Development of molecularly imprinted polymers as tailored templates for the solid-state [2+2] photodimerization

Biosensors and Bioelectronics 25 (2009) 640–646

Contents lists available at ScienceDirect

Biosensors and Bioelectronics journal homepage: www.elsevier.com/locate/bios

Development of molecularly imprinted polymers as tailored templates for the solid-state [2+2] photodimerization Xiangyang Wu, Ken D. Shimizu ∗ Department of Chemistry and Biochemistry, University of South Carolina, 631 Sumter Street, Columbia, SC 29208, USA

a r t i c l e

i n f o

Article history: Received 23 September 2008 Received in revised form 8 January 2009 Accepted 13 January 2009 Available online 4 February 2009 Keywords: Molecularly imprinted polymer Templated reaction [2+2] photodimerization

a b s t r a c t In this study, a molecularly imprinted polymer (MIP) was prepared to selectively template the [2+2] photodimerization of trans-1,2-bis(4-pyridyl)ethylene. First, an MIP selective for rctt-tetrakis(4pyridyl)cyclobutane, which is the [2+2] photodimerization product of trans-1,2-bis(4-pyridyl)ethylene, was prepared from methacrylic acid (MAA) and ethylene glycol dimethacrylate (EGDMA). The noncovalent MIP showed enhanced affinity for both the templating agent, rctt-tetrakis(4-pyridyl)cyclobutane, and the alkene precursor, trans-1,2-bis(4-pyridyl)ethylene. The solid-state photodimerization reaction proceeded in significantly higher yields in the presence of the MIP. Control reactions carried out in the absence of polymer gave no product, and reactions carried out in the presence of a non-imprinted polymer and an MIP imprinted with a different template, 3-hydroxymethylpyridine, gave much lower yields of the cyclobutane photodimerization product. The outcome of the MIP-templated photodimerization reaction was strongly influenced by the binding site heterogeneity of the non-covalently imprinted polymers. For example, higher yields were observed with decreasing olefin loadings levels on the MIPs. This binding site heterogeneity was characterized via application of the Freundlich binding model to the experimentally measured binding isotherms. These confirmed that the non-covalent MIPs had very few high-affinity binding sites, which greatly limits the capacity and ultimately the utility of these materials as templates in synthetic organic applications. © 2009 Elsevier B.V. All rights reserved.

1. Introduction The intermolecular [2+2] photodimerization of alkenes is a wellstudied reaction and is a key step in the synthesis of many important natural products and pharmaceutical candidates (Hopf et al., 1995). The distance and orientation of the two reactive olefins are important in determining the distribution of isomeric intramolecular and intermolecular photoproducts. For efficient formation of the [2+2] cyclobutane product, the distance between the double bonds should be less than 4.2 Å (Coates et al., 1998) and the double bonds should be in a parallel alignment. In order to catalyze and control the reaction, chemists have sought to design environments that can align alkenes into specific orientations prior to the photodimerization reaction both in solution and in the solid-state. For example, molecular receptors have been used to steer double bonds for reaction with high yield and selectivity in the solid-state (MacGillivray et al., 2000; Ramamurthy et al., 1992; Papaefstathiou et al., 2001) and in solution (Pattabiraman et al., 2005; Rao et al., 1999; Maddipatla et al., 2007). Porous inorganic and organic zeo-

∗ Corresponding author. Tel.: +1 803 777 6523; fax: +1 803 777 9521. E-mail address: [email protected] (K.D. Shimizu). 0956-5663/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.bios.2009.01.038

lites have also been used to guide the photodimerization reactions in the solid-state (Lalitha et al., 2000; Ramamurthy, 2000; Yang et al., 2006). Finally, self-assembled templates such as micelles and vesicles have been used as templates to control the photodimerization of olefins in solution (Nakamura et al., 1997; Lehnberger et al., 1997; Schütz and Wolff, 1997; Takagi et al., 1994). Each of these approaches has specific limitations. Templated [2+2] reactions in the solid-state often proceed with higher selectivities but were not readily adaptable to different substrates. Conversely, templated [2+2] reactions in solution were generally adaptable to a wider range of reactants but have lower selectivities. Therefore, there is need for a versatile template that can facilitate the [2+2] reaction in the solid-state. In this study, we have designed molecularly imprinted polymers (MIPs) that can act as templates for the [2+2] photoreaction in the solid-state. The advantage of using MIPs is that they can be efficiently prepared and tailored via the molecular imprinting process. The MIPs were specifically designed to facilitate the [2+2] photodimerization reaction (Scheme 1) of trans-1,2-bis(4-pyridyl)ethylene (trans-BPE) to form rctt-tetrakis(4pyridyl)cyclobutane (TPCB1). The requisite MIP was prepared using the product of the [2+2] reaction as a template to form the binding cavity (Visnjevski et al., 2004; Zhang et al., 2006). MIPs have

