Biosensors and Bioelectronics 23 (2007) 201209

Binding site characteristics of 17 -estradiol imprinted polymers

Shuting Wei, Boris Mizaikoff
School of Chemistry and Biochemistry, Georgia Institute of Technology, Atlanta, GA 30332-0400, USA Received 2 January 2007; received in revised form 22 March 2007; accepted 27 March 2007 Available online 12 April 2007

Abstract The variety of applications utilizing molecularly imprinted polymers (MIPs) requires synthetic strategies yielding different MIP formats including lms, irregular particles, or spheres, along with precise knowledge on the specic material characteristics, such as binding capacity and binding efciency of these materials. In response to this demand, MIPs are prepared in different formats by variation of the polymerization methodology. It is commonly agreed that micro- and sub-microspheres are particularly advantageous MIP formats, due to their monodispersity and facile synthesis procedures in contrast to conventional imprinted polymers prepared by bulk polymerization. However, the differences in actual rebinding characteristics of different MIP formats based on molecular interactions under a variety of binding/rebinding conditions have not been studied in detail to date. Consequently, the present work details an analytical strategy generically applicable to MIP systems for rebinding studies including equilibrium binding, non-equilibrium binding, and release experiments enabling more profound understanding on the molecular interactions between the imprinted materials and the template molecules. In this study, three MIP formats were considered for the same template molecule, 17 -estradiol: irregularly shaped particulate polymers prepared by bulk polymerization and grinding, microspheres, and sub-microspheres. The latter two formats were synthesized via precipitation polymerization using different processing strategies. The morphologies and porosities of the resulting imprinted materials were characterized by scanning electron microscopy (SEM) and BrunauerEmmettTeller (BET) analysis, respectively. The obtained results indicate that microspheres prepared by precipitation polymerization provide superior rebinding properties during equilibrium binding in contrast to bulk polymers and sub-microspheres, and that the rebinding properties are different during equilibrium binding versus non-equilibrium binding. The median binding afnity constant determined during non-equilibrium rebinding is higher than the values obtained from equilibrium rebinding. Furthermore, the binding site distribution appears more homogeneous thief derived from non-equilibrium rebinding, as reected in a heterogeneity index of m = 0.725. Moreover, it is hypothesized that the specic interactions between template and monomers are related to the porosity of the imprinted polymers, which implies that the amount of binding sites and the pore sized distribution of the imprinted materials are a critical factor in achieving the desired MIP performance in various analytical applications. The BET results indicate that particles prepared with lower cross-linker-to-template ratio have a reduced surface area. Furthermore, it can be expected that there are less specic binding sites available at particles with reduced surface area and pore volume given similar distribution of the binding sites, as conrmed by the equilibrium binding isotherm studies. The pore size distribution results reveal that control of the pore size in the range of 100180 A is essential to obtain the desired retention properties and Gaussian peak shape during HPLC analysis of small molecules. 2007 Published by Elsevier B.V.
Keywords: Molecular imprinting; Imprinted microspheres; Imprinted sub-microspheres; Freundlich adsorption isotherm; Porosity; 17 -Estradiol

1. Introduction The primary polymerization method for MIPs derived from literature is based on bulk polymerization in a porogenic solvent creating a block co-polymer (Arshady and Mosbach, 1981; Vlatakis et al., 1993; Wulff et al., 1977). The major drawback of

Corresponding author. Tel.: +1 404 894 4030; fax: +1 404 385 6447. E-mail address: boris.mizaikoff@chemistry.gatech.edu (B. Mizaikoff). URL: http://asl.chemistry.gatech.edu (B. Mizaikoff).

this method is the fact that particles with the desired dimensions, as e.g. for chromatographic applications serving as stationary phase sorbents, have to be obtained by grinding and sieving, which is a labor intensive and wasteful process with a considerable amount of ne particles due to the generated highly disperse particle size distribution. In general, the development of novel MIP sorbent materials requires considering the desired molecular recognition properties along with the material format, such as the appropriate particle size for the respective target application. Hence, synthesis of MIP materials for sensors (Greene and Shimizu, 2005), assay technology (Ansell, 2004), and separa-

0956-5663/$ see front matter 2007 Published by Elsevier B.V. doi:10.1016/j.bios.2007.03.031

