Synthesis and evaluation of a solid supported molecular tweezer type receptor for cholesterol

Article abstract of DOI:10.1039/b210427j

A new macroporous stationary phase bearing 'tweezer' receptors that exhibit specificity for cholesterol has been constructed from rigid multifunctional vinylic monomers derived from 3,5-dibromobenzoic acid, propargyl alcohol and cholesterol. The synthesis

Full text of DOI:10.1039/b210427j

View Article Online / Journal Homepage / Table of Contents for this issue
Synthesis and evaluation of a solid supported molecular tweezer type
receptor for cholesterol{
Louise Davidson,a Anton Blencowe,a Michael G. B. Drew,a Keith W. Freebairnb and
Wayne Hayes*a
aThe School of Chemistry, The University of Reading, Whiteknights, Reading, Berkshire, UK
RG6 6AD. E-mail: w.c.hayes@reading.ac.uk
bChemical Development, GlaxoSmithKline, Medicines Research Centre, Gunnels Wood Road,
Stevenage, Hertfordshire, UK SG1 2NY
Received 23rd October 2002, Accepted 6th February 2003
First published as an Advance Article on the web 21st February 2003
A new macroporous stationary phase bearing ‘tweezer’ receptors that exhibit specificity for cholesterol has been
constructed from rigid multifunctional vinylic monomers derived from 3,5-dibromobenzoic acid, propargyl
alcohol and cholesterol. The synthesis of the novel tweezer monomer that contains two cholesterol receptor
arms using palladium mediated Sonogashira methodologies and carbonate couplings is reported. The
subsequent co-polymerisation of this tweezer monomer with a range of cross-linking agents via a ‘pseudo’
molecular imprinting approach afforded a diverse set of macroporous materials. The selectivity and efficacy of
these materials for cholesterol binding was assessed using a chromatographic screening process. The optimum
macroporous stationary phase material composition was subsequently used to construct monolithic solid phase
extraction columns for use in the selective extraction of cholesterol from multi-component mixtures of
structurally related steroids.
and D’Souza reported18 a family of bile acid-based mole-
cular tweezers capable of binding electron-deficient aromatic
substrates such as trinitrofluorenone in organic media (Ka ~
220 M21 in CDCl3).
Introduction
In nature, complex molecular assemblies are formed by a
multitude of non-covalent interactions.1 When considered
individually, the strength of a single non-covalent interaction
is weak in comparison to a covalent bond. However, the
simultaneous action of multiple non-covalent bonds facilitates
the production of highly stable complexes, as demonstrated
elegantly by DNA helices. Attempts by synthetic chemists to
imitate nature have led to the generation of numerous synthetic
receptor systems including large macrocyclic hosts such as
crown ethers,2 cyclophanes,3 and calixarenes.4
In contrast to the synthetically challenging macrocyclic host
systems, Rebek has reported5 the synthesis of simpler
‘molecular cleft’ type synthetic receptors. Related studies by
Whitlock6 and Zimmerman7 and their co-workers enabled the
development of a subset of this receptor class that are referred
to as molecular tweezers. This particular receptor type has been
studied extensively and molecular tweezers that exhibit selec-
tive recognition for peptides,8 saccharides,9 adenine deriva-
tives,10 cations,11 anions,12 electron deficient aromatics,13 and
transition metal ions have been reported.14
In this paper, we report the synthesis of a cholesterol-based
molecular tweezer 1, (see Fig. 1) and its incorporation into a
macroporous stationary phase. A polymerisable unit was
incorporated onto the aromatic hub component of the mole-
cular tweezer thereby allowing the tweezer receptor to be
co-polymerised with cross-linkers to construct a macroporous
polymer network. A ‘pseudo’ molecular imprinting19 protocol
was used in this co-polymerisation process to produce highly
ordered pores within the polymer network capable of rebind-
ing cholesterol in a selective manner. An evaluation of the
selective binding characteristics of this new stationary phase is
presented.
Experimental
Materials
Commercial dry solvents were used in all preparations except in
the case of THF that was distilled from benzophenone and
˚
sodium. Triethylamine was pre-dried over 4 A molecular sieves
and then freshly distilled. All other reagents were purchased
from either the Aldrich Chemical Company or Acros Chimica
and were used as received without purification. Styrene was an
exception and was filtered through neutral alumina prior to
use.
Cholesterol is one of the most widely occurring steroids and
can be isolated by extraction from most animal tissue.15 There
is considerable interest in the reduction of high cholesterol
levels in humans which result from the consumption of dairy
products.16 Several studies have focused on receptors for
cholesterol in order to extract the sterol from the food source
and molecular imprinted polymers (MIPs) have been investi-
gated for this purpose.17
To date, molecular tweezers incorporating steroidal units
have not been used widely for the recognition of specific sterols,
although molecular tweezer type receptors have been devel-
oped using the bile acid family of steroids. For example, Maitra
Cholest-5-en-3-yl (3-{[(cholest-5-en-3-yloxy)carbonyl]oxy}prop-
1-ynyl)-5-[(4-vinylanilino)carbonyl]phenyl}prop-2-ynylcarbonate1.
To a cooled (ice bath) solution of 3,5-bis(3-hydroxyprop-1-
ynyl)-N-(4-vinylphenyl)benzamide 8 (0.12 g, 0.36 mmol) in dry
THF (1 cm3) and triethylamine (200 mL) was added dropwise a
solution of cholesterol chloroformate (0.41 g, 0.92 mmol) in dry
THF (1 cm3). The ice bath was removed, and the reaction
stirred at room temperature for 4 hours. The solvent was
{Electronic supplementary information (ESI) available: synthetic
procedures and analytical data for compounds 3–8. See http://
www.rsc.org/suppdata/jm/b2/b210427j/
758
J. Mater. Chem., 2003, 13, 758–766
This journal is # The Royal Society of Chemistry 2003
DOI: 10.1039/b210427j

View Article Online
Fig. 1 The cholesterol-based molecular tweezer system.
