of the ability of polymers imprinted with a single
enantiomer to resolve racemic mixtures in chromatography.22
The effects of additional functional groups in the template
molecule23 and the structures of polymerizable boronic acids24
on the selectivity of binding have also been studied in some
In this paper we describe the interaction of a series of
homologous sterols with their respective imprinted polymers
containing one or two boronic acid residues in the recognition
sites. The covalent method of imprinting was chosen as the most
suitable for the purpose of this investigation, as it does not
require functional monomers to be present in the polymerization
mixture in excess. This was important for achieving the
incorporation of functionality exclusively into the recognition
sites to enhance the selectivity in chemical modification
experiments on polymer-bound ligands. Sterols were selected
as model templates on account of their biomedical importance28
and because numerous structural analogues are commercially
available and/or can be synthesized according to published
procedures. Also, the regioselective modification of individual
hydroxyl groups in sterols relies on the extensive use of
protecting groups and is notoriously difficult to accomplish.
Hence, it was an ideal model system for testing the utility of
imprinted materials as protecting groups in polymer-directed
regioselective synthesis.
(14) (a) Beach, J. V.; Shea, K. J. J. Am. Chem. Soc. 1994, 116, 379-
380. (b) Muller, R.; Andersson, L. I.; Mosbach, K. Makromol. Chem. Rapid
Commun. 1993, 14, 637-641. (c) Wulff, G.; Gross, T.; Scho¨nfeld, R.
Angew. Chem., Int. Ed. Engl. 1997, 36, 1962-1964. (d) Matsui, J.; Nicholls,
I. A.; Karube, I.; Mosbach, K. J. Org. Chem. 1996, 61, 5414-5417. (e)
Liu, X.-C.; Mosbach, K. Macromol. Rapid Commun. 1997, 18, 609-615.
(f) Gamez, P.; Dunjic, B.; Pinel, C.; Lemaire, M. Tetrahedron Lett. 1995,
36, 8779-8782.
(15) (a) Yamamura, K.; Hatakeyama, H.; Naka, K.; Tabushi, I.; Kurihara,
K. J. Chem. Soc., Chem. Commun. 1988, 79-81. (b) Starodub, S.; Piletsky,
S. A.; Lavryk, N. V.; Elskaya, A. V. Sens. Actuators B 1994, 18-19, 629-
632. (c) Hedborg, E.; Winquist, F.; Lundstro¨m, I.; Andersson, L. I.;
Mosbach, K. Sens. Actuators A 1993, 37-38, 796-799. (d) Kriz, D.;
Ramstro¨m, O.; Mosbach, K. Anal. Chem, 1997, 69, A345-A349.
(16) (a) Whitcombe, M. J.; Alexander, C.; Vulfson, E. N. Trends Food
Sci. Technol. 1997, 8, 140-145. (b) Joshi, V. P.; Karode, S. K.; Kulkarni,
M. G.; Mashelkar, R. A. Chem. Eng. Sci. 1998, 53, 2271-2284.
(17) D’Souza, S. M.; Alexander, C.; Carr, S. W.; Waller, A. M.;
Whitcombe, M. J.; Vulfson, E. N. Nature 1999, 398, 312-316.
(18) (a) Mallik, S.; Plunkett, S. D.; Dhal, P. K.; Johnson, R. D.; Pack,
D.; Shnek, D.; Arnold, F. H. New J. Chem. 1994, 18, 299-304. (b) Shnek,
D. R.; Pack, D. W.; Sasaki, D. Y.; Arnold, F. H. Langmuir 1994, 10, 2382-
2388. (c) Hjerten, S.; Liao, J. L.; Nakazato, K.; Wang, Y.; Zamaratskaia,
G.; Zhang, H. X. Chromatographia 1997, 44, 227-234.
Results and Discussion
For the purpose of this study, two methods for imprinting
compounds such as sterols with single or spatially separated
hydroxyl groups were suitable.29 One relies on the formation
(24) Wulff, G.; Minarik, M. J. Liquid Chromatogr. 1990 13 (15), 2987-
3000. See also ref 4a above for the effect of flexibility in the polymerizable
boronic acid on the resolution of phenyl-R-D-mannopyranoside.
(25) Sasaki, D. Y.; Shea, K. J. J. Am. Chem. Soc. 1991, 113, 4109-
4120.
(26) (a) Wulff, G. Polymeric Reagents and Catalysts; Ford, W. T., Ed.;
ACS Symposium Series 308; American Chemical Society: Washington,
DC, 1986; pp 186-230. (b) Wulff, G.; Vieitmeier, J. Makromol. Chem.
1989, 190, 1717-1726. (c) Wulff, G.; Vieitmeier, J. Makromol. Chem. 1989,
190, 1727-1735.
(27) (a) Shea, K. J.; Thompson, E. A. J. Org. Chem. 1978, 43, 4253-
4255. (b) Shea, K. J.; Thompson, E. A.; Pandey, S. D.; Beauchamp, P. S.
J. Am. Chem. Soc. 1980, 102, 3149-3151. (c) Damen, J.; Neckers, D. C.
Tetrahedron Lett. 1980, 21, 1913-1916. (d) Damen, J.; Neckers, D. C. J.
Am. Chem. Soc. 1980, 102, 3265-3267.
(28) Hoberman, J. M.; Yesalis, C. E. Sci. Am. 1995, 272, 76-81.
(29) The noncovalent strategy has been used successfully for imprinting
steroids (see refs 7 and 8). However, it was not suitable for our investigation
owing to the necessity of working with an excess of functional monomer,
which would inevitably lead to the incorporation of some functional groups
“outside” the imprints, thus complicating any investigation of cooperative
interactions in the recognition sites.
(19) (a) Aherne, A.; Alexander, C.; Perez, N.; Payne, M. J.; Vulfson, E.
N. J. Am. Chem. Soc. 1996, 118, 8771-8772. (b) Alexander, C.; Vulfson,
E. N. AdV. Mater. 1997, 9, 751-755.
(20) Shea, K. J.; Sasaki, D. Y.; Stoddard, G. J. Macromolecules 1989,
22, 1722-1730.
(21) Wulff, G.; Kemmerer, R.; Vietmeier, J.; Poll, H.-G. NouV. J. Chim.
1982, 6, 681-687.
(22) (a) Wulff, G.; Vesper, W.; Grobe-Einsler, R.; Sarhan, A. Makromol.
Chem. 1977, 178, 2799-2816. (b) Wulff, G.; Grobe-Einsler, R.; Vesper,
W.; Sarhan, A. Makromol. Chem. 1977, 178, 2817-2825. (c) Wulff, G.;
Vesper, W. J. Chromatogr. 1978, 167, 171-186. (d) Wulff, G.; Oberko-
busch, D.; Minarik, M. React. Polym. Ion Exch. Sorbents 1985, 3, 261-
275.
(23) (a) Wulff, G.; Sarhan, A.; Zabrocki, K. Tetrahedron Lett. 1973,
4329-4332. (b) Wulff, G.; Schulze, I.; Zabrocki, K.; Vesper, W. Makromol.
Chem. 1980, 181, 531-544.

6642 J. Am. Chem. Soc., Vol. 121, No. 28, 1999
Alexander et al.
Figure 1. Synthesis of polymerizable sterol-boronophthalide complexes.
of labile boronate esters; this was introduced and extensively
studied with glycoside templates by Wulff and co-workers.21,22,30
The other is the sacrificial spacer approach developed in our
laboratory.31 Although both allow the introduction of functional
groups exclusively into the polymer’s recognition sites, the
former methodology was selected as much stronger binding of
ligands, in absolute terms, was expected in this case (more
energy is required for breaking the boronate ester bond
compared to a single hydrogen bond). This was perceived to
be an advantage in the context of this investigation over
noncovalent and sacrificial spacer methods because relatively
little “leakage” of polymer-bound ligands was expected during
chemical modification experiments. Boronophthalide (I) was
employed as a functional monomer rather than the more
commonly used 4-vinylphenyl boronic acid in order to ensure
that each polymerizable moiety could bind to only a single
hydroxyl group. This monomer was synthesized in two steps
from 5-nitroboronophthalide32 (Figure 1), using a variant of the
method of Dederichs.33 Initial experiments were performed on
a set of polymers imprinted with androst-5-ene derivatives 1-3
bearing either a single hydroxyl group at the 3â or 17â positions
(1 and 2, respectively) or hydroxy substituents at both positions
(3), as these compounds were easily accessible from dehydro-
isoandrosterone. Boronate esters of each sterol with I were
prepared and characterized by NMR spectroscopy;34 all were
highly moisture-sensitive solids and thus were prepared and
polymerized in situ.35
The polymers P1, P2, and P3 (Table 1) were prepared by
thermal polymerization of boronate-sterol esters (1-I, 2-I, and
3-I, respectively) at template loadings of 2.5 mol % with
divinylbenzene (DVB) as the cross-linker in chloroform. “Non-
imprinted” polymers were synthesized using the mono- and bis-
boronophthalide esters of methanol and ethyleneglycol (poly-
mers PNI-1 and PNI-2) rather than free I to overcome the low
solubility of the latter in the polymerization mixture. After
polymerization, the materials were ground to a fine powder,36
and the template was removed by Soxhlet extraction with
aqueous ethanol. The polymers were dried and shaken with
sterol solutions in the presence of CaH2, used as a mild drying
agent,37 to assess their binding properties.