202

S. Wei, B. Mizaikoff / Biosensors and Bioelectronics 23 (2007) 201209

tions (Remcho and Tan, 1999) is intimately related to controlling the synthesis parameters yielding the desired material properties. In response to this demand, MIPs have been prepared by conventional bulk polymerization, suspension polymerization in water (Lai et al., 2001), liquid peruorocarbon (Mayes and Mosbach, 1996), or mineral oil (Kempe and Kempe, 2004), dispersion polymerization (Say et al., 2003), aqueous two-step swelling polymerization (Piscopo et al., 2002), and precipitation polymerization (de Boer et al., 2002; Li et al., 2003; Ye et al., 2000; Zhang et al., 2003; Zhang et al., 2004). Among these methods, precipitation polymerization is probably the most facile synthetic strategy providing imprinted spherical particles with sufcient control on the desired characteristics. In the present study, MIP particles with dimensions ranging from 3 to 5 m were synthesized by precipitation polymerization according to a method previously reported by our research group (Wei et al., 2006). Smaller particles (<1 m) with narrow size distribution have also been prepared by precipitation polymerization with a lower amount of cross-linker (Ye et al., 2000). To evaluate the binding site properties of spherical particles and bulk polymers, equilibrium binding studies, non-equilibrium binding studies, and release studies were performed to understand the rebinding properties of the imprinted polymer matrices at different conditions. More widespread use of MIP materials is frequently limited by the achievable binding properties. In general, MIPs are characterized by comparatively low binding afnities and a high degree of binding site heterogeneity in contrast to natural receptors, which somewhat restricts their analytical application in e.g. solid phase extraction, chromatography, binding assays, and sensing. Hence, improving and thoroughly characterizing the binding properties of MIPs should be a driving force toward next-generation MIP technology. Batch rebinding studies are a characterization method providing rst insight into the binding properties of a specic MIP. The most common approach for estimating the binding parameters is to assume a bimodal distribution of binding sites with the binding parameters derived from a Scatchard plot (Matsui et al., 1995; Subat et al., 2004; Zhu et al., 2002). Thereby, two binding afnity constants are obtained from the tting parameters. However, MIPs are commonly characterized by low average binding afnities and a high degree of binding site heterogeneity (Andersson et al., 1996). The degree of heterogeneity is of particular importance for MIP applications in separation techniques is a main source of chromatographic peak asymmetry and peak tailing in the resulting chromatograms (Sellergren and Shea, 1995). Furthermore, binding site heterogeneity is among the main parameters responsible for cross-reactivity in sensing applications (Allender et al., 1997), and for applications in catalysis (Wulff, 2002). Consequently, in this study a more universal approach based on the Freundlich isotherm (Rampey et al., 2004) is applied to characterize the binding site heterogeneity providing a more generic model for MIP materials. The group of Shimizu has introduced the method of modeling the adsorption isotherms of non-covalent imprinted polymers (Umpleby et al., 2001). Compared to the 2-binding site model (Matsui et al., 1995), the Freundlich isotherm is a heteroge-

neous binding model, which can inherently accommodate and evaluate the heterogeneity of MIPs. Fitting of experimental isotherms to the Freundlich model provided excellent agreement for MIPs against l-phenylalanine anilide (Szabelski et al., 2002), aminoantipyrine (Yang et al., 2003), hemoglobin (Guo et al., 2004), and monocrotophos (Zhu et al., 2005). Furthermore, it was demonstrated that this model is applicable to isotherms measured at low concentration or sub-saturation levels (Rampey et al., 2004; Rushton et al., 2005). The empirical form of the Freundlich isotherm (Freundlich, 1926) has been widely used for modeling heterogeneous surfaces including activated carbon (Khan et al., 2000), silica (Burris et al., 1991), sediment materials (Brownawell et al., 1997), and -cyclodextrin polymer (Murai et al., 1998). The Freundlich binding isotherm describes the relationship of the concentration of bound (B) and free (F) guest molecules following B = aF m , (1)

where a is related to the median association constant K0 = a1/m , and m is the heterogeneity index (Jaroniec and Madey, 1988). Accordingly, the afnity distribution can be calculated from a and m. The tting constants a and m can be applied to the calculation of corresponding afnity distributions using N(K) = 2.303am(1 m2 ) e2.303m log K Kmin = 1 1 , Kmax = Fmax Fmin (2)

This equation is only valid within a certain range of binding afnities. Kmin and Kmax are determined by the free concentration (Fmax and Fmin ) during the binding experiments. In the present study, this strategy was adapted for comparing the binding parameters between different imprinted polymer formats. Furthermore, these results are complemented by pore size and surface area studies on the synthesized imprinted polymer materials, as provided by BrunauerEmmettTeller (BET) analysis (Karlsson et al., 2004; Piletsky et al., 2000; Wei et al., 2006). Finally, these comprehensive investigations enabled correlating the obtained porosities for different imprinted polymer formats (particulates, microspheres, and sub-microspheres) with differences observed during rebinding studies, which have not been reported to date. 2. Materials and methods 2.1. Materials Methacrylic acid (MAA), ethylene glycoldimethacrylate (EGDMA), divinylbenzene (DVB), and were purchased from SigmaAldrich (Milwaukee, WI, USA) and puried by distillation. 2,2 -Azobis(2-isobutyronitrile) (AIBN) was purchased from SigmaAldrich and puried by recrystallization. 17 Estradiol, 17 -estradiol, and estrone were purchased from SigmaAldrich and used as supplied. All solvents were of HPLC grade from VWR International Inc. (Suwannee, GA, USA).