X. Wu, K.D. Shimizu / Biosensors and Bioelectronics 25 (2009) 640–646

641

Scheme 1. Formation of a molecularly imprinted polymer to facilitate the [2+2] photodimerization of trans-1,2-bis(4-pyridyl)ethylene (trans-BPE) to form rctt-tetrakis(4pyridyl)cyclobutane (TPCB1).

recently been utilized as nanoreactors (Wulff, 2002), as templates for organic transformations (Shea et al., 1980; Byström et al., 1993; Alexander et al., 1999; Carboni et al., 2008), and as enzyme mimics (Liu and Wulff, 2004a,b; Svenson et al., 2004; Mosbach et al., 2001). These synthetic polymers have good stability against heat, chemical, and harsh environments (Svenson and Nicholls, 2001). 2. Experimental 2.1. Materials and instrumentation Methacrylic acid (MAA), 2-(trifluoromethyl)acrylic acid (TFM), itaconic acid (ITA), ethylene glycol dimethacrylate (EGDMA), 2,2-azobisisobutyronitrile (AIBN), resorcinol, 3-pyridylcarbinol (PDC), and trans-1,2-bis(4-pyridyl)ethylene were purchased from Sigma–Aldrich (USA). All solvents were purchased from Fisher. UV measurements were taken on a Jasco V-530 UV–vis spectrophotometer or/and a Varian Prostar HPLC system. UV irradiation and photoreaction were conducted with a Hanovia medium-pressure 450-W mercury arc lamp. Proton NMR data was measured using a Varian Mercury/VX 300 or 400 MHz. 2.2. Synthesis of the template TPCB1 The synthesis of rctt-tetrakis(4-pyridyl)cyclobutane template followed the procedure reported by MacGillivray et al., 2000. Resorcinol (1.36 mmol, 150 mg) and trans-1,2-bis(4pyridyl)ethylene (1.36 mmol, 250 mg) were dissolved in hot (ethanol/acetonitrile = 1:1) with sonication. Slow evaporation of the mixture over 5 days resulted in co-crystal formation. The co-crystal was ground to a uniform powder and spread evenly between two glass plates. The solid-state photoreaction was conducted under irradiation with a medium-pressure mercury arc lamp (450 W) at 20 ◦ C for 24 h on each side. The product was purified by flash chromatography on silica gel (40% ethyl acetate/methanol) to yield the product (TPCB1) as a light yellow solid (212 mg, 85%). The proton NMR of the product was consistent with that reported by MacGillivray et al. 1 H NMR (CD3 COCD3 , 300 MHz) ı ppm: 8.32 (d, 8 H), 7.26 (d, 8 H), 4.78 (s, 4 H). 2.3. Polymer synthesis TPCB1 (1 mmol, 336 mg), carboxylic acid monomer (12 mmol per carboxylic acid), EGDMA (30 mmol, 5.65 mL), AIBN (1 mmol,

164 mg), and CHCl3 (20 mL) were placed in a screw cap vial and vortexed until all the components were dissolved. The vials were sonicated under nitrogen for 5 min to purge any dissolved oxygen, capped, and polymerized at 0 ◦ C under irradiation by a mercury arc lamp (450 W) for 12 h. Control polymers were also prepared in the same manner with PDC as the template (4 mmol, 436 mg) and in the absence of template. The polymer monoliths were crushed and washed by soxhlet extraction for 12 h with toluene/acetic acid (72:28) and then with acetonitrile/methanol (4:1). Finally, the polymers were ground and sieved, and the particles in the range of 36–65 ␮m were collected. 2.4. Batch binding studies Solutions of varying concentration of TPCB1 and trans-BPE in acetonitrile were prepared and their concentrations were verified by their absorbance at 286 nm. The ground, washed, and sieved polymer (75 mg) and 5 mL of the TPCB1 or trans-BPE solution were placed in a vial and shaken for 8 h. The supernatant was removed and filtered, and the concentration of TPCB1 or trans-BPE remaining in solution was measured by UV–vis. In order to quantitatively compare the capacity and affinity of the polymers, the binding isotherms were measured using the above batch binding protocols. The binding isotherms were plotted in concentrations of log bound (B) versus log free (F) analyte and were fitted to the Freundlich model (Eq. (1)). Where m and a are binding parameters. The preexponential factor a is a measure of the binding affinity (K) and the number of binding site (N). The second fitting parameter m is the heterogeneity index, which varies from zero to one, with higher values corresponding to more homogeneous systems: log B = mlog F + log a

(1)

