Molecularly imprinted polymers in the drug discovery process

B

Daniel L. Rathbone *

School of Life and Health Sciences, Aston University, Birmingham B4 7ET, UK

Received 20 September 2004; accepted 29 July 2005

Available online 14 October 2005

Abstract

Since molecularly imprinted polymers (MIPs) are designed to have a memory for their molecular templates it is easy to draw

parallels with the affinity between biological receptors and their substrates. Could MIPs take the place of natural receptors in the
selection of potential drug molecules from synthetic compound libraries? To answer that question this review discusses the
results of MIP studies which attempt to emulate natural receptors. In addition the possible use of MIPs to guide a compound
library synthesis towards a desired biological activity is highlighted.
D 2005 Elsevier B.V. All rights reserved.

Keywords: Combinatorial chemistry; Molecular imprinting; Receptor mimics; Cross-reactivity; Multiple templates; Composite template

Contents

1.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1855

1.1.

A brief history of combinatorial chemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1855

1.2.

Desirable properties in an imprinted polymer for use in drug discovery . . . . . . . . . . . . . . . . .

1856

2.

Attempts to simulate the binding characteristics of biological receptors
by means of molecular imprinting. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1856

2.1.

Steroid receptor mimics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1856

2.2.

Folate receptor mimics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1857

2.3.

Adrenergic receptor binding mimics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1860

2.4.

Cinchona alkaloid receptor mimics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1864

2.5.

a2-Adrenoreceptor mimics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1865

2.6.

Attempts to make the template selection more relevant to drug discovery . . . . . . . . . . . . . . . .

1866

2.6.1.

The use of multiple templates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1867

2.6.2.

Design and implementation of composite templates . . . . . . . . . . . . . . . . . . . . . . .

1868

0169-409X/$ - see front matter

D 2005 Elsevier B.V. All rights reserved.

doi:10.1016/j.addr.2005.07.017

B

This review is part of the Advanced Drug Delivery Reviews theme issue on bMolecularly imprinted polymers: Technology and applicationsQ,

Vol. 57/12, 2005.

* Tel.: [phone]; fax: [phone].

E-mail address: [e-mail].

Advanced Drug Delivery Reviews 57 (2005) 1854 – 1874

www.elsevier.com/locate/addr

3.

The use of molecularly imprinted polymers to direct a synthesis to deliver
biologically active products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1868

4.

Concluding remarks: the relevance of MIPs to drug discovery . . . . . . . . . . . . . . . . . . . . . . . . .

1873

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1873

1. Introduction

Molecularly imprinted polymers (MIPs) have

potential to assist at various points in the drug dis-
covery process. At the time of writing this potential is
for the most part unrealised. This article will discuss
some of the areas relevant to drug discovery where the
properties of MIPs might be exploited in future drug
discovery programmes. Two areas will be considered:
the use of MIPs in uncovering biological activity in a
library of compounds (the biological activity can
mean the desired disease-curing activity or the toxicity
profiles); the use of MIPs to direct the chemical
synthesis towards the production of compounds of a
desired biological activity.

1.1. A brief history of combinatorial chemistry

The last two decades have seen a considerable

change in the approach taken by the pharmaceutical
industry in both the synthesis of new chemical entities
(NCEs) and in their screening. Formerly there existed
a balance between the synthesis and the screening
since both were essentially manual operations. The
requirement for a very high degree of purity and
extensive characterisation before a compound was
allowed to progress to biological evaluation effec-
tively constituted the rate-limiting step in compound
production. It is generally easy to create a crude
compound mixture but often most of the effort is
directed towards the isolation and purification of the
desired product. The advent of automated biological
screening protocols (high throughput screening) cre-
ated an imbalance in the production-assay equation.
This stimulated a more pragmatic approach to the
synthesis side since the crippling purity requirements
could no longer be imposed if enough compounds
were to be made to satisfy the voracious robotic
screening systems. Following on from the screening
side, automation began to be introduced into the
synthesis of NCEs

[1]

. This was assisted by the

sudden burgeoning in research into solid phase chem-
istry or polymer-supported synthesis

[2]

. The ability

to separate a polymer-bound reaction product from its
reaction mixture simply by filtration proved to be easy
to automate. This opened the door to automated multi-
step high throughput syntheses using a matrix of
reaction vessels on a numerical scale which could
match the demands of the screening systems.

At about the same time the concept of combinator-

ial chemistry was gaining ground. At its inception
combinatorial chemistry involved the production of
mixtures of compounds, sometimes containing thou-
sands of compounds in one reaction vessel

[3,4]

. It

was thought that time could be saved both on the
synthetic and the screening sides if a large number
of compounds (a compound library) could be made by
a condensed synthetic protocol and all screened as one
sample. Unfortunately this generally does not work.
There are problems in the quality assurance of a
complex mixture. For example, can one be sure that
each of the desired components of the mixture is
present at its target concentration or is indeed present
at all? If one were to screen such a mixture and no
biological activity was found, could the putative
library components really be dismissed from the dis-
covery programme? In addition, for many biological
assays it is simply not possible for, let us say, a single
sample comprising one thousand potential drug can-
didates, for any one active compound present therein
to be at the concentration necessary for it to exert its
biological effect. This means it is not generally useful
to screen complex mixtures of compounds. These
days it is necessary to produce large numbers of
compounds for screening but as discrete compounds
(automated parallel synthesis) rather than mixtures.
The term bcombinatorial chemistryQ has remained
but its meaning has expanded to include the high-
throughput parallel synthesis of discrete compounds.
This is usually achieved by means of automated sys-
tems managing arrays of reaction vessels. The purifi-
cation of large numbers of reaction products is still a

D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1855

problem and the purity requirement has been relaxed
relative

to

the

traditional

medicinal

chemistry

approach.

This constitutes the current situation in drug dis-

covery in which MIPs will have to make their mark.
Large numbers of compounds are being produced but
of moderate purity rather than very high purity. From
these it is necessary to identify the compounds which
have the necessary biological activity and pharmaceu-
tical properties and this has to be done quickly. In a
nutshell, the issues are mimicry of a biological system
(enzyme active site, receptor, etc.) and application in
high-throughput screening.

1.2. Desirable properties in an imprinted polymer for
use in drug discovery

How can MIPs help? To answer this question one

must consider the nature of MIPs. They are crosslinked
polymers containing bespoke functionalised cavities
arising from the inclusion of template molecules in
the polymerisation mixture and their later extraction.
Functional polymerisable monomers are included
which either associate with the templates in the pre-
polymerisation mixture (non-covalent imprinting) or
are temporarily covalently bound to the templates
(covalent imprinting). The functional monomers,
once they are part of the polymer matrix, serve to
decorate the cavities with functionality appropriate to
the template molecules. Overall, binding sites are cre-
ated which have a memory for the templates both in
terms of shape and matching functionality. It is tempt-
ing to say that the binding sites are specific for the
template molecules but absolute specificity is not what
is actually required for the use of MIPs in drug dis-
covery; rather it is appropriate cross-reactivity to
molecules of similar shape and functionality and not
necessarily from the same chemical class. Drug dis-
covery is mostly concerned with the uncovering of new
compounds which will bind and exert an effect at the
biological target but which are not the natural substrate.
To put it in medicinal chemistry terms, it is necessary
for the MIPs to be able to recognise molecules similar
in size and shape to the template and which also carry a
compatible pharmacophore.