S. Wei, B. Mizaikoff / Biosensors and Bioelectronics 23 (2007) 201209

203

2.2. Preparation of bulk imprinted polymers and micro-/sub-microspheres 17 -Estradiol imprinted polymers were synthesized according to methods previously reported by our research group (Wei et al., 2006). Typically, DVB (40 mmol) and MAA (8 mmol) were added to a solution of 17 -estradiol (1 mmol) and AIBN (0.96 mmol) in porogenic solvent (6 mL acetone for bulk polymer; 60 mL mixture of toluene and acetonitrile (1:3, v:v) for microspheres) in a screw-cap vial. For sub-microspheres, EGDMA (6.7 mmol) and MAA (8 mmol) were added to a solution of 17 -estradiol (1 mmol) and AIBN (0.29 mmol) in porogenic solvent (40 mL mixture of acetone and acetonitrile (1:3, v:v) in a screw-cap vial. The resulting solutions were sonicated and deoxygenated with nitrogen for 5 min, and then thermally polymerized at 60 C for 24 h (70 C for microspheres). For imprinted bulk polymers, the resulting materials were ground and wet-sieved with acetone through a 25 m sieve. Sedimentation in acetone was applied for removing the remaining nes. For the synthesis of microspheres and sub-microspheres, the obtained materials were separated from the reaction medium by ltration or centrifugation. All obtained MIP materials were successively washed three times with methanol:acetic acid (85:15, v:v), methanol, and acetonitrile, respectively. Each control polymer was prepared by exactly the same synthetic route as the imprinted material, however, in absence of the template. 2.3. Scanning electron microscopy and optical microscopy Imprinted micro- and sub-microspheres were deposited onto silicon slides, and coated with 15 nm of gold using a thermal evaporator (Denton DV-502A, Denton Vacuum, Moorestown, NJ, USA). Scanning electron micrographs were obtained at 25 kV (SEM 1530, thermally assisted FEG, LEO, Oberkochen, Germany). Optical micrographs were obtained by depositing the particles onto glass slides, and recording digital images with an optical microscope (BX41, Olympus Optical Co. Ltd., Tokyo, Japan). 2.4. Surface area analysis The porosity and surface area of the developed micro-/submicrospheres and bulk polymers was investigated by nitrogen adsorption/desorption analysis using a nitrogen surface area analyzer (SA3100, Beckman Coulter, Hialeah, FL, USA). 2.5. Equilibrium binding isotherm studies Ten different solutions of 17 -estradiol in acetonitrile covering the concentration range from 0.01 to 2 mM were prepared. Twenty milligrams of polymer were suspended in 1 mL of 17 estradiol solution in 1.5 mL polypropylene centrifuge tubes. The tubes were shaken for 24 h, and centrifuged at 10,000 rpm for 15 min using a VWR Galaxy 16 DH centrifuge (VWR Inter-

national, Atlanta). UV spectral data was collected at 280 nm from the supernatants using a HPLC separation system with a reverse phase column (Kromasil C18) for quantifying the concentration of free estradiol (F). The mobile phase was acetonitrile:water (77:23). The bound amount (B) was calculated by subtracting F from the initial estradiol concentration. Experimental binding isotherms were analyzed using the Freundlich isotherm-afnity distribution (FIAD) method (Rampey et al., 2004): (i) the experimental data were plotted as log B versus log F. (ii) The Freundlich isotherm was tted to the log-plot obtained for the experimental adsorption isotherms using a spreadsheet calculation software (Microsoft Excel 2002, Microsoft Corporation, Redmond, WA, USA) by optimizing the coefcient of determination (R2 ) based on variation of the tting parameters a and m. (iii) An afnity distribution was calculated by inserting the obtained parameters m and a for the best t into Eq. (2). The valid afnity range is dictated by the minimum (Fmin ) and maximum (Fmax ) free concentration during the experimental binding isotherm studies. 2.6. Chromatographic analysis Bulk polymer particles and microspheres were suspended in acetone and packed into 150 mm 4.6 mm stainless steel HPLC columns with a slurry packer (Alltech 1666, Deereld, IL, USA) using acetone as the packing solvent. Chromatographic analysis was performed using a HPLC system equipped with a UVvis diode array detector (Dionex P580 pump, UVD 340S Detector, Sunnyvale, CA, USA). 2.7. Non-equilibrium binding isotherm studies The polymer particles (20 mg) were packed in an empty solid-phase extraction cartridge. Polyethylene frits (Argonaut Technologies, Foster City, CA, USA) were placed at both ends. Prior to and between uses, the cartridges were conditioned successively with 20 mL of methanol containing 15% of acetic acid, 20 mL of methanol, 20 mL of acetonitrile, and 20 mL of de-ionized water. One milliliter of 17 -estradiol solution with a concentration ranging from 10 M to 2 mM in acetonitrile was loaded onto the cartridge, and the polymer was washed with 0.6 mL of acetonitrile to remove and 17 -estradiol that has non-specically bound to the polymeric matrix. Then, vacuum was applied for 30 min to dry the cartridges. Finally, the analytes were quantitatively eluted with 1.5 mL of methanol. The obtained fraction was evaporated to dryness and re-dissolved in 0.5 mL of acetonitrile:water (77:23). The loading, washing, and eluting steps were performed at a constant ow rate of 2.5 mL/min. The analyte concentrations in the nal fraction, which represent the amount of analyte specically bound to the polymer (B), were determined by HPLC-UV analysis, as described in Section 2.6. The amount of unbound analyte (F) was obtained by subtracting B from that of initial analyte concentration of the samples loaded onto the polymers.