evaporated, and the residue dissolved in dichloromethane
(2 cm3). The organic phase was washed with water (3 6 1 cm3),
dried over sodium sulfate, filtered and evaporated to yield a
white solid that was purified by column chromatography
(dichloromethane : EtOAc 9 : 1) to yield the desired tweezer 1
(0.29 g, 67%) as a sticky cream solid; mp 73–74 uC; nmax (film)/
cm21 843, 1162, 1255, 1594, 1677, 1748, 2237, 2950, 3360 and
3753; dH (250 MHz; CDCl3; Me4Si) 0.58 (12H, d, 4 6 CH3),
0.60 (6H, s, 2 6 CH3), 1.02 (6H, s, 2 6 CH3), 1.04 (4H, m, 2 6
CH2), 1.12 (4H, m, 2 6 CH2), 1.15 (4H, m, 2 6 CH2), 1.29
(2H, m, 2 6 CH), 1.32 (4H, m, 2 6 CH2), 1.44 (4H, m, 2 6
CH2), 1.48 (4H, m, 2 6 CH2), 1.51 (6H, s, 2 6 CH3), 1.57 (4H,
m, 2 6 CH2), 1.62 (4H, m, 2 6 CH2), 1.77 (2H, m, 2 6 CH),
1.81 (2H, m, 2 6 CH), 1.84 (4H, m, 2 6 CH2), 1.96 (4H, m,
2 6 CH2), 2.01 (2H, m, 2 6 CH), 2.15 (2H, m, 2 6 CH), 2.19
(2H, m, 2 6 CH), 3.15 (2H, m, 2 6 CH), 4.30 (2H, m, 2 6
CH), 4.95 (4H, m, 2 6 CH2), 5.24 (4H, m, 2 6 CH2). 5.35 (1H,
dd, J ~ 18 Hz, 6 Hz, CHLCH2), 5.86 (1H, dd, J ~ 18 Hz, 6 Hz,
CHLCH2), 6.55 (1H, t, J ~ 6 Hz, CHLCH2), 6.99 (1H, s, NH),
7.14 (2H, AA’XX’, 2 6 ArCH), 7.30 (2H, AA’XX’, 2 6
ArCH), 7.80 (1H, appt. t, J ~ 3 Hz, ArCH), 7.99 (2H, appt. d,
J ~ 3 Hz, 2 6 ArCH); dC (67.5 MHz; CDCl3; Me4Si) 10.9 (2 6
CH3), 17.7 (2 6 CH3), 18.3 (2 6 CH3), 20.0 (2 6 CH2), 21.6
(2 6 CH2), 21.8 (2 6 CH3), 22.8 (2 6 CH3), 23.3 (2 6 CH2), 26.7
(2 6 CH2), 27.0 (2 6 CH2), 27.2 (2 6 CH2), 28.7 (2 6 CH),
29.3 (2 6 CH), 30.8 (2 6 CH2), 30.9 (2 6 CH), 34.8 (2 6
CH2), 35.2 (2 6 CH2), 35.81 (2 6 C–C), 36.9 (2 6 C(CH3)),
38.5 (2 6 C(CH3)), 38.7 (2 6 C(CH3)), 41.30 (2 6 C(CH3)),
48.9 (2 6 CH2), 54.5 (2 6 CH2), 55.7 (2 6 CLC), 55.8 (2 6
CH2), 73.2 (2 6 CMC), 77.7 (2 6 CMC), 83.6 (2 6 C–O), 112.4
(CHLCH2), 121.3 (2C, 2 6 ArCH), 122.1 (2 6 ArCH), 122.3
(ArCH), 124.5 (2 6 –CLC), 125.9 (2C, 2 6 ArCH), 129.4 (2 6
ArC–CMC), 133.4 (ArC–CHLCH2), 134.0 (ArC–CLO), 135.0
(ArCN), 136.6 (CHLCH2), 138.2 (CLO), 152.9 (CLO); m/z
(MALDI-TOF) calculated for C77H106O7N: 1156.37, found:
1195.48 [M 1 K]1.
polymerise for 14 hours in a thermostatically maintained water
bath at 60 uC, yielding a thin (ca. 2 mm) polymeric film on the
base of each vial.
Extraction and rebinding experiments
To each of the solid cross-linked polymer networks was added
a 1 cm3 volume of THF and the vials sonicated for one hour
without heating. Then, 1 cm3 of a 4 : 1 mixture of THF : acetic
acid (AcOH) was added to each vial, and the vials sonicated for
a further hour without heating. The amount of cholesterol
released from the polymers was quantified in each extract by
HPLC analysis (Table 1). The vials were allowed to stand
overnight, and subsequently sonicated for a further hour at
40 uC. No further cholesterol was released from the polymers—
as determined by HPLC analysis of the washing solvent. The
washing solvent was removed from each polymer by means of a
syringe, and a further 1 cm3 volume of a THF : acetic acid
mixture (4 : 1) was added to each polymer. The vials were
sonicated for a further 3 hours at 40 uC—no further cholesterol
was released from the polymers.
Rebinding experiments were performed by addition of 1 cm3
of a 5 mg cm23 solution of cholesterol. The vials were sonicated
for 1 hour, and then shaken overnight. The amount of
cholesterol present in the supernatant liquid was quantified by
HPLC and the amount of cholesterol uptake by the polymers
was calculated (see Table 1).
General protocol for the selectivity studies
Following the uptake experiments, each polymer was repeat-
edly washed with 1 cm3 of a THF : AcOH mixture (4 : 1) to
remove the bound cholesterol. Once HPLC analysis showed
that the removal of bound cholesterol had been achieved to an
appreciable extent, under these detection conditions, the
polymers (MIPs and controls) were exposed to 1 cm3 of a
5 mg cm23 solution of stigmasterol. The polymers were shaken
overnight to allow equilibration, before quantification of the
residual stigmasterol via HPLC analysis (see Fig. 6). This
selectivity experiment was repeated using cholesteryl acetate.