The positive aspects of MIPs compared to their

biological counterparts include their chemical and
physical robustness, long shelf life, ease of prepara-

tion and resulting low cost relative to, for example,
antibodies. In some cases the target receptor is not
available in sufficient quantities for widespread
screening. Here, if the receptor structure is known
then a suitable template for imprinting may be con-
cocted and a MIP equivalent prepared. Sometimes the
structure of the putative target receptor is unknown
and has to be inferred from the structures of the
known agonists or antagonists. In such a situation it
might be possible to design and synthesise a template
molecule which is effectively a summation of the
known agonists thereby enabling the production of
plastic receptor mimics where no biological equiva-
lents are available.

This review will discuss examples of attempts to

prepare MIP-based receptor mimics from the stand-
point of their potential use in drug discovery both at
the post-synthetic compound library screening phase
and at the level of reagent selection during compound
library synthesis. Priority has been given to articles
which provide a wide enough range of test com-
pounds for cross-reactivity patterns to be discerned.

2. Attempts to simulate the binding characteristics
of biological receptors by means of molecular
imprinting

2.1. Steroid receptor mimics

Several accounts of MIP-based steroid mimics may

be found in the literature. Often it has been found that
these MIPs are able to bind preferentially steroids of a
particular class which in turn equates to the selection
of particular classes of biological activity.

In a study by Cheong et al.

[5]

a testosterone

receptor binding mimic was created by preparing a
series of MIPs templated with testosterone and using
ethylene glycol dimethacrylate as the crosslinker and
methacrylic acid as the functional monomer. Generally
the MIPs recognised their template whilst showing
some cross-reactivity towards four other related ster-
oids especially h-estradiol but also progesterone, tes-
tosterone propionate and estrone. The strength with
which the steroidal test compounds bind in the
imprinted sites is a measure of their molecular simi-
larity to the template testosterone in terms of size,
shape and of functional group disposition. In this

D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1856

particular study it was the presence or absence of the
17-hydroxyl (present in testosterone and h-estradiol
but not the others) which was dominant in determining
the affinity of the substrates for the receptor. It also was
noted that when propylene glycol dimethacrylate was
used in place of the ethylene glycol dimethacrylate as
the crosslinker the MIP characteristics changed signif-
icantly. This new MIP bound h-estradiol more strongly
than the template testosterone. In addition the elution
order of the steroids was changed.

In a related study, Piscopo et al.

[6]

created MIPs

templated against 17h-estradiol using ethylene glycol
dimethacrylate as the crosslinker but this time 4-vinyl-
pyridine as the functional monomer. The MIPs were
challenged

chromatographically

with

four

other

related estrogens (17a-estradiol, estrone, estriol and
17a-ethynylestradiol). Some selectivity towards tem-
plate was observed along with considerable cross-
reactivity towards the other estrogens.

In a variation by Tarbin and Sharman

[7]

a series of

polymers was imprinted with hexesterol, a non-ster-
oidal estrogenic compound. The retention index
values (see footnote to

Table 1

for definition) obtained

after the MIP columns were probed chromatographi-
cally with a series of estrogenic compounds indicated
that when acrylamide was used as the functional mono-
mer a relatively sharp selectivity was observed for the
non-steroidal estrogenic compounds most closely
related to the template (diethylstilbestrol and dienes-
trol) although some small cross reactivity was observed
towards some of the steroidal estrogenic compounds.
The distinction was less pronounced when 2-(diethy-
lamino)ethyl methacrylate was employed as the func-
tional monomer (

Table 1

).

In a wider ranging study Ye et al.

[8]

constructed

two MIPs against 17h-estradiol and dienestrol using
methacrylic acid as the functional monomer and ethy-
lene glycol dimethacrylate as the bulk crosslinking
agent. Both templates are estrogenic. The former is
steroidal, whereas the latter is non-steroidal. The two
MIPs were packed into HPLC columns and chal-
lenged chromatographically with the set of steroids
and related compounds shown in

Fig. 1

.

The capacity factors (kV) for the compounds on

each MIP column and control columns were deter-
mined and used to calculate the retention index (RI).
By definition the template RI is 100%. It can be seen
from the results in

Table 2

that both of the MIP

columns, each of which had been imprinted with
estrogenic substances, were able to recognise their
templates and also showed considerable cross-reactiv-
ity towards the other estrogenic compounds. It was
also stated that the anti-17h-estradiol MIP exhibited
smaller RI values for all of the non-estrogen steroids
than any of the estrogen compounds. Unfortunately
these latter data are not given so the reader cannot
make a judgment on how significant the effect is.
Notwithstanding this the study represents an example
of a MIP-based estrogen receptor which is able to
tease out a subset of estrogenic compounds from a
set comprising closely related steroids and some more
distantly related non-steroidal compounds. The prin-
ciple of using MIPs to screen for biologically active
compounds is thus established.

2.2. Folate receptor mimics

In an elegant report by the Sellergren group

[9]

substantial progress was made towards MIP receptors
for folic acid. Such receptors have obvious potential
medicinal chemistry uses in the screening of com-

Table 1
Retention indices for a set of estrogenic compounds

[7]

Test compound

Retention index

a

Acrylamide MIP

2-(Diethylamino)ethyl
methacrylate MIP

Hexestrol

1.00

1.00

Diethylstilbestrol

0.70

0.90

Dienestrol

0.51

0.70

Zeranol

0.12

0.49

a-Estradiol

0.11

0.31

h

-Estradiol

0.12

0.31

Estradiol-3-benzoate

0.08

Estradiol-17h-acetate

0.12

Ethynylestradiol-

3-monomethyl ether

0.08

Estrone

0.11

Genistein

0.12

Formononetin

0.09

Biochanin A

0.09

Daidzein

0.12

a

The RI represents the retention of the test substance relative to

the retention of the template and is calculated from the equation:

RI

¼ 100 kV

analyte

MIP

ð

Þ=kV

analyte

control

ð

Þ





=

kV

template

MIP

ð

Þ=k

template

control

ð

Þ





where kV is the chromatographic capacity factor.

D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1857

pound libraries for new inhibitors of the enzyme
dihydrofolate reductase (DHFR). The structures of
the compounds used in this study are given in

Fig.

2

. Methotrexate (MTX), with its close resemblance to

folic acid and being a very potent strong-binding
inhibitor of DHFR, was initially investigated as a
template for creating the folic acid receptor. This
served to illuminate the problems with imprinting
relatively large multifunctional biologically important
compounds. MTX, having two carboxyl functions and
two primary amino groups, is highly polar and
amphoteric. Following the usual imprinting approach
one would prefer to manipulate the template into a

hydrogen bonding based association with appropriate
functional monomers in a non-polar aprotic solvent.
This is difficult with MTX. The two ends of the
molecule have different needs. For instance the pter-
idine and tertiary aliphatic amine moieties may be best
targeted with acidic functional monomers whilst the
glutamic acid grouping may fare better with basic
functional monomers. To crown it all MTX is poorly
soluble in non-polar aprotic solvents. After initial MIP
trials with a range of functional monomers a folate
receptor was obtained but its chromatographical use-
fulness was limited. The density of imprinted sites
was low owing to the solubility-limiting concentration

Androstenone types

O

R

Estrogen types

O

H

R

11

α

-Hydroxyprogesterone

11 -Hydroxyprogesterone

17

α

-Hydroxyprogesterone

Progesterone

4-Androsten-3,17-dione

1,4-Androstadiene-3,17-dione

17 -Estradiol

17

α

-Estradiol

Estrone

17

α

-Ethynylestradiol

Estriol

Corticosterone

Cortexone

11-Deoxycortisol

Cortisone

Cortisone 21-acetate

Cortisol 21-acetate

Deoxycorticosterone

Cortisol

Additional estrogenic compounds

O

H

OH

O

H

OH

O

H

OH

Cl

O

O

O

OH

O

Diethylstilbestrol

Dienstrol

Hexestrol

Chlorotrianisene

Mestranol

β

β

Fig. 1. Test compounds applied to MIP-based estrogen receptors

[8]

.