In addition, the number of binding sites (N) and association constant (K) can be estimated for the subset of binding sites that are accessed within the concentration limits of the binding isotherm. The weighted average affinity constant (K¯ K1 −K2 ) and number of binding sites (NK1 −K2 ) for the polymers was calculated using Eqs. (2) and (3) (Umpleby et al., 2004). The Freundlich isotherm models only a portion of the complete binding isotherm within a specific concentration range, and thus, the average binding constants (K¯ K1 −K2 ), and number of binding sites (NK1 −K2 ) can only be estimated for a specific subset of the binding sites that have binding constants within the range of K1 –K2 . This range is set by the experimental

642

X. Wu, K.D. Shimizu / Biosensors and Bioelectronics 25 (2009) 640–646

binding isotherm (F1 –F2 ) as defined by Eq. (4): K¯ K1 −K2 =

m m−1

K11−m K1−m

− K21−m − K2−m

NK1 −K2 = a(1 − m2 )(K1−m − K2−m ) K1 =

1 , F1

K2 =

1 F2

(2) (3) (4)

2.5. Photoreaction of trans-BPE in the MIPs The polymers were loaded with trans-BPE by shaking with the polymer particles (200 mg) in an acetonitrile solution of trans-BPE for 8 h. The mixtures were either dried in vacuo or were filtered. The trans-BPE loaded polymer particles were spread evenly between two glass plates (10 cm × 20 cm). The photoreaction was conducted under irradiation with a mercury arc lamp for 8–72 h at 20 ◦ C. Then the polymers were collected and washed six times with methanol. The supernatants were combined and dried in vacuo. The yields and selectivities were characterized by integration of the 1 H NMR spectra of the reaction products in acetone-d6. The yields of the photodimers were calculated from the cyclobutane protons: 4.78 ppm (s, 4 H, TPCB1) and 3.95 ppm (s, 4 H, TPCB2). The yields of cis-BPE were calculated from the olefin protons: 6.83 ppm (s, 2 H, cis-BPE). And the yields of the intramolecular photoproduct was calculated from the measurement of the aryl proton: 9.12 (s, 2 H, PAH). The unreacted trans-BPE was estimated from the olefin protons: 7.52 ppm (s, 2 H, trans-BPE). The yields of the photodimerization products (TPCB1 and TPCB2) were calculated using Eq. (5). The selectivity (˛) was defined as the TPCB1 to TPCB2 ratio (Eq. (6)): TPCB1 + TPCB2 2(trans − BPE) + 2(cis − BPE) + 2(PAH) + TPCB1 + TPCB2 (5) TPCB1 ˛= (6) TPCB2 yield =

3. Results and discussion 3.1. Experimental design of the [2+2] photodimerization in an MIP UV irradiation of the olefin trans-BPE can yield a mixture of products (Scheme 2) including the photoisomerization product (cisBPE), the intramolecular photoproduct (PAH), and the isomeric [2+2] products (TPCB1 and TPCB2) (Maddipatla et al., 2007). Therefore, MIPs tailored to form TPCB1 were prepared using TPCB1 as the template (Scheme 1). After washing the MIP to remove the template molecule, the binding sites were reloaded with trans-BPE. Irradiation of the polymer bound trans-BPE was expected to give enhanced yields the desired cyclobutane product, TPCB1.

was selected for investigation because it had the potential to preorganize the carboxylic acids in the binding cavity. The two carboxylic acid groups in ITA are positioned at the proper distance to bind to two molecules of trans-BPE and to orient them in a favorable geometry for photodimerization. ITA has two carboxylic acids whereas MAA and TFM only have one. Thus, the ITA polymers were prepared with half as much ITA. 3.3. Batch binding studies for TPCB1 and trans-BPE The imprinting capabilities of each polymer matrix were evaluated by comparison of the binding capacities of the MIPs and NIPs for TPCB1. The binding studies were carried out by shaking the polymers in a 0.5 mM TPCB1 solution in acetonitrile. The unbound TPCB1 remaining in the supernantant was measured using a UV–vis spectrometer (286 nm). The relative binding capacities of the TMF, ITA, and MAA polymers are shown in Fig. 1. Only the imprinted polymer prepared with MAA showed a significant imprinting effect. This evident from the higher binding capacity of the MAA TPCB1 imprinted polymer (Fig. 1, white bar on the right) in comparison to the MAA NIP and MAA MIP-PDC. The magnitude of the increased binding capacity of MIP-TPCB1 was consistent with that generally observed in non-covalently imprinted MAA polymers. The MAA MIP-TPCB1 bound more than 50% more TPCB1 than the NIP in a 0.5 mM TPCB1 solution. The MAA MIP-TPCB1 also had the highest binding capacity of all the imprinted and non-imprinted polymers, which was surprising as MAA is less acidic than TFM and also has half the number of carboxylic acids of ITA. Both the ITA and TFM polymers did not show any significant imprinting effect. The ITA polymers showed significantly lower binding capacities and a slight imprinting effect as the ITA MIPTPCB1 (Fig. 1, white bar in the middle) had a similar capacity to the ITA NIP. The lower capacity of the ITA polymers suggests that both carboxylic acids may not be able to form effective hydrogen bonding interactions with the template, possibly because of steric reasons. The TFM matrix showed the highest binding affinity for TPCB1 as the TFM NIP (Fig. 1, black bar on the left) had a greater capacity than the ITA or MAA NIPs. However, the increased affinity of the TFM monomer did not result in an enhanced imprinting effect. Both the TFM MIP-TPCB1 and MIP-PDC had lower capacity than the TFM NIP. This lack of imprinting effect with the more acidic TFM monomer has been seen elsewhere with multidentate basic templates (Matsui et al., 1998). This may be due to the higher acid-

