A

ltmf

sHepdmip©

KM

1

gttbtosnc

0d

Analytica Chimica Acta 578 (2006) 50–58

Capturing molecules with templated materials—Analysisand rational design of molecularly imprinted polymers

Shuting Wei a, Michael Jakusch b, Boris Mizaikoff a,∗a School of Chemistry and Biochemistry, Georgia Institute of Technology, Atlanta, GA 30332-0400, USA

b Austrian Research Centers, A-2444 Seibersdorf, Austria

Received 7 March 2006; received in revised form 19 June 2006; accepted 24 June 2006Available online 7 July 2006

bstract

The creation of synthetic tailor-made receptors capable of recognizing desired molecular targets with high affinity and selectivity is a persistentong-term goal for researchers in the fields of chemical, biological, and pharmaceutical research. Compared to biomacromolecular receptors,hese synthetic receptors promise simplified production and processing, less costs, and more robust receptor architectures. During recent decades,

olecularly imprinted polymers (MIPs) are widely considered mimics of natural molecular receptors suitable for a diversity of applications rangingrom biomimetic sensors, to separations and biocatalysis.

A remaining challenge for the next generation of MIPs is the synthesis of deliberately designed and highly efficient receptor architecturesuitable for recognizing biologically relevant molecules, for which natural receptors are either not prevalent, or difficult to isolate and utilize.ence, this review discusses recent advances in synthetic receptor technology for biomolecules (e.g. drugs, amino acids, steroids, proteins,

ntire cells, etc.) via molecular imprinting techniques. Surface imprinting methods and epitope imprinting approaches have been introduced forrotein recognition at imprinted surfaces. Imprinting techniques in aqueous solution or organic-water co-solvents have been introduced avoidingenaturation of biomolecules during MIP synthesis. In addition, improved bioreactivity of entire enzyme or active site mimics generated by

olecular imprinting will be highlighted. Finally, the emerging importance of molecular modeling and molecular dynamics studies detailing the

ntermolecular interactions between the template species, the porogenic solvent molecules, and the involved monomer and cross-linker in there-polymerization solution will be addressed yielding a rational approach toward next-generation MIP technology.

2006 Elsevier B.V. All rights reserved.

eywords: Molecularly imprinted polymers (MIPs); Biomolecular recognition; Biomimetic receptors; Protein recognition; Synthetic enzymes; Rational design;

[ibis(

pp

olecular modeling

. Introduction

The selective recognition of biologically relevant moleculesoverns many essential biological interactions. Yet, mimickinghese processes with artificial receptors for, e.g. peptides, pro-eins, cells, etc. remains a major challenge in chemistry andiology. Among the variety of approaches for generating syn-hetic receptors, molecular imprinting techniques offer a numberf distinct advantages [1]. These include the – comparative –

traightforwardness of MIP preparation, the inherent robust-ess of the recognition element, and the chemical, mechani-al and thermal stability of the obtained templated materials

∗ Corresponding author. Tel.: +1 404 894 4030.E-mail address: boris.mizaikoff@chemistry.gatech.edu (B. Mizaikoff).

pspttBb

003-2670/$ – see front matter © 2006 Elsevier B.V. All rights reserved.oi:10.1016/j.aca.2006.06.077

2]. Many relatively small molecules have successfully beenmprinted for a variety of applications aiming at the unam-iguous detection and quantification of the template moleculesncluding biomimetic sensors [3–5], affinity chromatographyupports [6–8], solid phase extraction membranes [9–16], andimmuno)assays [17–19].

Piletsky et al. [20] recently summarized the results on thereparation of molecular imprinted membranes with specificermeability and separation properties. Some guidelines wererovided for understanding the relationship of the moleculartructure of the imprinted membranes with the observed trans-ort properties. However, more physical/chemical factors need

o be considered on the preparation of biomacromolecular recep-ors in comparison to molecular receptors for smaller targets.ecause of this growing interest, there is a distinct need inetter understanding the principles of design and optimization

S. Wei et al. / Analytica Chimica Acta 578 (2006) 50–58 51

Table 1Examples for molecular imprinting of biomolecules by novel functional/cross-linking monomers, or improved methods other than conventional Poly(MAA-co-EGDMA) based bulk polymerization techniques commonly applied for MIP synthesis

Type Template Imprinting method Functional interaction Polymer composition

Amino acids d-Phenylalanine [76] Bulk polymerization Hydrophobic interactions AMPSA, CD, DAPa

l-Glutamine [77] Phase inversion Hydrogen bonding Nylon-6N-Benzyloxycarbonyl-glutamic acid[36]

Surface imprinting atthe oil–water interface

Ionic interactions DVB, benzyldimethyl-n-tetradecylammoniumchloride

Steroids Cholesterol, stigmasterol[78]

Bulk polymerization Hydrogen bonding �-Cyclodextrin, toluene2,4-diisocyanate

Estradiol [18,79] Precipitationpolymerization

Hydrogen bonding MAA, DVB/TRIM

Peptides His-Ala [21] Bulk polymerization Metal complexation PolyacrylamideBoc-D-Ala-L-Ala-pNA[80]

Bulk polymerization Hydrogen bonding Butanamide (amino acidderivatives), EGDMA

Fmoc-Phe-Gly-OH [71] Solid phase synthesis;bulk polymerization

Hydrogen bonding Aminopropylsilica, MAA,EGDMA

[Sar1, Ala8]angiotensin II[37]

Bulk polymerization Ionic interactions Sodium acrylate, poly(ethyleneglycol) diacrylate

Enzymes �-Amylase [60] Phase inversion Hydrogen bonding Clarene, dextran

Proteins Bovine serum albumin,immunoglobulin G,fibrinogen [26]

Surface imprinting Hydrogen bonding Disaccharides,hexafluoropropylene

Dengue virus [33] Epitope imprinting Size and charge Acrylic acid/acrylamide/N-benzylacrylamide

Hemoglobin [81] Core-shell particles Shape, charge, hydrophobicity Aminopropyl silica (core), APTMSand PTMS (shell)b

