Anal. Chem. 2001, 73, 5281-5286
Surface-Grafted Molecularly Imprinted Polymers for Protein Recognition Alessandra Bossi,*,† Sergey A. Piletsky,‡,§ Elena V. Piletska,‡ Pier Giorgio Righetti,† and Anthony P. F. Turner‡
Institute of BioScience and Technology, Cranfield University, Bedfordshire, MK43 0AL, U.K., and Dipartimento Scientifico e Tecnologico, University of Verona, Strada le Grazie 15, 37134 Verona, Italy
A technique for coating microplate wells with molecularly imprinted polymers (MIPs) specific for proteins is presented. 3-Aminophenylboronic acid was polymerized in the presence of the following templates: microperoxidase, horseradish peroxidase, lactoperoxidase, and hemoglobin, via oxidation of the monomer by ammonium persulfate. This process resulted in the grafting of a thin polymer layer to the polystyrene surface of the microplates. Imprinting resulted in an increased affinity of the polymer toward the corresponding templates. The influence of the washing procedure, template concentration, and buffer pH on the polymer affinity was analyzed. It was shown that the stabilizing function of the support and spatial orientation of the polymer chains and template functional groups are the major factors affecting the imprint formation and template recognition. Easy preparation of the MIPs, their high stability, and their ability to recognize small and large proteins, as well as to discriminate molecules with small variations in charge, make this approach attractive and broadly applicable in biotechnology, assays and sensors. Affinity matrixes for the selective recognition of proteins are highly desirable in biotechnology for separation and purification, in biosensor technology, and in the field of diagnostics.1-6 Nowadays, affinity matrixes are mainly prepared using cell receptors or mono- or polyclonal antibodies. Despite the success achieved with biomaterials, their application in reusable affinity sorbents, membranes, and sensors suffers from limited stability of the antibodies and receptors and from labor-intensive and expensive production. The chemical alternative to biological ligands is synthesis of polymers carrying specific functional * Corresponding author. Fax:+39-045-8027929; E-mail:
[email protected]. † University of Verona. ‡ Cranfield University. § Fax: 01234 753562. E-mail:
[email protected]. (1) Sii, D.; Sadana, A. J, Biotechnol. 1991, 19, 83-98. (2) Scouten, W. H. Curr. Opin. Biotechnol. 1991, 2, 37-43. (3) Gupta, M. N.; Kaul, R.; Guoqiang, D.; Dissing, U.; Mattiasson, B. J. Mol. Recognit. 1996, 9, 356-9. (4) Niven, G. W.; Scurlock, P. G. J. Biotechnol. 1993, 31, 179-90. (5) Nieba, L.; Nieba-Axmann, S. E.; Persson, A.; Hamalainen, M.; Edebratt, F.; Hansson, A.; Lidholm, J.; Magnusson, K.; Karlsson, A. F.; Pluckthun, A. Anal. Biochem. 1997, 252, 217-28. (6) Sakamoto, N.; Shioya, T.; Serizawa, T.; Akashi, M. Bioconjugate Chem. 1999, 10, 538-43. 10.1021/ac0006526 CCC: $20.00 Published on Web 09/29/2001
© 2001 American Chemical Society
groups, e.g., dye-based and boronate resins.7,8 Unfortunately, the matrixes capable of a mere chemical recognition have a limited application for a few classes of proteins (e.g., glycoproteins) due to the small number of functional groups available and to the homogeneous, random-spaced distribution of the functional groups onto the polymeric support. As result of the complexity of the protein structures, the variety of their sequences, and folding motives, a high level of specificity in protein recognition could be achieved only through a three-dimensional distribution of functional groups and by the introduction of a large number of weak complementary interactions. This can be achieved through the self-assembling of the functional monomer around a template molecule in an ordered, low-energy configuration, followed by the polymerization leading to the formation of molecularly imprinted polymers (MIP).9-12 The resulting polymer possesses sites with the shape and the orientation of the functional groups complementary to those of the template molecule. The MIP’s approach has already been used successfully for mimicking natural receptors and for the synthesis of polymers carrying binding sites with high affinity toward drugs, small analytes, peptides, and sugars. Two different strategies were proposed for the design of MIPs, specific for high molecular mass compounds such as proteins. The first approach is based on the proper placement of few functional groups able to form strong interactions with the template.13 Predominantly metal-chelating or strong electrostatic interactions were explored for template recognition.14-16 The second approach is based on MIPs ability to recognize a template by using a combination of shape complementarity and multipoint weak interactions provided by the monomers able to form hydrophobic interactions or hydrogen bonds.17 The major problem with imprinting of large templates, (7) Lowe, C. R.; Burton, S. J.; Burton, N. P.; Alderton, W. K.; Pitts, J. M.; Thomas, J. A. Trends Biotechnol. 