3.2. Optimization of the imprinting matrix First, a variety of functional monomers were tested for their abilities to effectively imprint the cyclobutane template, TPCB1. MIPs and non-imprinted polymers (NIPs) were prepared using different functional monomers (TFM, ITA, or MAA) (Fig. 1). Control MIPs were also prepared using a mismatched template, PDC, which has only one basic pyridine, in comparison to TPCB1 that has four basic pyridines. Therefore, four times as much PDC was used to prepare MIP-PDC. All the polymers were synthesized using EGDMA as the crosslinker and were polymerized under free radical conditions (AIBN, 0 ◦ C, UV). MAA was investigated because it is the most popular and widely utilized functional monomer for the preparation of non-covalently imprinted polymers. TFM was investigated due to greater acidity and thus, TFM had a greater probablility of forming complexes with the basic template, TPCB1 (Yilmaz et al., 1999). ITA

Fig. 1. Binding capacities of NIP, MIP-TPCB1, and MIP-PDC prepared with different functional monomers, TFM, ITA, and MAA, for 0.5 mM TPCB1 in acetonitrile. Standard errors in the measured binding percentages were ±5%.

X. Wu, K.D. Shimizu / Biosensors and Bioelectronics 25 (2009) 640–646

643

Scheme 2. (a) Products for the photoreaction of trans-BPE; (b) monomers (MAA, TFM, ITA), crosslinkers (EGDMA), and control molecular template (PDC) used to make the MIPs, NIPs, and control MIPs.

ity of TFM, which leads to proton transfer that leads to a stronger but less directional electrostatic monomer-template interactions (Hannon et al., 1997). Alternatively, the stronger acidity of TFM may catalyze the decomposition of the template molecule or the EGDMA crosslinker. Regardless of its origins, only the MAA polymers showed an appreciable imprinting effect for TPCB1. Therefore, the MAA polymers were chosen for use as polymeric templates in the photodimerization studies. The MAA polymers were first tested for their relative binding capacities for the alkene starting material, trans-BPE. MAA MIP-TPCB1 showed twice the binding capacity for trans-BPE (49% bound) than the corresponding NIP (23%) and also higher binding capacity than the MIP-PDC (39%). These binding studies with trans-BPE were carried out at lower concentrations (0.1 mM) because the polymers displayed lower affinity for transBPE than for the template TPCB1. This is not surprising because TPCB1 has twice as many basic pyridine groups as trans-BPE. 3.4. Photoreaction of trans-BPE in the MIP cavities Next, the ability of MIP-TPCB1 to facilitate the [2+2] photodimerization of trans-BPE was examined. Initially, the photodimerization reactions were carried out in solution in the presence of MIPTPCB1 but no significant enhancements in yields and selectivities was observed. We hypothesized that this might be due to the high background of photoisomerization products formed in the solution-phase. The problem was that the MIPs bind only a fraction of the trans-BPE. Thus, the majority of the reaction was occurring via the untemplated reaction in solution. This was confirmed by the similarity in yields and selectivities of the photodimer products in the absence and presence of the MIP. Therefore, the next series of studies examined the photodimerization reaction with the MIPs in the solid-state, where the olefin reactants were preadsorbed onto the polymer surfaces. Preliminary studies demonstrated that the MAA polymers could enhance the photodimerization reaction in the solid-state. Neat trans-BPE was photostable and did not yield any photoproducts even after irradiation for 48 h. However, when trans-BPE was loaded onto the MIP particles, high yields of the photodimer products were observed. However, the yields and selectivities of the photodimerization reaction in the solid-state were initially highly variable due to the heterogeneous nature and opacity of the MIP particles. The reproducibility of the solid-state reactions improved greatly by sieving the MIP particles and using only the fine particles in the