Hemoglobin [82] Adsorption Hydrophilicity Polyacrylamide, chitosan beadsMyoglobin [40] Cross-linked gels Variety of weak interactions PolyacrylamideGlycoprotein (AFP) [83] Conjugation, gelation Antigen-antibody interaction Lectin, AFP-antibody, acrylamide,

N,N-methylenebisacrylamide

xtrin;

olocmbwsTfftvdafotcaliam

dilsmetapaapiopssta

a AMPSA: 2-acryloylamido-2-methylpropane sulphonic acid; CD: �-cyclodeb APTMS: 3-aminopropyl trimethoxysilane; PTMS: propyl trimethoxysilane.

f imprinted polymers for biomacromolecules. The main chal-enge during the synthesis of MIPs selective to biomoleculesr macromolecules is the fact that they need to be imprinted atonditions close to their natural environment ensuring confor-ational integrity. Moreover, MIPs for large molecular weight

iomolecules need to provide access to the binding pockets,hich is different from conventional MIPs expecting diffu-

ion of the template into the polymer matrix during rebinding.able 1 lists some prevalent examples of molecular imprintingor biomolecules utilizing a variety of imprinting methods dif-erent from traditional MIP synthesis, in particular in terms ofhe involved functional monomers (methacrylic acid (MAA), 4-inyl pyridine (4-VP), etc.) and cross-linkers (ethylene glycolimethacrylate (EGDMA), divinylbenzene (DVB), etc.) usu-lly applied. Some progress in this area has been achievedocusing on the design of MIPs via pronounced arrangementf functional interactions such as metal-chelation [19,21] withhe template, which is however limited by the structure andhemical properties of the selected templates. A more generallypplicable approach appears to be designing MIP strategies uti-

izing the shape complementarity along with multi-point weaknteractions such as hydrogen bonding or hydrophobic inter-ctions for providing molecularly selective binding sites foracromolecules.

tabm

DAP: N,N′-diacryloylpiperazine.

Most recently, the introduction of molecular modeling proce-ures to the field of molecular imprinting has significantly facil-tated more in depth understanding on the fundamental molecu-ar mechanisms governing molecular imprinting procedures bytudying the intermolecular interactions of template, functionalonomers, cross-linkers, and the solvent on the imprinting strat-

gy [22–25]. These molecular modeling strategies are analogouso the design and understanding of biological receptor systemsnd host-guest chemistry, and therefore hold substantial promiserogressing the field of molecular imprinting to the same extent,s molecular modeling and molecular dynamics simulationsre advancing the fundamental understanding of biorecognitionrocesses. Given the current state-of-the-art in rationally design-ng MIPs, such simulations may enable selecting combinationsf template, functional monomers, cross-linkers, and solventsroviding the most stable complex in the pre-polymerizationolution, which will then be selected for molecular dynamicsimulations investigating the interaction and conformation ofhe pre-polymerization complex. In the more distant future, it isnticipated that these findings will further enable simulations of

he polymerization procedure for determining the most appropri-te polymerization strategy yielding a maximum of high-specificinding sites in the resulting molecularly imprinted polymeratrix [25].

5 himica Acta 578 (2006) 50–58

2i

olmmliHif

famionfcoaadhtafidpsiSs(wvc5iafsgnpctdimt

ma

Fig. 1. Surface imprinted proteins. A freshly cleaved mica surface was usedas substrate adsorbing the template protein, which was decorated with dis-accharides forming a sugar overlayer. The sugar membrane was incorporatedinto a plasma-polymerized film of C3F6. The resulting polymer membrane wasadt

iodNnatt

hwbmpopiwwtcdtppet

2 S. Wei et al. / Analytica C

. Macromolecules and other larger template entities:mprinting at surfaces

Creating MIPs selective for large molecules such as proteinsr even entire cells remains a prevalent challenge in molecu-ar imprinting. The major problem during templating of large

olecules is the poor efficiency of specific binding due to theultitude of functional sites at these molecules, as well as the

imited mobility of the template molecules into and out of anmprinted polymer network with high degree of cross-linking.ence, new monomers addressing the high density of functional-

ties at large biomolecules, and new synthetic protocols adaptedor molecular imprinting of large molecules are demanded.

Surface imprinting strategies provide substantial advantagesor large molecule (e.g. protein) recognition: the binding sitesre more readily accessible at the surface of the imprintedatrix, and the mass transfer and binding kinetics are less lim-

ted given that template rebinding only requires the presencef the molecule at or close to the recognition surface. Rat-er and co-workers developed a surface imprinting procedureor protein recognition at a quartz crystal microbalance (QCM)hip surface [26]. Firstly, the template protein was adsorbednto an atomically flat mica surface. Disaccharides were thenpplied decorating the protein surface as a protective layergainst denaturation and degradation. Moreover, the hydrophilicisaccharides arranged at the protein surface by recognizingydrophilic moieties resulting in mapping of the functionali-ies. This “sugar shell” was coated with a fluoropolymer filmpplied by plasma deposition. Finally the outer surface of thatlm was attached to a glass substrate by adhesion facilitatingetachment of the mica substrate. After the mica substrate waseeled off, the template protein was extracted from the hemi-pherically shaped binding pockets with basic solution, provid-ng shape-complementary protein imprinted cavities (Fig. 1).elective recognition matrices have been prepared for bovineerum albumin (BSA), immunoglobulin G (IgG), lysozomeLSZ) and ribonuclease A (RNase). Competitive adsorptionsere assayed on binary protein mixtures (BSA versus IgG; LSZersus RNase). The competition ratios R50 (the ratio required toause a 50% reduction in the maximum adsorption) increase± 10-fold for the BSA imprint and 4 ± 7-fold for the IgG

mprint. For the lysozyme/RNase mixture, there was a 20-foldnd 26-fold increase in selectivity for the RNase imprint andor the lysozyme imprint, respectively. The specificity of theseurfaces is predominantly based on shape selectivity and hydro-en bonding interactions. While the achieved selectivity wasot utterly convincing, this promising strategy certainly hasotential for further optimization utilizing different recognitionhemistries for functional mapping of protein surfaces. Fur-hermore, this strategy assumes that the conformation of theeposited protein used as a template is similar to the free proteinn solution, which is rebound to the surface. The disparity of the

olecular structures may also contribute to the low discrimina-

ory power of these recognition surfaces.