1992, 10, 442-8. (8) Garg, N.; Galaev, I. Yu; Mattiasson, B. J. Mol. Recognit. 1996, 9, 259-74. (9) Mosbach, K. Trends Biochem. Sci. 1994, 19, 9-14. (10) Kriz, D.; Ramstroem, O.; Mosbach, K. Biotechnology 1996, 14, 163-70. (11) Wulff, G. Angew. Chem., Int. Ed. Engl. 1995, 34, 1812-32. (12) Piletsky, S. A.; Piletskaya, E. V.; Panasyuk, T. L.; El’skaya, A. V.; Levi, R.; Karube, I.; Wulff, G. Macromolecules 1998, 31, 2137-40. (13) Mallik, S.; Plunkett, S. D.; Dhal, P. K.; Johnson, R. D.; Pack, D.; Shnek, D.; Arnodl, F. H. New J. Chem. 1994, 18, 299-304. (14) Mallik, S.; Johnson, R. D.; Arnold, F. H. J. Am. Chem. Soc. 1994, 116, 890211. (15) Kempe, M.; Glad, M.; Mosbach, K. J. Mol. Recogn. 1995, 8, 35-9. (16) Kempe, M.; Mosbach, K. J Chromatogr., A 1995, 691, 317-23. (17) Hjerte´n, S.; Liao, J. L.; Nakazato, K.; Wang, Y.; Zamaratskaia, G.; Zhang, H. X. Chromatographia 1997, 44, 227-34.
Analytical Chemistry, Vol. 73, No. 21, November 1, 2001 5281
such as proteins, lies in the restricted mobility of those molecules within highly cross-linked polymer networks and the poor reversibility and efficiency in binding. Despite the laborious procedure, polymer grinding also can cause the destruction of affinity sites. Surface grafting seems to be the most promising way to overcome such difficulties. Here we propose an approach to the preparation of affinity matrixes for proteins based on the surface coating of polystyrene microtiter plates with a thin layer of a stable conjugated polymer polymerized in the presence of various protein templates. Such a polymer is easily synthesized by chemical oxidation of 3-aminophenylboronic acid (APBA). The APBA polymer is grafted tightly to the surface of the plate by aromatic ring electron-pairing interactions. A few proteins (horseradish peroxidase, lactoperoxidase, hemoglobin, microperoxidase), differing in mass and charge, were selected as templates, to investigate the broadness of applicability of the imprinted poly-APBA matrixes. The best conditions for polymerization and washing out of the various templates were assessed. Then MIPs were tested for their ability to rebind templates and to discriminate between the template and similar molecules. The results obtained allow us to draw some hypotheses about the factors playing the essential role in the MIP-template interactions and indicate that the synthetic materials with selective recognition, here proposed, might be attractive for further applications in the development of ELISA-like assays. MATERIALS AND METHODS Ammonium persulfate [(NH4)2S2O8], cytochrome c from horse heart muscle, and glutaric dialdehyde were purchased from Aldrich (Dorset, U.K.). Microperoxidase was purchased from Biozyme laboratories (Blaenavon, South Wales, U.K.). Lactoperoxidase from bovine milk was from Fluka (Buchs, Switzerland). 2,2′-Azino-bis(3-ethylbenzthiazoline-6-sulfonic acid (ABTS), horseradish peroxidase (EC 1.11.1.7 from horseradish roots, MW 44 000), Tween-20, and acetic acid were purchased from Sigma (St. Louis, MO). Hemoglobin and glycated hemoglobin were purified according to the procedure of Koskinen.18 Polystyrene microplates were from Corning (Corning, NY). All other chemicals and solvents (HPLC grade) were obtained from commercial sources and used as received. The 50 mM sodium phosphate buffers (PB), containing 0.1% Tween-20, were prepared and their pHs were adjusted by adding phosphoric acid to pH 5.0, 6.0, 7.0, and 8.0. Washing buffer was 3% acetic acid containing 0.1% Tween-20. Polymerization of MIP Films onto the Microtiter Plates and Washing Procedure. Blank polymers were prepared by placing 50 µL of a 100 mM water solution of APBA in each well and adding 50 µL of ammonium persulfate (50 mM). Various MIPs were prepared by adding proteins to the APBA solution to a final concentration ranging between 0.025 and 2.5 mg/mL. The polymerization was carried out at 22 °C for 30 min (time was increased to 50 min in case of 2.5 mg/mL template). Polymerized microplates were first washed thoroughly with distilled water, then 7 times with the washing buffer, and finally conditioned for 30 min with PB at the chosen pH. Rebinding and Measure of the Dissociation Constants for the Template Protein to the MIP. Protein solutions with (18) Koskinen, L. K. Clin. Chim. Acta 1996, 253, 159-69.