204

S. Wei, B. Mizaikoff / Biosensors and Bioelectronics 23 (2007) 201209

2.8. Release studies The polymer particles (20 mg) were incubated with 1 mL of 2 mM 17 -estradiol solution for 24 h. Unbound 17 -estradiol was separated from the polymer material by centrifugation at 13,000 rpm for 5 min. In the following, 1 mL of fresh acetonitrile was incubated with the bound polymer particles, and the amount of 17 -estradiol released from the imprinted polymer matrices was determined during time-resolved release experiments quantied by HPLC-UV analysis, as described in Section 2.6. 3. Results and discussions 3.1. Polymer synthesis and characterization In order to facilitate the interpretation of the results reported in this study, the polymerization methods have been adapted for achieving suitable conditions aiming at the most objective comparison between the different MIP materials. For instance, the type and ratio of functional monomer and cross-linker used for the preparation of microshperes was similar to the preparation of bulk polymers. However, toluene/acetonitrile instead of acetone had to be used as porogenic solvent for MIPs synthesized by the precipitation method facilitating the creation of monodisperse microspheres. To obtain spheres with size of 400 nm using the poly(MAA-co-DVB) system, the ratio of MAA:DVB should be reduced to 4:1 (Bai et al., 2006). However, the resulting exibility of the polymer chains corrupts the rigidity of the polymer matrix and does not maintain binding afnity in the nally synthesized MIP (Yilmaz et al., 1999) considering that DVB has only one vinyl group to ensure the backbone rigidity. For the synthesis of spherical particles and as there is only considerably weak interaction between MAA and EGDMA (Ni and Kawaguchi, 2004), poly(MAA-co-EGDMA) tends to form particles with smaller dimensions, while poly(MAA-co-DVB) tends toward particles of larger dimensions if the same monomer:cross-linker ratio is applied. Therefore, poly(MAA-co-EGDMA) was selected for the preparation of sub-microspheres in the present work. The rebinding properties of imprinted materials largely rely on template-functional monomer interactions, while the cross-

linker contributes only in a limited fashion to the specicity for the template; consequently, the observed binding behavior discussed in the present study is largely independent of the type of cross-linker. Accordingly, a lower amount of crosslinker had to be used for the preparation of sub-microspheres by the precipitation method. Evidently, these differences during MIP synthesis could affect the nally obtained polymer morphology and the specic interactions between template and functional monomers during the generation of imprinted and control bulk/particulate, microsphere, and sub-microsphere materials. The bulk materials were crushed, sieved, and sedimented for obtaining particles with dimensions <25 m. The morphologies of the polymer materials obtained via different synthetic routes were assessed by optical microscopy and SEM studies. The crushed bulk polymers provide irregularly shaped particles, as shown in Fig. 1A. Precipitation polymerization results in a nearly monodisperse population of spherical particles with diameters in a range of 35 m (Fig. 1B). Performing the precipitation polymerization with a reduced amount of crosslinker and a decreased monomer concentration yielded spherical particles with average diameters of approximately 400 nm (Fig. 1C). 3.2. Equilibrium binding studies For equilibrium binding assays, xed amounts of imprinted polymers (20 mg) were incubated with different concentrations of 17 -estradiol in acetonitrile. After 24 h, supernatants were obtained by centrifugation, and the amount of 17 -estradiol was determined by reversed phase HPLC. Binding isotherms were derived from the amounts of 17 -estradiol bound to the imprinted polymer material by plotting against the initial concentration of the incubation solution. All equilibrium binding isotherms obtained from imprinted polymers were characterized by linear loglog binding relationships. It is evident that signicantly less 17 -estradiol is bound to the control polymers (Fig. 2). Furthermore, the control microspheres ( ) absorbed less 17 -estradiol than the controls for the particulate material and for the sub-microspheres. This result indicates that there is less non-specic rebinding to the MIP microsphere material, which may result from the formation of porous microspheres

Fig. 1. Optical and SEM micrographs of imprinted polymers obtained by different polymerization methods. (A) Bulk polymer, optical micrograph; (B) microspheres, SEM; (C) sub-microspheres, SEM.

S. Wei, B. Mizaikoff / Biosensors and Bioelectronics 23 (2007) 201209

205

Fig. 2. Equilibrium binding isotherms for imprinted particulate bulk polymer ( ), microspheres ( ), and sub-microspheres ( ), as well as control particulate bulk polymer ( ), microspheres ( ), and sub-microspheres ( ) in loglog format (symbols , , , , , and represent the experimental data; lines are Freundlich tting functions).

with a large number of micropores, thereby reducing the binding site accessibility. Besides the mechanical forces incident at the bulk material, the preparation strategy for the bulk polymers and the microspheres were very similar (i.e., similar type of functional monomer and cross-linker, and similar ratio tempi ate: functional monomer:cross-linker), and should therefore yield comparable results during binding studies. The experimental data for MIP particles resulting from the bulk polymer, and the microsphere data ts well to the Freundlich model. The goodness of the t (R2 value) is used as quantitative descriptor for characterizing the difference between the model and the experimental data following R2 = 1 sum[(log B log aF m )2 ] sum[(log B average log B)2 ] (3)