(34) A preliminary 1H NMR assignment of these complexes has been
published (Smith, C. R.; Whitcombe, M. J.; Vulfson, E. N. In Separations
for Biotechnology 3; Pyle, D. L., Ed.: Royal Society of Chemistry:
Cambridge, UK, 1994; pp 482-488).
(30) Wulff, G.; Haarer, J. Makromol. Chem. 1991, 192, 1329.
(31) Whitcombe, M. J.; Rodriguez, M. E.; Villar, P.; Vulfson, E. N. J.
Am. Chem. Soc. 1995, 117, 7105-7111.
(35) 1H NMR spectra of the polymerization mixture in CDCl3, monitoring
the 3R and 17R protons at 4.2 and 4.3 ppm, respectively, indicated that the
complexes remained intact up to the point of gelation, when signal
broadening rendered spectra difficult to interpret. IR spectra of the polymers
prior to washing did not display free hydroxyl bands, further suggesting
that the boronophthalide esters were not decomposed under the polymer-
ization conditions.
(32) Lennarz, W. J.; Snyder, H. R. J. Am. Chem. Soc. 1960, 82, 2172-
2175.
(33) This compound was first reported in a Ph.D. Thesis (Dederichs,
W. Ph.D. Dissertation, University of Dusseldorf, Germany, 1983), and its
potential as a binding group for monoalcohols was alluded to in reviews
(see: Wulff, G. Pure Appl. Chem. 1982, 11, 2093-2102. Wulff, G.;
Dederichs, W.; Grotstollen, R.; Jupe, C. In Affinity Chromatography and
Related Techniques; Gibnau, T. C. J., Visser, J., Nivard, R. J. F., Eds.;
Elsevier: Amsterdam, 1982; pp 207-216).
(36) Ground polymers were not subject to any particle sizing. In view
of the slow kinetics of binding and the fact that polymers were used at
equilibrium, it was not considered necessary.

Imprinted Polymers as Protecting Groups
J. Am. Chem. Soc., Vol. 121, No. 28, 1999 6643
Table 1. Composition and Characterization Data for DVB-Based Polymersa
polymer
template
yield, %b
BET surface area, m2/g
template recovery, %
boronophthalide groups,d µmol/g
P1
P2
P3
P4
P5
PNI-1
PNI-2
1
2
3
7
8
93
94
91
89
94
93
85
27
35
16
212
240
5
92
95
90
91
90
ndc
ndc
168
173
330
329
328
182e
362e
MeOH
HOCH2CH2OH
111
a Polymers were prepared with 55% technical grade divinylbenzene at 2.5 mol % template loading and polymerized thermally at 60 °C (see
Experimental Section for details). b Yield of polymer after drying, before template removal. c Not determined. d Calculated from template recovery.
e Calculated assuming all template was removed.
Extensive binding studies with all the combinations of
templates and polymers yielded several conclusions. First,
polymers imprinted with androst-5-ene-3â-ol (1) and androst-
5-ene-17â-ol (2), when challenged with these two ligands,
showed little discrimination (less than 10%) in binding. Similar
results were obtained with polymers synthesized with different
cross-linking monomers38 (EDGMA and trimethylolpropane-
trimethacrylate (TMPTMA)) and at template loadings from 2
to 5 mol %. Second, less than 2% binding of sterols was detected
with control polymers prepared in DVB and the methacrylates
(EGDMA and TMPTMA) in the absence of boronophthalide.
This experiment proved that the binding observed was due to
interaction between the boronophthalide moiety and the hy-
droxyl groups of the ligand. A further comparison between
polymers imprinted with boronophthalide esters of 1, 2, and a
short-chain primary alcohol (PNI-1) was carried out. Imprinted
polymers (P1 and P2) showed 14% and 12% higher uptake,
respectively, of the sterols compared to PNI-1, suggesting that,
at least in this case, the binding of sterols by only one point
was insufficient to engender significant specificity.
We then turned to the polymer imprinted with androst-5-ene-
3â,17â-diol (P3) and investigated the binding of template 3 as
well as that of ligands 1 and 2. In accordance with earlier
studies,39 the introduction of the second functional group in the
recognition site of the imprinted polymer had a very pronounced
effect on the selectivity of binding. It is evident from Figure 2a
that the uptake of 3 was much better than that of 1 and 2 (85%
uptake compared to 40-50%), clearly suggesting a high degree
of cooperativity between the boronophthalide residues in the
site. The same pattern of binding was observed in chloroform
and two other solvents, THF and ethyl acetate.40 Furthermore,
Figure 2. Uptake of sterols by imprinted polymers. (a) Uptake of
androst-5-ene-3â,17â-diol (O), androst-5-ene-3â-ol (0), and androst-
5-ene-17â-ol (4) by androst-5-ene-3â,17â-diol-imprinted polymer in
CHCl3. (b) Uptake of androst-5-ene-3â,17â-diol by androst-5-ene-3â,-
17â-diol-imprinted (b), androst-5-ene-3â-ol-imprinted (9), androst-5-
ene-17â-ol-imprinted (2), and ethyleneglycol-imprinted (1) polymers
in CHCl3.
the uptake of the diol 3 to polymers imprinted with 1 and 2
and to a polymer “imprinted” with ethyleneglycol (PNI-2) was
far inferior (up to 50% reduction in binding; see Figure 2b),
which is significant as the ethyleneglycol-imprinted polymer
(37) In the absence of CaH2, uptake of templates was exceedingly slow.
No binding of sterol to drying agent alone was observed.
was prepared with exactly the same template loading as P3 and
differed only in the spatial distribution of boronophthalide
functionality in the matrix. Data shown are for DVB polymers
P3 and PNI-2, P1, P2, and P3 at 2.5 mol % template loading,
but similar results were also obtained at template loadings of 5
mol % and in EGDMA and TRIM polymers.41
(38) Previous reports (Wulff, G.; Vietmeier, J.; Poll, H. G. Makromol.
Chem. 1987, 188, 731-740) have suggested that EGDMA-based imprinted
polymers were superior in performance to those made with DVB, due
possibly to the greater flexibility of the polymer chains, thus allowing facile
removal and rebinding of templates. However, in this work, no problems
were encountered in template removal with DVB polymers, which was
essentially quantitative in all cases, and in uptake experiments these polymers
performed as well as those prepared with EGDMA.
Inevitably, some binding of 3 to nonimprinted boronophtha-
lide-containing polymers PNI-1 and PNI-2 was observed in all
the experiments because, regardless of the distribution of I, the
formation of ester bonds between steroid hydroxyls and bo-
ronophthalide residues should have occurred whenever the
groups could come into close enough proximity to react. It
appeared, therefore, that a cooperative interaction between at
least two functional groups in the polymer sites was required
for specific binding, with the “pocket” itself playing a lesser
(39) The improvement in selectivity engendered by the introduction of
a second binding group in the imprinted site has been well-documented by
Wulff, (see ref 1d, pp 1819 and references therein).
(40) The best uptake of templates was shown in the solvent used for
polymerization, chloroform, which accords well with previous observations
(Spivak, D.; Gilmore, M. A.; Shea, K. J. J. Am. Chem. Soc. 1997, 119,
4388-4393), suggesting that the imprinted site returns to its original
conformation only if swollen by the correct solvent. Nevertheless, some
binding of templates was observed in tetrahydrofuran and ethyl acetate,
with the pattern of sterol binding being rather similar to that seen in
chloroform.

6644 J. Am. Chem. Soc., Vol. 121, No. 28, 1999
Alexander et al.
similarities to the results of Shea and Sasaki, who observed low
reaction rates for bis-ketalization in binding sites even of
relatively small diketone templates.43 However, attempts to
accelerate boronate ester formation by adding electron donors
such as pyridine and methylpiperidine failed to effect any
improvement. Thus, although the initial binding of the sterol to
the polymer was rapid, covalent template-polymer bonds were
formed extremely slowly.44
To provide further evidence for slow formation of covalent
bonds in the polymer binding sites, a series of “washing”
experiments was carried out. We reasoned that, if the sterols
were not bound in the sites covalently, it would be possible to
remove them by washing the polymer with an appropriate
anhydrous aprotic solvent. Subsequent washes with a hydroxylic
solvent such as methanol would then release the covalently
bound fraction of the ligand, thus allowing quantitative com-
parison. Indeed, it was found that only 50% of androst-5-ene-
3â,17â-diol (3) was bound covalently after 1 h reflux in
chloroform (>90% overall binding), and this increased to 80%
after 36 h. Qualitatively similar results (30% covalently bound
after 1 h, 70% after 36 h) were obtained with mono-alcohols 1
and 2, although the proportion of covalently bound ligand was
somewhat lower. This was to be expected because the formation
of just one ester bond in the case of 3 would be sufficient to
prevent the release of this ligand by an aprotic solvent wash.