General procedure for the synthesis of ‘mini’ molecular imprinted
polymers
Two solutions were prepared (one with the template and one
without) by mixing ethylene glycol dimethylacrylate (EDMA)
(142.5 mL, 750 mmol), and AIBN (1.5 mg, 9 mmol) in THF
(210 mL). To the template solution was also added tweezer 1
(21.6 mg, 18.75 mmol), and cholesterol (7 mg, 18.75 mmol). A
total of 70 mL of each mother solution was dispensed into
1.5 cm3 HPLC clear glass vials. To each vial was added the
required volume of functional co-monomer, and subsequently
each vial was sealed under argon. The vials and contents were
sonicated at room temperature for 1 hour and purged with
argon for a further 5 minutes. The vial contents were allowed to
General procedure for the synthesis of monolithic molecular
imprinted polymers
A mixture of the tweezer receptor 1 (0.116 g, 0.1 mmol),
cholesterol (0.039 g, 0.1 mmol), EDMA (750 mL, 4 mmol),
methacrylic acid (68 mL, 0.8 mmol) and AIBN (0.008 g,
0.048 mmol) in dry THF (1.12 cm3) was sonicated for 30
minutes at room temperature. A stainless steel column (50 mm 6
4.6 mm id) was filled with the mixture and placed in a water
bath at 60 uC. After 14 hours, the column was removed, and
connected to a HPLC pump and washed with THF, followed
J. Mater. Chem., 2003, 13, 758–766
759

View Article Online
Table 1 Composition of MIPs prepared for the screening process
Ratio template :
functional
co-monomer
Functional
co-monomerb
% Cholesterol
recovery
% Cholesterol
binding
Polymera
Cross-linkerb
1
3
5
7
9
MAA
MAA
MAA
MAA
MAA
MAA
MAA
MAA
MAA
MAA
MAA
MAA
MAA
MAA
MAA
MAA
MAA
MAA
MAA
MAA
MMA
MMA
MMA
tBuMA
tBuMA
tBuMA
tBuMA
tBuMA
tBuMA
Styrene
Styrene
Styrene
1 : 2
1 : 3
1 : 3.5
1 : 4
1 : 4.5
1 : 5
1 : 6
1 : 8
1 : 2
1 : 3
1 : 4
1 : 5
1 : 6
1 : 8
1 : 4
1 : 6
1 : 8
1 : 4
1 : 6
1 : 8
1 : 4
1 : 6
1 : 8
1 : 4
1 : 6
1 : 8
1 : 4
1 : 6
1 : 8
1 : 4
1 : 6
1 : 8
73% EDMA
73% EDMA
73% EDMA
73% EDMA
73% EDMA
73% EDMA
73% EDMA
73% EDMA
50% EDMA
50% EDMA
50% EDMA
50% EDMA
50% EDMA
50% EDMA
73% DVB
73% DVB
73% DVB
TRIM
TRIM
TRIM
73% EDMA
73% EDMA
73% EDMA
73% EDMA
73% EDMA
73% EDMA
50% EDMA
50% EDMA
50% EDMA
73% DVB
15
29
20
31
30
12
10
24
17
27
25
25
17
19
10
9
16
36
28
20
0
28
15
20
16
13
38
54
53
28
24
7
60
48
30
74
61
41
55
39
46
19
54
27
45
51
42
52
47
0
27
10
0
22
0
30
0
0
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
43
45
47
49
51
53
55
57
59
61
63
41
0
26
13
49
51
73% DVB
73% DVB
aNote: for each MIP a subsequent control polymer was prepared in the absence of the tweezer receptor and cholesterol guest (for example, in
the case of polymer 1, polymer 2 represents the corresponding control). The concentration of the cholesterol guest in each of the above poly-
mers was 0.012 mmol dm23. The porogenic solvent used in the preparation of the polymers was THF. bAbbreviations: MAA ~ methacrylic
acid; tBuMA ~ tert-butyl methacrylate; EDMA ~ ethylene glycol dimethylacrylate; DVB ~ divinylbenzene; TRIM ~ trimethylolpropane tri-
methacrylate.
by a THF : AcOH mixture (4 : 1) to elute the cholesterol guest.
Non-imprinted control ‘blank’ monoliths were prepared in the
same way, but with omission of the tweezer host and
cholesterol guest.
The procedures outlined above were performed in tripli-
cate for each imprinted and blank polymeric film or mono-
lithic column. The chromatographic results were reproducible
between replicate samples with errors ¡5%.
left to air-dry prior to analysis. 1H Nuclear magnetic resonance
(NMR) spectra were recorded on a Bruker AC250 (250 MHz)
spectrometer (using the solvent proton signal or tetramethyl-
silane as internal reference). 13C Nuclear magnetic resonance
(NMR) spectra were recorded on a Bruker AC250 (62.5 MHz)
spectrometer. Infrared (IR) spectroscopic analyses were per-
formed on a Perkin Elmer 1720-X Infrared Fourier Transform
spectrometer.
Characterisation
X-Ray diffraction analysis. X-Ray diffraction patterns were
obtained using the MARresearch Image Plate System equipped
with a Mo-Ka radiation source. The crystals were positioned at
70 mm from the Image Plate and 3948 reflections were
measured. 100 frames were measured at 2u intervals with a
counting time of 2 minutes. Data analysis was carried out using
the XDS program20 to provide 1613 independent reflections
(Rint ~ 0.0835). The structure was solved using direct methods
with the SHELX86 program.21 The non-hydrogen atoms were
refined with anisotropic thermal parameters. The hydrogen
atoms bonded to carbon were included in geometric positions
and given thermal parameters equivalent to 1.2 times those of
the atom to which they were attached. The structure was
refined on F2 using SHELXL22 to R1 0.0764 and wR2 0.2277
for 941 reflections with I w 2s(I).
General. Thin-layer chromatography (TLC) was performed
on aluminium sheets coated with Merck 5735 Kieselgel
60F. Developed TLC plates were air-dried and scrutinized
under a UV lamp. Sorbsil 60 (0.040–0.063 mm mesh, Merck
9385) was used to perform column chromatography. Melting
points were determined on an Electrothermal digital melting
point apparatus and are uncorrected. Mass spectra (MS) of
lower molecular weight materials were obtained using a VG
Autospec mass spectrometer operating in the chemical
ionisation mode employing ammonia as the impact gas.
Accurate mass spectra (MS) were obtained using a Micromass
Q-Tof1 LC-MS-MS spectrometer operating in the electrospray
mode. MALDI-TOF mass spectra of higher molecular weight
materials were obtained on a SAI LT3 LaserTof using trans-3-
indoleacrylic acid as the matrix. A typical sample preparation
for MALDI-TOF mass spectrometric analysis is described as
Data for 5: C13H10O4, M ~ 230.21, orthorhombic, a ~
3
˚
˚
15.47(2), b ~ 7.797(14), c ~ 18.86(3) A, U ~ 2275 A , space
group Pbca, dcalc ~ 1.344 g cm23, Z ~ 8.