Another strategy for protein recognition by surface graftingethods was reported by Bossi et al. [27]. In this approach,

ffinity matrices for proteins were obtained by grafting an

pmct

ttached to a glass slide with epoxy resin. The initial mica substrate was thenetached, and the sample was rinsed with NaOH/NaClO to dissolve and removehe template protein leaving selective binding sites at the surface.

mprinted 3-aminophenylboronic acid polymer at the surfacef polystyrene microtiterplates. The resulting polymers provideiscrimination for proteins with different shapes and charges.otably, experiments demonstrating improved template recog-ition with modified microtiterplates have not been shown,lthough the results discussed in this contribution indicate thathe reported MIP is characterized by enhanced affinity to theemplate molecule in contrast to bulk MIPs.

Molecularly imprinted polymers are usually prepared in aighly cross-linked matrix with limited control on the pore size,hich could hinder the access of guest molecules to the hostinding sites. Titirici et al. [28] reported a surface imprintingethod based on immobilizing the template onto a porous dis-

osable silica substrate prior to polymerization with a mixturef MAA and EGDMA. The silica mold was dissolved after theolymerization resulting in imprinted polymer beads with bind-ng sites at the surface. The retention factor of MIPs imprintedith immobilized templates was higher than for MIPs imprintedith free template molecules. However, the benefits of confining

he binding sites at the pore surface need to be augmented byonsiderable mechanistic insight into the governing interactionsuring binding site formation for controlling this process. Fur-hermore, experimental studies on the porosity of the resultingolymer beads, and an in depth evaluation on the impact of theore size of the silica mold on the obtained recognition prop-rties would support the hypothesis that higher accessibility ofhe binding sites is created with this synthesis strategy.

From these examples it is evident that the exploration of novel

olymer matrices by surface imprinting strategies will renderolecular imprinting more attractive and more broadly appli-

able for applications in biotechnology, biomedicine, and assayechnology.

S. Wei et al. / Analytica Chimica Acta 578 (2006) 50–58 53

F ndronT temp

iraocammtaHi“ttetMo

iwrpmitvtcilie

c

smcwwataiwtrpTtsa

sbstsfvplatp

3i

ig. 2. Schematic illustration of monomolecular imprinting with dendrimers. Dehe end groups of the dendrons were cross-linked forming the MIP matrix. The

In contrast to natural receptors, the application of MIPsn bioanalysis is limited by the usually incomplete templateemoval, comparatively broad distribution of guest affinities,nd restricted selectivity. A remarkably elegant approach basedn dendrimers ensuring nearly homogeneous binding sites, andomplete template removal, was reported by Zimmerman etl. [29,30]. As schematically shown in Fig. 2, the templateolecule acts as a core with several dendrons attached to the coreolecule similar to functional monomers assembling around

he template molecule. Next, the end groups of the dendrimersre cross-linked and the template is removed by hydrolysis.ence, synthetic receptor pockets containing individual bind-

ng sites for each template molecule are created, also calledmonomolecular imprinting” process. The monodispersity ofhis dendrimer-based imprinting approach with excellent con-rol on the properties of the polymer matrix also facilitatesxperimental analysis of the obtained matrices. For example,he number of cross-links can be determined by MALDI-TOF-

S [31], which provides detailed information on the structuref such monomolecular imprinted dendrimers.

Another strategy toward more homogeneous molecularmprints was described by Whitcombe and co-workers [32],herein a combination of covalent imprinting and non-covalent

ebinding techniques was developed. The template was pre-ared by covalently linking the target peptide with a functionalolecule (2-methacryloyloxybenzoyl chloride). Upon polymer-

zation and chemical cleavage, the carboxyl group is left inhe cavities with precise spatial arrangement and the essentialoid space facilitating removal and rebinding of the target pep-ide. The selected approach shows strong template binding byovalent interactions, and efficient rebinding by non-covalentnteraction. This promising strategy indicates that the target ana-ogue provides highly selective recognition properties, which

s a useful concept if a target constituent is highly toxic orxpensive.

Beyond creating highly effective MIPs for biomolecules,onsiderable effort has focused on developing novel detection

ma

s were covalently attached to the template forming a template–core dendrimer.late was removed by hydrolysis.

ystem for proteins with high accuracy, short response time, andinimum sample preparation and processing. Most recently, a

ombination of several existing imprinting strategies for proteinsas reported involving surface imprinting techniques combinedith epitope-mediated techniques for target protein imprinting

t the surface of a QCM chip [33]. A solution of monomers andemplate, which was the linear epitope of the target protein, wasdded on top of the (N-Acr-l-Cys-NHBn)-gold electrode andrradiated with UV light. After the polymerization, the film wasashed leaving peptide-imprinted cavities at the surface. Fur-

hermore, it was shown that this MIP-QCM chip was capable ofecognizing the template peptide and its mother protein in a com-lex matrix without prior separation or purification procedures.he dissociation constant Kd (0.04 nM) of the purified protein to

he MIP-QCM chip is comparable with the Kd (0.05 nM) of theame purified protein to the chip immobilized with monoclonalntibodies [34].

In conclusion, the generation of molecular imprints at sub-trate surfaces is an obvious solution to the problems of accessi-ility and mass transfer for large (bio)molecules. Furthermore,urface molecular imprinting provides more control on orienta-ion and density of the binding sites by tuning the porosity of theupport. In addition, considerable efforts in surface imprintingocus on the generation of more homogenous binding sites pro-iding rapid and accurate analyte detection. However, a genericrotocol for surface molecular imprinting has not yet been estab-ished given the multitude of currently explored strategies, whichre usually less straightforward than conventional synthetic pro-ocols for bulk MIPs, and require adaptation to individual tem-lates and template properties.