To verify further the formation of covalent bond(s) at the
recognition site, chemical modification of the polymer-bound
sterol was carried out. Clearly, no hydroxyl groups would be
free to react if cooperative covalent binding of the androst-5-
ene-3â,17â-diol (3) in the recognition site of the polymer
occurred. In contrast, binding of diol 3 to a mono-alcohol
-imprinted polymer should result in boronate ester formation
at one end of the molecule, with the remaining hydroxyl
accessible to derivatization. Accordingly, 3 was bound to
polymers P1, P2, and P3 (imprinted with androst-5-ene-3â-ol,
androst-5-ene-17â-ol, and androst-5-ene-3â,17â-diol, respec-
tively), with stoichiometries chosen such that the excess of
steroid hydroxyl relative to boronophthalide was the same in
each case. The sterol was reacted with the polymer for at least
72 h under reflux to ensure sufficient time for the formation of
covalent bonds (sterol concentration in solution and hydroxyl
IR absorption on the polymers were monitored until no further
change occurred), and exhaustive extraction with an aprotic
solvent was carried out to remove unbound 3. The polymer-
sterol complex was then refluxed with a large excess of acetic
anhydride and pyridine.45 After a second extraction stage to
remove excess reagents, the polymer was washed with THF/
Figure 3. Infrared spectra of androst-5-ene-3â,17â-diol (3) and
polymer P3 (a) before reflux, (b) after 1 h reflux, (c) after 24 h reflux,
and (d) after 36 h reflux
role in recognition. It should also be stressed that it was possible
to achieve virtually quantitative binding (over 90% mol equiv
relative to boronophthalide) of the sterols to imprinted polymers
by carrying out the experiments under more vigorous dehydrat-
ing conditions, to favor a shift of the equilibrium toward
esterified products. However, the results of these latter experi-
ments showed that the binding of the sterols to the polymers
was surprisingly slow, compared to the rate of the same
boronophthalide-sterol reactions in solution. Was this because
of hindered access to recognition sites within the matrix or due
to some unanticipated chemical reason? To shed some light on
the actual mechanism, the binding in DVB-based polymers was
studied by FT-IR (Figure 3), following hydroxyl absorptions
between 3400 and 3650 cm-1
.
Examination of diagnostic OH stretch absorptions at 3644
and 3450 cm-1 (due to the steroid and boronophthalide hydroxyl
moieties) for imprinted polymers loaded with ligands after 1
and 36 h under reflux (>90% uptake in both cases) revealed a
significant difference between the two samples: a large propor-
tion of hydroxyl groups still remained free after 1 h, despite
the presence of the ligand in the active site. Even after prolonged
reflux (36 h), with azeotropic removal of water via CaH2 in a
Soxhlet thimble suspended above the solution, complete sup-
pression of the hydroxyl absorption was not achieved, in contrast
to solution-state reactions between boronophthalides and the
respective sterols, which were complete as indicated by 1H NMR
within 30 min at reflux.42 This behavior, in terms of rebinding
at both points in a bifunctional imprint site, has certain
(42) The formation of the boronophthalide-sterol esters was followed
by IR and NMR: representative spectra for starting materials and the product
of reaction between 3 and I are appended as Supporting Information.
(43) Shea and Sasaki, in their detailed study (ref 25) of acetal-forming
reaction between benzylic alcohols and diketones, also observed that, despite
extensive one-point binding taking place after 24 h in the imprinting sites,
even after a further 24 h reaction time only 60% of the difunctional template
was bound covalently at both positions. The authors suggested that this
was largely due to a rate-limiting segmented chain motion of the polymer
backbone supporting the binding groups: this reasoning should apply in
our case, too.
(44) The presence of a residual broadened OH stretching band at 3550
cm-1, even after 36 h reflux, may indicate that a number of sterol hydroxyl
groups were attached to the polymer via hydrogen bonds rather than as
esters: however, the low intensity of this band suggests that this number
was small.
(41) The increase in levels of template loading from 2.5 and 5 mol %
for DVB polymers (and for 2-5 mol % imprinted polymers prepared from
EGDMA and TRIM) did not lead to a difference in the selectivity of ligand
binding, as might be anticipated if the sites became less isolated. Clearly,
if the sites were not sufficiently isolated, the difunctional template 3 might
“bridge” boronophthalide residues from adjacent sites, and the level of
uptake would have decreased relative to that of 1 and 2. In addition, if the
sites were “poorly defined”, reaction of excess monoalcohol 1 and diol 3,
respectively, with androst-5-ene-3â,17â-diol-imprinted polymer should lead
to substantially higher level of binding of 1 as compared to 3 (i.e., two
one-point bindings of the mono-alcohol as opposed to one two-point binding
of the diol). In fact, only 1 mol of androst-5-ene-3â-ol bound per bifunctional
site, suggesting that binding of androst-5-ene-3â-ol in the cavity rendered
the second residue of boronophthalide completely unreactive, presumably
due to steric shielding: once one molecule was bound, a second could not
enter the site.
(45) Under the reactions conditions used (300-fold excess of acetic
anhydride in pyridine/CHCl3 at reflux), both the 3â and 17â hydroxyl groups
were rapidly acylated in solution in the absence of polymer. The same
acylation procedure carried out on polymer P3 prior to hydrolysis (i.e.,
with androst-5-ene-3â,17â-diol remaining bound to the polymer) yielded
only the unmodified diol in the protic wash fraction.

Imprinted Polymers as Protecting Groups
J. Am. Chem. Soc., Vol. 121, No. 28, 1999 6645
Table 2. Modification of Androst-5-ene-3â,17â-diol (3) on
strongly hindered by the angular methyl groups at C10 and C13
and was further shielded by the polymer backbone. However,
the ratio of 11â-acetoxy to 3â-acetoxy products was at least
1:1 by integration, and no evidence for modification at the 17â
position was obtained, whereas in solution the 3â,17â-diacetate
was readily formed, while large excesses of reagents and long
reaction times were required in order to achieve complete
acylation of all three hydroxyl groups on 4.51 It was concluded
that the low reactivity of the target group, combined perhaps
with poor accessibility of imprinted sites to relatively bulky
acetic anhydride/pyridine complexes, was responsible for the
poor yield. As 4 was the only trihydroxyandrost-5-ene available
and the quantities were insufficient for a more detailed
investigation,52 we decided to adopt another sterol-based system,
with the particular aim of addressing the issue of reagent
accessibility.
Owing to the difficulties inherent in modification of many
sterols (and their high costs), we employed more readily
available and functionally amenable hydroxy steroids in further
investigations. The isomeric bile compounds deoxycholic acid
(5) and chenodeoxycholic acid (6) were chosen for their
accessible chemistry and availability in gram quantities. To
imprint these sterols, the carboxylic acid groups were first
converted to tert-butyl esters (7 and 8) and (via the triols 9 and
10) the trityl ethers (11 and 12), to prevent the formation of
mixed boronic anhydrides with I during the imprinting stage
and, at the same time, to create an extra pocket to facilitate
access by acylating reagents (Figure 5).
Polymers were prepared with both tert-butyldeoxycholate-
and tert-butylchenodeoxycholate-bis(boronophthalide) esters53
in chloroform at template loadings of 2.5 and 5 mol %. The
cross-linkers used were DVB (both 80% and 55% tech grades),
EGDMA, and TMPTMA. Almost quantitative (>90%) template
removal was achieved for tert-butyl esters from DVB 55
polymers, but the recovery fell to 70-80% for polymers
prepared at higher cross-linking density, as well as for those
imprinted with bulkier trityl ether derivatives. In the experiments
described below, DVB 55 materials polymerized at 2.5 mol %
template loading were used (polymers P4 and P5, Table 1),
although qualitatively the same results were obtained for
modifications carried out on EGDMA- and TMPTMA-based
polymers.
For the modification experiments, isomeric trihydroxysterols
were prepared by reduction of the acid functionality of deoxy-
cholic acid and chenodeoxycholic acid using LiAlH4 in THF,
which generated the corresponding deoxycholan-24-ol (9) and
chenodeoxycholan-24-ol (10). The triols were refluxed with
imprinted polymers in chloroform under dehydrating conditions
as carried out for the androstene templates. After removal of
unbound template by a chloroform wash,54 the attached sterols
were reacted with excess acetic anhydride and pyridine at
reflux55 and washed successively with aprotic and hydroxylic
solvents as before. Two sets of reactions were carried out:
modifications of deoxycholan-24-ol (9) or chenodeoxycholan-
24-ol (10) attached to tert-butyldeoxycholate-imprinted polymer
Imprinted Polymers
yielda of 3 yielda of mono-acetates
ratio of
polymer
(µmol)
(µmol)
diol:mono-acetateb
P1
P2
P3
9.0
5.2
20.7
3.2
4.4
0.6
2.8:1
1.2:1
34.5:1
a Yields are for products bound to polymers and recovered from the
protic wash fraction. Results are averaged from at least duplicate
measurements (typical variation (0.2 µmol). b Determined by GC
analysis..
methanol/water, and the products were recovered by conven-
tional workup. Although the major component was unchanged
starting material 3 in all cases (Table 2), up to 40% mono-
acetates46 were observed in the products from 3 bound to
polymers P1 and P2. By contrast, less than 3% mono-acetate
was recovered from the reaction products of diol 3 bound to
polymer P3. This strongly supported the conclusion that most
of the diol was covalently attached at both “ends” in the diol-
imprinted polymer.47 The two mono-acetates (identified by TLC
and GC from authentic chemically prepared standards) were
formed in similar quantities, indicating that discrimination
between ends of the molecule was poor. The same pattern of
products was observed independent of the solvent used for sterol
loading or modification and for template loadings of 2.5-5 mol
%.48
Having established that the two boronophthalide residues were
acting cooperatively in the binding site of polymer P3 and that
covalent boronate ester bonds with the 3â and 17â hydroxyls
of 3 were re-formed (albeit after long reaction times and drastic
conditions), we proceeded to investigate the feasibility of using
imprinted polymers as “protecting groups” for regioselective
chemical modification.49 To this end, the polymer imprinted with
androst-5-ene-3â,17â-diol (P3) was reacted with the trihydroxy-
steroid androst-5-ene-3â,11â,17â-triol (4), and modification with
acetic anhydride (Figure 4) was carried out as described above.