D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1858

of MTX used as template. Rather than targeting the
whole of MTX it was found to be better to home in
one part, namely the pteridine moiety. Fortunately
there exists a series of lipophilic basic inhibitors of

microbial DHFR; trimethoprim (TMP), 2,4-diamino-
6,7-diisopropylpteridine

(DIP)

and

trimetrexate

(TRX). Since they have much higher solubility in
organic solvents compared with MTX these three
inhibitors were imprinted in dichloromethane using
methacrylic acid as functional monomer. All three
MIPs exhibited an imprinting effect for their templates
and for the folic acid analogues chromatographically,
eluting with MeCN/AcOH/H

2

O (92.5:5:2.5, v/v/v).

The MIP templated with DIP exhibited the poorest
selectivity. In contrast, when aqueous mobile phases
were employed, this MIP was found have a much
enhanced retention of folic acid and its analogues
Leu and MTX. It was suggested that the hydrophobic
isopropyl substituents of the template DIP might have
lead to the formation of a hydrophobic region in the
MIP cavity which gave rise to an additional hydro-
phobic driving force upon folate binding in water. In
other words the aqueous chromatography suggested
that bthe recognition was driven by a combination of

N

N

N

H

N

O

N

H

2

N

H

N

H

O

OH

O

OH

O

Folic acid

O

N

N

H

N

H

N

O

N

H

2

N

H

N

H

O

OH

O

OH

O

Leucovorine (Leu)

N

N

N

N

NH

2

N

H

2

N
CH

3

N

H

O

OH

O

OH

O

N

N

NH

2

N

H

2

O

O

O

Methotrexate (MTX)

Trimethoprim (TMP)

N

N

NH

2

N

H

2

N

H

O

O

O

N

N

N

N

NH

2

N

H

2

Trimetrexate (TRX)

2,4-diamino-6,7-diisopropylpteridine (DIP)

Fig. 2. Templates and test compounds used by Quaglia et al.

[9]

.

Table 2
Retention indices for a set of estrogenic compounds

[8]

MIP templated with
17h-estradiol

MIP templated with
dienestrol

Retention index (%)

Retention index (%)

Progesterone

29

57

17h-Estradiol

100

56

17a-Estradiol

39

55

Estrone

31

56

17a-Ethynylestradiol

58

55

Estriol

42

59

Diethylstilbestrol

23

78

Dienestrol

31

100

Hexestrol

31

81

Chlorotrianisene

31

59

Mestranol

33

59

D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1859

specific hydrophobic and ionic forces.Q This is an
important statement when one considers what a
mimic of a biological receptor acting in water must
necessarily attempt to simulate: hydrophilic interac-
tions (ionic, acid–base, hydrogen bonding) and hydro-
phobic interactions. At first sight it may seem
surprising that the relatively small template, TMP,
could create a MIP cavity capable of accommodating
folic acid or Leu or MTX. It is, however, to be born in
mind that in this case both the template and the test
compounds have significant flexibility. In the former
case the trimethoxy benzyl portion with its two major
rotatable bonds is probably able to carve out a sig-
nificant void during MIP formation. In the latter case,
the test compounds have many rotatable bonds and
may have many low energy conformations available
which would allow their insertion into the imprinted
sites without undue bond strain or steric clashes. It is
difficult, however, to extend this argument to the MIP
derived from the small and inflexible template DIP.
Overall, this study successfully demonstrated the uti-
lity of engaging a substructure of a complex target
molecule where the target itself proves to be unco-
operative in the imprinting process. This is reminis-
cent of the epitope approach for the imprinting of
peptides where the peptide will be recognised by a
polymer imprinted with a relatively short section of a
terminal sequence

[10]

. Working with a substructure

of the main target obviated the use of awkward
amphoteric templates and allowed the use of more
tractable organic-soluble templates.

2.3. Adrenergic receptor binding mimics

The binding sites found in h-adrenergic receptors

of the autonomous nervous system comprise seven
transmembrane helices arranged around a hydropho-
bic core

[11]

. There is a conserved aspartic acid

residue which is thought to form ionic bonds with
the basic amino function of the ligands whilst the
hydrophobic core interacts with the aromatic part of
the ligands by means of a suitably placed phenylala-
nine residue. The ligands are based upon a h-phe-
nethylamine skeleton containing an essential hydroxy
group at the h-position. The known antagonists are
generally more hydrophobic and have an increased
spacing between the aromatic part and the amino
group. Ramstrom et al.

[12]

have attempted to use

molecular imprinting to construct polymers capable of
mimicking the binding characteristics of h-adrenergic
receptors. Structures of the agonists and antagonists to
the receptor used in their study are given in

Fig. 3

.

The obvious choice for the template in the creation

of the artificial receptors would have been the endo-
genous ligand epinephrine. Unfortunately its low
solubility in reasonable imprinting solvents meant
that the closely related and more soluble compounds,
(1S,2R)-(+)-ephedrine and (1S,2S)-(+)-pseudoephe-
drine, had to be used instead. In order to encourage
saturation of the ligand in the pre-polymerisation
mixture a functional monomer to template ratio of
4:1 was employed. Photolytically initiated polymer-
isation in the presence of trimethylolpropane trimetha-
crylate delivered the MIPs containing, it was hoped,
functionalised cavities as shown in

Fig. 4

. It was

further hoped that the particular ionic attraction
between the carboxyl derived from the methacrylic
acid and the templates’ amino function, together with
the remainder of the template resting in an otherwise
hydrophobic cavity, would simulate the known bind-
ing features of the target adrenergic receptors.

The two imprinted polymers were probed chroma-

tographically. The capacity factors (kV) for the various
test compounds are given in

Table 3

.

Each polymer recognised its own template and was

able to discriminate against the enantiomers of their
own templates but exhibited poorer enantiodiscrimi-
nation between diastereoisomers of the template
(

Table 3

). From a consideration of the capacity factors

it was discovered that the h-carbon of ephedrine is the
more important of the two chiral centres for determin-
ing the chiral recognition since the 1-S compounds
were more strongly retained than the 1-R isomers. The
much higher capacity factors observed for the isomers
of epinephrine were rationalised on the basis that the
catechol moiety present resulted in an increased non-
specific binding to the polymers. The relatively high
enantioselectivity of both of the MIPs towards the
isomers of the endogenous ligand epinephrine corre-
late well with the natural receptor and goes some way
towards validating these MIPs as potential adrenergic
receptor mimics. Upon moving further away from the
structures of the templates to those of the h-adrenergic
receptor antagonists, with their increased distance
between the amino function and the aromatic portion,
the enantiodiscriminatory powers of the MIPs dec-

D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1860

reased. In the case of timolol, where there is no enantio-
discrimination, the aromatic portion is a polar five-mem-
bered ring heterocycle which bears little resemblance to
the aryl portions of the templates in both shape and
polarity.

Taking a different approach to the format of the

imprinted polymers, Piletsky et al.