. The holy grail of MIPs: non-covalent molecularmprinting in aqueous solutions

To date, the majority of reports on MIPs are based on poly-ers synthesized in organic solvents, as hydrogen bonding inter-

ction between template and functional monomers considered

5 himica Acta 578 (2006) 50–58

asodmoit

poiiaIwia4aMnptoo

bwfiaempgb

ulsBdfgbicslmslsati

tirgaitftccftdoiemr

haauitrpwd

s

4 S. Wei et al. / Analytica C

mong the dominant forces in non-covalent imprinting is largelyuppressed at aqueous conditions. However, the recognitionf biomolecules is most relevant in aqueous solution avoidingenaturing or degradation of the template. Hence, non-covalentolecular imprinting in aqueous solutions has to take advantage

f other types of interaction such as �–� stacking [35], ionicnteractions [36,37], and hydrophobic interactions [38] facili-ating molecular recognition.

Selective and reversible recognition of 2,4-dichloro-henoxyacetic acid by accordingly synthesized MIPs wasbserved in aqueous solution [4,39]. Both hydrogen bondingnteractions and aggregation of aromatic moieties have beendentified as governing forces for complex formation of 2,4-Dnd 4-vinylpyridine in the pre-polymerization solution by FT-R and NMR spectroscopy [25]. Moreover, ionic interactions asell as hydrophobic interactions can play an important role dur-

ng the selective recognition of 4-nitrophenol (4-NP) by MIPs inqueous solution [39]. Janotta et al. demonstrated that MIPs for-NP can discriminate 4-NP from other phenolic analogues suchs 2- and 3-NP in both organic and aqueous media. Interestingly,IPs against the other nitrophenols (2-NP, 3-NP) show less pro-

ounced imprinting effects comparing to the 4-NP imprintedolymer MIP. The decrease of the imprinting effect from 4-NPo 2-NP may be attributed to the fact that the functional moietiesf 2- and 3-NP are in closer proximity preventing the formationf discernable selective binding sites.

An attractive approach to imprinting in aqueous solutions isased on preparing water-soluble imprinted polymers, whichould ensure that the binding sites are readily accessible

or biomolecules in aqueous solution. Polyacrylamide gelsmprinted with proteins were reported to show preferentialffinities for their templates in aqueous solutions by Hjertent al. [40–42] and other groups [43,44], however, the imprintingechanism in aqueous solutions remains uncertain. While it is

roposed that the shape selectivity and multi-point weak hydro-en bonding contribute to specific recognition, the latter has toe doubted given the surrounding aqueous matrix.

Another promising strategy is the combination of molec-lar imprinting with hydrogels [43,45]. Hydrogels are cross-inked, three-dimensional hydrophilic polymer networks, whichwell but do not dissolve when brought into contact with water.ecause of their significant water content, hydrogels provide aegree of flexibility very similar to natural tissues. Therefore,rom a biological point of view molecularly imprinted hydro-els would present an ideal matrix for recognizing and bindingiomolecules. The challenge of this methodology substantiatesn the creation of effective imprinting structures providing suffi-ient rigidity/integrity of the binding pocket to achieve adequatepecificity within hydrogel system usually characterized by aow degree of cross-linking. Hiratani et al. [46,47] reported thatolecular imprinted hydrogels can be prepared by dissolving

mall amounts of MAA and EGDMA in hydroxyethylmethacry-ate or N,N-diethylacrylamide, which are the materials to prepare

oft contact lenses. The resulting soft matrix showed higherffinity (increased uptake) to timolol (the target biomolecule)han the corresponding non-imprinted systems. Beyond thenfluence of functional monomers and cross-linkers, it was found

aapo

Fig. 3. Molecular structure of chitosan.

hat the nature of the backbone monomer also affects their capac-ty to bind drug molecules, and the release characteristics. Mostecently, a semi-interpenetrating molecularly imprinted hydro-el (MIH) for hemoglobin (Hb) was prepared based on poly-crylamide/chitosan in aqueous solution [48]. Using chitosann the polymer network increases the affinity and selectivity toarget proteins due to its high content of amino and hydroxylunctional groups (Fig. 3). Weak hydrogen bonding interac-ions between chitosan and polyacrylamide further increase thehain entanglement. The –NH3

+ and the hydroxyl groups ofhitosan will interact with the target protein by electrostaticorces rather than hydrogen bonding, which are suspected ashe main recognition mechanism at high water contents. Theegree of specificity (Kd is defined as the concentration of Hbn MIH over the concentration of Hb in the incubation solution)s 83.63 for Hb and not detectable for BSA when the MIH wasquilibrated in the mixture of Hb and BSA. The control poly-er hydrogel has a Kd value of 1.7 and 1.4 for Hb and BSA,

espectively.In contrast to weak interactions like hydrogen bonding and

ydrophobic interactions in water, metal complexation [21,49]nd ionic interaction [50] would provide much stronger inter-ctions for molecular recognition. Consequently, macromolec-lar receptors for peptides were developed based on strongnteractions between the N-terminal histidine of peptides, andhe Ni(II)-polymerizable NTA ligand, which was incorpo-ated in water-soluble polyacrylamides (Fig. 4) [21]. Theseolyacrylamide-based MIPs show adequate water solubility,hile providing strong histidine binding sites attracting theipeptides to the polymer surface.

In summary, the creation of selective binding sites in aqueousolution involves strong interactions such as metal complex-tion, ionic interaction, or multi-point hydrogen bonding. In

ddition, utilizing water-soluble matrices such as hydrogels is aotentially useful strategy toward selective recognition in aque-us media.

S. Wei et al. / Analytica Chimic

Ff

4b

o[t

aeasopslieewltdmomi

spbticstwc(ba(tTic

cgowtbamtwtWcth

btwwl

F(

ig. 4. Structure of the histidine imprinted cross-linked polyacrylamide (adaptedrom reference [21]); Pep: amino acid sequence append to the histidine.

. Imprinting of biomolecules: improvement ofioreactivity

Enzyme imprinting has been introduced to attain high levelsf catalytic activity with synthetic enzymes in organic solvents51–54], or for designing efficient synthetic analogs mimickinghe active site of enzymes [55–57].

