Cross-Linked Surface-Grafted Glycopolymer for Multivalent

Oct 31, 2007 - lectin recognition, additional selectivity may be achieved ... avidity of the interaction by about 100-fold or even higher.1 ..... that...
1 downloads 0 Views 426KB Size
Download PDF
Anal. Chem. 2007, 79, 8979-8986

Cross-Linked Surface-Grafted Glycopolymer for Multivalent Recognition of Lectin Lei Yu,† Mingchuan Huang,‡ Peng G. Wang,‡ and Xiangqun Zeng*,†

Department of Chemistry, Oakland University, Rochester, Michigan 48309, and Department of Biochemistry and Chemistry, The Ohio State University, Columbus, Ohio 43210

An r-link mannose-conjugated acrylamide monomer was synthesized. This monomer was polymerized by free radical polymerization with acrylamide, a cross-linker, and a surface linker directly on the gold surface. The surface linker, with an active carbon-carbon double bond, was preimmobilized on the gold surface by the thiol anchor. Thus, a cross-linked mannose-conjugated polymer thin layer was grafted onto a gold surface. This thin layer of polymer showed high binding sensitivity and excellent selectivity to its target lectin, concanavalin A (Con A), surpassing the formerly used linear glycopolymer and selfassembled glycol monolayers, validated by the techniques of quartz crystal microbalance, atomic force microscopy, and surface plasmon resonance. Remarkable response was observed to Con A at a concentration as low as 5 × 10-10 M. The response is proportional to the Con A concentration up to 10-7 M in phosphate-buffered saline. The use of cross-linked polymer decreased the flexibility of the polymer backbone between the carbohydrate binding sites. Therefore, the cost of conformational entropy for multivalent binding was minimized. The binding constants of the so-prepared cross-linked polymer with Con A were measured to be between 2.5 × 106 and 3.2 × 106 M-1. These values are significantly larger than that obtained in our early study with a carbohydrate selfassembled monolayer. In addition to the carbohydratelectin recognition, additional selectivity may be achieved by controlling the degree of cross-linking. Carbohydrates cover cell surfaces. The interaction of carbohydrates with their surrounding environment is one of the most fundamental molecular recognition events. The interaction of carbohydrates (also called ligands or epitopes) with their corresponding proteins (also called receptors) is involved in cell recognition, antibody recognition, invasion of virus, bacteria, or toxins, hormonal action, and many other physiological and pathological processes. Therefore, carbohydrate ligands in principle could be used as sensing elements to detect or monitor a variety of receptors. However, the development of carbohydratebased biosensors is slow due to several major challenges. First, carbohydrate-protein interactions are essentially weak, often * Corresponding author. Phone: 248-370-2881. Fax: 248-370-2312. E-mail: [email protected]. † Oakland University. ‡ The Ohio State University. 10.1021/ac071453q CCC: $37.00 Published on Web 10/31/2007