Although the majority of product (90%) recovered in the protic
solvent wash under these conditions was unmodified triol, NMR
analysis suggested the formation of a small amount (5-10%
by integration) of the 11-acetoxy derivative, as judged by a peak
at 5.25 ppm in the 1H NMR spectrum, assigned to the geminal
11R proton.50 The low degree of modification of the 11â-
hydroxyl group was perhaps not too surprising, given that it is
(46) Although the two isomeric mono-acetates were resolvable by TLC,
GC analysis under a range of conditions failed to effect baseline separation.
(47) As a control, monomer-template complexes 1-I, 2-I, and 3-I were
prepared in CDCl3 and subjected to the same acetylating conditions as 3
1
bound to polymers P1, P2, and P3. H NMR (monitoring the 3R and 17R
protons at 4.2 and 4.3 ppm, respectively) indicated that the complexes did
not break down under these conditions, and no acetylation of the steroid
hydroxyl groups took place while esterified with boronophthalide.
(48) This provided further evidence for site isolation: reaction of androst-
5-ene-3â,17â-diol (added in excess) with P3 (250 mg, theoretical borono-
phthalide content 366 µmol/g) led to the binding of 46 ( 2 µmol (188 (
8 µmol/g) of sterol, corresponding to 100-104% occupation of available
binding sites (based on theoretical content and template removal, respec-
tively) or 0.51-0.52 mol of sterol/mol of available boronophthalide group
rather than the theoretical value of 0.5. If this was a “coincidence” with
the ligand forming “random” covalent bonds in bridging sites, acylation of
the polymer bound diol should have yielded a significant amount of
monoacetates, rather than the 3% observed.
(51) Conditions for acylation were 300-fold excess of acetic anhydride
in pyridine/CHCl3 at reflux for 4 h.
(52) Androst-5-ene-3â,11â,17â-triol was formerly obtainable in milligram
quantities from Sigma, but production was discontinued before the end of
the project.
(53) Both compounds were prepared and polymerized in situ to prevent
hydrolysis. 1H NMR spectroscopy, following proton resonances at C3â (4.5
ppm) and C7â (4.8 ppm) or C12â (4.7 ppm), verified the complete formation
of the complexes.
(54) The removal of adsorbed template in this way ensured that the
product distribution reflected only modifications of templates bound at or
in the sites.
(49) For a discussion of polymeric supports used as protecting groups,
see: Hodge, P. In Syntheses and Separations Using Functional Polymers;
Sherrington, D. C., Hodge, P., Eds.; John Wiley & Sons: Chichester, 1988;
pp 43-122.
(50) No previous NMR data were available for this compound; however,
the 11R-proton in a structurally similar 11-acetoxyandrost-5-ene sterol was
reported with a chemical shift of 5.2-5.3 ppm (Moon, S.; Stuhmiller, L.
M.; Chadha, R. K.; McMorris, T. C. Tetrahedron 1990, 46, 2287-2306).

6646 J. Am. Chem. Soc., Vol. 121, No. 28, 1999
Alexander et al.
Figure 4. Modification of androst-5-ene-3â,11â,17â-triol (4) on polymer P3.
i.e., the 24-acetoxysterol. Similarly, one should expect the
formation of a mixture of monoesters57 in the unmatched sites.
In accordance with this prediction, we found (Table 3) that
the major component (up to 95% of acetylated products; >50%
of total) recovered by the protic solvent wash from the matched
sites for both trihydroxysterols was the 24-acetoxy derivative
(any 24-acetoxy derivatives formed in solution or not bound
covalently to the polymer would have been removed in the
aprotic solvent wash, and therefore should have been discarded
prior to analysis). This confirmed that most of the ligand was,
indeed, bound in the “correct” orientation. However, a signifi-
cant amount of unreacted material (40-45%) was also recov-
ered, which indicated poor access of acylpyridinium interme-
diates once a molecule of sterol was in place. The use of bulkier
trityl ethers as templates or alternative methacrylate cross-linkers
did not lead to noticeable improvements in yields. The fact that
acetylation took place predominantly at C24 in the matched
cases suggested that an efficient “molecular sorting” must have
taken place in the imprinted sites during template loading,
because one might expect the 24-hydroxyl moiety to be the first
to react with boronophthalide residues, as it is a primary
hydroxyl on a flexible chain. This would lead to a kinetic
preference for the “wrong” orientation. However, the correct
mode of binding should be thermodynamically more stable due
to the formation of two ester bonds as opposed to one, and the
regioselectivity of modification provided evidence for this
manner of binding. The relative proportions of 3-acetoxy
products were also indicative of a better fit in the “matched”
sites. For both sets of templates, the ratio of 24-acetoxy to
3-acetoxy products was much lower (4:1 rather than 20:1) in
the “unmatched” sterol-polymer complexes, indicating that,
where the template could not fit exactly, it bound via the most
reactive, i.e., the 24-hydroxyl group, leaving secondary hy-
droxyls accessible for modification. As a further test of this
hypothesis, acetylation of deoxycholan-24-ol (9) bound to the
ethyleneglycol-imprinted polymer PNI-2 was carried out. In this
Figure 5. Structures of bile salt derivatives.
P4, and chenodeoxycholan-24-ol (10) or deoxycholan-24-ol (9)
bound to tert-butylchenodeoxycholate-imprinted polymer P5.
In this way, the polymer-sterol complexes were classed as
“matched” or “unmatched”, dependent on whether the template
was of the correct theoretical fit to the polymer site, as
illustrated in Figure 6 for tert-butyldeoxycholate-imprinted
polymer P4.
It is evident from Figure 6 that binding of a triol in the
“correct” mode in the matched sites should lead to the formation
of two ester bonds between the boronophthalide residues and
hydroxyl groups at C3 and C12 (or C3 and C7 for the case of
polymer P5). This would leave the primary 24-hydroxyl group
available for modification. However, binding of the sterol to
the unmatched site can only result in the formation of a single
ester bond, most likely at the more reactive 3- or 24-positions.56
Thus, if the binding in the matched sites was 100% correct,
modification with acetic anhydride would give a single product,
(55) Under the reactions conditions used (300-fold excess of acetic
anhydride in pyridine/CHCl3 at reflux), in solution in the absence of polymer
all the steroid hydroxyl groups were acylated. Thus, any difference in the
pattern of acetate products was due to the effect of the polymer imprint
site, rather than that of the intrinsic reactivity of the sterols.
(56) The differences in the reactivity of the various hydroxyl groups on
the cholic acid framework have been considered thoroughly elsewhere
(Baker, J. F.; Blickenstaff, R. T. J. Org. Chem. 1975, 40, 1579) (and by
Bonar-Law et al. in ref 57); however, in this case, binding of the sterol to
the polymer was more likely to occur via the reactive primary hydroxyl at
C24 rather than the secondary hydroxyl groups at C3, C7, and C12.
(57) The formation of diesters might also be expected for “unmatched”
template-polymer complexes, but none were observed, due possibly to the
difficulties in fitting two acetate groups into the sites.

Imprinted Polymers as Protecting Groups
J. Am. Chem. Soc., Vol. 121, No. 28, 1999 6647
Figure 6. Modification of triols 9 and 10 on polymer P4.
Table 3. Modification of Sterols 9 and 10 on Imprinted Polymers
polymera
sterol
sterol bound, µmolb
fraction unmodified, %
fraction mono-acetates, %
ratio 24-acetoxy:3-acetoxy
P4 (matched)
P5 (matched)
P5 (unmatched)
PNI-2
9
10
9
29.0
26.1
25.5
8.3
36
35
49
87
64
65
51
13
10.5:1
23.1:1
5.4:1
9
<1:100
a
“Matched” refers to deoxycholan-24-ol (9) bound to tert-butyldeoxycholate-imprinted polymer or chenodeoxycholan-24-ol (10) bound to tert-
butylchenodeoxycholate-imprinted polymer. “Unmatched” refers to deoxycholan-24-ol bound to tert-butylchenodeoxycholate-imprinted polymer.
b Yield quoted is for the total amount of sterol and acetylated derivatives recovered from the polymer after the protic solvent wash. Results are
averaged from at least duplicate measurements (typical variation in sterol bound, (0.15 µmol).
case, almost no 24-acetoxy product was formed, with the
predominant component being unreacted 9 and the remainder
(∼13%) shown by GC to be the 3-acetoxydeoxycholan-24-ol.
This suggested a more random binding process, with attachment
to the polymer occurring primarily via the 24-hydroxyl
position. Virtually no modification at the 7- or 12-positions
was observed58 in any of the above cases, probably due to
much lower reactivity of these hydroxyls56 compared to C3 and
C24 and/or hindered access to the acetylating reagents in the
sites.