In order to maintain the conformation and activity of thective site of an enzyme, a so-called imprinter was added to thenzyme solution followed by a freeze-drying procedure to formn enzyme–imprinter complex. The imprinter is a ligand used toelectively mask the active site of the target enzyme. Removalf the imprinter was achieved by consecutive rinses with appro-riate solvents. The obtained active site of the enzyme shouldtill provide acceptable catalytic activity (Fig. 5) [54]. The chal-enge of this method is the usually poor solubility of imprintersn water, which limits the applicability of this approach. Richt al. developed a strategy of including water-insoluble imprint-rs with enzymes by using derivatives of the imprinter moleculeith sufficient water solubility. The enhancement of the acy-

ation rates following imprinting were ≤20-fold comparing tohe native enzymes. In an alternative approach, enzymes wereissolved in aqueous solution along with organic co-solvents

ediating the solubility of the imprinter. The rate enhancements

f such imprinted enzymes in co-solvents were of a similaragnitude as those obtained by imprinting with water-soluble

mprinters.

twta

ig. 5. Schematic representation of tuning the enzyme active site by molecular imprimprinter) followed by a freeze-drying process. The resulting lyophilized complex c

a Acta 578 (2006) 50–58 55

High activity could also be obtained by mimics of the activeite of enzymes with a stable transition-state analogue as tem-late, and a functional monomer containing a triamine forinding the metal ion, and an amidimium ion for binding theemplate, as recently shown by Liu and Wulff [57]. This methods particularly attractive due to the straightforward adoption ofonventional bulk polymer preparation methods as applied fortandard MIPs. After removal of the template, the activity ofhese mimics was sampled by a catalytic hydrolysis process,hich revealed a substantial enhancement of the reaction rate

ompared to control polymers. The ratio of Kcat/Kuncat = Ksolnthe rate constant with catalyst and without catalyst) from car-onate hydrolysis was used to express the catalytic activity ofntibodies and natural enzymes. This value was 110,000 for di-2-pyridyl)-carbonate for the MIP catalyst, which is much higherhan the value for catalytic antibodies with Kcat/Kuncat = 810.his is surprising considering the fact that the polymers are

nsoluble and have a rigid structure with embedded “polyclonal”avities.

Another example for mimics of the active site of enzymesatalyzing carbon–carbon bond formation was reported by theroup of Mosbach [58,59]. Unlike hydrolysis, the processf carbon–carbon bond formation is entropically unfavorable,hich is a major challenge in organic synthesis. In their work,

he imprinting strategy is based on the formation of the complexetween the reaction intermediate analogue (4-vinylpyridine)nd Co2+. The resulting MIP shows catalytic activity for the for-ation of chalcon by acetophenone and benzaldehyde, however,

he reaction time was on the order of 3 weeks, and the polymeras slowly degraded during that process. The polymer increases

he reaction rate eight-fold relative to the solution reaction.hile the catalytic activity of imprinted materials apparently

annot yet compete with that of the natural enzyme analogues,heir low cost and high chemical stability render this strategy aighly promising approach.

In the area of biological applications, improvements in theiocompatibility of MIPs are essential. An example of main-aining enzyme functionality during the imprinting procedureas reported by Silvestri et al. [60]. Imprinted membranesere prepared from Clarene®/Dextran/�-amylase solutions fol-

owed by freeze-drying with a lyophilizer. It was expected that

he structure of the enzyme can be maintained, as the enzymeas physically entrapped in the polymer network. Interestingly,

he imprinted membrane retained more template moleculeslthough it had a larger pore size than the control matrix. The

inting. The target enzyme is dissolved in the solution containing the substrateontains the active site for the target imprinter.

5 himic

cctc

chredolrb

5u

db

cip(ptocw

mipSdtclttItpgetb

lmsetsa

[saifscics

aaiibiomtipftIe4lwere generated from simulated annealing runs after energyminimization and equilibration. This procedure yields possi-ble conformations of the entire pre-polymerization system afterannealing, as exemplarily shown in Fig. 6. It is anticipated that

6 S. Wei et al. / Analytica C

atalytic activity of thus immobilized enzymes was tested andompared with the activity of the free enzyme indicating thathe immobilized enzymes maintained adequate performance foratalytic conversions.

Examples for the modification of natural enzymes and for thereation of enzyme-like catalysts by molecular imprinting wereighlighted. Enhanced catalytic activity and selectivity wereeported due to molecular imprinting, however, quantitative ben-fits provided by molecular imprinting for catalysis are stillifficult to assess as the comparison of reported catalytic ratesf molecularly imprinted catalyst and the natural enzyme ana-ogues has yet to be established on a more profound basis, as theate enhancement reported from MIPs may be largely affectedy tuned reaction conditions rather than by the imprint itself.

. Molecular modeling: a route to rationalnderstanding and design of molecular imprints

In order to better understand the intermolecular interactionsuring molecular imprinting, computational approaches haveeen introduced for more rational design of MIPs.

Takeuchi et al. [61] reported on an estimation of the complexonformation by docking of the most stable conformers using anntermolecular Monte Carlo-based conformational search withossibilities of rotating and translating the functional monomerMAA) toward a template molecule. In the simulation, the com-lexes were allowed to rotate randomly at their mass center, ando translate in three-dimensional space. While the conformationf the most stable snapshot was discussed, any cluster analysis orlassification addressing the multitude of possible interactionsas not reported.The group of Piletsky used a virtual library of functional

onomers to screen against the template molecules. By compar-ng the binding energies of each virtual pair of monomer and tem-late, the best monomer for the template was selected [23,62,63].ode and co-workers [64] further developed this method for theesign of molecularly imprinted catalysts. A virtual library ofhe intermediates of a lipase-catalyzed transesterification pro-ess was screened, and the most stable intermediate expected toead to a higher turn over rate was selected. In the following,he optimized molecularly imprinted catalyst showed enhancedurnover rates by a factor of 3 compared to the control polymer.n order to provide closer similarity between the simulation andhe real polymers, clusters including up to 10 monomers, orolymer chains were used [22,24]. Various intermolecular ener-ies were extracted from the total energy, and it was found thatlectrostatic energy provides the most significant contributiono the total energy, which could be contributions from hydrogenonding between template and monomers.