CCDC reference number 203506.
follows: 3 mL of a solution of the analyte in THF (10 mg cm23
was combined with 20 mL of the freshly prepared matrix (0.2 M
in THF) in a mini-vial, and from the mixture was taken a 2 mL
aliquot which was carefully transferred onto a sample plate and
)
See http://www.rsc.org/suppdata/jm/b2/b210427j/ for crys-
tallographic data in CIF or other electronic format.
760
J. Mater. Chem., 2003, 13, 758–766

View Article Online
Liquid chromatography. Chromatographic analyses were
performed on a Beckman Gold HPLC System equipped with
programmable solvent module 126, a Rheodyne injector
(injection loop ~ 20 mL) and a UV/vis programmable detector
module 166 operating at 210 nm. During the ‘pseudo-
combinatorial’ molecular imprinted study, the cholesterol
uptake and recoveries were quantified using a Luna 3u
C18(2), with a 150 6 4.60 mm HPLC column, mobile phase
Following the publication of this procedure, an increase in the
yield of diol 4 was achieved via optimisation of conditions
reported by Moore and co-workers.24 Use of Pd(dba)2 and
triphenylphosphine in conjunction with copper(I) iodide and
triethylamine in refluxing THF afforded excellent yields (70–
76%) of product 4 in comparison to our published procedure.
This enhancement in yield could be attributed to the longer
lifetime of this active catalytic system. Low donating ligands
such as tri-2-furylphosphine (TFP) are known to have only
short-lived catalytic activities25 (albeit longer than the triphenyl-
arsine ligand), and hence it is likely that the TFP ligand did
not remain sufficiently active to afford good yield of the
bis-acetylenated product. The electron donating triphenyl-
phosphine ligand, in comparison, is likely to undertake more
frequent catalytic cycles, and hence an enhancement in yield
was observed. The main drawback in this case was, however,
the long reaction times and harsh conditions that were neces-
sary to force this palladium catalytic system to approach
reaction completion. At least 24 hours under continuous reflux
were essential for attainment of a 76% yield.
of 50 : 50 acetonitrile : isopropyl alcohol, flow rate 1 cm3 min21
,
and column temperature 20 uC.
Capacity factors for the molecular imprinted monoliths were
calculated from the equation ka ~ (tR 2 tm)/tm, where tR is the
retention time of the analyte and tm the dead volume
determined from the void marker (toluene). The separation
factor a was calculated from the equation a ~ Ka(MIP)
Ka(CONTROL)
/
.
Results and discussion
Synthesis of a novel molecular tweezer for cholesterol
The ethyl ester of free diol 4 was hydrolysed smoothly to the
free carboxylic acid 5 in basic media. Subsequent TBDMS
protection of each hydroxy moiety under standard condi-
tions26 afforded the required acid 6 in acceptable yield. Use of
the coupling agent 1-(3-dimethylaminopropyl)-3-ethylcarbodi-
imide hydrochloride (EDCI) in combination with HOBt27
proved highly effective for the amide bond formation in con-
junction with 4-vinylaniline, and a good yield of the TBDMS
protected amide 7 was realised if DMF was employed as
the solvent. Subsequent cleavage of the TBDMS protecting
The synthetic approach employed to afford the cholesterol-
based molecular tweezer 1 is illustrated in Scheme 1.
We have recently reported the synthesis of 3,5-bis(3-
hydroxyprop-1-ynyl)-N-(4-vinylphenyl)benzamide 8, which is
referred to in this paper as the ‘tweezer hub’ unit, and the
precursor to 1 (Scheme 1).23 Synthesis of this compound was
realised via a six-step synthetic strategy,{ which included a
synthetically challenging Sonogashira coupling between pro-
pargyl alcohol and 3,5-dibromobenzoic acid ethyl ester 3.
Scheme 1 Reagents: i. EtOH, DCC, DMAP, Et2O; ii. propargyl alcohol, Pd(dba)2, PPh3, CuI, Et3N, THF; iii. LiOH, H2O, (CH3)2CO; iv.
TBDMSCl, imidazole, DMF; v. 4-vinylaniline, EDCI, HOBt?H2O, DMF A 7 then TBAF, THF, AcOH; vi. cholesterol chloroformate, Et3N, THF.
J. Mater. Chem., 2003, 13, 758–766
761

View Article Online
groups28–30 afforded the free diol 8 in good yield. The required
molecular tweezer 1 was obtained in acceptable yield via
reaction of the tweezer hub unit 8 with cholesteryl chloro-
formate under basic conditions.17
cholesterol ‘tweezer arms’. Evaluation of the receptor–choles-
terol guest binding by the steroidal molecular tweezer receptor
1 in the solution state indicated negligible binding between the
molecular tweezer host and cholesterol guest. The solution
state binding assay was hindered severely by the solubility of
both the tweezer host and the cholesterol guest in the solvents
employed. An NMR spectroscopic binding assay was
performed in neat CDCl3 as this solvent proved to be the
only suitable medium to obtain solubility of both host and
guest. Negligible binding was detected between the tweezer host
and cholesterol guest. It was decided to pursue the evaluation
of the binding properties of this receptor following polymer-
isation since it has been noted31 by other research groups that
other tweezer-type receptors exhibit enhanced binding proper-
ties in the solid phase in comparison to solution phase analysis.
Subsequently, despite the poor binding efficiencies demon-
strated in the solution state in an apolar solvent, it was
predicted that by synthesising molecular imprinted polymers
using a 1 : 1 mixture of molecular tweezer host : cholesterol
guest, effective binding would occur in the solid-state environ-
ment. Hence this detrimental solubility factor would not be
so significant, and use of polar solvents known to promote
hydrophobic binding (such as acetonitrile and isopropyl
alcohol) would be feasible.
X-Ray crystallographic analysis of 3,5-bis(3-hydroxyprop-1-
ynyl)benzoic acid 5
The structure of 5 consists of discrete molecules as shown in
Fig. 2a together with the atomic numbering scheme. The acid
diol molecule 5 is essentially planar with slight deviations
observable in the acetylenic substituents. Thus the acid group is
rotated by 4.6u from the plane of the benzene ring and atoms
˚
C(41), C(42), C(43), O(44) are 20.06, 20.14, 20.24, 20.91 A
˚
and C(61), C(62), C(63), O(64) 20.03, 20.08, 20.15, 20.68 A
from this ring plane, respectively. The molecules are held
together in the crystal via three strong intermolecular hydrogen
…
bonds (Fig. 2b) with dimensions (O O, H O distances/A, O–
…
˚
…
…
…
…
H
0.5 2 z) 2.68, 1.86, 171; O(44)–H(44) O(22) (x 1 0.5, 1.5 2 y,
O angle/u) as follows: O(23)–H(23) O(64) (1 2 x, y 2 0.5,
2z) 2.76, 1.98, 159 and O(64)–H(64) O(44) (x 2 0.5, y, 0.5 2
z) 2.72, 1.91, 166, respectively.