[13]

were able to

create imprinted polymers using the endogenous sub-

strate epinephrine as template. An aqueous solution
containing 3-aminophenylboronic acid and epinephr-
ine was placed in microtitre plate wells and polymer-
isation was initiated with ammonium persulphate.
This resulted in the deposition of an imprinted poly-
meric layer on the well surfaces. After extraction of
the template (epinephrine) the polymers were chal-
lenged with a conjugate of horseradish peroxidase

N

H

R

1

R

3

R

4

R

5

R

2

R

6

β

-Adrenergic receptor agonists

R

1

R

2

R

3

R

4

R

5

R

6

(1S,2R)-(+)-Ephedrine

(1R,2S)-(-)-Ephedrine

(1S,2S)-(+)-Pseudoephedrine

(1R,2R)-(-)-Pseudoephedrine

(S)-(+)-Halostachine

(R)-(-)-Halostachine

(S)-(+)-Epinephrine

(R)-(-)-Epinephrine

β

-Adrenergic receptor antagonists

OH

N

H

O

OH

OH

H

H

H

H

H

H

H

H

OH

OH

H

H

H

H

H

H

OH

OH

H

H

H

H

H

CH

3

CH

3

CH

3

CH

3

H

H

H

H

H

H

H

H

H

OH

OH

OH

OH

OH

H

OH

H

OH

H

H

O

N

H

O

Propranolol

Metoprolol

N

H

2

O

OH

N

H

O

O

N

N

N

OH

N

H

O

S

Atenolol

Timolol

Fig. 3. h-Adrenergic receptor agonists and antagonists used by Ramstrom et al.

[12]

.

D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1861

and norepinephrine. The norepinephrine, conjugated
via its primary amino function was able to interact
with the polymer by presentation of that part of itself
which was identical to the template epinephrine,
namely the catechol and the h-hydroxyethyl portion.
As mentioned in an earlier section, this is essentially
the epitope approach wherein a large molecule is
recognised by a polymer which has been imprinted
with only a small part of the whole target molecule

[10]

. Addition of 2,2V-azino-bis(3-ethyl)benzthiazo-

line-6-sulfonic acid and hydrogen peroxide generated
a colour change in solution, quantifiable at 450 nm
and proportional to the amount of horseradish perox-
idase bound to the polymer. The quantification of the

binding of the horseradish peroxidase–norepinephrine
conjugate formed the basis for competitive binding
experiments to establish the affinity of the polymers
towards the various h-adrenergic receptor agonists
and antagonists (structures given in

Fig. 5

). The

fact of imprinting was supported by the observation
that the imprinted polymers had much higher affinity
for the horseradish peroxidase–norepinephrine conju-
gate than untemplated control polymers. The polymer
imprinted with epinephrine exhibited strong affinity
for its template (K

D

9.2 AM and an affinity factor of

16.3 relative to blank polymer). Small but discern-
able cross-reactivities were observed for the related
ligands isoproterenol and norepinephrine as judged

O

O

H

H

2

N

O

H

O

O

+ Crosslinker

Polymerisation

Extraction of template

Rebinding

O

O

H

H

2

N

O

H

O

O

OH

O

H

O

O

+

+

Fig. 4. Formation of imprinted sites mimicking adrenergic receptors (adapted from Ramstrom et al.

[12]

).

Table 3
Chromatographic data adapted from Ramstrom et al.

[12]

Ligand

(S,R)-Ephedrine MIP

(S,S)-Pseudoephedrine MIP

kV

Enantio-discrimination

a

kV

Enantio-discrimination

(S,R)-Ephedrine

8.84 (temp.)

3.41

1.62

1

(R,S)-Ephedrine

2.59

1.62

(S,S)-Pseudoephedrine

2.29

1.14

8.82 (temp.)

3.42

(R,R)-Pseudoephedrine

2.01

2.58

(S)-Halostachine

7.30

1.93

7.44

2.12

(R)-Halostachine

3.79

3.51

(S)-Epinephrine

25.9

1.60

28.1

1.76

(R)-Epinephrine

16.2

16.0

(S)-Propranolol

2.67

1.31

3.22

1.48

(R)-Propranolol

3.50

4.76

(S)-Metoprolol

2.35

1.28

2.08

1.51

(R)-Metoprolol

3.02

3.14

(S)-Atenolol

11.6

1.16

12.3

1.29

(R)-Atenolol

13.5

15.9

(S)-Timolol

1.27

1

1.12

1.20

(R)-Timolol

1.27

1.34

a

Ratio of the capacity factors for the two enantiomers. A value of 1 indicates no chiral discrimination.

D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1862

by calculated similarity indices (Selected data given
in

Table 4

). In addition there was essentially no

recognition of catechol or phenylephrine. A polymer
templated with phenylephrine exhibited higher dis-
sociation constants across the board and also a more
catholic cross-reactivity profile towards epinephrine,
isoproterenol and norepinephrine. Challenging the
epinephrine imprinted polymer with the antagonists
propranolol, clenbuterol and pindolol revealed much

higher dissociation constants (157, 300, 464 AM
respectively) which were essentially the same as for
the non-imprinted polymer. The major difference
between the natural ligands and the antagonists is
the lack of a catechol or phenol in the latter. Thus,
although it can be said that the imprinted polymer
was able to distinguish between agonists and antago-
nists, it would not be of any use in a drug screening
programme aimed at discovering more antagonists. It

N

H

O

H

O

H

OH

NH

2

O

H

O

H

OH

N

O

H

O

H

OH

Epinephrine Norepinephrine

Isoproterenol

N

H

O

H

OH

O

H

O

H

N

H

2

Cl

N

H

OH

Phenylephedrine Catechol

Clenbuterol

N

H

N

H

O

OH

N

H

O

OH

Pindolol Propranolol

Fig. 5. h-Adrenergic receptor agonists and antagonists used by Piletsky et al.

[13]

.

Table 4
Selected dissociation constant data for imprinted polymers, blank polymers, and h-adrenergic receptor agonists, arising from a competitive
assay at pH 6.0

[13]

MIP templated with

Dissociation constants [AM] (Affinity factor

a

, Similarity index

b

)

(

)-Epinephrine

(

)-Isoproterenol

(

)-Norepinephrine

(

)-Phenylephrine

Epinephrine

9.2 (16.3, 1)

21 (1.29, 0.08)

46 (1.96, 0.13)

N1000 (b0.1,b0.01)

Phenylephrine

46 (3.62, 0.91)

35 (0.77, 0.42)

107 (0.84, 0.21)

25 (4.00, 1)

Blank

150

27

90

100

a

Affinity factor defined here as K

D

(blank) / K

D

(imprinted).

b

Similarity index defined here as Affinity factor (test compound)/Affinity factor (template).

D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1863

is, however, encouraging that a selective imprinted
polymer was constructed in a format which is far
more amenable to a high throughput screening pro-
tocol than any HPLC-based approach reported to
date.

Allender et al.

[14]

prepared a MIP templated with

(R, S)-ephedrine using methacrylic acid as the func-
tional monomer and ethyleneglycol dimethacrylate as
the crosslinking agent. This was probed chromatogra-
phically with a set of h-adrenergic receptor agonists
and antagonists. The template was retained well and
could easily be separated from its enantiomer. Sig-
nificant cross-reactivity was observed towards the
antagonists R- and S-propranolol and naldolol as
judged by the normalised retention indices (

Table

5

). This indicates that this approach may have a use

in screening for such activity within a compound
library.

2.4. Cinchona alkaloid receptor mimics

Matsui et al.