Recently, our research group has introduced a molecu-ar modeling approach studying the interactions of template

olecules and functional monomer building blocks in explicitolvent using AMBER, a suite of programs for molecular mod-

ling and molecular dynamics simulations, which is particularlyailored for biomolecular interactions. AMBER comprises a sub-tructure database, a comprehensive set of force field parameters,nd a variety of preparatory, simulation, and analysis programs

Fsewt

a Acta 578 (2006) 50–58

65]. In the latest version of the software, parameters for mostmall organic molecules can be conveniently generated by anutomated procedure. We have applied AMBER to simulat-ng the interactions between small molecules (i.e. template andunctional monomer) confirming hydrogen bonding and �–�tacking interactions during molecular dynamics simulation inhloroform and water [25]. The presence of these particularnteractions between template and functional monomers wereonfirmed by NMR and IR spectroscopic studies on the corre-ponding pre-polymerization solutions.

However, the molecular modeling approaches discussedbove are not fully comprehensive, as the effects of cross-linkersnd the correct ratio of the involved molecules are not yet takennto account. It can be expected that establishing models includ-ng all the reaction species at the correct ratios will provideetter understanding on the role of the different constituents dur-ng the imprinting process. Our ongoing research efforts focusn including the effects of cross-linker, solvent, and functionalonomer at their experimental ratio into the simulation sys-

em, which provides a more realistic simulation environmentn closer analogy to the experimental conditions present in there-polymerization solution. First results have been obtainedor simulating a pre-polymerization solution for the prepara-ion of a 17�-estradiol MIP, which was reported elsewhere [18].n these simulations, the interactions between 1 molecule 17�-stradiol, 8 molecules methacrylic acid (or 4-vinylpyridine), and0 molecules divinylbenzene (or ethylene glycol dimethacry-ate) were observed in explicit solvent. One thousand snapshots

ig. 6. Models of pre-polymerization mixture for 17�-estradiol generated byimulated annealing (gray: divinylbenzene, blue: methacrylic acid, orange: 17�-stradiol; for clarity of the representation the display of the solvent moleculesas omitted). (For interpretation of the references to colour in this figure legend,

he reader is referred to the web version of the article.)

himic

fitt

ottafinatiomp[aicbmtttoto[rvseritpFigttMettp

6

sirT

barabedmoar

tnrtm

A

FLl#

R

[[

[

[[

[[[[[[

S. Wei et al. / Analytica C

rom these simulations a variety of parameters describing thenteractions between template and monomers such as bond dis-ances can be extracted providing fundamental understand onhe binding mechanisms in these complex systems.

While these simulation approaches provide detailed insightn molecular interactions, the accuracy of these results needso be confirmed by thorough experimental analysis establishinghe validity of the developed model. Nicholls [66] suggestedthermodynamic model to identify factors influencing the site

ormation for the design of MIPs including the contributions ofndividual enthalpic terms and entropic factors. The thermody-amic aspects involve the studies of the binding site distributionnd the binding equlibria. Template–monomer complex forma-ion has been investigated with a series of analytical techniquesdentifying binding ratio and binding constants upon the analysisf proton signal shifts resulting from the complex formation. Aodel of the speciation of individual monomers–template com-

lexes by 1H NMR studies was published by Sellergren et al.67]. A more extensive theoretical model considering the selfssociation of the monomers provides a more accurate deter-mination of the binding constant of the monomer–templateomplex [68]. UV/Vis spectroscopy provides information of theinding events by probing the electronic structure of the involvedolecules independent of the nature of the molecular interac-

ion [69]. IR spectroscopy probing the vibrational signatures ofhe molecules provides additional complementary informationo UV/Vis and NMR studies on the nature and stoichiometryf the pre-polymerization complexes [25,68]. Isothermal titra-ion calorimetry (ITC) enables insight into the thermodynamicsf the binding properties of molecularly imprinted polymers70,71], or of the pre-polymerization complexes [72]. Despiteecent successes in modeling of molecular interactions rele-ant to pre-polymerization solutions as the governing precur-or step for MIP synthesis, it is evident that validation of thestablished models with a broad range of analytical techniquesemains a necessary prerequisite. The modeling of molecularnteractions involves the studies of the binding properties ofhe pre-polymerization complex as well as the ligand rebindingroperties of the resultant MIPs. Shimizu and co-workers usedreundlich [73] isotherms (FI) and Langmuir–Freundlich [74]

sotherms (LFI) to predict the rebinding constant and hetero-eneity of MIPs. They also proposed the affinity spectrum fromhe binding parameters obtained from FI or LFI methods forhe investigation of the distribution of binding site affinities in

IPs [75]. The FI or LFI methods can provide valuable data forvaluation and optimization of MIPs. It should be noticed thathe application of these methods is restricted by the concentra-ion region of the binding assays and the heterogeneity of theolymer materials.

. Conclusions

In this review, we summarize the current state-of-the-art in

ynthetic macromolecular receptors synthesized via molecularmprinting methods for the selective recognition of biologicallyelevant molecules such as amino acids, peptides, and proteins.he critical steps for the advancement in MIP research for

[[[

a Acta 578 (2006) 50–58 57

iomolecular targets is the preparation of MIPs providing highffinity to the macromolecular template molecule without dis-upting the bioreactivity or biofunctionality of the target analyte,nd a homogeneous binding site distribution. Finally, applica-ility of such synthetic receptors in aqueous environments isssential for biomolecular recognition. Furthermore, it is evi-ent that better understanding of MIP synthesis by molecularodeling and analytical techniques will enable the development

f more effective MIP-based receptors, in analogy to the currentdvances in understanding the molecular mechanisms of naturaleceptors and host–guest systems.

Consequently, the combination of molecular modeling withhe orchestrated application of sophisticated analytical tech-iques has the potential of providing the information needed forational design of next-generation synthetic receptors based onemplated materials suitable for capturing biologically relevant

olecules.

cknowledgements

This work is supported by the European Commission’s 5thramework Programme (Quality of Life and Management ofiving Resources, #QLK4-CT2002-02323) and the U.S. Geo-

ogical Survey (National Water Quality Assessment Program,2002GA30G).

eferences

[1] J. O’Mahony, A. Molinelli, K. Nolan, M.R. Smyth, B. Mizaikoff, Anal.Chim. Acta 534 (2005) 31.