Conclusions
We have shown that the incorporation of two boronophthalide
residues into the polymer’s recognition site results in cooperative
interactions with target dihydroxytemplates. This is evident from
the higher uptake of the diol 3 compared to that of monohy-
droxyandrost-5-enes (1 and 2) by P3 as well as from the
experiments on the comparative binding of 3 to other polymers
(P1, P2, and PNI-2). This conclusion is also supported by the
results of the acylation experiments, where low amounts (<3%)
of monoester products were observed in the modification of 3
loaded on P3 compared to P1 and P2. These data, combined
with other observations (FT-IR and washing the loaded polymers
with THF/MeOH/H2O), strongly suggest that both (rather than
(58) Characterization of reaction products was based on NMR assign-
ments. See: (a) Bonar-Law, R. P.; Davis, A. P. Tetrahedron 1993, 49,
9845-9854. (b) Bonar-Law, R. P.; Davis, A. P.; Dorgan, B. J. Tetrahedron
1993, 49, 9855-9866 and references therein.

6648 J. Am. Chem. Soc., Vol. 121, No. 28, 1999
Alexander et al.
min up to 275 °C. For analysis of cholic acid derivatives, the column
was held at 275 °C for a further 10 min. Thin-layer chromatography
(TLC) was used for the qualitative analysis of sterol derivatives, using
silica gel plates and hexane/ethyl acetate/methanol (55:40:5 v/v) as the
eluent. TLC plates were developed by spraying with phosphomolybdic
acid and heating at 100 °C for 1 min. Flash chromatography was
performed using Silica gel 60 (230-400 mesh) from Aldrich. All
standard reagents were purchased from Aldrich or BDH and used as
received. Solvents used for chromatography were purchased from Fisher
Scientific and were at least HPLC grade. Anhydrous solvents were
prepared by standard methods.61
one) covalent bonds were re-formed in the majority of recogni-
tion sites on binding. The same conclusion can be drawn from
the results of acetylation of polymer-bound androst-5-ene-3â,-
11â,17â-triol (4) and (cheno)deoxycholan-24-ol (9 and 10). In
the former case, the formation of the sterically hindered 11-
acetoxy product was predicted and, indeed, observed, while in
the latter experiment the difference in the product distribution
obtained can best be explained by preferential binding of the
matched ligands in the correct orientation to allow the formation
of two boronate ester bonds.
Starting Materials. Dehydroisoandrosterone, androst-5-ene-3â,11â,-
17â-triol, deoxycholic acid, and chenodeoxycholic acid were purchased
from Aldrich. Androst-5-ene-3â-ol (1) was prepared by reduction of
dehydroisoandrosterone with hydrazine hydrate/KOH/diethyleneglycol
and recrystallized from ethyl acetate to constant melting point 133 °C
(lit.62 mp 135 °C). Other spectroscopic data were consistent with the
desired product.63 Androst-5-ene-17â-ol (2) was synthesized in two steps
by tosylation of the 3-hydroxyl group with p-toluene sulfonyl chloride
in CHCl3/pyridine, followed by reduction with LiAlH4 and successive
crystallization from hexane and ethyl acetate: melting point 156 °C
(lit.64 mp 158-165). Androst-5-ene-3â,17â-ol (3) was prepared by
direct reduction of dehydroisoandrosterone with L-Selectride (lithium
tri(sec-butyl)borohydride, Aldrich) in diglyme. Spectroscopic data were
in accordance with literature values.65 3â-Acetoxyandrost-5-ene-17â-
ol was purchased from Sigma and used as received. 17â-Acetoxy-
androst-5-ene-3â-ol was prepared by literature methods66 and recrys-
tallized from hexane/acetone.
5-Aminoboronophthalide. From 5-nitroboronophthalide, prepared
by the method of Lennarz and Snyder,32 the unstable amino derivative
was synthesized by reduction with palladium/carbon and ammonium
formate.67 (See Supporting Information for full details.)
Preparation of (5-Methacryloylamino)boronophthalide (I). A
variation of the procedure of Dederich33 was followed. Spectroscopic
and analytical data were in accordance with the desired structure. (See
Supporting Information for full details.)
Another question which arises from this investigation is the
scope for using imprinted polymers as protecting groups for
regioselective modifications of multifunctional compounds.59
Our results, as well as those of others using imprinted polymer
binding sites as “microreactors”,26,27 suggest that this is feasible,
although several issues must be resolved to make it attractive
from a practical standpoint. First, it seems that access of reagents
to the imprinted sites is hampered by the high density of the
cross-linked polymeric network. The creation of a special pocket
for the reagent may alleviate the problem to a certain extent
but is unlikely to solve it completely. Presumably, using
polymers with significantly lower degrees of cross-linking would
be beneficial and, provided that very strong, preferably covalent,
interactions between the functionality of the target and polymer’s
recognition site are employed, there is no reason this cannot be
achieved. The second, and perhaps more elegant, approach
would be to have imprinted sites positioned exclusively on the
surface of polymeric particles or films, and we are currently
studying this possibility. Among potential applications of this
approach, one would envisage imprinted polymer-directed
regioselective modifications of complex natural products or
drugs such as, for example, taxol60 and the synthesis of
combinatorial libraries. The latter is especially attractive because
the polymer can be used for a number of different modifications
of the same functional group in a target molecule and, in most
cases, very small quantities of compounds are required for the
evaluation of biological activity.
Preparation of tert-Butyl-3r,12r-dihydroxy-5â-cholan-24-oate
(tert-Butyldeoxycholate) (7) and tert-Butyl-3r,7r-dihydroxy-5â-
cholan-24-oate (tert-Butylchenodeoxycholate) (8). The procedure
followed for both 7 and 8 was a modification of that used by Bonar-
Law et al.68 for the preparation of tert-butylcholate. (See Supporting
Information for full details.)
Experimental Section
tert-Butyl-3r,12r-dihydroxy-5â-cholan-24-oate (tert-Butyldeoxy-
cholate) (7). The crude compound yield was 19 g, and it was shown
by NMR to be the desired compound. TLC indicated only one
component, but crystallization from nonpolar solvents could not be
effected. Slow crystallization from water/EtOH yielded the pure
compound: mp 72-76 °C. Overall yield was 18.4 g, (82%). NMR
assignments were based on those by Bonar-Law and Davis for cholic
acid derivatives.58
General Methods. Nuclear magnetic resonance (NMR) spectra were
recorded on a JEOL EX 270 Fourier transform spectrometer at 67.8
(13C) and 270.05 MHz (1H). All chemical shifts (δ) are reported in
parts per million (ppm) relative to tetramethylsilane (TMS). Chemical
shift values (δ) and peak assignments were determined by a combination
of one-dimensional 1H and 13C NMR spectroscopic analysis (including
DEPT 90 and DEPT 135) and two-dimensional H-1H homonuclear
1
1H NMR (δ/ppm, CDCl3): 0.61 (3H, s, 18-Me), 0.84 (3H, s, 19-
Me), 0.89 (3H, d, J ) 4 Hz, 21-Me), 1.37 (9H, s, C-(CH3)3), 2.15 (1H,
m, J ) 13 Hz, CH2CO2-t-Bu), 2.22 (1H, m, J ) 12 Hz, CH2CO2-t-
Bu), 3.52 (1H, m, tt, J ) 11, 4 Hz, 3â-H), 3.92 (1H, t, J ) 4 Hz,
12â-H).
COSY and 1H-13C heteronuclear COSY experiments. IR spectra were
recorded on a Perkin-Elmer 1600 series spectrometer by the diffuse
reflectance method using KBr as dispersant. Negative fast atom
bombardment mass spectrometry (FAB-MS) spectra were obtained on
a Kratos MS9/50TC spectrometer, using Xenon at 5-8 keV. Accurate
mass measurements were recorded at 1.0 mamu resolution, using PEG
600 ions as reference. GC analysis was undertaken on a Hewlett-
Packard 5890 gas chromatograph equipped with FID. Trimethylsilyl
derivatives (1 µL), prepared by adding a sample to a 1:1 mixture of
bis(trimethylsilyl)trifluoroacetamide (BSTFA):pyridine at 50 °C for 15
min, were applied to a BPX-5 capillary column using helium as the
carrier gas. The oven temperature gradient began at 50 °C and increased
at 5 °C/min to 120 °C, followed by 10 °C/min to 250 °C, and 4 °C/
13C NMR (δ/ppm, CDCl3): 12.60 (C18), 17.16 (C21), 23.15 (C19),
28.02 (C-(CH3)3), 71.50 (C3), 72.96 (C12), 79.78 (O-C(CH3)3), 173.60
(C24).
(61) Perrin, D. D.; Armanego, W. F.; Perrin, D. R. Purification of
Laboratory Chemicals, 2nd ed.; Pergamon: Oxford 1980.
(62) Halsall, T. G.; Jones, E. R. H.; Tan, E. L.; Chaudhry, G. R. J. Chem.
Soc. C. 1966, 1374-1383.
(63) Crabb, T. A.; Dawson, P. J.; Williams, R. O. J. Chem. Soc., Perkin
Trans. 1 1980, 2535-2541.
(64) Shoppee, C. W.; Killick, R. W. J. Chem. Soc. C 1970, 1513.
(65) Suginome, H.; Nakayama, Y. J. Chem. Soc., Perkin Trans. 1 1992,
1843-1847.