© 2007 American Chemical Society

weaker than protein-protein interactions by a factor of approximately 102-103. Second, there are limited methods for carbohydrate immobilization. Third, the carbohydrates have low specificity, and cross-reaction cannot be effectively avoided without additional selection mechanisms. Therefore, the optimal conditions for their applications as sensing elements have not been discovered. In nature, the interaction of carbohydrates with target proteins is not a simple monomeric binding. It has been shown that the occurrence of two simultaneous binding events can increase the avidity of the interaction by about 100-fold or even higher.1 Glycoconjugated materials, possessing synthetic flexibility and structural diversity, have been explored to achieve multivalent interactions including glycoclusters,2 cyclodextrins and calixarenes,3 dendrimers,4 neoglycoproteins,5 self-assembled monolayers/ bilayers,6 and oligomers and polymers.7 Synthetic polymers with polyacrylamide, polylysine, polystyrene, and other backbones8 containing sugar moieties as pendant groups or terminal groups were developed to mimic natural glycoconjugates to bind to receptors for drug development and recently for sensing. These artificial glycopolymers are often linear and have high flexibility. The high flexibility is not desired since the carbohydrate ligands (1) (a) Kitov, P. I.; Sadowska, J. M.; Mulvey, G.; Armstrong, G. D.; Ling, H.; Pannu, N. S.; Read, R. J.; Bundle, D. R. Nature 2000, 403, 669-672. (b) Shinohara, Y.; Kim, F.; Shimizu, M.; Goto, M.; Tosu, M.; Hasegawa, Y. Eur. J. Biochem. 1994, 223, 189-194. (c) Mammen, M.; Choi, S. K.; Whitesides, G. M. Angew. Chem., Int. Ed. 1998, 37, 2755-2794. (2) (a) Zanini, D.; Roy, R. Architechtonic Neoglycoconjugates: Effects of Shapes and Valencies in Multiple Carbohydrate-Protein Interactions. In Carbohydrate Mimics: Concepts and Methods; Chapleur, Y., Ed.; Wiley-VCH: New York, 1998; p 385. (b) Uchiyama, T.; Vassilev, V. P.; Kajimoto, T.; Wong, W. C.; Huang, H. M.; Lin, C. C.; Wong, C. H. J. Am. Chem. Soc. 1995, 117, 5395. (c) Wittmann, V.; Takayama, S.; Gong, S. W.; Wetiz-Schmidt, G.; Wong, C. H. J. Org. Chem. 1998, 63, 5137. (d) Saisch, G.; Ohrlein, R. Angew. Chem., Int. Ed. 1996, 35, 1812. (3) (a) Furuike, T.; Aiba, S.; Nashimura, S. I. Tetrahedron 2000, 56, 9909. (b) Baussanne, I.; Benito, J. M.; Ortiz-Mettel, C.; Garcia-Fernandez, J. M.; Law, H.; Defaye, J. Chem. Commun. 2000, 1489. (c) Roy, R.; Kim, J. M. Angew. Chem., Int. Ed. 1999, 38, 369. (d) Fujimoto, T.; Miyata, T.; Aoyama, Y. J. Am. Chem. Soc. 2000, 122, 3558. (4) Turnbull, W. B.; Pease, A. R.; Stoddart, J. F. ChemBioChem 2000, 1, 70. (5) Roy, R. Top. Curr. Chem. 1997, 187, 242. (6) (a) Pan, J. J.; Charych, D. Langmuir 1997, 13, 1365. (b) Mann, D. A.; Kanai, M.; Maly, D. J.; Kiessling, L. L. J. Am. Chem. Soc. 1998, 120, 575. (c) Plant, A. L. Langmuir 1999, 15, 5128. (d) Mrksich, M. Chem. Soc. Rev. 2000, 29, 267. (e) Zhang, Y.; Telyatnikov, V.; Sathe, M.; Zeng, X.; Wang, P. G. J. Am. Chem. Soc. 2003, 125, 9292-9293. (f) Zhang, Y.; Luo, S.; Tang, Y.; Yu, L.; Hou, K.; Cheng, J.; Zeng, X.; Wang, P. G. Anal. Chem. 2006, 78, 2001-2008. (g) Love, K. R.; Seeberger, P. H. Angew. Chem., Int. Ed. 2002, 41, 3583-3586.

Analytical Chemistry, Vol. 79, No. 23, December 1, 2007 8979

Scheme 1. Schematic View of the Multivalent Binding with Cross-Linked Glycopolymer