The ‘pseudo-combinatorial’ technique pioneered by Lanza
and Sellergren32 was used to generate a large array of molecular
imprinted polymers on a small scale in order to assay the
cholesterol binding potential of these polymeric receptors. The
overall aim of this ‘pseudo’-MIP study was the polymerisation
of a 1 : 1 tweezer host–cholesterol complex (the ‘template’),{ in
the presence of various functional co-monomers and an excess
of the cross-linking agent. This polymerisation would be
performed within 1.5 cm3 HPLC vials, and it was predicted that
a 1–2 mm thick polymeric film would be formed upon the base
of each vial. Following the polymerisation, it was envisaged
that the predominantly hydrophobic interactions could be
disrupted by washing the polymers with a good solvent for
cholesterol such as THF with addition of a small quantity of
acetic acid. After HPLC analysis of the washing solvent
revealed that extraction of cholesterol from the polymers was
sufficient, it was perceived that ‘imprinted’ polymeric receptors
would result featuring the tweezer receptor 1 within their
structure. Following the addition of a solution of known
concentration (12.9 mM) of cholesterol to each polymer, the
uptake of cholesterol was quantified by HPLC analysis of the
supernatant following a suitable period of equilibration (24
hours) of the MIP host and cholesterol guest.
The initial variables that were considered for the composi-
tion of the MIPs were as follows: the functional co-monomer,
the ratio of template : functional co-monomer, the choice of
cross-linker and cross-linking density and porogen selection.
Each of these variables was considered individually during the
screening process. The composition of a selection of the poly-
mers synthesised is illustrated in Table 1. In all cases the
polymerisations were initiated by AIBN at 60 uC.
Choice of functional co-monomer. In this screening assay,
various functional co-monomers were tested: methacrylic acid
(MAA), styrene, 4-vinylbenzoic acid (VBA), 4-vinylphenol
(VP), ethylene glycol dimethylacrylate phosphate (EGDMP),33
tert-butyl methacrylate (tBuMA),34 methyl methacrylate
(MMA), and a 1 : 1 mixture of methacrylic acid and styrene.
The key results from the study of the influence of the
functional co-monomer on the imprinting efficiency are
illustrated in Fig. 3.
Fig. 2 a) The structure of 5 with ellipsoids at 15% probability and b)
the hydrogen bonded array of 5 in the solid state.
Evaluation of cholesterol binding properties of molecular tweezer
receptor 1
The optimum results for cholesterol rebinding were achieved
The molecular tweezer receptor 1 was designed so that the rigid
aromatic bis-acetylene arms created a cavity of suitable dimen-
sions for cholesterol uptake between the two hydrophobic
{A binding stoichiometry of 1 : 1 was assumed to occur under these
conditions between the molecular tweezer and the cholesterol guest
system.
762
J. Mater. Chem., 2003, 13, 758–766

View Article Online
Fig. 3 Effect of choice of functional co-monomer on cholesterol uptake.
using methacrylic acid as the functional co-monomer in a 1 : 4
template : monomer ratio, in conjunction with a 73% EDMA
density and using THF as the porogen (solvent). The large
difference in the percentage binding of cholesterol by the MIP
and control polymers indicates that an ‘imprinting’ effect has
occurred. A degree of non-specific binding occurred within
these resins, as indicated by the binding of cholesterol by the
control polymers albeit to a lesser degree. Furthermore,
addition of an extra quantity of cholesterol to these resins
did not result in any extended uptake of cholesterol, suggesting
that the maximum capacity of these resins had been realised.
Methacrylic acid derivatives proved to be not so efficient as
functional co-monomers. For example, tert-butyl methacrylate
yielded inefficient results overall both in terms of cholesterol
uptake and recovery. In the case of this functional co-
monomer, poor cholesterol recoveries were achieved when
the EDMA density was 73%. Furthermore, the uptake of
cholesterol by these MIPs proved very inefficient in the case of
the higher template : co-monomer ratios. Only in the case of the
1 : 4 template : co-monomer ratio was any significant uptake of
cholesterol observed, however, greater uptake was noted for
the corresponding control polymer, thereby indicating the
presence of only non-specific binding. Decreasing the size of the
ester group from the tert-butyl to the methyl ester in a 1 : 4
template : monomer ratio yielded slightly higher cholesterol
uptake from the MIPs, although higher non-specific binding
from the corresponding control was also prevalent.
In the absence of definitive NMR spectroscopic data, we can
only speculate upon the mode of binding of the polymers
imprinted with the tweezer receptor with the cholesterol sub-
strate. In the case of the methacrylic acid imprinted polymers, it
is probable that polar residues (arising from the methacrylic
acid) surround the binding cavities created by the tweezer host,
from the methacrylic acid. Consequently, selective inclusion of
cholesterol into the apolar binding cavities is more probable
than non-selective binding to the surrounding polar residues.
In the case of hydrophobic co-monomers such as styrene, there
are likely to be apolar hydrophobic residues arising from the
styrene surrounding the apolar-binding site. Consequently in
this situation, there is now competition for the inclusion of the
apolar cholesterol guest into the imprinted binding cavities,
and a certain degree of non-specific interaction of the chole-
sterol with these additional hydrophobic residues will occur. In
fact, equal binding of cholesterol by the imprinted and blank
polymers was observed, thereby supporting this hypothesis. In
the case of the methacrylate esters, the hydroxy moiety of the
carboxylic acid residues on the polymer is blocked, increasing
the apolar nature of these sites, subsequently multiplying the
possibility for non-specific binding. Furthermore, there is the
possibility of a detrimental steric factor placed upon the
cholesterol binding from the bulky tert-butyl ester group. It is
thought that the overall binding mechanism of cholesterol to
the imprinted polymers is a delicate balance between simple
hydrogen bonding of the cholesterol hydroxy at C-3 of the
steroid backbone with the methacrylic carboxylic acid
functionality and the tweezer binding mechanism proposed
at the onset of this work.