[15]

have imprinted two antimalarial

alkaloids (

)-cinchonidine and (+)-cinchonine using

methacrylic acid as the functional monomer and ethy-
leneglycol dimethacrylate as the crosslinker. For the
template structures and those of other compounds
used in this study see

Fig. 6

.

The two MIPs had a pronounced selectivity for

their templates and also exhibited high stereosepara-
tion factors. It was postulated that the rigidity of the
templates was responsible for the fidelity of the
imprinted sites obtained. The normalised retention
indices (RI) in

Table 6

may be viewed as the MIPs’

assessment of the molecular similarity between the
templates and the other test compounds. In moving
from the templates to quinine and quinidine there are
two potential changes, namely the introduction of a
methoxy substituent and the inversion of two chiral
centres. When one of these changes is made the RI

Table 5
Chromatographic data from an ephedrine imprinted MIP

[14]

kV MIP

kV Blank

Imprinting
factor

Retention
index

(1R,2S)-Ephedrine

25.7

0.10

257

1

(2R,1S)-Ephedrine

3.01

0.11

27.4

0.11

Salbutamol

3.57

0.24

14.9

0.06

Metaprolol

3.65

0.17

21.5

0.08

(R)-Propranolol

3.57

0.11

32.5

0.12

(S)-Propranolol

5.28

0.11

48.0

0.19

Nadolol

3.61

0.09

40.1

0.16

Labetolol

2.91

0.17

17.1

0.07

Adrenaline

3.34

0.49

6.94

0.03

Noradrenaline

3.08

0.24

12.8

0.05

Pseudoephedrine

4.03

0.24

16.8

0.07

Tyramine

0.68

0.06

11.3

0.04

N

N

O

H

H

H

R

H

N

N

O

H

H

R

H

N

H

O

O

H

R = H; (-)-Cinchonidine

R = CH

3

O; Quinine

R = H; (+)-Cinchonine

R = CH

3

O; Quinidine

Propranolol

Fig. 6. Alkaloid structures used by Matsui et al.

[15]

.

Table 6
Chromatographic capacity factors, calculated imprinting factors and
retention indices from Matsui et al.

[15]

Test compound

MIP imprinted with

(

)-Cinchonidine

(+)-Cinchonine

Blank

kV

K

RI

kV

K

RI

kV

(

)-Cinchonidine 34.24 118.1 1

1.46

5.0 0.05 0.29

(+)-Cinchonine

1.08

16.6

0.14

33.65 99.0 1

0.34

Quinine

5.35

15.7

0.13

0.97

2.9 0.03 0.34

Quinidine

1.55

9.1

0.08

4.70 27.6 0.28 0.17

kV: Chromatographic capacity factor.
K: Imprinting factor (kV(imprinted) / kV(blank)).
RI: Retention index (K(analyte) / K(template)).

D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1864

drops to 0.14 or 0.34 and when both alterations are
implemented these values are halved. In other words,
the closer the structure of the test compound to that of
the template the more strongly it will interact with the
MIP. It was also demonstrated that it was possible to
separate partially the enantiomers of the h-adrenergic
antagonist, propranolol. An inspection of its structure
reveals some molecular similarities to the templates,
namely, a bicyclic aromatic part connected to a h-
hydroxy ethylamine thus further highlighting the pos-
sibilities for using MIPs for detection of molecular
similarity or pharmacophore patterns.

2.5. a

2

-Adrenoreceptor mimics

a

2

-Adrenoreceptors represent a situation where it is

difficult to obtain sufficiently pure material for con-
ducting receptor–ligand binding studies upon combi-

natorial or high-throughput matrix compound libraries.
Here it would be a great advantage to use MIPs to
simulate these elusive receptors. In a key paper by
Berglund et al.

[16]

monolithic MIPs were prepared

using the a

2

-adrenergic antagonist yohimbine or cor-

ynanthine as templates (

Fig. 7

), methacrylic acid as the

functional monomer and ethylene glycol dimethacry-
late as the crosslinker.

A phage display library was used to probe the MIP

cavities with a view to identifying particular short
peptide sequences which could display the same phar-
macophore pattern as yohimbine. A schematic repre-
sentation of this approach is given in

Scheme 1

. The

particular phages used expressed a randomised hexa-
peptide library externally which could potentially bind
to the imprinted cavities of the MIPs. This operation
was performed and the unbound phages were washed
away. The bound phages from the primary library were
then used to infect Escherichia coli in order to produce
a second phage library which would be enriched with
hexapeptides having some selectivity towards the
MIPs. This selection procedure was repeated five
times with the bound phages from one selection
round being used as the input for the following selec-
tion round. The final phage library produced was
found to have an overall three-fold higher affinity for
the yohimbine imprinted polymer compared with the
primary library. Unfortunately the same phage affinity

H

OH

N

N

H

H

H

H

3

COOC

H

OH

N

N

H

H

H

H

3

COOC

Yohimbine

Corynanthine

Fig. 7. Structures of the templates used by Berglund et al.

[16]

.

MIP

Library of bacteriophages with randomly
expressed external hexapeptides

MIP

Selective binding of bacteriophages by
interaction of their hexapeptides with
the imprinted MIP cavities

Scheme 1. Phage display library selection using a MIP as the selector.

D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1865

was observed for the blank polymer and similar results
were obtained for a MIP templated with corynanthine.
This indicated that the phages had been selected for a
structure type common to both the imprinted polymers
and the blank polymer rather than for the imprinted
cavities. The dominant hexapeptide arising from
phages with high affinity for the yohimbine MIP was
the hydrophilic, positively charged sequence Lys–
Lys–His–Val–His–Arg. It was postulated that the affi-
nity arose from ionic interactions with the negatively
charged carboxylic acid groups distributed throughout
the polymer network. Two possibilities stand out for
potential improvement on these results. Firstly, there is
the question of adequate steric access to the imprinted
sites by the peptide chains which are themselves
attached to a truly vast (on a molecular scale) payload,
namely the bacteriophage. Perhaps MIPs in a surface

mode thin film format (discussed elsewhere in this
special edition) would deliver the steric access
required in this case. Secondly, in retrospect, perhaps
it would have been better to choose a non-ionisable
functional monomer to create the yohimbine imprinted
polymer. This would have obviated the ionic-based
non-(MIP)-specific affinity. It is obvious from these
preliminary results that the potential exists with this
approach to perform truly combinatorial selection of
potential drug candidates.

2.6. Attempts to make the template selection more
relevant to drug discovery

The relevance of compounds selected by a MIP

based screening approach is entirely dependant upon
the template molecule chosen. An existing drug or

Table 7
Chromatographic data reported by Bowman et al.