appear more homogeneous than bulk polymer particles, as derived by comparing their heterogeneity indices (0.676 for microspheres and 0.605 for bulk polymer, respectively). The imprinted sub-microspheres were prepared with EGDMA as cross-linker using a reduced ratio for the crosslinker in the pre-polymerization solution. The equilibrium binding studies reveal a lower median binding afnity constant (0.0025 mM1 ) and heterogeneity index (0.492), which indicates that the binding sites provide by imprinted submicrospheres have a lower afnity and a higher heterogeneity in comparison to binding sites available in microspheres and particles generated from bulk polymer. The corresponding control polymers for each synthetic route were analyzed by precisely the same equilibrium binding procedure. The heterogeneity index (Table 1) reveals that the control polymers are generically more heterogeneous than the imprinted polymers, which is in contrast to results reported by Rampey et al. (2004). In their studies, the imprinted polymers appear more heterogeneous than the control polymers. This circumstance conrms the importance of thorough analytical studies for each MIP approach, as templated materials prepared by different synthetic routes and for different templates apparently yield different results. Moreover, the binding results also indicate that the control polymers have very small binding afnity, with only minimal uctuations (ranging from 1.7 103 to 4.1 103 mM) if the initial concentration of the binding solution changes from 0.01 to 2 mM, which indicates that the control polymers are saturated within this concentration range. The experimental data obtained for the control polymers does not t well to a Freundlich binding isotherms, as indicated by a R2 value <0.9. This result is consistent with the expected deviation of the experimental binding isotherm from the Freundlich isotherm at higher analyte-to-polymer ratios, as this model is only valid at low concentrations or at sub-saturation levels (Rampey et al., 2004; Rushton et al., 2005). 3.3. Porosity studies The porosities of the developed micro-/sub-microspheres and particulate bulk polymers were determined by nitrogen adsorption/desorption analysis of BrunauerEmmettTeller multi-point adsorption isotherms (Fig. 3). The substantial difference in porosity between poly(MAA-co-DVB) (bulk and microspheres), and poly(MAA-co-EGDMA) (submicrospheres) is mainly attributed to the reduced effective cross-linker level (Li et al., 1999). For low surface areas, as characteristic for sub-microspheres, the amount of gas adsorbed

The tting data for the afnity constants and heterogeneity indices are provided in Table 1. The obtained R2 value for the particulate imprinted polymer is 0.969, and 0.973 for the imprinted microspheres, respectively. Following the binding isotherm, it is evident that the microsphere MIP is characterized by a higher median binding afnity constant (1.3 102 4.8 104 mM1 ), than the particulate MIP generated from bulk polymer synthesis (9.0 103 5.3 104 mM1 ). Moreover, the microspheres
Table 1 Freundlich tting parameters for the investigated imprinted and control polymers Fitting data a m (STD)a R2 K0 (mM1 )
a b

Bulk MIP 0.060 (0.003) 0.606 (0.001) 0.969 (0.021) 0.0090 (0.00053)

Bulk control 0.0042 0.157 (NM)b 0.693 (NM)b 7.3E16 (NM)b (NM)b

Microspheres MIP 0.053 (0.002) 0.673 (0.004) 0.973 (0.023) 0.013 (0.00048)

Microspheres control 0.0025 0.000 (NM)b 0.00280 (NM)b (NM)b

Sub-microspheres MIP 0.052 (0.004) 0.492 (0.008) 0.940 (0.053) 0.0025 (0.00045)

Sub-microspheres control 0.0072 (NM)b 0.157 (NM)b 0.811 (NM)b 2.3E14 (NM)b

(STD)a

STD: Standard deviation. NM: Not measured.

206

S. Wei, B. Mizaikoff / Biosensors and Bioelectronics 23 (2007) 201209

Fig. 3. Nitrogen adsorption BET isotherms of imprinted sub-microspheres ( ), control sub-microspheres ( ), imprinted microspheres ( ), control microspheres ( ), particulate bulk imprinted polymer ( ), and particulate bulk control polymer ( ).

Fig. 4. Afnity distribution of sub-microspheres (black solid line), microspheres (red dashed line), and particulate bulk polymer (blue dotted line). (For interpretation of the references to color in this gure legend, the reader is referred to the web version of the article.)

onto the material is relatively low. For the adsorption isotherms recorded at microspheres, small pores apparently slow down the diffusion of gas molecules into the material. The shape of the adsorption isotherm of the particulate bulk polymers indicates that the sorbent is dominated by a mesoporous structure (Sellergren, 2001). Table 2 summarizes the pore volume and specic surface area of the polymer particles investigated during this study. It is clearly evident that sub-microspheres are characterized by a much smaller pore volume and surface area, as compared to bulk polymers and micropsheres. Based on the studies reported here, we infer a correlation between the binding site properties determined from the Freundlich model based on equilibrium binding assays, and the porosities of the imprinted materials, which has not been reported to date. Interestingly, Fig. 4 shows that the binding site distribution generated within differently synthesized materials is apparently quite similar. If the specic binding sites are predominantly located at the surface of the imprinted materials, it is expected that there are less specic binding sites available at particles with smaller pore volume and surface area, given similar distribution of the binding sites, as conrmed by the equilibrium binding isotherm studies. To support this hypothesis, the binding afnity constants of the imprinted polymers were compared. As expected, the sub-microspheres provide a lower median binding afnity constant (2.5 103 4.5 104 mM1 ) than the bulk polymers (9.0 103 5.3 104 mM1 ), than the microspheres

(1.3 102 4.8 104 mM1 ), as derived from the tting parameters of the Freundlich isotherm. Accordingly, the surface areas of these imprinted particles can be ranked by increasing surface area as sub-microspheres (10.30 m2 /g), particulates generated from bulk polymer (552.89 m2 /g), and microspheres (706.98 m2 /g), which is consistent with the hypothesis that less specic binding sites are available at particles with smaller surface area considering the binding site distributions are similar. As much less cross linker was used during the synthesis of the templated sub-microsphere materials, these results suggest that the concentration of the cross linker is crucial to the pore structure of imprinted polymer materials, and to their binding site distribution. With increasing amounts of cross-linker, apparently less pore volume were created during the polymerization process. However, it is essential to provide sufcient fractions of crosslinker for maintaining the integrity of the generated binding sites. These ndings clearly conrm that optimizing the fraction of cross-linker is essential for maintaining the balance between sufcient rigidity of the material maintaining the structural integrity of binding sites, and sufcient porosity of the resulting polymer material during rational design of next-generation MIP technology. The pore size and surface area are of particular importance for the application of MIPs as stationary phase materials, as shown by Quaglia et al. (2003) the porosity of the MIP affects the stationary phase performance in HPLC experiments. Schmidt and