Inefficient binding of cholesterol was noted for both MIPs
and controls synthesised from 4-vinylbenzoic acid, suggesting
that this monomer is unsuited to this imprinting protocol.
Similarly inferior results were also obtained from a 1 : 1 mixture
of methacrylic acid : styrene. These results indicate that a
combination of hydrophobic and hydrogen-bonding moieties
in the functional co-monomer yield an insufficient imprinting
performance. This result was further confirmed by the low
cholesterol uptake observed using 4-vinylphenol as the
functional co-monomer. EGDMP showed poor binding
affinities with cholesterol, in contrast to the literature reports.33
Ineffective cholesterol recoveries were obtained initially,
suggesting a lower number of possible cholesterol binding
sites and in general poor cholesterol binding was observed by
the MIPs (possibly linked to the lower cholesterol recoveries).
Although more moderate cholesterol binding was achieved at
higher template : monomer ratios, the binding mode was most
likely non-specific, since identical cholesterol binding was
observed with the MIPs and the corresponding control
polymers.
Choice of cross-linker and cross-linking density. Initially
ethylene glycol dimethylacrylate (EDMA) was chosen as the
cross-linker. EDMA is the most common cross-linker
employed in imprinting protocols,19 and was found to be
more effective than divinylbenzene (DVB) used in the
imprinting of cholesterol by Whitcombe and co-workers.17
DVB and trimethylolpropane trimethacrylate (TRIM) were
also tested in this study. The key results from the study of the
influence of the cross-linking agent on the imprinting efficiency
are illustrated in Fig. 4.
In most cases, the cross-linking density was maintained19
at 73% in order to assess the effect of variation of functional
J. Mater. Chem., 2003, 13, 758–766
763

View Article Online
Fig. 4 Effect of choice of cross-linking agent and density on cholesterol uptake.
co-monomer on cholesterol rebinding. It was observed, how-
ever, that recoveries of cholesterol from the polymers after
completion of the polymerisation were not optimal. These low
recoveries are attributed to the high cross-linking density of the
polymer hindering the release of the template, despite the use of
relatively harsh washing conditions (4 : 1 THF : AcOH). Hence,
it was decided to decrease the cross-linking density from 73% to
50% for selected polymers and assess the effect upon the
cholesterol recovery. In most cases, this reduction in cross-
linking density served only to reduce the overall uptake of
cholesterol by the MIPs, thereby suggesting that a more rigid
polymer network is a pre-requisite for efficient molecular
imprinting techniques.
From this study, it was apparent that EDMA used at 73%
cross-linking density in conjunction with a 1 : 4 ratio of
template : MAA represented the optimum imprinting reagent
combination. The two other cross-linkers tested—DVB and
TRIM—both proved less efficient in the production of efficient
imprinted materials. In particular, extremely poor results were
obtained using the trifunctional cross-linker TRIM. For DVB,
in the majority of cases very similar uptakes of cholesterol were
noted for the MIPs and their corresponding controls. This
further confirms the likely presence of non-specific binding sites
present within these MIPs. This situation is particularly
significant for the DVB cross-linked polymers based on
styrene, thereby emphasising that aromatic functional co-
monomers and cross-linkers participate predominantly in non-
specific hydrophobic interactions with the cholesterol guest and
are hence unsuited to this imprinting protocol.
porogen, the results obtained from the screening assays
indicate that the optimum porogen for this pseudo-imprinting
protocol was THF. Combinations of THF and acetonitrile
afforded mixed results, and higher uptakes of cholesterol were
observed from polymers derived from the use of a neat THF
porogen. In general, addition of isopropyl alcohol to the
porogen proved to have a particularly detrimental effect on
cholesterol uptake. This effect was pronounced particularly in
the case of the methacrylic acid-based MIPs. It is proposed that
the isopropyl alcohol was acting as a hydrogen-bonding
competitor with the cholesterol guest. This result further
emphasises the importance of secondary hydrogen bonding
interactions between the template and the functional co-
monomer in this imprinting protocol.
Selectivity studies. Although certain polymers revealed high
uptake of cholesterol in comparison to their corresponding
blank control polymers, the nature of this cholesterol binding is
not certain. It was decided, therefore, to study the uptake of
certain cholesterol steroidal analogues for comparison with
cholesterol itself. The steroids chosen in this case were
cholesteryl acetate and stigmasterol (Fig. 5).
Ideally, the MIPs should exhibit a higher selectivity for
cholesterol in comparison to its structural analogues. The MIPs
Choice of porogen. The porogen selection was hindered
severely by the limited solubility range of both the tweezer host
and cholesterol guest. Both substrates presented sufficient
solubility in THF and hence this solvent was the porogen of
choice at the onset of this study. However, efficient promotion
of hydrophobic binding required use of a porogen such as
acetonitrile. The solubility of the tweezer and cholesterol was
not sufficient in acetonitrile to permit its use alone. A mixture,
however, of 50 : 50 THF : acetonitrile could be employed.
Likewise isopropyl alcohol is another suitable porogen as
this has been shown to be a highly effective solvent for hydro-
phobic binding involving steroids.33 Further solubility studies
revealed that a combination of 50 : 25 : 25 of THF : aceto-
nitrile : isopropyl alcohol was suitable for use in the imprint-
ing assays.
Fig. 5 Structures of the steroids employed in the selectivity study.
Despite the use of mixed solvent systems as the polymerisation
764
J. Mater. Chem., 2003, 13, 758–766

View Article Online
subsequently to construct monolithic solid phase extraction
(SPE) columns for use in the selective extraction of cholesterol
in the presence of the previously studied structural analogues.
In recent years, monolithic stationary phases35 have been used
to overcome the problems associated with conventional
particle-packed HPLC columns, for example, issues relating
to permeability. Furthermore, monolithic stationary phases are
used to accelerate mass transfer processes yielding increased
performance at elevated flow rates.34 Monolithic materials of
this type also have the advantage over their more conventional
counterparts in the speed of their preparation. A typical
protocol for the synthesis of a macroporous monolithic
stationary phase involves simply mixing the polymerisation
components in an empty column, and then heating this column
until polymerisation is complete. This approach compares
favourably with the more conventional stationary phase
synthesis, in which the material is first polymerised, then
crushed and ground before packing into the analysis column
under high pressure. Although this process often affords highly
efficient stationary phases, it is time-consuming.