5282 Analytical Chemistry, Vol. 73, No. 21, November 1, 2001
Figure 1. Template (HRP) washing with 3% acetic acid/0.1% Tween-20.
concentrations of 100, 30, 10, 3, 1, 0.3, and 0.1 µg/mL were prepared in PB-0.1% Tween-20 at the desired pH and added in 100-µL aliquots to the microplate wells. After a 90-min incubation, plates were washed with PB-0.1% Tween-20 to remove the nonadsorbed protein and the quantity of bound protein was measured as function of its peroxidase activity using ABTS substrate. A 6-mg aliquot of ABTS solution was dissolved in 10 mL of sodium citrate buffer (0.1 M, pH 6.0) and mixed with 3 µL of 30 wt % H2O2. A 100-µL solution was added to microplate wells containing the membrane sample and incubated for 10 min at room temperature. Next, the optical density of the solution was measured at 450 nm using a MR700 microplate reader (Dynatech Laboratories, Billinghurst, Sussex, U.K.). All the measurements were made in triplicate. The data obtained were plotted as they were with the saturation equation for specific binding (B ) Bmaxc/ Kd + c, where c is the concentration of protein and B the fraction of sites bound). Data were linearized with a variation of the Scatchard-Rosenthal plot, using the equation, Bmax/B ) Kd1/c + 1. RESULTS AND DISCUSSION A thin film of polymeric APBA imprinted for the selective recognition of proteins was deposited onto polystyrene microplates, by chemical oxidation of the monomers with ammonium persulfate. On the basis of previous experiments, a monomer-tocatalyst 2:1 molar ratio was chosen for the polymer preparation as being the most appropriate in order to create a thin and stable polymer coating onto the microplate surface with an average thickness of 100 nm.18 MIP plates were polymerized in the presence of 0.5 mg/mL concentration of horseradish peroxidase (HRP) and treated either with alkaline, neutral, or acidic solutions in order to select the most effective washing procedure. The quantity of remaining HRP in the matrix was evaluated by measuring a residual enzymatic activity, as a function of the number of washing cycles undergone by the polymer. No template release was observed for alkaline washing, while washing with acid in the presence of a neutral detergent gave a logarithmic decrease of the template contents in the polymer. After five to six cycles, only minute amounts (10 000 >10 000
0.082 ( 0.004 0.24 ( 0.07
0.056 ( 0.005 >10
0.036 ( 0.003 0.540 ( 0.01
a
Binding was performed in 50 mM sodium-phosphate buffer, pH 7.0.