(59) Bystro¨m et al. (ref 13) successfully demonstrated that regioselective
modifications could be mediated by imprinted polymers: in this case,
reduction of steroidal ketones using LiAlH4 introduced into imprinted sites
after template removal. This is, however, different from carrying out
regioselective modifications with added reagents using imprinted polymers
as “self-assembled” protecting groups.
(66) Johns, W. F.; Salomon, K. W. J. Org. Chem. 1971, 36, 1952-
1960.
(67) Ram, S.; Ehrenkaufer, R. E. Tetrahedron Lett. 1984, 25, 3415-
3417.
(60) Khmelnitsky, Y. L.; Budde, C.; Arnold, J. M.; Usyatinsky, A.; Clark,
D. S.; Dordick, J. S. J. Am. Chem. Soc. 1997, 119, 11554-11555.
(68) Bonar-Law, R. P.; Davis, A. P.; Sanders, J. K. M. J. Chem. Soc.,
Perkin Trans. 1 1990, 2245-2250.

Imprinted Polymers as Protecting Groups
J. Am. Chem. Soc., Vol. 121, No. 28, 1999 6649
IR (νmax/cm-1, KBr): 3388 (O-H stretch), 2934 (C-H stretch), 1728
(CdO stretch), 1450 (CH3 asymm bend).
Mass spectra EI (m/z, relative intensity): 448.4 (M+, 0.1), 391.30
(M - C(CH3)3, 11.3), 373.31 (-OH, 49), 356.30 (-H2O, 88), 273.29
(steroid ring backbone, 100).
Accurate mass: calculated for C28H44O4, 448.3552; found, 448.3599.
tert-Butyl-3r,7r-dihydroxy-5â-cholan-24-oate (tert-Butylcheno-
deoxycholate) (8). TLC (CH2Cl2/EtOAc 1:1) indicated the presence
of two components, the first at Rf 0.3 (shown to be the desired product
by NMR) and a faster moving component (Rf 0.75), which was not
identified. Column chromatography on silica gel (CH2Cl2/EtOAc)
followed by evaporation of solvent yielded the tert-butylchenodeoxy-
cholate as a colorless foam, which was crystallized from water/MeOH
and again from hexane: mp 76-80 °C. Overall yield after column
chromatography was 13.9 g (62%).
1H NMR (δ/ppm, CDCl3): 0.56 (3H, s, 18-Me), 0.82 (3H, s, 19-
Me), 0.86 (3H, d, J ) 6 Hz, 21-Me), 2.93 (2H, m, CH2-O-CPh3,), 3.51
(1H, m, tt, J ) 11, 4 Hz, 3â-H), 3.90 (1H, t, J ) 4 Hz, 12â-H), 7.15,
(10H, m, J ) 5, 5 Hz, ArCH), 7.34 (5H, m, J ) 5, 5 Hz, ArCH).
13C NMR (δ/ppm, CDCl3): 12.65 (C18), 17.56 (C21), 23.08 (C19),
63.99 (C24), 71.52 (C3), 73.70 (C12), 86.14 (O-CPh3), 126.38, 127.57,
128.57 (Ar-CH), 144.40 (Ar-C).
IR: (νmax/cm-1, KBr): 3565, 3401 (O-H stretch), 3048 (ArC-H
stretch), 2920 (C-H stretch), 1595 (ArCdC stretch).
Mass spectra EI (m/z, relative intensity): 620.4 (M+, 0.1), 602.5
(M - H2O, 0.2), 543.5 (-Ph, 2.1), 243.2 (CPh3), 100).
Accurate mass: calculated for C43H56O3, 620.4229; found, 620.4231.
3r,7r-Dihydroxy-5â-cholan-24-oxytriphenylmethane (12). The
yield of product was 520 mg (84%), and it was shown (TLC) to consist
of one component. Crystallization from hexane/ethyl acetate yielded
colorless plates: mp 84-86 °C.
1H NMR (δ/ppm, CDCl3): 0.59 (3H, s, 18-Me), 0.83 (3H, s, 19-
Me), 0.86 (3H, d, J ) 4 Hz, 21-Me), 1.37 (9H, s, C-(CH3)3), 2.10 (1H,
m, J ) 13 Hz, CH2CO2-t-Bu), 2.17 (1H, m, J ) 10 Hz, CH2CO2-t-
Bu), 3.40 (1H, m, tt, J ) 11, 4 Hz, 3â-H), 3.78 (1H, q, J ) 4 Hz,
7â-H).
1H NMR (δ/ppm, CDCl3): 0.56 (3H, s, 18-Me), 0.83 (3H, s, 19-
Me), 0.88 (3H, d, J ) 6 Hz, 21-Me), 2.95 (2H, m, CH2-O-CPh3,), 3.44
(1H, m, tt, J ) 11, 4 Hz, 3â-H), 3.79 (1H, q, J ) 4, 4 Hz, 7â-H), 7.18,
(10H, m, J ) 5, 5 Hz, ArCH), 7.32 (5H, m, J ) 5, 5 Hz, ArCH).
13C NMR (δ/ppm, CDCl3): 11.77 (C18), 18.62 (C21), 22.77 (C19),
64.13 (C24), 68.57 (C7), 72.02 (C3), 86.20 (O-CPh3), 126.77, 127.67,
128.68 (Ar-CH), 144.53 (Ar-C).
13C NMR (δ/ppm, CDCl3): 11.72 (C18), 18.24 (C21), 22.73 (C19),
28.07 (C-(CH3)3), 68.43 (C7), 71.92 (C3), 79.86 (O-C(CH3)3), 173.69
(C24).
IR (νmax/cm-1, KBr): 3380 (O-H stretch), 2920 (C-H stretch), 1726
(CdO stretch), 1450 (CH3 asymm bend).
IR (νmax/cm-1, KBr): 3378 (O-H stretch), 3046 (ArC-H stretch),
2930 (C-H stretch), 1596 (ArCdC stretch).
Mass spectra EI (m/z, relative intensity): 448.3615 (M+, 0.2), 392.43
(MH - C(CH3)3), 12.6), 374.42 (-OH, 55), 356.41 (-H2O, 100).
Accurate mass: calculated for C28H44O4, 448.3552; found, 448.3615.
Preparation of 3r,12r,24-Trihydroxy-5â-cholane (Deoxycholan-
24-ol) (9) and 3r,7r,24-Trihydroxy-5â-cholane (Chenodeoxycholan-
24-ol) (10). A solution of deoxycholic acid or chenodeoxycholic acid
(3.92 g, 10 mmol) in dry THF (350 mL) was stirred rapidly as lithium
tetrahydroaluminate (2.5 g, excess) was added carefully in small
portions under nitrogen. The reaction was continued overnight at room
temperature before the suspension was poured carefully onto a mixture
of cooled ethyl acetate (400 mL) and methanol (50 mL). The suspension
was filtered, and the filtrate was shaken with 1 M HCl (3 × 100 mL)
and water (3 × 50 mL). The residual solids were washed with aqueous
methanol (4 × 100 mL) and ethyl acetate (250 mL) to extract any
adsorbed organics. The combined organic layers were then washed with
1 M NaOH (500 mL) and water (2 × 500 mL) before drying over
(MgSO4) and removal of solvent in vacuo.
3r,12r,24-Trihydroxy-5â-cholane (9). The resultant white solid
(3.02 g, 80%) was pure by TLC (hexane/ethyl acetate/methanol 30:
20:5 v/v). Crystallization from hexane/ethyl acetate yielded colorless
plates: mp 134-136 °C (lit.69 mp 122-124 °C). Spectroscopic and
other analytical data were in agreement with literature values.
3r,7r,24-Trihydroxy-5â-cholane (10). The product was shown by
TLC to contain a small amount of starting material and two faster-
moving components. Column chromatography (silica gel, elution with
ethyl acetate) yielded the desired compound (2.87 g, 75%) as colorless
needles:mp 148-152 °C (lit.70 mp 150 °C). Analytical data were in
accordance with the expected structure.
Preparation of 3r,12r-Dihydroxy-5â-cholan-24-oxytriphenyl-
methane (11) and 3r,7r-Dihydroxy-5â-cholan-24-oxytriphenyl-
methane (12). A solution of deoxycholan-24-ol (9) or chenodeoxycholan-
24-ol (10) (378 mg, 1 mmol) in dry pyridine (20 mL) was stirred rapidly
under nitrogen as triphenylmethyl chloride (750 mg, excess) was added
in one portion. The reaction was continued overnight at room
temperature before the solvent was removed in vacuo. The resultant
oily solid was purified by column chromatography (silica; elution with
CHCl3 and CHCl3/MeOH (10:1)).
3r,12r-Dihydroxy-5â-cholan-24-oxytriphenylmethane (11). The
yield of product was 526 mg (85%) and was shown by TLC to consist
of one component. Crystallization from hexane/ethyl acetate yielded
colorless needles: mp 188-189 °C.
Mass spectra EI (m/z, relative intensity): 620.4 (M+, 0.3), 543.2
(-Ph, 4.2), 243.0 (CPh3), 100).
Accurate mass: calculated for C43H56O3, 620.4229; found, 620.4237.
Preparation of Sterol-Boronophthalide-Imprinted Polymers. A
standard method was used for the synthesis of all the imprinted
polymers in this study: details of polymer composition are given in
Table 1.