For synthesis of monoliths featuring the steroid tweezer
receptor, it was decided to employ clean, short (50 mm 6
4.6 mm id) stainless-steel columns both to minimise analysis
times and reduce the back pressures during the chromato-
graphic analysis. The optimum molecular imprinting condi-
tions described above were applied to the synthesis of two sets
of monolithic columns (one MIP and one corresponding
control/blank column per set). The results from this initial
study are illustrated in Table 2.
Encouragingly, the imprinted monoliths exhibited similar
properties to their corresponding polymers in the pseudo-
imprinting screening assay, both in terms of cholesterol affinity
and selectivity. In both cases, the templated monoliths revealed
acceptable capacity factors for cholesterol in comparison to the
relevant blank columns, prepared in the absence of the tweezer
receptor and cholesterol guest (average Ka ~ 1.36). This is
reflected in the observed separation factor a of 5.44 for the
imprinted monoliths in comparison to the blank untemplated
polymers. 100% THF was employed as the mobile phase to
prevent undesirable swelling of the polymeric resins. Use of low
chromatographic flow rates (0.05 cm3 min21) was desirable as a
result of the short column length—when higher flow rates were
employed there was insufficient time to enable the establish-
ment of an efficient binding equilibration between cholesterol
and the imprinted stationary phase. Consequently the good
uptake of cholesterol observed at such low flow rates may be
attributed to the realisation of effective thermodynamically
controlled binding equilibration thereby indicating that uptake
maybe predominantly occurring under kinetic control. It is,
however, also possible that kinetics were inhibiting mass
transfer at such higher flow rates.
Fig. 6 Substrate selectivity assays carried out using the cholesterol-
based molecular tweezer imprinted polymers.
that exhibited the most significant selective binding of
cholesterol in comparison to the relevant control and the
highest levels of cholesterol recovery following the imprinting
protocol were chosen for this selectivity study (see Table 1).
The results from the selectivity studies are illustrated in Fig. 6.
The polymers showed a negligible uptake of stigmasterol,
indicating minimal affinity for this steroid by these MIPs.
Cholesteryl acetate was bound to a small extent by all polymers
screened in this separate study. This result was predicted, given
the extremely close structural similarities of this steroid to the
cholesterol template. Cholesteryl acetate differs from choles-
terol only in the modification of the C-3 hydroxy of the steroid
backbone; hence a potential hydrogen-bonding site is blocked,
thereby further supporting evidence of selective hydrogen
bonding in the polymers.
Given the overall results obtained from this chromato-
graphic screening process, it is apparent that the optimum
conditions for the synthesis of molecular imprinted polymers
featuring the steroid molecular tweezer are the use of
methacrylic acid as the functional co-monomer in a ratio of
four equivalents to the template with a cross-linking density of
73% EDMA in a THF porogen, initiated by AIBN at 60 uC.
Significance of the tweezer cavity. To assess the influence of
the tweezer-binding cavity on cholesterol uptake, the monomer
cholesteryl (4-vinyl)phenyl carbonate used by Whitcombe
and co-workers for the imprinting of cholesterol was syn-
thesised according to the published procedure.17 Cholesteryl
(4-vinyl)phenyl carbonate can be considered a ‘single-armed’
analogue of tweezer receptor 1, and consequently differs from
receptor 1 as a result of the lack of a formal ‘binding cavity’.
Molecular imprinted polymers were synthesised for cholesteryl
(4-vinyl)phenyl carbonate using the conditions deemed to be
optimum for receptor 1 (4 equivalents methacrylic acid, 73%
EDMA, THF), and the binding capacities of cholesterol
compared for each receptor.
Promisingly, imprinted polymers based on cholesteryl
(4-vinyl)phenyl carbonate revealed minimal uptake of choles-
terol (average 8% from solution). This result compares favour-
ably with the average cholesterol uptake of 74% exhibited by
the two-armed tweezer receptor 1, thereby highlighting the
need for a well-defined cavity such as that created by the two
arms of receptor 1.
Furthermore the imprinted monoliths revealed excellent
selectivity for cholesterol, in that the structurally related sterols
stigmasterol and cholesteryl acetate were retained on the
column to a far lesser extent. Indeed, negligible retention times
were noted for stigmasterol on the column, and this is
consistent with the selectivity study performed during the
screening assays.
Table 2 Performance of molecular imprinted monoliths for application
in SPEa
Ka(MIP)
Ka(CONTROL)
Application of the imprinting protocol to molecular imprinted
monoliths
Cholesterol
1.36
0.04
0
0.25
0.10
0.08
—
Cholesterol acetate
Stigmasterol
Selectivity factor a
Having developed the optimum molecular imprinting protocol
for efficient cholesterol uptake, it was decided to apply these
conditions to the synthesis of novel monoliths35 featuring our
steroidal tweezer receptor. The optimum macroporous sta-
tionary phase material composition thus determined was used
5.44
aAverage values are quoted for the capacity factor ka and the selec-
tivity factor a.
J. Mater. Chem., 2003, 13, 758–766
765

View Article Online
J. D. Kilburn, Chem. Eur. J., 2001, 7, 1889; R. Arienzo and
J. D. Kilburn, Tetrahedron, 2002, 58, 711.
M. Takeuchi, T. Imada and S. Shinkai, J. Am. Chem. Soc., 1996,
118, 10658.
Conclusions
9
A novel macroporous stationary phase material bearing
tweezer receptors that are specific for cholesterol has been
constructed from rigid multifunctional vinylic monomers
derived from 3,5-dibromobenzoic acid, propargyl alcohol
and cholesterol. Co-polymerisation of this tweezer monomer
in conjunction with methacrylic acid and a large excess of
10 D. Mink and G. Deslongchamps, Tetrahedron Lett., 1996, 37,
7035.
11 S. Pohl, R. Goddard and S. Kubik, Tetrahedron Lett., 2001, 42,
7555.
12 M. Albrecht, J. Zauner, R. Burgert, H. Ro¨ttele and R. Fro¨hlich,
Mater. Sci. Eng., C, 2001, 18, 185.