36–65 ␮m range. This variability of the solid-state reaction, however, did not allow accurate measurement of the reaction kinetics. Thus to achieve reproducible results, the trans-BPE loaded polymers were irradiated for 48 h until the products had reached a photostationary state. The solid-state reaction was also found to be highly sensitive to the method of loading the trans-BPE onto the polymer particles. Initially, the same weight percent of trans-BPE was loaded onto the MAA MIP-TPCB1 and NIP particles. The polymers were shaken in acetonitrile solutions of trans-BPE for 8 h, and then the solvent was removed in vacuo. After photo-irradiation, high yields of the photodimer products of up to 80% were observed. In addition, the yields incrementally increased with decreasing loading levels, suggesting that the polymer was the source of the increased yields. For example, the polymers with the highest loadings (46 mg trans-BPE) had a yield of only 20%, and the polymers with lowest loadings (3 mg trans-BPE) had a yield of 80%. At lower loadings, the trans-BPE has a higher probability of being bound to the binding sites in the polymer matrix as opposed to being bound to non-specific sites on the surface of the polymer particles. However, no specific enhancement in the photodimerization reaction was observed with MIP-TPCB1. The photoproducts of the olefin adsorbed to MIP-TPCB1 and NIP were formed in similar yields and relative ratios for all the loadings measured. We hypothesized that this was due to the method that was used to load the trans-BPE onto the polymer particles. Binding studies showed that the MIPs and NIPs had very different binding capacities. However, this first method loaded the same weight of trans-BPE onto both polymers. Therefore, a second protocol for loading the polymers with transBPE was examined that accentuated the differences in loading capacities of the MIP and NIP. The polymers (200 mg) were first shaken in a solution of trans-BPE in acetonitrile (2 mM) for 8 h. Then the solutions were filtered in order to remove the unbound trans-BPE in solution. Under these new loading conditions, a significant enhancement in yield of the photodimer was observed with MIP-TPCB1 (Fig. 2, white bars). The photodimerization reaction carried out in the presence of MIP-TPCB1 gave more than twice the yield in comparison to the reaction in the presence of the NIP. This suggests that MIP-TPCB1 contains more cavities that can accommodate two starting olefins to carry out the [2+2] intermolecular photodimerization than the other two polymers. The yields (<20%) of the photodimer products (TPCB1 and TPCB2) were also much lower than that observed with the first loading method (20–80%). This was attributed to the problems asso-

644

X. Wu, K.D. Shimizu / Biosensors and Bioelectronics 25 (2009) 640–646

Fig. 2. Effect of the polymers on the yields of the photoreaction of trans-BPE (the polymers were equilibrated in 2 mM and 0.5 mM solutions of trans-BPE in acetonitrile for 8 h. The solvent and unbound olefins were filtered off from the polymers. Then the polymers were dried in vacuo to run the photoreaction).

Fig. 3. Binding isotherms of the MAA polymers (MIP-TPCB1, MIP-PDC, and NIP) for trans-BPE in acetonitrile. Isotherms were all fit to a Freundlich model (solid lines).

too few high-affinity binding sites in the non-covalently imprinted polymers due to the exponential heterogeneous distribution. ciated with the bimolecular nature of the reaction. For the reaction to proceed in the MIP, the binding cavities must be occupied by two molecules of trans-BPE. However, when trans-BPE was selectively loaded into the TPCB1 selective sites using this second loading method, the binding sites were not fully saturated and thus, the majority of binding sites contained only a single molecule of transBPE. In contrast, when trans-BPE was loaded using the first method, the evaporation of the solution forced the trans-BPE into the binding cavities, yielding a higher percentage of binding cavities with two olefin molecules. The yields with MIP-TPCB1 were also only slightly higher than with MIP-PDC when the polymers were loaded in 2 mM trans-BPE solutions. Larger differences in binding capacity were observed at lower concentrations, and thus, trans-BPE was loaded onto the polymers at a lower concentration (0.5 mM) and the polymers were irradiated for 48 h. Significant differences were now observed between MIP-TPCB1 and MIP-PDC (Fig. 2, grey bars). The yield of photodimer on MIP-TPCB1 was about six times higher than the corresponding NIP and MIP-PDC. The higher yields observed at lower reactant concentrations were consistent with the severe binding site heterogeneity commonly observed in non-covalent MIPs (Rampey et al., 2004). At lower concentrations, the analyte is bound selectively to the highest affinity binding sites. However, at higher concentrations, the few high-affinity binding sites are saturated and thus the majority of analyte is bound to the unselective low-affinity binding sites. Although higher yields were observed for the solid-state reaction with MIP-TPCB1, no significant changes in the selectivities were observed as measured by the TPCB1/TPCB2 product ratios. Even at the lower loading concentrations (0.5 mM), the selectivities of the reactions were similar for all three polymers (MIP-TPCB1, MIP-PDC, and NIP). One possible explanation is that the isomeric photodimerization products TPCB1 and TPCB2 are structurally very similar. Both TPCB1 and TPCB2 have two pyridines either face of the cyclobutane ring. Therefore, binding cavities formed with TPCB1 as template can accommodate both the TPCB1 and TPCB2 transition states. One method to improve the selectivity of the reaction is to load the transBPE onto the polymers at even lower ␮M concentrations. Evidence for the higher product selectivities at lower concentrations can be found in the literature. Ye and co-workers were able to develop noncovalent MIPs that were able to control the selectivity of a [3+2] cycloaddition reaction (Zhang et al., 2006). These studies were carried out at 80 ␮M of the limiting reagent. Therefore, we reduced the loading concentrations of trans-BPE to 30 ␮M. However, the concentrations of products produced under these conditions were too low to be accurately detected using NMR. Effectively, there were