A mixture of sterol (0.25 mmol) and (5-methacryloylamino)-
boronophthalide (54.25 mg, 0.25 mmol per sterol hydroxyl) in CHCl3
(2 mL) was heated to reflux in a tube fitted with a suspended CaH2
drying thimble. Removal of water by azeotropic distillation through
the drying agent resulted in the production of a clear solution. Dry
divinylbenzene, 55% (1.38 mL ) 1.2675 g, 9.75 mmol), was added
and reflux maintained for a further 60 min before the addition of AIBN
(10 mg). The solution was degassed three times with freeze-thaw
cycles before polymerization at 60 °C for 24 h. Removal of solvent in
vacuo yielded a polymer which was ground to a fine powder in an
agate mortar on a Fritsch Pulverisette “O” grinding mill and extracted
with chloroform, followed by aqueous ethanol in a Soxhlet apparatus
for 12 h. The polymer was dried at 80 °C in vacuo for 24 h prior to
use.
Batch Binding Experiments of Androst-5-ene Derivatives to
Polymers. Solutions of steroid ligand (2 mL, 2mM) in CHCl3 were
shaken with imprinted polymers (40 mg) and CaH2 (10 mg) for 24 h.
Aliquots (100 µL) were taken at intervals and added to BSTFA/pyridine
(1:1 v/v, 100 µL); after 30 min, ethyl acetate (500 µL) was added to
each aliquot for GC analysis. Where filtration was necessary, aliquots
were increased to 150 µL (sample and BSTFA). To correct for
adsorption of template to the dehydrating reagent, control samples were
prepared with template solutions and CaH2 (10 mg) without polymer.
Peak areas were normalized to tetradecane internal standard (5 µL
mL-1). Results shown are the average of at least two measurements.
Binding of Androst-5-ene Derivatives to Polymers: Elevated
Temperature Studies. A solution of template (10-100 mg) in CHCl3
(5.0 mL) was refluxed with imprinted polymer (250 mg) under a dry
nitrogen atmosphere in a tube with a suspended Soxhlet thimble
containing CaH2 as drying agent. Aliquots (100 µL) were taken at
intervals and added to BSTFA/pyridine (1:1 v/v, 100µL), followed by
addition of ethyl acetate (500µL) as above. Peak areas were normalized
to tetradecane internal standard (5 µL mL-1). For solvent wash studies
to probe template binding, after a set time (1-36 h), the polymers were
washed with dry chloroform (3 × 10 mL, fraction 1) and MeOH/THF/
H2O (3 × 10 mL, fraction 2). For analysis, the solvent was removed
from each fraction and the yield of recovered template recorded either
gravimetrically or by GC.
(69) Kihira, K.; Mikami, T.; Ikawa, S.; Okamoto, A.; Yoshii, M.; Miki,
S.; Mosbach, E. H.; Hoshita, T. Steroids 1992, 57, 193-198.
(70) Ahmed, S.; Alauddin, M.; Caddy, B.; Martin-Smith, M.; Sidwell,
W. T. L.; Watson, T. R. Aust. J. Chem. 1971, 24, 521-547.

6650 J. Am. Chem. Soc., Vol. 121, No. 28, 1999
Alexander et al.
Polymer-Directed Synthesis. (a) Modification of Androst-5-ene-3,-
17-diol (3).
13C NMR (δ/ppm, CDCl3): 12.65 (C18), 17.5 (C21), 20.95 (24-
OCOCH3), 21.38 (3-O-COCH3), 23.04 (C19), 64.94 (C24), 73.03 (C12),
74.22 (C3), 170.66 (3-OCOCH3), 171.19 (24-OCOCH3).
IR (νmax/cm-1, KBr): 3550 (O-H stretch), 2931, 2860 (C-H stretch)
1731 (CdO stretch), 1445 (CH3 asymm bend), 1243 (C-O acetate
stretch).
Mass spectra EI (m/z, relative intensity): 462.3361 (M+, 0.4), 444.4
(M - H2O, 5.0), 402.3 (M - CH3CO2H, 27), 384.0 (-H2O, 75), 255.0
(steroid backbone).
3r-Acetoxy-12r,24-dihydroxy-5â-cholane (15). Crystallization
from hexane/ethyl acetate yielded colorless plates: mp 34-42 °C.
1H NMR (δ/ppm, CDCl3): 0.62 (3H, s, 18-Me), 0.85 (3H, s, 19-
Me), 0.92 (3H, d, 21-Me), 1.96 (3H, s, CH3CdO), 3.55 (3H, s overlying
m, C24 CH2-OH), 3.94 (1H, m, 12â H), 4.62 (1H, tt, J ) 11, 5 Hz, 3â
H).
A solution of 3 (100 mg, 0.35 mmol) in CHCl3 (5.0 mL) was refluxed
with imprinted polymer (250 mg) for 72 h under a dry nitrogen
atmosphere using CaH2 as drying agent in a thimble suspended above
the solution. After the template loading was finished, the polymers were
washed with dry CHCl3 (5 × 10 mL) and dried in vacuo. Dry CHCl3
(5.0 mL) was added to the template-polymer complex, and the contents
were brought to reflux under a dry nitrogen atmosphere before acetic
anhydride (2 mL) and pyridine (2.5 mL) were added. After reaction
was complete (12 h, reflux), the polymers were again washed with
CHCl3 (5 × 10 mL, into methanol (10 mL)) and aqueous THF/methanol
(5 × 10 mL), collecting each fraction separately. Following solvent
removal, ethyl acetate (5.0 mL) was added and the product composition
determined by GC as above.
13C NMR (δ/ppm, CDCl3): 12.56 (C18), 17.51 (C21), 21.28 (3-O-
COCH3), 23.02 (C19), 63.53 (C24), 73.02 (C12), 74.09 (C3), 170.0
(3-OCOCH3).
IR (νmax/cm-1, KBr): 3450 (br O-H stretch), 2931, 2860 (C-H
stretch) 1737 (CdO stretch), 1448 (CH3 asymm bend), 1244 (C-O
acetate stretch).
Mass spectra EI (m/z, relative intensity): 420.3258 (M+, 0.1), 402.2
(M - H2O, 2.2), 384.2 (M - 2H2O, 9.1) 360.2 (M - CH3CO2H, 10.9),
342.2 (-H2O, 31.1), 255.2 (steroid backbone).
24-Acetoxy-3r,12r-dihydroxy-5â-cholane (16). Crystallization
from hexane/ethyl acetate yielded colorless plates. A broad melting
range was observed (82-116 °C), indicative of a mesophase.
1H NMR (δ/ppm, CDCl3): 0.61 (3H, s, 18-Me), 0.84 (3H, s, 19-
Me), 0.91 (3H, d, 21-Me), 1.97 (3H, s, CH3CdO), 3.54 (1H, tt, J )
11, 5 Hz, 3â H), 3.96 (3H, overlying m, CH2OAc, 12â H).
13C NMR (δ/ppm, CDCl3): 12.56 (C18), 17.34 (C21), 20.98 (24-
O-COCH3), 23.00 (C19), 64.93 (C24), 71.41 (C12), 72.93 (C3), 171.17
(24-OCOCH3).
(b) Modification of 3r,12r,24-Trihydroxy-5â-cholane (9) and
3r,7r,24-Trihydroxy-5â-cholane (10). A solution of 9 (100 mg, 0.26
mmol) or 10 (100 mg, 0.26 mmol) in CHCl3 (5.0 mL) was refluxed
with imprinted polymer P4 or P5 (250 mg) for 72 h under a dry nitrogen
atmosphere using CaH2 as drying agent in a thimble suspended above
the solution. After the template loading was finished, the polymers were
washed with dry CHCl3 (5 × 10 mL) and dried in vacuo. Dry CHCl3
(5.0 mL) was added to the template-polymer complex, and the contents
were brought to reflux under a dry nitrogen atmosphere before acetic
anhydride (2 mL) and pyridine (2.5 mL) were added. After reaction
was complete (12 h, reflux), the polymers were again washed with
CHCl3 (5 × 10 mL, into methanol (10 mL)) and aqueous THF/methanol
(5 × 10 mL), collecting each fraction separately. Following solvent
removal, ethyl acetate (5.0 mL) was added and the product composition
again determined by GC.
Preparation of Reference 3r,12r,24-Trihydroxy-5â-cholane and
3r,7r,24-Trihydroxy-5â-cholane Acetates. A solution of trihydroxy-
5â-cholane 9 or 10 (378 mg, 1 mmol) and DMAP (4 mg) in dry pyridine
(0.5 mL) and dry CHCl3 (5 mL) was vigorously stirred as acetyl chloride
(240 mg ) 217 µL) was added rapidly. The reaction was continued
overnight before the contents were poured onto ice/HCl. Extraction of
the aqueous layer with ethyl acetate (3 × 50 mL), washing of the
organic layer with aqueous acid (3 × 50 mL), aqueous base (NH4OH,
3 × 50 mL), and water (3 × 50 mL), and drying over MgSO4 yielded,
after removal of solvent, a white solid in each case. Flash chromatog-
raphy (CH2Cl2 . CH2Cl2/EtOAc/MeOH (10:2:1) was carried out for
both sets of sterol derivatives, and the products were identified by NMR.