Table 2 Porosities of molecularly imprinted polymers prepared as particulate bulk material, microspheres, and sub-microspheres as determined by BET analysis Sample Total pore volume (mL g1 ) BET surface area (m2 g1 ) Micropore volume (mL g1 ) Micropore specic surface area (m2 g1 )
a

Bulk 0.639 552.89 0.0527 131.40

Microspheres 0.382 706.98 0.137 320.04

Sub-microspheres 0.0382 10.30 a a

At pore diameters of 720 A, the micropore volume and specic surface area is not detectable by gas adsorption measurements.

S. Wei, B. Mizaikoff / Biosensors and Bioelectronics 23 (2007) 201209

207

Fig. 5. (A) Pore sized distribution of particulate bulk imprinted polymer ( ), imprinted microspheres ( ), and imprinted sub-microspheres ( ). (B) Chromatographic separation of estrone (peak 1), 17 -estradiol (peak 2) and 17 -estradiol (peak 3) using the particulate bulk imprinted polymer (red) and the imprinted microspheres (black) as the stationary phase. Acetonitrile containing 0.5% acetic acid was used as the mobile phase at a ow rate of 0.6 mL/min. Analytes were monitored at 280 nm. (For interpretation of the references to color in this gure legend, the reader is referred to the web version of the article.)

Haupt (2005) studied the ability of MIP lms to rebind target analytes as function of the concentration and molecular weight of the polymer porogen. Fig. 5A shows the pore size distribution of the developed micro-/sub-microspheres and particulate bulk polymers. The pore size distribution of the synthesized microspheres (Fig. 5A) reveals that the pore dimensions predominantly cluster in the range of 06 nm. In comparison, the pore size distribution of the synthesized bulk polymers shows higher portion (815%) in the range of 820 nm than that in the microspheres (46%). It can be seen from Table 2 that the total pore volume of bulk polymer is higher than that of the microspheres, so it can be concluded that the bulk polymer has much more pores in the range of 820 nm comparing to the microspheres. The particle size of the sub-microspheres is not suitable for normal HPLC analysis, so the performance of the sub-microspheres in HPLC is not discussed here. As we know, to have strong retentivity and good peak shape in HPLC analysis, the pore size in the range of 100180 A (Supelco HPLC Column Selection Guide) is usually chose for the HPLC analysis of small molecules and biomolecules, suggesting the synthesized imprinted bulk polymer is more suitable to use as HPLC stationary phase. Fig. 5B shows the chromatographic separation of estrone, 17 -estradiol and 17 -estradiol using the bulk imprinted polymer and the imprinted microspheres as the stationary phase. On the bulkimprinted polymer, all analytes get fast elution and complete separation. The small pores (less than 6 nm) in the imprinted microspheres show excellent retention properties for the target analytes, however, the separation of three structurally related compounds is inadequate due to the slow elution. These ndings indicate that the application of MIPs as stationary phase materials should thoroughly consider the pore size distribution within the MIP material for obtaining the desired HPLC performance. 3.4. Release studies The kinetics of binding experiments were studied by release experiments. The polymer materials, which have been equilibrated with 17 -estradiol, were incubated under shaking with fresh acetonitrile to release the re-bound template, and the released amount was temporally monitored. Fig. 6 shows

the release of 17 -estradiol from imprinted particulate bulk polymer, microspheres, and sub-microspheres. There is little evident difference in release kinetics between different polymer particle dimensions at these conditions. Microspheres and sub-microspheres release more 17 -estradiol than bulk polymer particles, which may results from an increased mobility of the smaller spherical particles in solution providing more advantageous extraction conditions. Furthermore, it is evident for all materials that during the rst 2 h, 7080% of 17 -estradiol were released, which relates to target molecules that have been bound with lowest afnity to the polymer matrix. The more tightly bound molecules were released at a signicantly longer timescale after continuous incubation, until complete extraction was achieved after 40 h. From the calculation of binding site distribution (Fig. 4), it is evident that the number of low afnity binding sites (K = 0.028 mM) is 20 times higher than the number of high afnity binding sites (K = 2.13 mM) for particulates generated from bulk MIPs, which is consistent with the conclusion that there is a dominating amount of low afnity binding sites, as observed from the release experiment. As equal amounts of polymer and solvent were used for the kinetic studies at all MIP formats, it can be concluded that the binding site distribution for the three MIP particle formats studied here are comparable. This conclusion is supported by the results obtained during the equilibrium binding studies, which are shown in Fig. 4 and conrm

Fig. 6. Percentage of released 17 -estradiol (R/R0 ) from microspheres ( ), sub-microspheres ( ), and particulate bulk polymer ( ).