[17]

Blank

First eluted stereoisomer

Second eluted stereoisomer

kV

kV

K

RI

kV

K

RI

Timolol [Template 1]

0.34

13.8

40.6

1.7

Pindolol

0.4

7.7

19.3

0.81

10.15

25.4

1.06

Atenolol [Template 2]

0.48

8.64

18

0.75

Propranolol [Template 3]

0.42

5.54

13.2

0.55

7.61

18.1

0.77

Naldolol

0.44

5.39

12.3

0.51

9.33

21.2

0.89

Adrenaline

0.48

5.39

11.2

0.47

Oxprenolol

0.32

3.51

11

0.46

4.63

14.5

0.61

Ephedrine

0.31

3.3

10.6

0.45

Atropine

0.19

0.93

4.9

0.2

Chlorpheniramine

0.23

1.12

4.9

0.2

Cinchonidine

1.01

4.27

4.2

0.18

Promethazine

0.32

1.29

4

0.17

Cinchonine

0.97

3.74

3.9

0.16

Metropolol

0.39

1.45

3.7

0.16

Proguanil

0.38

1.23

3.2

0.14

Quinidine

0.9

2.56

2.8

0.12

Quinine

0.88

2.52

2.9

0.12

Acebutolol

0.36

0.94

2.6

0.11

Labetalol

0.59

1.45

2.5

0.1

Boc-Tyrosine

0.17

0.2

1.2

0.05

Hydrocortesone acetate

0.42

0.37

0.9

0.04

Indapamide

0.28

0.25

0.9

0.04

Ketoprofen

0.21

0.19

0.9

0.04

Theophylline

0.4

0.38

1

0.04

Ibuprophen

0.19

0.13

0.7

0.03

Boc-Tryptophan

0.23

0.15

0.7

0.03

kV: Chromatographic capacity factor.
K: Imprinting factor (kV(imprinted) / kV(blank)).
RI: Retention index (K(analyte) / K(mean for the three templates)).

D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1866

known substrate/antagonist used as a template may fill
only part of the receptor to which it binds. Indeed, it
may be sub-optimal in its binding characteristics.
Compounds selected from a library by the resulting
MIP are likely to be similarly sub-optimal. Thus for
MIPs to have a significant future in drug discovery a
great deal more effort will have to be put into template
design. Two reports are discussed in Section 2.6.1
where several known substrates for a given receptor
have been used as templates in an attempt to develop a
more complete MIP-based receptor-binding mimic.
As a step further along this line of thinking an exam-
ple is discussed in Section 2.6.2

where composite

templates, designed to represent as much of the recep-

tor as possible

in a single molecule, have been

designed, synthesised and incorporated into a series
of MIPs.

2.6.1. The use of multiple templates

Bowman et al.

[17]

reported a MIP based h-

adrenergic receptor mimic composed from three
separate MIPs templated with three high-affinity
ligands: F-atenolol, F-propranolol and S-timolol.
An HPLC column was charged with a mixture of
the three MIPs and probed with a range of h-block-
ers. This hybrid MIP column was selective for the
individual templates and also retained naldolol and
pindolol more strongly than two of the templates

N

N

N

N

N

O

O

NH

2

NH

N

N

N

O

N

O

O

O

N

S

Cl

N

OH

O

O

N

OH

O

H

H

H

N

H

N

O

1

2

3

Fluorescent monomer

4

Verapamil

Chlorpromazine

Ibuprofen

Doxepin

Nortriptyline

Codeine

Fig. 8. Structures of the fluorescent functional monomer, the templates and the test compounds used by Rathbone et al.

[19]

.

D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1867

atenolol and propranolol. It would appear that mix-
ing the individual distinct MIPs resulted in a blend-
ing of their properties. The column was then
explored with a larger set of 25 structurally diverse
compounds. The results are presented in

Table 7

along with retention indices (calculated here by tak-
ing the ratio of the imprinting factor of the analyte
and the mean of the imprinting factors of the three
templates). There was high selectivity for several
compounds having a motif common to the h-blocker
templates, namely the h-aminoalcohol moiety thus
indicating that such hybrid column could be used to
select for molecules with a particular biological
activity amongst a general set of compounds.

In a slight variation and using a much smaller

dataset, Sreenivasan

[18]

constructed an imprinted

polymer wherein three templates were used simulta-
neously to generate the hybrid MIP directly. That
polymer was shown to be able to absorb all three of
the templates used in its construction (cholesterol,
testosterone, and hydrocortisone).

2.6.2. Design and implementation of composite
templates

Rathbone et al.

[19]

have reported the design and

implementation of a set of composite templates in a
quest to derive a MIP-based receptor mimic for the
cytochrome P450 isoform CYP2D6. The X-ray crystal
coordinates for this enzyme have yet to be reported.
Nevertheless, on the basis of a pre-existing in silico
model, several of template molecules were designed
which were intended to encompass most of the com-
bined volume of the known substrates of the enzyme.
The substrates usually have at least one basic nitrogen
atom which interacts with the carboxylic acid present
in the side chain of Asp301 or Glu216. The synthetic
templates for this study (compounds 1–3,

Fig. 8

) thus

also incorporated a basic nitrogen atom. Codeine was
also chosen as a template since it is a large and virtually
rigid member of the known CYP2D6 substrate set. The
functional monomer incorporated a fluorphore such
that the resulting MIPs exhibited binding-sensitive
fluorescence and entry of a test compound caused
fluorescence quenching. These observations were qua-
lified by investigations into the relative quenching
abilities of the test compounds towards the fluorophore
in an unrestricted non-imprinted environment. This
allowed an informed judgment on whether or not a

given test compound had been recognised by the
imprinted polymer.

The polymers were exposed to a panel of drugs,

some members of which are known to bind to
CYP2D6 and some which are bound by other iso-
forms. The results are given in

Table 8

.

The MIP templated with compound 1 was unable

to accept any of the drug test compounds. In contrast,
the MIP derived from the largest of the templates,
compound 3, was indiscriminate in its acceptance of
the test compounds. The polymer templated with
codeine recognised half of the test drugs but was
not specific to CYP2D6 substrates. The polymer
imprinted with compound 2, however, appears to be
the most promising as a starting point for further
progress towards a binding mimic for CYP2D6. It
accepted the CYP2D6 substrates/inhibitors nortripty-
line, doxepine, chlorpromazine and codeine but
rejected the CYP3A4 substrate verapamil. Ibuprofen,
however, a CYP2C9 substrate, was accepted. Overall
template 1 is too small, 2 is approximately the correct
size and shape, 3 is too large and codeine results in the
wrong recognition profile.

3. The use of molecularly imprinted polymers
to direct a synthesis to deliver biologically active
products

Perhaps the most exciting recent development in

the field of molecularly imprinted polymers has been

Table 8
Interpretation of the MIP fluorescence quenching observations; (+)
test substance recognised by the MIP; (

) test substance not recog-

nised by the MIP

Test substance

Imprinted polymers
templated with:

1

2

3

Codeine

1

+

+

+



2

+

+

+

+

3

+



+

+

4

+



+



Nortriptyline

CYP2D6 substrate



+

+



Doxepin

CYP2D6 substrate



+

+



Chlorpromazine

CYP2D6 inhibitor



+

+

+

Codeine

CYP2D6 substrate



+

+

+

Verapamil

CYP3A4 substrate





+

+

Ibuprofen

CYP2C9 substrate



+

+



D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1868

christened banti-idiotypic imprinting.Q Two papers
from the Mosbach group report the use of imprinted
polymers as microreactors

[20,21]

. The principle of

this approach is portrayed in

Scheme 2

.