3.5. Characterization of the binding properties of the MAA polymers (NIP, MIP-PDC, and MIP-TPCB1) for trans-BPE The trends observed in the above templation studies suggest that the binding site heterogeneity of non-covalently imprinted polymers has a profound influence on their utility as selective reaction environments. Therefore, the binding site heterogeneity and the potential numbers of high-affinity binding sites in the MIPs were characterized. Binding isotherms for trans-BPE were measured and analyzed using the Freundlich model. The binding isotherms are shown in Fig. 3 in log B versus log F format. Qualitative comparisons of the binding isotherms were consistent with the single point binding studies described in Section 3.3. The MIPTPCB1 showed the highest capacity for trans-BPE. The NIP showed the lowest capacity, and MIP-PDC showed an intermediate capacity over the entire concentration range of the binding isotherm. The log–log format was chosen because it has been shown to be particularly helpful in choosing the appropriate binding model (Rampey et al., 2004). In this format, binding models that can simulate saturation behaviour, such as the Langmuir, bi-Langmuir, and Langmuir–Freundlich isotherms, will all be curved. On the other hand, binding isotherms that can simulate non-saturation behaviour, such as the Freundlich isotherm, will be linear. The binding isotherms for the MIPs and NIPs in this study were all relatively straight in log–log format, which allowed the application of the Freundlich model for quantitative analysis (Umpleby et al., 2001). The applicability of the Freundlich isotherm is also consistent with an exponentially decaying heterogeneous distribution in the MIPs and NIPs in which there are exponentially fewer high-affinity sites than low-affinity sites. The precise distribution was not calculated, as it was easier to quantitatively compare the Freundlich binding parameters. The heterogeneity indexes (m), average association constant (K¯ K1 −K2 ) and numbers of binding sites (NK1 −K2 ) were calculated for a similar range of binding sites (K1 = 1 × 103 M−1 to K2 = 2 × 104 M−1 ). The calculated binding parameters are shown in Table 1. MIP-TPCB1 Table 1 Binding parameters of the MAA polymers for trans-BPE calculated from Freundlich isotherms (K1 = 1 × 103 M−1 and K2 = 2 × 104 M−1 ). m MIP-TPCB1 MIP-PDC NIP

0.307 0.310 0.341

a (mM)

K¯ K1 −K2 (M−1 )

NK1 −K2 (␮mol/g)