208

S. Wei, B. Mizaikoff / Biosensors and Bioelectronics 23 (2007) 201209

that the three types of particles have similar binding site afnity distributions during equilibrium binding. 3.5. Non-equilibrium binding studies The analysis of equilibrium binding isotherms provides useful information for binding assays, however, is of limited value for non-equilibrium applications of MIPs, such as chromatographic separations including HPLC and SPE. Hence, non-equilibrium binding isotherms have to be established for characterizing the templated material behavior at these conditions. Here, non-equilibrium binding analysis was performed using a molecularly imprinted solid phase extraction (MISPE) set-up. As the lters closing off SPE cartridges have a pore size of 10 m ensuring rapid ow of solvent through the stationary phase material, only particulates generated from bulk polymers were tested in these experiments. Hundred milligrams of polymer particles were packed into an empty SPE cartridge, and sequentially conditioned with methanol, water, and acetonitrile, respectively. In the following, different concentrations of 17 -estradiol solutions were percolated through the MISPE cartridge. 17 -Estradiol in the eluate was considered the bound fraction (B) while the free concentration (F) was calculated by subtracting B from the initial 17 -estradiol concentration in the sample solution. Fig. 7 shows the Freundlich adsorption isotherms of these non-equilibrium binding experiments. It is evident that the Freundlich model does not t the observed binding behavior (R2 = 0.924) with the same delity determined during equilibrium binding. As a rst approximation, information on the binding parameters was still derived from these ts. The median binding afnity constant determined during these non-equilibrium experiments (0.061 mM1 ) is higher than the afnity constant obtained from the equilibrium experiments (0.010 mM1 ). Also, the binding site distribution appears more homogeneous, as reected in a heterogeneity index of m = 0.725. A possible explanation for this difference in equilibrium versus non-equilibrium binding is related to the observed binding kinetics. It is hypothesized that the high afnity binding sites are occupied rst, followed by the low afnity binding sites.

Hence, it is derived that during the non-equilibrium binding conditions a signicant fraction of low afnity binding sites remains unoccupied, which explains the higher median binding afnity constant and the higher heterogeneity index derived from these experiments. 4. Conclusions In summary, a comparison of the specic binding properties of imprinted polymers for 17 -estradiol prepared by different synthetic routes yielding particulates generated from bulk polymers, microspheres, and sub-microspheres was obtained. The relationships between the particle porosity and rebinding properties were detailed, providing useful guidelines for controlling the particle properties for the desired application including, SPE pre-concentration, HPLC separations, and biomimetic binding assays. While microspheres prepared by precipitation polymerization revealed excellent rebinding properties at equilibrium binding conditions, it was also conrmed that the porosity of imprinted polymers plays an important role in their rebinding characteristics. Particulate polymer materials with larger surface area provide superior binding, if the binding site distribution is similar between several generate MIP formats. Furthermore, the importance of an optimized degree of cross-linking was conrmed for maintaining the balance between structural rigidity of the binding sites, and porosity of the resultant polymers. Exemplarily, it was shown that application of MIPs as HPLC stationary phase material requires particle pore dimensions the range of 100180 A for maintaining the desired retention properties at acceptable elution times. While these particulate bulk imprinted polymers revealed excellent HPLC performance, their binding afnity was less compared to imprinted microspheres, as determined during equilibrium binding studies. In general, the performed release experiments and the derived binding site distribution revealed that all MIP formats are characterized by a majority of low afnity binding sites. Furthermore, it has been found that the rebinding properties of the synthesized materials are different at equilibrium versus non-equilibrium conditions conrming that thorough analytical characterization of each MIP system is essential to the nal application conditions. As the

Fig. 7. Non-equilibrium binding isotherms for particulate bulk imprinted polymer (A, : experimental data, red line: Freundlich tting), and control polymer (B, : experimental data) in loglog format. (For interpretation of the references to color in this gure legend, the reader is referred to the web version of the article.)

S. Wei, B. Mizaikoff / Biosensors and Bioelectronics 23 (2007) 201209

209

optimization of imprinted materials is based on fundamental understanding of the binding site properties, the investigations presented here will aid in establishing a more rational basis for further tailoring imprinted materials to the desired analytical application. Acknowledgements This work was in part supported by the European Union (Quality of Life and Management of Living Resources, #QLK4CT2002-02323), and the U.S. Geological Survey (National Water Quality Assessment Program, #2002GA30G). References
Allender, C.J., Brain, K.R., Heard, C.M., 1997. Chirality 9 (3), 233237. Andersson, H.S., Koch-Schmidt, A.C., Ohlson, S., Mosbach, K., 1996. J. Mol. Recognit. 9 (56), 675682. Ansell, R.J., 2004. J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 804 (1), 151165. Arshady, R., Mosbach, K., 1981. Macromol. Chem. Phys. 182 (2), 687692. Bai, F., Yang, X., Li, R., Huang, B., Huang, W., 2006. Polymer 47 (16), 57755784. Brownawell, B.J., Chen, H., Zhang, W.J., Westall, J.C., 1997. Environ. Sci. Technol. 31, 1735. Burris, D.R., Antworth, C.P., Stauffer, T.B., Macintyre, W.G., 1991. Environ. Toxicol. Chem. 10, 433. de Boer, T., Mol, R., de Zeeuw, R.A., de Jong, G.J., Sherrington, D.C., Cormack, P.A.G., Ensing, K., 2002. Electrophoresis 23 (9), 12961300. Freundlich, H., 1926. Colloid and Capillary Chemistry. Methuen, London. Greene, N.T., Shimizu, K.D., 2005. J. Am. Chem. Soc. 127 (15), 56955700. Guo, T.Y., Xia, Y.Q., Hao, G.J., Song, M.D., Zhang, B.H., 2004. Biomaterials 25 (27), 59055912. Jaroniec, M., Madey, R., 1988. Physical Adsorption on Heterogeneous Solids. Elsevier, New York. Karlsson, J.G., Karlsson, B., Andersson, L.I., Nicholls, I.A., 2004. Analyst (Cambridge, United Kingdom) 129 (5), 456462. Kempe, H., Kempe, M., 2004. Macromol. Rapid Commun. 25, 315320. Khan, A.R., Riazi, M.R., Al-Roomi, Y.A., 2000. Sep. Purif. Technol. 18, 237. Lai, J.P., Lu, X.Y., Lu, C.Y., Ju, H.F., He, X.W., 2001. Anal. Chim. Ada. 442 (1), 105111. Li, P., Rong, F., Yuan, C., 2003. Polym. Int. 52 (12), 17991806. Li, W.-H., Li, K., Stover, H.D.H., 1999. J. Polym. Sci. Polym. Chem. 37 (14), 22952303. Matsui, J., Miyoshi, Y., Doblhoff-Dier, O., Takeuchi, T., 1995. Anal. Chem. 67 (23), 44044408. Mayes, A.G., Mosbach, K., 1996. Anal. Chem. 68 (21), 37693774.