prepared in the presence of different amounts of HRP (ranging from 0.025 to 2.5 mg/mL protein). The readsorption was performed with 0.1 mg/mL HRP in 50 mM sodium phosphate buffer, pH 7.0. Figure 3 shows the protein rebinding as a function of the concentration of template used for MIP preparation. The saturation course indicates that the polymer’s ability to rebind the template increases steeply for MIPs prepared within 0.05-0.5 mg/mL template, while it rests unchanged for MIPs prepared in the presence of a template concentration above 0.5 mg/mL. The saturation could be partially explained by the stoichiometry of the template/monomer complexation at these grafting conditions, which is optimal at 0.5-1 mg/mL concentration of protein. The effect of the pH on the rebinding of the template was studied for MIP and blank polymers, prepared in the presence of 0.5 mg/mL HRP. Their ability to readsorb the template protein at pH 5.0, 6.0, 7.0, and 8.0 was measured and the affinity of the polymer to template assessed for each buffer (see Figure 4). The values of the dissociation constants (Kds) are shown in Table 1. All Kds are in the nanomolar range, indicating a strong polymertemplate interaction resulting from the multipoint attachment of the macromolecular ligand to the matrix. The readsorption of HRP onto blank polymer showed no significant dependence on the pH. Kd values vary from 210 nM at 5284
Analytical Chemistry, Vol. 73, No. 21, November 1, 2001
Figure 6. Competitive adsorption of the template protein (HbA0) in respect to a nontemplate protein. Rebinding was done onto MIPs and onto nonspecific polymers, at pH 7 for 2 h, with a fixed concentration of HbA0 and increasing concentrations of competing protein. Competing proteins: bovine serum albumin (BSA) and penicillin G acylase (PGA). Rebinding was quantified in pg/cm2 HbA0 adsorbed onto the selective polymer.
Figure 7. Ability of selective rebinding of MIP-HbA0 and of MIP-HbA1c for HbA1c (1) and HbA0 (O).
pH 6.0 to 260 nM at pH 8.0. Surprisingly, MIP demonstrated a marked 10 times enhancement in its affinity at high pHs with a decrease in Kd from 450 nM at pH 5 to 31 nM at pH 8. By comparing blank polymers and MIPs, the optimum pH for a binding assay was assessed between pH 7 and 8, due to the fact that the affinity displayed by HRP for MIP was several times higher than that for blank polymer. Finally, imprinted poly-APBA was prepared with different macromolecular templates to investigate the broadness of applicability of such affinity matrix. MIPs prepared with lactoperoxidase, hemoglobin, HRP, cytochrome c, and microperoxidase were checked for their ability to rebind templates at pH 7; plots are displayed in Figure 5 and the resulting Kds are shown in Table 2. The chosen proteins possess large differences in molecular mass and isoelectric points (pI).20 From the rebinding experiments, it was evident that the size (shape) of the molecule and its charge affect the imprinting efficiency. For a small protein such as microperoxidase, the dissociation constant was rather high: Kd ) 1.5 µM. Large proteins, such as lactoperoxidase (Mr 77 000), hemoglobin A0 (Mr 65 000), and horseradish peroxidase (Mr 44 000), exhibited much lower Kds, 36, 56, and 82 nM, respectively. Charge influence also affects significantly the formation of binding sites and the polymer affinity. Thus, highly positively charged proteins such as cytochrome c (pI 10.6) are not able to form imprints in the APBA polymer, which has a net positive charge. Competitive adsorption of a template protein in respect to a nontemplate was studied (see Figure 6). The selective polymers were prepared with HbA0, as template, and assayed for the rebinding with a mixture made of a fixed concentration of the template and an increasing concentration of the competing protein. The competitors chosen were bovine serum albumin (BSA), which has approximately the same size of hemoglobin (Mr 66.000) and is the most abundant protein in plasma and penicillin G acylase (PGA), which is larger than hemoglobins86.000sand has a more acidic pI (pI 5.8). Figure 6 shows the quantity of template (HbA0) adsorbed onto the selective polymer plotted in dependence of the concentration of competitor. It appears clearly that the addition of the competitor displaced the binding of HbA0: the more competitor was added, the lower was the quantity of HbA0 rebound (19) Piletsky, S. A.; Piletska, E. V.; Bossi, A.; Karim, K.; Lowe, P.; Turner, A. P. F. Biosens. Bioelectron., in press. (20) Righetti, P. G.; Tudor, G.; Ek, K. J. Chromatogr. 1981, 220, 115-94.