0.155 0.098 0.067

5.1 × 10 5.1 × 103 5.0 × 103

6.8 4.2 2.4

3

X. Wu, K.D. Shimizu / Biosensors and Bioelectronics 25 (2009) 640–646

displayed twice the average number of binding sites (NK1 −K2 ) than the NIP and 50% higher NK1 −K2 than the MIP-PDC. The average association constants (K¯ K1 −K2 ) for these three polymers were very similar using this analysis. This was because the Freundlich analysis was applied to the same concentration window for each polymer, and therefore, the average association constant is expected to be very similar from polymer to polymer by this analysis. Finally, all three polymers contained a highly heterogeneous distribution of binding sites with heterogeneity indexes (m) ranging from 0.307 to 0.341. The MIP-TPCB1 was the most heterogeneous and the NIP was the least heterogeneous that is consistent with previous studies, which found that non-covalent MIPs generally tend to be more heterogeneous than NIPs (Umpleby et al., 2001). This severe binding site heterogeneity in MIP-TPCB1 has a number of consequences for its utility as a polymeric template for the [2+2] photodimerization reaction. First, it confirms that there is an exponentially decaying population of binding sites with increasing binding energy. This has the positive consequence that the polymers do contain very high affinity binding sites with binding affinities that probably greatly exceed those that were observed in this binding isotherm. The polymer imprinted with the photodimer product also has a higher percentage of templated binding sites. However, the numbers of binding sites per gram even in the MIP are extremely small as can be seen in Table 1. MIPTPCB1 has only 6.8 ␮mol/g binding sites. This value was consistent with observation that enhancements in yield were only observed for loadings using the 0.5 mM solutions, which calculates to a loading of approximately 2.5 ␮mol/g. This value was also consistent with the poor imprinting efficiencies of the non-covalent imprinting process used in this study, as <5% of the template molecules appear to have formed effective binding sites. This is an extremely low number of sites per gram of polymer. For example, assuming 100% efficiency of each binding site, one would need half a kilogram of polymer to template the photodimerization of 1 g of trans-BPE. A key for future studies will be to identify MIP systems and reactions that are better suited to one another. For example, the major source of these limitations is probably the binding site heterogeneity. The exponentially decaying populations of binding sites with increasing binding energy in non-covalent MIPs leads to a very small fraction and population of high-affinity, high-selectivity binding sites that are effective in templating the reaction. This was probably the source of the lack of selectivity observed in these studies as the number of effective sites was below the concentration range that could be monitored by NMR analysis. Also, the polymers had a very low loading of binding sites, which limits their practical utility. Thus, future studies will focus on stoichiometric imprinted polymers (Wulff and Knorr, 2001; Lubke et al., 2000), which are formed using templates that form stoichiometric complexes with the functional monomers and thus have very low binding site heterogeneity. In addition, the choice of reaction is critical. Bimolecular reactions such as [2+2] photodimerization reactions are particularly challenging because two molecules of reactant must be loaded into each binding site. Instead, unimolecular reactions such as an intramolecular cyclization, or isomerization reactions should be targeted as only a single reactant molecule needs to be present in each binding site. Alternatively, reactions that have only single reactant such as elimination, fragmentation, or hydrolysis (Sellergren et al., 2000; Strikovsky et al., 2000) reaction are also possible candidates. For example, we are currently examining the intramolecular [2+2] photocyclization of a diene. Finally, this MIP system was limited to acting as a polymeric template as opposed to a catalyst due to product inhibition. MIPs were imprinted with the product (TPCB1) and thus, the MIP binds the product much more strongly than the starting material (trans-BPE). A better choice would be a reaction in which the product is composed of multiple units

645

such as a hydrolysis or fragmentation reaction. The multiple units would not bind as strongly as the starting materials due to entropy and thus, would not inhibit the turn over of the binding site. For example, the highest rate accelerations observed with MIPs have been systems by Wulff that fulfil all of the above characteristics (Liu and Wulff, 2004b; Liu and Wulff, 2008). Wulff’s system is a stoichiometrically imprinted polymer was tailored to catalyze the hydrolysis reaction of a carbonate, which is a unimolecular reaction. This system demonstrates the potential of MIP systems in organic transformations as it was able to catalyze the reaction with rate accelerations approaching those of biological systems (Kcat /Kuncat ) of 105 . 4. Conclusions This study demonstrates that non-covalent MIPs can be readily prepared that act as polymer templates for the solid-state [2+2] photodimerization reaction of trans-BPE. The MAA polymers imprinted with the photodimer product (TPCB1) showed enhanced affinity and capacity not only for the original template but also for the olefin precursor trans-BPE. Thus, trans-BPE could be loaded into the MAA MIP-TPCB1 and then irradiated to give the photodimerization products. The MIP-TPCB1 appears to facilitate the photodimerization reaction leading to higher yields of the photodimer in comparison to the reactions in control polymers. However, the study also highlights some of the limitations of non-covalent MIPs in templating and catalytic applications. The exponentially decaying distribution of affinities in non-covalent MIPs limits the overall capacity, as only a small percentage of binding sites are complementary to the transition state analog. A second consequence of binding site heterogeneity is that the capacity and selectivity of the non-covalent MIPs will be highly concentration dependent. Higher yields are observed at lower reactant loadings but these benefits are offset by the exponentially decreasing loading capacities at lower concentrations. Acknowledgement The author would like to thank the National Science Foundation (CBET 0828897) for the financial support. References Alexander, C., Smith, C.R., Whitcombe, M.J., Vulfson, E.N., 1999. J. Am. Chem. Soc. 121, 6640–6651. Byström, S.E., Börje, A., Akermark, B., 1993. J. Am. Chem. Soc. 115, 2081–2083. Carboni, D., Flavin, K., Servant, A., Gouverneur, V., Resmini, M., 2008. Chem. Eur. J. 14, 7059–7065. Coates, G.W., Dunn, A.R., Henling, L.M., Ziller, J.W., Lobkovsky, E.B., Grubbs, R.H., 1998. J. Am. Chem. Soc. 120, 3641–3649. Hannon, C.L., Bell, D.A., Kellyrowley, A.M., Cabell, L.A., Anslyn, E.V., 1997. J. Phys. Org. Chem. 10, 396–404. Hopf, H., Greiving, H., Jones, P., Bubenitschek, P., 1995. Angew. Chem. Int. Ed. 34, 685–687. Lalitha, A., Pitchumani, K., Srinivasan, C., 2000. J. Photochem. Photobiol. A: Chem. 134, 193–197. Lehnberger, C., Scheller, D., Wolff, T., 1997. Heterocycles 45, 2033–2039. Liu, J.Q., Wulff, G., 2004a. Angew. Chem. Int. Ed. 43, 1287–1290. Liu, J.Q., Wulff, G., 2004b. J. Am. Chem. Soc. 126, 7452–7453. Liu, J.Q., Wulff, G., 2008. J. Am. Chem. Soc. 130, 8044–8054. Lubke, C., Lubke, M., Whitcombe, M.J., Vulfson, E.N., 2000. Macromolecules 1433, 5098–5105. MacGillivray, L.R., Reid, J.L., Ripmeester, J.A., 2000. J. Am. Chem. Soc. 122, 7817–7818. Maddipatla, M.V.S.N., Kaanumalle, L.S., Natarajan, A., Pattabiraman, M., Ramamurthy, V., 2007. Langmuir 23, 7545–7554. Matsui, J., Kubo, H., Takeuchi, T., 1998. Anal. Sci. 14, 699–702. Mosbach, K., Yu, Y.H., Andersch, J., Ye, L., 2001. J. Am. Chem. Soc. 123, 12420–12421. Nakamura, T., Takagi, K., Itoh, M., Fujita, K., Katsu, H., Imae, T., Sawaki, Y., 1997. J. Chem. Soc., Perkin Trans. 2 (12), 2751–2755. Papaefstathiou, G.S., Kipp, A.J., MacGillivray, L.R., 2001. Chem. Commun. 23, 2462–2463. Pattabiraman, M., Natarajan, A., Kaanumalle, L.S., Ramamurthy, V., 2005. Org. Lett. 7, 529–532.