Murai, S., Imajo, S., Takasu, Y., Takahashi, K., Hattori, K., 1998. Environ. Sci. Technol. 32, 782. Ni, H., Kawaguchi, H., 2004. J. Polym. Sci. Part A: Polym. Chem. 42 (11), 28332844. Piletsky, S.A., Matuschewski, H., Schedler, U., Wilpert, A., Piletska, E.V., Thiele, T.A., Ulbricht, M., 2000. Macromolecules 33 (8), 30923098. Piscopo, L., Prandi, C., Coppa, M., Sparnacci, K., Laus, M., Lagana, A., Curini, R., DAscenzo, G., 2002. Macromol. Chem. Phys. 203 (10/11), 15321538. Quaglia, M., De Lorenzi, E., Sulitzky, C., Caccialanza, G., Sellergren, B., 2003. Electrophoresis 24 (6), 952957. Rampey, A.M., Umpleby II, R.J., Rushton, G.T., Iseman, J.C., Shah, R.N., Shimizu, K.D., 2004. Anal. Chem. 76 (4), 11231133. Remcho, V.T., Tan, Z.J., 1999. Anal. Chem. 71 (7), 248A255A. Rushton, G.T., Karns, C.L., Shimizu, K.D., 2005. Anal. Chim. Acta 528 (1), 107113. Say, R., Birlik, E., Ersoz, A., Yilmaz, F., Gedikbey, T., Denizli, A., 2003. Anal. Chim. Acta 480 (2), 251258. Schmidt, R.H., Haupt, K., 2005. Chem. Mater. 17 (5), 10071016. Sellergren, B., 2001. Molecularly Imprinted Polymers: Man-Made Mimics of Antibodies and Their Applications in Analytical Chemistry (Techniques and Instrumentation in Analytical Chemistry, vol. 23. Elsevier, Amsterdam; New York, p. 52. Sellergren, B., Shea, K.J., 1995. J. Chromatogr. A 690, 2939. Subat, M., Borovik Andrew, S., Konig, B., 2004. J. Am. Chem. Soc. 126 (10), 31853190. Supelco HPLC Column Selection Guide, http://www.sigmaaldrich.com/ supelco/general information/t405089 l.pdf. Szabelski, P., Kaczmarski, K., Cavazzini, A., Chen, Y.B., Sellergren, B., Guiochon, G., 2002. J. Chromatogr. A 964 (12), 99111. Umpleby, R.J., Baxter, S.C., Bode, M., Berch, J.K., Shah, R.N., Shimizu, K.D., 2001. Anal. Chim. Acta 435 (1), 3542. Vlatakis, G., Andersson, L.I., Muller, R., Mosbach, K., 1993. Nature 361 (6413), 645647. Wei, S., Molinelli, A., Mizaikoff, B., 2006. Biosens. Bioelectron. 21 (10), 19431951. Wulff, G., 2002. Chem. Rev. (Washington, DC) 102 (1), 127. Wulff, G., Grobe-Einsler, R., Vesper, W., Sarhan, A., 1977. Makromol. Chem. 178 (10), 28172825. Yang, G., Wang, D., Li, Z., Zhou, S., Chen, Y., 2003. Chromatographia 58 (1/2), 5358. Ye, L., Weiss, R., Mosbach, K., 2000. Macromolecules 33 (22), 82398245. Yilmaz, E., Mosbach, K., Haupt, K., 1999. Anal. Commun. 36 (5), 167170. Zhang, Z.-H., Long, Y.-M., Liu, Y.-J., Yao, S.-Z., 2003. Chin. J. Chem. 21 (5), 550555. Zhang, Z., Liu, Y., Long, Y., Me, L., Yao, S., 2004. Anal. Sci. 20 (2), 291295. Zhu, Q.-Z., Haupt, K., Knopp, D., Niessner, R., 2002. Anal. Chim. Acta 468 (2), 217227. Zhu, X., Yang, J., Su, Q., Cai, J., Gao, Y., 2005. Anal. Chim. 95 (1112), 877884.