Table 3. Analysis of the Hemoglobins Binding to Imprinted Polymers Prepared from Poly-APBAa Kd, nM
a
MIP
HbA1C
HbAo
MIP [HbA1c] MIP [HbAo]
65 ( 2 188 ( 10
162 ( 9 58 (2
Binding was performed in 50 mM sodium-phosphate buffer, pH
7.0.
onto the grafted polymer. Surprisingly, the displacement effect was more pronounced when PGA was used as competitor, whereas the binding of HbA0 was not seriously compromised by the presence of the BSA competitor. In both cases, the high selectivity of the polymer toward its template was demonstrated. Again, the charge effect seems to play a crucial role in the selection of the binding species by the matrix of APBA, favoring PGA bindingsnegatively chargedsand repelling BSAspositively charged. Moreover, the poor competition shown by BSA might be of interest in the perspective of creating synthetic affinity APBA polymers for assaying hemoglobins or other clinically relevant plasma components. In this respect, it is interesting to note that MIP was able to discriminate between hemoglobin HbA0 and its glycosylated derivative HbA1C,21 confirming the high specificity of the imprinted polymer (Figure 7, Table 3). The Kds observed for template recognition on the specific polymer (HbA0 on HbA0MIP and HbA1c on HbA1c-MIP) were around 50-60 nM, whereas Kds for cross-rebinding (HbA0 on HbA1c-MIP and HbA1c on HbA0-MIP) were 3 times higher. Surely, a reason for this remarkably high selection should lay in a combination of charge and shape effect. Specific HbA0-MIP polymer was formed at a pH close to the pI of the HbA0 (pI 6.93), while HbA1c has a slightly lower pI (pI 6.84). In such conditions, HbA0 possesses an equal number of positive and negative residues; thus, a better complementary matching with the monomers could be envisaged, while the steric hindrance of the glycated variant could limit the access to the binding sites created for HbA0. On the other hand, the binding sites of the specific HbA1c-MIP polymer should have been structured for embedding and recognizing the only different structure present on this hemoglobin variant: the glucose moiety (21) Bunn, H. F.; Haney, D. V.; Kamin, S.; Gabbay, K. H.; Gallop, P. M. J. Clin. Invest. 1976, 57, 1652-9.
Analytical Chemistry, Vol. 73, No. 21, November 1, 2001
5285
on the N-terminal valine of the β-chain of HbA1c.Thus, the preferential binding of HbA1c over HbA0 is justified. In any case, it is difficult to envisage a difference in recognition between the two hemoglobin variants of more than 3-4 times, due to the high similarity of these proteins. To evaluate the possible role of the microplate surface in stabilizing the structure of the specific binding sites and template recognition, free-standing MIP, synthesized under the same conditions, was tested for its ability to rebind template. It was found that the affinities of MIP and blank particles were similar and 100 times smaller than the affinity of the MIP grafted to the polystyrene surface. Supports would thus appear to play a critical role in template recognition by grafted polymer in contrast to highly cross-linked MIPs synthesized by the conventional approach. The MIPs were repeatedly (5 times) regenerated, and their affinity was measured during 30 days. Over this period of time, the polymers lost 20% of their affinity on average. The decrease in affinity resulted mainly from the partial destruction of the polymer binding sites since no evident decrease was observed for polymers that were kept intact after washing and used one month later.
5286
Analytical Chemistry, Vol. 73, No. 21, November 1, 2001
In conclusion, the results presented here demonstrate the efficiency of the grafting procedure for generating surfaces with high affinity toward template molecules and the ability of imprinted poly-APBA to recognize the target proteins. The origin of the polymer specificity lies in the formation of the specific binding sites (imprints). The stabilizing function of the support and spatial orientation of the polymer chains and template functional groups are the major factors affecting the template recognition. Easy preparation of the MIPs, their high stability, and their ability to recognize small and large proteins as well as to discriminate the molecules with small variations in charge make this approach attractive and broadly applicable in biotechnology, assays, and sensors. ACKNOWLEDGMENT Supported in part by grants from P.F. Biotecnologie No. 49, CNR, and Agenzia 2000 CNR-G0078F9, Roma, Italy.
Received for review June 5, 2000. Accepted July 19, 2001. AC0006526