As with all forms of molecular imprinting, anti-

idiotypic imprinting relies upon the availability of a
suitable template in order to generate MIPs appro-
priate to the targeted biological receptor. In addition,
this approach requires suitable chemistry and chemi-
cal building blocks for use inside the MIP cavity. If
the MIP is a reasonable approximation of the biolo-
gical receptor then products formed inside the MIP
cavity will have shape and disposition of function-
ality which may enable the products to bind with,
and even inhibit, the receptor. In essence this repre-
sents the possibility of undertaking breceptor-guided
synthesisQ of potential drug candidates. Thus, a MIP

was prepared against template 1, a known inhibitor
of the enzyme kallikrein (

Fig. 9

). The template was

removed and replaced by the related reactive build-
ing block (2-(4-amidinophenylamino)-4,6-dichloro-s-
triazine). This compound reacts with nucleophilic
amines to give the aminotriazine products as shown
generically in

Scheme 3

. Exposure of the pre-loaded

MIP to an excess of phenethylamine gave the corre-
sponding product at a rate four times higher than
when a control polymer was used. In addition, no
product was observed when the reaction was
attempted in solution at the same concentration.
The pre-loaded MIP was also allowed to interact
individually with three other related amines. The
product yields relative to the template forming reac-
tion are given in

Table 9

. The template forming

reaction was clearly the best. The low yield of

Y

B2

Y

B1

Y

B4

Y

B3

X

Template

+
Functional monomers
+
Crosslinkers

Polymerisation

Core building block
loaded into the
polymer cavities

Exposure to building blocks B1 - Bn

Dissociation of product

+

Product

MIP

Template

Bulding
block A

Template
extraction

X and Y are complementary reactive functionalities.

Scheme 2. The process of anti-idiotypic imprinting.

D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1869

product 2 was attributed to steric discrimination by
the polymer against the corresponding amine nucleo-
phile. It may also be true that the hydrophobic
portion of the MIP cavity associated with the phe-
nethylamine part of the imprinting template 1 bound
less strongly to the more hydrophilic phenolic por-
tion of tyramine compared with the more hydropho-
bic phenethylamine. Product 3, when tested against
the enzyme, turned out to have an inhibitory activity
approaching that of the template molecule. The
absence of product 4 was ascribed to steric hindrance
in the nucleophile by the carboxymethyl ester, in
other words, discrimination by the MIP. When the
larger template 2 was used to create the MIP it was
then possible to form product 4 as well as the
smaller products 1 and 3 owing to the larger cavity
size. Template 2 is chiral and this imparted a weak
stereoselectivity to the MIP-facilitated reaction. The

template (product 5) was formed preferentially over
its enantiomer (product 6) equivalent to an enantio-
meric excess of 26% (100  difference between
enantiomers / sum of enantiomers).

Anti-idiotypic imprinting has the potential for use

in bcombinatorially selective synthesis.Q After control-
ling for the relative reactivities of the reagents, a MIP
loaded with one building block could be presented
with a large number of reagents in the same reaction
vessel. The subset reagents able to enter the MIP
cavities would lead to preferentially to the production
of those products which would have a good chance of
adopting the same shape, size and pharmacophore
presentation as the template used to create the MIP.
This would be especially useful in the case where the
reagents represent molecular components with a rela-
tively large conformational flexibility. Compounds
with several rotatable bonds have many millions of

N

H

N

N

N

N

H

N

H

N

H

2

Cl

O

O

N

H

N

N

N

N

H

N

H

Cl

Template 2

MIP2

N

N

N

N

H

N

H

N

H

2

N

H

2

Cl

Cl

Reactive building block

→

Template 1

MIP1

→

Fig. 9. Templates and core building block used in anti-idiotypic imprinting

[20,21]

.

N

N

N

N

H

N

H

N

H

2

Cl

Cl

N

H

2

R

N

H

N

N

N

N

H

N

H

N

H

2

Cl

R

+

Reactive building block

Scheme 3. Aminotriazine formation.

D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1870

equivalent low energy conformations available to
them. At present it is difficult and time consuming
to predict computationally which flexible compounds
may adopt low energy conformations suitable for
gaining access to a given receptor. The use of MIPs
either in the anti-idiotypic mode or the traditional
sensing mode is a kind of bphysical computationQ
wherein the test compounds or reagents are able to
explore their full conformational potential. An exam-
ple of relatively large flexible substrates gaining
access to surprisingly small MIP cavities can be

seen in the work of Rathbone and Ge

[22]

. The

structures of the template, fluorescent functional
monomer and test compounds are given in

Fig. 10

.

The template and test compounds are all N

1

-benzyli-

dene pyridine-2-carboxamidrazones, the variation
being found in the arylidene part. The arylidene por-
tion of the template comprised the inflexible 1-
naphthyl moiety. Owing to the presence of the fluor-
ophore from the functional monomer the binding
events were registered by quenching of the fluores-
cence of the polymer. The resulting MIP was able to

Table 9
Normalised product yields and associated kallikrein inhibition data from compounds arising from MIP assisted synthesis

[20,21]

Product yield relative to template
forming reaction

Amine nucleophile

Product
number

MIP1

MIP2

Kallikrein
inhibition

K

i

(

µ

M)

NH

2

1

100%
(Template)

114%

4.5

NH

2

HO

2

21%

40

NH

2

3

34%

133%

5.2

O

O

NH

2

4

0%

33

O

O

NH

2

5

100%

(Template)

32.7

O

O

NH

2

HO

6

59%

47.1

D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1871

discriminate against the larger inflexible test com-
pounds containing anthryl, phenanthryl, substituted
naphthyl or a simple tert-butyl phenyl groups. In

contrast, the more flexible alkoxy-substituted benzy-
lidene compounds were able to enter the MIP cavities
and quench the fluorophore present therein.

Template

N

N

NH

2

N

Fluorescent

functional

monomer

N

H

N

O

O

S

S

O

Test compounds with inflexible arylidene moieties

N

N

NH

2

N

N

N

NH

2

N

N

N

NH

2

N

N

N

NH

2

N

O

N

N

NH

2

N

O

N

N

NH

2

N

Test compounds with flexible arylidene moieties

N

N

NH

2

N

O

N

N

NH

2

N

O

O

N

N

NH

2

N

O

O

N

N

NH

2

N

O

O

Fig. 10. Template, flexible substrates and inflexible substrates used by Rathbone and Ge

[22]

.

D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1872

4. Concluding remarks: the relevance of MIPs to
drug discovery

Nearly all of the examples discussed in this review

use chromatography, in particular HPLC, to probe the
behaviour of the imprinted polymer. This approach,
however, is simply too slow for the routine screening
of hundreds of thousands of NCEs for their biological
activity. If it is left to HPLC, MIPs will never achieve
their potential in drug discovery. Either the productivity
of the HPLC experiment must be increased by several
orders of magnitude or the MIP–guest interaction must
be registered by some other means which is more
amenable to large scale automation. It may be possible
to use MIP chromatography coupled to mass spectro-
metry to perform multiple simultaneous assays, provid-
ing, of course, that care is taken to organise the sample
mixtures such that each component has a unique mole-
cular mass. One example was discussed

[14]

where the

MIP was presented as a thin film in a microtitre plate.
All the manipulations required are easily automated
and such technology is the backbone of high through-
put screening. It is my opinion that the techniques
which will enable the uptake of MIPs in drug discovery
will involve the immobilisation of the MIP, perhaps as
thin layers in a multi-well format, together with a
molecular interrogation which is quantified by a spec-
trophotometric or spectrofluorimetric measurement.

Examples have been presented in the areas of ster-

oid, folate, alkaloid, adrenergic and cytochrome P450
receptor mimicry. Often the MIPs have exhibited
appropriate cross-reactivity in chromatographic mode
such that subsets of compounds with molecular simi-
larity to the template may be highlighted within a larger
set of test compounds. The data sets, however, are
relatively small. Further proof of concept is required
where MIP-based receptor mimics are challenged with
much larger compound libraries covering a wider diver-
sity space. Furthermore, it would be useful to know
how well MIP-based screening assays will perform
when presented with the type of crude products gener-
ated in high throughput syntheses. The vast majority of
MIP studies to date use pure compounds as their test
substances. In contrast, products from compound
library syntheses are likely to contain impurities closely
related to the structures of the target compounds and
therefore likely to contain the same key recognition
fragments which would be used to construct the MIP.