[2] A. Molinelli, M. Janotta, B. Mizaikoff, Molecularly imprinted polymersfor biomolecular recognition, in: V.D. Tuan (Ed.), Protein Nanotechnol-ogy: Protocols, Instrumentation and Applications, Methods in MolecularBiology, vol. 300, Humana Press, Totowa, NJ, 2005, p. 243.

[3] D. Kriz, O. Ramstroem, A. Svensson, K. Mosbach, Anal. Chem. 67 (1995)2142.

[4] M. Jakusch, M. Janotta, B. Mizaikoff, K. Mosbach, K. Haupt, Anal. Chem.71 (1999) 4786.

[5] N.T. Greene, K.D. Shimizu, J. Am. Chem. Soc. 127 (2005) 5695.[6] M. Kempe, K. Mosbach, J. Chromatogr. A 691 (1995) 317.[7] J. Haginaka, H. Sanbe, J. Chromatogr. A 913 (2001) 141.[8] Y. Watabe, K. Hosoya, N. Tanaka, T. Kubo, T. Kondo, M. Morita, J. Chro-

matogr. A 1073 (2005) 363.[9] B. Sellergren, Anal. Chem. 66 (1994) 1578.10] M.T. Muldoon, L.H. Stanker, Anal. Chem. 69 (1997) 803.11] T. Stroink, E. Paarlberg, J.C. Waterval, A. Bult, W.J. Underberg, Elec-

trophoresis 22 (2001) 2375.12] Q.-Z. Zhu, P. Degelmann, R. Niessner, D. Knopp, Environ. Sci. Technol.

36 (2002) 5411.13] A. Molinelli, R. Weiss, B. Mizaikoff, J. Agric. Food Chem. 50 (2002) 1804.14] E. Caro, R.M. Marce, P.A.G. Cormack, D.C. Sherrington, F. Borrull, J.

Chromatogr. A 995 (2003) 233.15] X. Zhu, J. Yang, Q. Su, J. Cai, Y. Gao, J. Chromatogr. A 1092 (2005) 161.16] N. Perez-Moral, A.G. Mayes, Biosens. Bioelectron. 21 (2006) 1798.17] M.-C. Hennion, V. Pichon, J. Chromatogr. A 1000 (2003) 29.18] S. Wei, A. Molinelli, B. Mizaikoff, Biosens. Bioelectron. 21 (2006) 1943.19] S. Vidyasankar, M. Ru, F.H. Arnold, J. Chromatogr. A 775 (1997) 51.20] S.A. Piletsky, T.L. Panasyuk, E.V. Piletskaya, I.A. Nicholls, M. Ulbricht,

J. Membr. Sci. 157 (1999) 263.21] B.R. Hart, K.J. Shea, Macromolecules 35 (2002) 6192.22] D. Pavel, J. Lagowski, Polymer 46 (2005) 7543.23] S.A. Piletsky, K. Karim, E.V. Piletska, C.J. Day, K.W. Freebairn, C. Legge,

A.P.F. Turner, Analyst (Cambridge, UK) 126 (2001) 1826.

5 himic

[[

[

[

[[

[[

[

[

[

[[[

[

[

[

[[

[[

[[[

[

[[

[[

[[

[

[[[

[

[

[[

[

[[

[[

[[[

[

[

[

[

[

[

[

[

[[

[

8 S. Wei et al. / Analytica C

24] D. Pavel, J. Lagowski, Polymer 46 (2005) 7528.25] A. Molinelli, J. O’Mahony, K. Nolan, M.R. Smyth, M. Jakusch, B.

Mizaikoff, Anal. Chem. 77 (2005) 5196.26] H. Shi, W.-B. Tsai, D.M. Garrison, S. Ferrari, D.B. Ratner, Nature 398

(1999) 593.27] A. Bossi, S.A. Piletsky, E.V. Piletska, P.G. Righetti, A.P. Turner, Anal.

Chem. 73 (2001) 5281.28] M.M. Titirici, A.J. Hall, B. Sellergren, Chem. Mater. 15 (2003) 822.29] S.C. Zimmerman, M.S. Wendland, N.A. Rakow, I. Zharov, K.S. Suslick,

Nature 418 (2002) 399.30] E. Mertz, S.C. Zimmerman, J. Am. Chem. Soc. 125 (2003) 3424.31] S.C. Zimmerman, I. Zharov, M.S. Wendland, N.A. Rakow, K.S. Suslick, J.

Am. Chem. Soc. 125 (2003) 13504.32] J.U. Klein, M.J. Whitcombe, F. Mulholland, E.N. Vulfson, Angew. Chem.

Int. Ed. 38 (1999) 2057.33] D.-F. Tai, L. Chung-Yin, W. Tzong-Zeng, L.-K. Chen, Anal. Chem. 77

(2005) 5140.34] C.C. Su, T.Z. Wu, L.C. Chen, H.H. Yang, D.F. Tai, Anal. Chim. Acta 479

(2002) 117.35] S. Marx, A. Zaltsman, Int. J. Environ. Anal. Chem. 83 (2003) 671.36] K. Araki, M. Goto, S. Furusaki, Anal. Chim. Acta 469 (2002) 173.37] A. Rachkov, M. Hu, E. Bulgarevich, T. Matsumoto, N. Minoura, Anal.