13 F. G. Kla¨rner, U. Burkert, M. Kamieth, R. Boese and J. Benet-
Buchholz, Chem. Eur. J., 1999, 5, 1700.
14 B. Ko¨nig, M. Nimtz and H. Zieg, Tetrahedron, 1995, 51, 6267.
15 E. Heftmann, Steroid Biochemistry, Academic Press, New York,
London, 1970; F. J. Zeelan, Medicinal Chemistry of Steroids,
Elsevier, The Netherlands, 1990.
16 D. G. Oakenfull, R. J. Pearce and G. S. Sidhu, Aust. J. Dairy
Technol., 1991, 46, 110.
17 M. J. Whitcombe, M. E. Rodriguez, P. Villar and E. Vulfson,
J. Am. Chem. Soc., 1995, 117, 7105.
18 U. Maitra and L. J. D’Souza, J. Chem. Soc., Chem. Commun.,
1994, 2793; L. J. D’Souza and U. Maitra, J. Org. Chem., 1996, 61,
9494.
cross-linking agent via
a ‘pseudo’ molecular imprinting
approach afforded a diverse set of macroporous materials
whose selectivity and efficacy for cholesterol binding was
assessed using a chromatographic screening process. The
optimum macroporous stationary phase material composition
was found to be a methacrylic acid-based imprinted polymer,
prepared using a cross-linking density of 73% EDMA, in
conjunction with THF as the porogen under AIBN-initiated
thermal conditions. Furthermore, the efficiency of the formal
binding cavity created by the tweezer receptor was assessed by
comparison to an analogous ‘one-armed’ tweezer receptor. The
MIP composition thus determined was subsequently used to
construct monolithic solid phase extraction cartridges for use in
the selective extraction of cholesterol in comparison to certain
structurally related steroids.
19 For comprehensive reviews of molecular imprinted polymers and
their potential applications, see: G. Wulff, Angew. Chem., Int. Ed.
Engl, 1995, 34, 1812; M. J. Whitcombe and E. N. Vulfson, Adv.
Mater., 2001, 13, 467; J. Steinke, D. C. Sherrington and
I. R. Dunkin, Adv. Polym. Sci., 1995, 123, 81; M. J. Whitcombe,
C. Alexander and E. N. Vulfson, Synlett, 2000, 6, 911; L. Davidson
and W. Hayes, Curr. Org. Chem., 2002, 6, 265.
Acknowledgements
20 W. Kabsch, J. Appl. Cryst., 1988, 21, 916.
L. D. would like to thank the BBSRC and GlaxoSmithKline
for funding. M. G. B. D. would like to thank the EPSRC and
the University of Reading for providing funds for the Image
Plate system. In addition, the authors would like to thank Dr
Douglas Philp from the University of St. Andrews for
constructive comments on certain aspects of this project.
21 G. M. Sheldrick, Acta Crystallogr., Sect. A, 1990, 46, 467.
22 G. M. Sheldrick, SHELXL, program for refinement of crystal
structures, University of Gottingen, Germany, 1993.
23 L. Davidson, K. W. Freebairn, A. T. Russell, H. S. Trivedi and
W. Hayes, Synlett, 2002, 2, 251.
24 Z. Xu and J. S. Moore, Acta Polym., 1994, 45, 83; C. Devadoss,
P. Bharathi and J. S. Moore, J. Am. Chem. Soc., 1996, 118, 9635.
25 N. G. Andersen and B. A. Keay, Chem. Rev., 2001, 101, 997.
26 T. W. Greene and P. G. M. Wuts, Protective Groups in Organic
Synthesis, Wiley Interscience, New York/Chichester, 3rd edn.,
1999.
27 D. L. Boger, T. Honda, R. F. Menezes, S. L. Colletti, Q. Dang and
W. Yang, J. Am. Chem. Soc., 1994, 116, 82.
28 C. M. Haywood, D. Yohannes and S. J. Danishefsky, J. Am.
Chem. Soc., 1993, 115, 9345.
29 A. B. Smith, III, S. S. Y. Chen, F. C. Nelson, J. M. Reichert and
B. A. Salvatore, J. Am. Chem. Soc., 1995, 117, 12013.
30 A. B. Smith, III and G. R. Ott, J. Am. Chem. Soc., 1996, 118,
13095.
31 A. J. Brouwer, H. J. van der Linden and R. M. J. Liskamp, J. Org.
Chem., 2000, 65, 1750.
References
1
2
G. M. Whitesides, Angew. Chem., Int. Ed. Engl., 1990, 29, 1209.
G. W. Gokel, in Large Ring Molecules, ed. J. A. Semlyen, John
Wiley & Sons Ltd, Chichester, 1996, Ch. 7, p. 263.
Y. Murakami, J. Kikuchi, T. Ohno, T. Hirayama, Y. Hisaeda,
H. Nishimura, J. P. Snyder and K. Steliou, J. Am. Chem. Soc.,
1991, 113, 1705.
C. D. Gutsche, in Large Ring Molecules, ed. J. A. Semlyen, John
Wiley & Sons Ltd, Chichester, 1996, Ch. 8, p. 309.
J. Rebek Jr., Top. Curr. Chem., 1988, 149, 190.
3
4
5
6
C. W. Chen and H. W. Whitlock, J. Am. Chem. Soc., 1978, 100,
7921; K. M. Nedar and H. W. Whitlock, J. Am. Chem. Soc., 1990,
112, 7269.
32 F. Lanza and B. Sellergren, Anal. Chem., 1999, 71, 2092.
33 A. Kugimiya, Y. Kuwada and T. Takeuchi, J. Chromatogr., A,
2001, 938, 131.
34 J. Jodlbauer, N. M. Maier and W. Lindner, J. Chromatogr., A,
2002, 945, 45.
35 S. Ghose and S. M. Cramer, J. Chromatogr., A, 2001, 928, 13.
7
8
S. C. Zimmerman, Z. Zeng, W. Wu and D. E. Reichert, J. Am.
Chem. Soc., 1991, 113, 183; S. C. Zimmerman, W. Wu and Z. Zeng,
J. Am. Chem. Soc., 1991, 113, 196; S. C. Zimmerman, Bioorganic
Chemistry Frontiers, Vol. 2, Springer-Verlag, New York, 1991.
T. Braxmeier, M. Demarcus, T. Fessmann, S. McAteer and
766
J. Mater. Chem., 2003, 13, 758–766