646

X. Wu, K.D. Shimizu / Biosensors and Bioelectronics 25 (2009) 640–646

Ramamurthy, V., 2000. J. Photochem. Photobiol. C 1, 145–166. Ramamurthy, V., Eaton, D.F., Caspar, J.V., 1992. Acc. Chem. Res. 25, 299–307. Rampey, A.M., Umpleby, R.J., Rushton, G.T., Iseman, J.C., Shah, R.N., Shimizu, K.D., 2004. Anal. Chem. 76, 1123–1133. Rao, K.S., Hubig, S.M., Moorthy, J.N., Kochi, J.K., 1999. J. Org. Chem. 64, 8098–8104. Schütz, A., Wolff, T., 1997. J. Photochem. Photobiol. A 109, 251–258. Sellergren, B., Karmalkar, R.N., Shea, K.J., 2000. J. Org. Chem. 65, 4009–4027. Shea, K.J., Thompson, E.A., Pandey, S.D., Beauchamp, P.S., 1980. J. Am. Chem. Soc. 102, 3149–3155. Strikovsky, A.G., Kasper, D., Grun, M., Green, B.S., Hradil, J., Wulff, G., 2000. J. Am. Chem. Soc. 122, 6295–6296. Svenson, J., Nicholls, I.A., 2001. Anal. Chim. Acta 435, 19–24. Svenson, J., Zheng, N., Nicholls, I.A., 2004. J. Am. Chem. Soc. 126, 8554–8560.

Takagi, K., Itoh, M., Usami, H., Imae, T., Sawaki, Y., 1994. J. Chem. Soc., Perkin Trans. 2 (5), 1003–1009. Umpleby, R.J., Baxter, S.C., Bode, M., Berch, J.K., Shah, R.N., Shimizu, K.D., 2001. Anal. Chim. Acta 435, 35–42. Umpleby II, R.J., Rampey, A.M., Baxter, S.C., Rushton, G.T., Shah, R.N., Bradshaw, J.C., Shimizu, K.D., 2004. J. Chromatogr. B 804, 141–149. Visnjevski, A., Yilmaz, E., Bruggemann, O., 2004. Appl. Catal. A: Gen. 260, 169–174. Wulff, G., 2002. Chem. Rev. 102, 1–27. Wulff, G., Knorr, K., 2001. Bioseparation 10, 257–276. Yang, J., Dewal, M.B., Shimizu, L.S., 2006. J. Am. Chem. Soc. 128, 8122–8123. Yilmaz, E., Mosbach, K., Haupt, K., 1999. Anal. Commun. 36, 167–170. Zhang, H., Piacham, T., Drew, M., Patek, M., Mosbach, K., Ye, L., 2006. J. Am. Chem. Soc. 128, 4178–4179.