IR (νmax/cm-1, KBr): 3500 (br O-H stretch), 2930, 2860 (C-H
stretch) 1735 (CdO stretch), 1450 (CH3 asymm bend), 1245 (C-O
acetate stretch).
Mass spectra EI (m/z, relative intensity): 420.3232 (M+, 0.1), 402.4
(M - H2O, 16.0), 384.4 (M - 2H2O, 53.9) 360.4 (M - CH3CO2H,
2.7), 342.4 (-H2O, 18), 273.8, 255.4 (steroid backbone, 83.8, 100).
3r,7r,24-Tris(acetoxy)-5â-cholane (17). Crystallization from aque-
ous ethanol yielded colorless plates: mp 92-98 °C.
1H NMR (δ/ppm, CDCl3): 0.68 (3H, s, 18-Me), 0.82 (3H, d, 21-
Me), 0.86 (3H, s, 19-Me), 1.97 (3H, s, CH3CdO), 1.98 (3H, s, CH3Cd
O), 2.04 (3H, s, CH3CdO), 3.95 (2H, m, CH2OAc), 4.61 (1H, tt, J )
11, 5 Hz, 3â H), 4.82 (1H, q, J ) 4, 4 Hz, 7â H).
3r,12r,24-Tris(acetoxy)-5â-cholane (13). Crystallization from
aqueous ethanol yielded colorless plates: mp 75-78 °C (lit.71 mp 79.5-
80.5).
1H NMR (δ/ppm, CDCl3): 0.66 (3H, s, 18-Me), 0.74 (3H, d, 21-
Me), 0.84 (3H, s, 19-Me), 1.97 (3H, s, CH3CdO), 1.98 (3H, s, CH3Cd
O), 2.04 (3H, s, CH3CdO), 3.95 (2H, m, CH2OAc), 4.62 (1H, tt, J )
11, 5 Hz, 3â H), 5.02 (1H, m, 12â H).
13C NMR (δ/ppm, CDCl3): 11.61 (C18), 18.45 (C21), 20.95 (24-
OCOCH3), 21.42 (7-OCOCH3), 21.52 (3-O-COCH3), 23.00 (C19),
64.96 (C24), 71.16 (C7), 74.08 (C3), 170.39 (7-OCOCH3), 170.57 (3-
OCOCH3), 171.14 (24-OCOCH3).
IR (νmax/cm-1, KBr): 2931, 2848 (C-H stretch) 1731 (CdO stretch),
1449 (CH3 asymm bend), 1238 (C-O acetate stretch).
Mass spectra EI (m/z, relative intensity): 504 3441 (M+, 0.1%) 444.5
(M - CH3CO2H, 6.2), 384.4 (M - 2CH3CO2H, 100).
3r,24-Bis(acetoxy)-7r-hydroxy-5â-cholane (18). Crystallization
from hexane/ethyl acetate yielded colorless needles: mp 84-88 °C.
1H NMR (δ/ppm, CDCl3): 0.60 (3H, s, 18-Me), 0.86 (3H, s, 19-
Me), 0.92 (3H, d, 21-Me), 1.93 (3H, s, CH3CdO), 1.96 (3H, s, CH3Cd
O), 3.79 (1H, q, J ) 3, 3 Hz, 7â H), 3.96 (2H, m, CH2OAc), (1H, tt,
J ) 11, 5 Hz, 3â H).
13C NMR (δ/ppm, CDCl3): 12.40 (C18), 17.70 (C21), 20.95 (24-
OCOCH3), 21.33 (12-OCOCH3), 21.40 (3-O-COCH3), 23.03 (C19),
64.89 (C24), 74.13 (C3), 75.86 (C12), 170.42 (12-OCOCH3), 170.49
(3-OCOCH3), 171.14 (24-OCOCH3).
IR (νmax/cm-1, KBr): 2931, 2848 (C-H stretch) 1731 (CdO stretch),
1449 (CH3 asymm bend), 1241 (C-O acetate stretch).
Mass spectra EI (m/z, relative intensity): 444.490 (M - CH3CO2H,
1.3), 384.4 (M - 2CH3CO2H, 65), 255.0 (steroid backbone).
3r,24-Bis(acetoxy)-12r-hydroxy-5â-cholane (14). Spectroscopic
and other data accorded with previous data for this compound; however,
as reported by Hammann and Habermehl,72 attempted crystallization
from a range of solvents gave only a viscous gum.
13C NMR (δ/ppm, CDCl3): 11.61 (C18), 18.44 (C21), 20.75 (24-
OCOCH3), 21.18 (3-O-COCH3), 23.00 (C19), 64.91 (C24), 68.45 (C7),
74.32 (C3), 170.60 (3-OCOCH3), 171.10 (24-OCOCH3).
IR (νmax/cm-1, KBr): 3550 (O-H stretch), 2920, 2860 (C-H stretch)
1726 (CdO stretch), 1449 (CH3 asymm bend), 1243 (C-O acetate
stretch).
Mass spectra EI (m/z, relative intensity): 462.3361 (M+, 4.8), 444.2
(M - H2O, 9.5), 402.2 (M - CH3CO2H, 13.1), 384.2 (-H2O, 100).
3r-Acetoxy-7r,24-dihydroxy-5â-cholane (19). Attempted crystal-
lization from hexane/ethyl acetate yielded a colorless oil which could
not be crystallized.
1H NMR (δ/ppm, CDCl3): 0.66 (3H, s, 18-Me), 0.88 (3H, s, 19-
Me), 0.95 (3H, d, 21-Me), 1.99 (3H, s, CH3CdO), 2.01 (3H, s, CH3Cd
O), 3.96 (3H, overlying m, CH2OAc, 12â H), 4.62 (1H, tt, J ) 11, 5
Hz, 3â H).
(71) Spero, G. B.; McIntosh, A. V.; Levin, R. H. J. Am. Chem. Soc.
1948, 70, 1907-1910.
(72) Hammann, P. E.; Habermehl, G. G. Z. Naturforsch. B 1987, 42,
781-782.

Imprinted Polymers as Protecting Groups
J. Am. Chem. Soc., Vol. 121, No. 28, 1999 6651
1H NMR (δ/ppm, CDCl3): 0.60 (3H, s, 18-Me), 0.85 (3H, s, 19-
Me), 0.89 (3H, d, 21-Me), 1.95 (3H, s, CH3CdO), 3.55 (2H, m, C24
CH2-OH), 3.79 (1H, q, J ) 3, 3 Hz, 7â H), 4.52 (1H, tt, J ) 12, 5 Hz,
3â H).
IR (νmax/cm-1, KBr): 3450 (br O-H stretch), 2930, 2860 (C-H
stretch) 1735 (CdO stretch), 1450 (CH3 asymm bend), 1245 (C-O
acetate stretch).
Mass spectra EI (m/z, relative intensity): 420.3253 (M+, 3.5), 402.4
(M - H2O, 28.1), 384.4 (M - 2H2O, 100).
13C NMR (δ/ppm, CDCl3): 11.75 (C18), 18.60 (C21), 20.54 (3-O-
COCH3), 22.72 (C19), 63.56 (C24), 68.82 (C7), 74.36 (C3), 170.80
(3-OCOCH3).
IR (νmax/cm-1, KBr): 3400 (br O-H stretch), 2931, 2861 (C-H
stretch) 1731 (CdO stretch), 1449 (CH3 asymm bend), 1245 (C-O
acetate stretch).
Acknowledgment. We thank the Biotechnology and Bio-
logical Sciences Research Council for financial support. We
also thank Rooma Patel and Mark Barnard for technical
assistance, John Eagles for recording mass spectra, and David
Gleeson and Professor Robert Burch, Department of Chemistry,
University of Reading, for surface area measurements.
Mass spectra EI (m/z, relative intensity): 420.3232 (M+, 0.1), 402.4
(M - H2O, 9.5), 384.4 (M - 2H2O, 13.6) 360.4 (M - CH3CO2H,
12.7), 342.4 (-H2O, 100).
24-Acetoxy-3r,7r-dihydroxy-5â-cholane (20). Crystallization from
hexane/ethyl acetate yielded colorless plates: mp 48-68 °C. Analytical
data were in agreement with the expected structure.73
1H NMR (δ/ppm, CDCl3): 0.59 (3H, s, 18-Me), 0.85 (3H, s, 19-
Me), 0.89 (3H, d, 21-Me), 1.98 (3H, s, CH3CdO), 3.50 (1H, tt, J )
11, 5 Hz, 3â H), 3.79 (1H, q, J ) 3, 3 Hz, 7â H), 3.96 (2H, m, CH2-
OAc).
Supporting Information Available: IR and 1H NMR
spectra and experimental details of the synthesis of 5-aminobo-
ronophthalide, 5-methacryloylaminoboronophthalide, tert-bu-
tyldecoxycholate, and tert-butylchenodeoxycholate (PDF). This
material is available free of charge via the Internet at
http://pubs.acs.org.
13C NMR (δ/ppm, CDCl3): 11.66 (C18), 18.43 (C21), 20.87 (24-
O-COCH3), 23.10 (C19), 65.37 (C24), 68.32 (C7), 71.68 (C3), 171.21
(24-OCOCH3).
(73) Posner, G. H.; Oda, M. Tetrahedron Lett. 1981, 22, 5003-5006.
JA982238H