In order for MIPs to function with a greater adher-

ence to the biological receptor profile further strides
forward are needed in the areas of functional monomer
and template design. When one considers the complex-
ity of natural receptors it is obvious that current MIP
design is far behind nature in the matters of subtlety,
complexity and homogeneity. Traditional MIP design
mostly focuses upon polar ionic or hydrogen bonding
attractive forces between the template and functional
monomer. It is unusual for a MIP to have built into it
explicitly both specific polar interactions and specific
hydrophobic interactions; mostly one, rarely both. If a
hydrogen bonding functional monomer is being used it
would appear that generally the hydrophobic interac-
tions are optimistically left to look after themselves in
the form of the ill-defined brest of the polymer.Q In this
context there is great scope for the use of functional
monomers of more complexity than can simply be
bought from the mainline chemical suppliers; for
example, bespoke functional monomers tailored to a
particular template and capable of the full range of
biomimetic molecular interactions. It may be that mole-
cular imprinting will have to sacrifice the generality of
its approach in order to achieve the monoclonality and
very high fidelity to the specific target receptor which is
lacking at the moment.

Another limiting factor in MIP design for drug

discovery is the choice of template, as discussed in
Section 2.6. It is my opinion that a much greater effort
will have to be put into template design and synthesis
than has hitherto been the case if we are to see an
uptake of MIPs in high throughput drug screening. All
of the preceding remarks apply also to the very new
field of anti-idiotypic molecular imprinting. The rele-
vance of the specific compounds selected for synth-
esis by the MIP microreactors is in direct proportion
to the fidelity of the MIP cavity to the biological
target.

Overall, there are encouraging indications to be

found in the literature that molecularly imprinted
polymers will in the end be able to make a significant
contribution to the drug discovery process.

References

[1] L.V. Hijfte, G. Marciniak, N. Froloff, Combinatorial chemis-

try, automation and molecular diversity: new trends in the
pharmaceutical industry, J. Chromatogr., B 725 (1999) 3 – 15.

D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1873

[2] P.H.H. Hermkens, H.C.J. Ottenheijm, D. Rees, Solid-phase

organic reactions: a review of the recent literature, Tetrahedron
52 (1996) 4527 – 4554.

[3] H. Maehr, Combinatorial chemistry in drug research from a

new vantage point, Bioorg. Med. Chem. 5 (1997) 473 – 491.

[4] X. Williard, I. Pop, L. Bourel, D. Horvath, R. Baudelle, P.

Melnyk, B. Deprez, A. Tartar, Combinatorial chemistry: a
rational approach to chemical diversity, Eur. J. Med. Chem.
31 (1996) 87 – 98.

[5] S.H. Cheong, S. McNiven, A. Rachkov, R. Levi, K. Yano, I.

Karube, Testosterone receptor binding mimic constructed
using molecular imprinting, Macromolecules 30 (1997)
1317 – 1322.

[6] L. Piscopo, C. Prandi, M. Coppa, K. Sparnacci, M. Laus, A.

Lagana, R. Curini, G. D’Ascenzo, Uniformly sized molecu-
larly imprinted polymers (MIPs) for 17 h-estradiol, Macromol.
Chem. Phys. 203 (2002) 1532 – 1538.

[7] J.A. Tarbin, M. Sharman, Development of molecularly

imprinted phase for the selective retention of stilbene-type
estrogenic compounds, Anal. Chim. Acta (2001) 71 – 79.

[8] L. Ye, Y. Yu, K. Mosbach, Towards the development of

molecularly imprinted artificial receptors for the screening of
estrogenic chemicals, Analyst 126 (2001) 760 – 765.

[9] M. Quaglia, K. Chenon, A.J. Hall, E. De Lorenzi, B. Sell-

ergren, Target analogue imprinted polymers with affinity for
folic acid and related compounds, J. Am. Chem. Soc. 123
(2001) 2146 – 2154.

[10] A. Rachkov, N. Minoura, Recognition of oxytocin and oxyto-

cin-related peptides in aqueous media using a molecularly
imprinted polymer synthesized by the epitope approach, J.
Chromatogr., A 889 (2000) 111 – 118.

[11] J. Ostrowski, M.A. Kjelsberg, M.C. Caron, R.J. Lefkowitz,

Mutagenesis of the h

2

-adrenergic receptor: how structure elu-

cidates function, Annu. Rev. Pharmacol. Toxicol. 32 (1992)
167 – 183.

[12] O. Ramstrom, C. Yu, K. Mosbach, Chiral recognition in

adrenergic receptor binding mimics prepared by molecular
imprinting, J. Mol. Recognit. 9 (1996) 691 – 696.

[13] S.A. Piletsky, E.V. Piletska, B. Chen, K. Karim, D. Weston, G.

Barrett, P. Lowe, A.P.F. Turner, Chemical grafting of molecu-
larly imprinted homopolymers to the surface of microplates.
Application of artificial adrenergic receptor in enzyme-linked
assay for h-agonists determination, Anal. Chem. 72 (2000)
4381 – 4385.

[14] C.J. Allender, C. Richardson, B. Woodhouse, C.M. Heard,

K.R. Brain, Pharmaceutical applications for molecularly
imprinted polymers, Int. J. Pharm. 195 (2000) 39 – 43.

[15] J. Matsui, I.A. Nicholls, T. Takeuchi, Highly selective mole-

cularly imprinted polymer synthetic receptors for cinchona
alkaloids, Tetrahedron: Asymmetry 7 (1996) 1357 – 1361.

[16] J. Berglund, C. Lindbladh, I.A. Nicholls, K. Mosbach, Selec-

tion of phage display combinatorial library peptides with
affinity for a yohimbine imprinted methacrylate polymer,
Anal. Commun. 35 (1998) 3 – 7.

[17] M. Bowman, C. Allender, C. Heard, K. Brain, A high-through-

put screening technique employing molecularly imprinted
polymers as biomimetic sensors, Methodolog. Surv. Bioanal.
Drugs 25 (1998) 37 – 43.

[18] K. Sreenivasan, Molecularly imprinted polyacrylic acid con-

taining multiple recognition sites for steroids, J. Appl. Polym.
Sci. 82 (2001) 889 – 893.

[19] D.L. Rathbone, A. Ali, P. Antonaki, S. Cheek, Towards a

polymeric binding mimic for cytochrome CYP2D6, Biosens.
Bioelectron. 20 (2005) 2353 – 2363.

[20] K. Mosbach, Y. Yu, J. Andersch, L. Ye, Generation of new

enzyme inhibitors using imprinted binding sites: the anti-idio-
typic approach, a step toward the next generation of molecular
imprinting, J. Am. Chem. Soc. 123 (2001) 12420 – 12421.

[21] Y. Yu, L. Ye, K. Haupt, K. Mosbach, Formation of a class

of enzyme inhibitors (drugs), including a chiral compound,
by using imprinted polymers or biomolecules as molecular-
scale reaction vessels, Angew. Chem., Int. Ed. 41 (2002)
4460 – 4463.

[22] D.L. Rathbone, Y. Ge, Selectivity of response in fluorescent

polymers imprinted with N

1

-benzylidene pyridine-2-carboxa-

midrazones, Anal. Chim. Acta 435 (2001) 129 – 136.

D.L. Rathbone / Advanced Drug Delivery Reviews 57 (2005) 1854–1874

1874