Chim. Acta 504 (2004) 191.38] S.A. Piletsky, H.S. Andersson, I.A. Nicholls, J. Mol. Recognit. 11 (1998)

94.39] M. Janotta, R. Weiss, B. Mizaikoff, O. Bruggemann, L. Ye, K. Mosbach,

Int. J. Environ. Anal. Chem. 80 (2001) 75.40] S. Hjerten, J.L. Liao, K. Nakazato, Y. Wang, G. Zamaratskaia, H.X. Zhang,

Chromatographia 44 (1997) 227.41] J.L. Liao, Y. Wang, S. Hjerten, Chromatographia 42 (1996) 259.42] D. Tong, C. Hetenyi, Z. Bikadi, J.P. Gao, S. Hjerten, Chromatographia 54

(2001) 7.43] S.H. Ou, M.C. Wu, T.C. Chou, C.C. Liu, Anal. Chim. Acta 504 (2004) 163.44] X. Pang, G. Cheng, R. Li, S. Lu, Y. Zhang, Anal. Chim. Acta 550 (2005)

13.45] M.E. Byrne, K. Park, N.A. Peppas, Adv. Drug Deliv. Rev. 54 (2002) 149.46] H. Hiratani, C. Alvarez-Lorenzo, J. Control. Release 83 (2002) 223.47] H. Hiratani, A. Fujiwara, Y. Tamiya, Y. Mizutani, C. Alvarez-Lorenzo,

Biomaterials 26 (2004) 1293.48] Y.-Q. Xia, T.-Y. Guo, M.-D. Song, B.-H. Zhang, B.-L. Zhang, Biomacro-

molecules 6 (2005) 2601.49] S. Striegler, Macromolecules 36 (2003) 1310.50] T. Kubo, K. Hosoya, M. Nomachi, N. Tanaka, K. Kaya, Anal. Bioanal.

Chem. 382 (2005) 1698.51] K. Dabulis, A.M. Klibanov, Biotechnol. Bioeng. 41 (1993) 566.52] P. Mishra, K. Griebenow, A.M. Klibanov, Biotechnol. Bioeng. 52 (1996)

609.

53] J.O. Rich, J.S. Dordick, J. Am. Chem. Soc. 119 (1997) 3245.54] J.O. Rich, V.V. Mozhaev, J.S. Dordick, D.S. Clark, Y.L. Khmelnitsky, J.

Am. Chem. Soc. 124 (2002) 5254.55] A.G. Strikovsky, D. Kasper, M. Gruen, B.S. Green, J. Hradil, G. Wulff, J.

Am. Chem. Soc. 122 (2000) 6295.

[

[

a Acta 578 (2006) 50–58

56] J.-q. Liu, G. Wulff, Angew. Chem. Int. Ed. 43 (2004) 1287.57] J.-q. Liu, G. Wulff, J. Am. Chem. Soc. 126 (2004) 7452.58] J. Matsui, I.A. Nicholls, I. Karube, K. Mosbach, J. Org. Chem. 61 (1996)

5414.59] I.A. Nicholls, J. Matsui, M. Krook, K. Mosbach, J. Mol. Recognit. 9 (1996)

652.60] D. Silvestri, C. Cristallini, G. Ciardelli, P. Giusti, N. Barbani, J. Biomater.

Sci. Polym. Ed. 15 (2004) 255.61] T. Takeuchi, A. Dobashi, K. Kimura, Anal. Chem. 72 (2000) 2418.62] S. Subrahmanyam, S.A. Piletsky, E.V. Piletska, B. Chen, K. Karim, A.P.

Turner, Biosens. Bioelectron. 16 (2001) 631.63] I. Chianella, M. Lotierzo, S.A. Piletsky, I.E. Tothill, B. Chen, K. Karim,

A.P.F. Turner, Anal. Chem. 74 (2002) 1288.64] Z. Meng, T. Yamazaki, K. Sode, Biosens. Bioelectron. 20 (2004) 1068.65] D.A. Case, T.A. Darden, T.E. Cheatham, III, C.L. Simmerling, J. Wang,

R.E. Duke, R. Luo, K.M. Merz, B. Wang, D.A. Pearlman, M. Crowley, S.Brozell, V. Tsui, H. Gohlke, J. Mongan, V. Hornak, G. Cui, P. Beroza, C.Schafmeister, J.W. Caldwell, W.S. Ross, P.A. Kollman, (2004). AMBER8, University of California, San Francisco.

66] I.A. Nicholls, J. Mol. Recognit. 11 (1998) 79.67] B. Sellergren, M. Lepistoe, K. Mosbach, J. Am. Chem. Soc. 110 (1988)

5853.68] P. Manesiotis, A.J. Hall, B. Sellergren, J. Org. Chem. 70 (2005) 2729.69] H.S. Andersson, I.A. Nicholls, Bioorg. Chem. 25 (1997) 203.70] A. Weber, M. Dettling, H. Brunner, G.E.M. Tovar, Macromol. Rapid Com-

mun. 23 (2002) 824.71] R. Kirchner, J. Seidel, G. Wolf, G. Wulff, J. Inclusion Phenom. Macrocyclic

Chem. 43 (2002) 279.72] W.P. Fish, J. Ferreira, R.D. Sheardy, N.H. Snow, T.P. O’Brien, J. Liq. Chro-

matogr. Relat. Technol. 28 (2005) 1.73] R.J. Umpleby, S.C. Baxter, M. Bode, J.K. Berch, R.N. Shah, K.D. Shimizu,

Anal. Chim. Acta 435 (2001) 35.74] R. Umpleby, S. Baxter, Y. Chen, R. Shah, K. Shimizu, Anal. Chem. 73

(2001) 4584.75] R.J. Umpleby, S.C. Baxter, A.M. Rampey, G.T. Rushton, Y. Chen, K.D.

Shimizu, J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 804 (2004)141.

76] S.A. Piletsky, H.S. Andersson, I.A. Nicholls, Macromolecules 32 (1999)633.

77] P.S. Reddy, T. Kobayashi, M. Abe, N. Fujii, Eur. Polym. J. 38 (2002)521.

78] T. Hishiya, M. Shibata, M. Kakazu, H. Asanuma, M. Komiyama, Macro-molecules 32 (1999) 2265.

79] L. Ye, R. Weiss, K. Mosbach, Macromolecules 33 (1999) 8239.80] K. Yano, T. Nakagiri, T. Takeuchi, J. Matsui, K. Ikebukuro, I. Karube, Anal.

Chim. Acta 357 (1997) 91.81] T. Shiomi, M. Matsui, F. Mizukami, K. Sakaguchi, Biomaterials 26 (2005)

5564.82] T.Y. Guo, Y.Q. Xia, G.J. Hao, M.D. Song, B.H. Zhang, Biomaterials 25

(2004) 5905.83] T. Miyata, M. Jige, T. Nakaminami, T. Uragami, Proc. Natl. Acad. Sci.

U.S.A. 103 (2006) 1190.