are in a random, undefined environments and the number of multicarbohydrate-lectin interactions is often unknown. In order to increase the rigidity and reduce the flexibility, glycodendrimers composed of a core and a highly branched polymer were made and have been proved to possess high density of carbohydrate functional groups.9 However, despite this improvement in design, linear polymers and dendrimers sometimes fail to display large enhancements in binding affinity, probably because of an improper spacing or nonideal geometry of the carbohydrates presented. In this report, cross-linked glycopolymers were directly synthesized on the gold surface of a quartz crystal microbalance (QCM) transducer for label-free detection of lectin (Scheme 1). The cross-linked carbohydrate polymer creates unique multivalent recognition sites that complement the analytes in chemical functionality and in size and shape. The combination of such a layer with an appropriate QCM transducer yields sensor devices that are highly suitable for fast, low-cost, and straightforward detection of lectin analytes. The cross-linked carbohydrate polymer could substantially minimize possible cross-reactivity and significantly enhance the specificity and sensitivity of detection through both the specific multivalent carbohydrate-target interactions and the size and shape of the cross-linked cavity. Lectins are proteins that interact with carbohydrates selectively. Most lectins are oligomeric with several sugar binding sites. Various bacterial toxins are lectins and bind to glycolipid determinants.10 For example, Vibrio cholera toxin consists of A and B subunits,11 in the ratio of AB5, in which the B subunits bind to GM1 ganglioside receptors through the carbohydrate recognition domain.12 Concanavalin A (Con A), which recognizes a commonly (7) (a) Sigal, G. B.; Mammen, M.; Dahmann, G.; Whitesides, G. M. J. Am. Chem. Soc. 1996, 118, 3789-3800. (b) Kobayashi, K.; Tsuchida, A.; Usui, T.; Akaike, T. Macromolecules 1997, 30, 2016-2020. (c) Xue, C.; Jog, S. P.; Murthy, P.; Liu, H. Biomacromolecules 2006, 7, 2470-2474. (d) Truchida, A.; Matruura, K.; Kobayashi, K. Macromol Chem. Phys. 2000, 201, 22452250. (e) Dohi, H.; Nishida, Y.; Mizuno, M.; Shinkai, M.; Kobayashi, T.; Takeda, T.; Uzawa, H.; Kobayashi, K. Bioorg. Med. Chem. 1999, 7, 20532062. (f) Uzawa, H.; Kamiya, S.; Minoura, N.; Dohi, H.; Nishida, Y.; Taguchi, K.; Yokoyama, S.; Mori, H.; Shimizu, T.; Kobayashi, K. Biomacromolecules. 2002, 3, 411-414. (g) Disney, M.; Zheng, J.; Swager, T. M.; Seeberger, P. H. J. Am. Chem. Soc. 2004, 126, 13343-13346. (h) Disney, M. D.; Seeberger, P. H. Chem. Biol. 2004, 11, 1701-1707. (8) (a) Thoma, G.; Patton, J. T.; Magnani, J. L.; Ernst, B.; Ohrlein, R.; Duthaler, R. O. J. Am. Chem. Soc. 1999, 121, 5919. (b) Matruura, K.; Ketakouji, H.; Sawada, N.; Ishida, H.; Kiso, M.; Kitajima, K.; Kobayashi, K. J. Am. Chem. Soc. 2000, 1222, 7406. (9) (a) Aoi, K.; Ito, K.; Okada, M. Macromolecules 1995, 28, 5391-5393. (b) Ashton, P. R.; Boyd, S. E.; Brown, C. L.; Jaryaraman, N.; Stoddart, J. F. Angew. Chem., Int. Ed. 1997, 36, 732-735. (10) (a) Karlsson, K. A. Annu. Rev. Biochem. 1989, 58, 309-350. (b) Dubreuil, J. D. In The Comprehensive Sourcebook of Bacterial Protein Toxins; Louf, J. E. A., Freer, J. H., Eds.; Academic Press: San Diego, CA, 1999; p 525. (11) Merritt, E. A.; Hol, W. G. Curr. Opin. Struct. Biol. 1995, 5, 165-171. (12) Zhang, R. G.; Scott, D. L.; Westbrook, M. L.; Nance, S.; Spangler, B. D.; Shipley, G. G.; Westbrook, E. M. J. Mol. Biol. 1995, 251, 563-573.

8980

Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

occurring sugar structure, R-linked mannose, was selected as the representative lectin to prove the concept. Polymers obtained by free radical polymerization are the substance of choice because of their chemically robust carbon backbone and the wide diversity of monomers available to construct them. Soluble polyacrylamide (PAAm) is a broadly used hydrophilic polymer, which has been reported as a model glycoconjugated polymer for investigation of lectin-carbohydrate interactions.13 The well-established polymerization procedures made it possible to control the molecular weight of the polymer and the density of functional carbohydrate groups on the polymer frame, as well as the physicochemical properties of the polymer itself.14 We use cross-linked polyacrylamide (c-PAAm) as an immobilized polyvalent ligand material to bind the lectin in solution to achieve a high-affinity binding and thus highly sensitive detection. When the PAAm is cross-linked, much of the conformational changes of the polymer are frozen. The cross-linker makes the polymer chains bend to make a threedimensional (3-D) network, Scheme 1. Hence, the conformational enthalpy changed, and the conformational entropy cost for the multivalent binding could be reduced. By controlling the ratio of the cross-linker, the mesh size of the cross-linked polymer could be controlled to “fit” the geometric size of each individual lectin that is detected.15 c-PAAm is much more stable than soluble linear PAAm. It is not soluble in water or organic solvents. By crosslinking and “decorating” the polymer backbone with the specific carbohydrate ligand, we achieved the needed rigidity for a surfacesensitive QCM or surface plasmon resonance (SPR) transducer.16 The mannose ligand that is presented in high concentration throughout the cross-linked polymer backbone facilitated the specific multivalent interactions with the target lectin Con A and significantly enhanced the sensitivity and specificity of carbohydrate-lectin recognition. EXPERIMENTAL SECTION Chemicals. Acrylamide (AAm), 1-thiol-2-propene, N,N′-methylenebis(acrylamide) (Bis), N,N,N′,N′-tetramethylethylenediamine (TMEDA), (NH4)2S2O8, Con A, Erythrina cristagalli lectin (ECL), and all other reagents are analytical reagents obtained from SigmaAldrich Inc. and used as received. The mannose-conjugated acrylamide monomer, N-acryloylaminophenyl-R-D-mannopyranosyl 5, is synthesized following the procedures showed in Scheme 2. Mannose donor 1 was prepared by reacting peracetylated mannose with HBr. Mannose conjugate 2 was afforded via the glycosylation reaction between 1 and p-nitrophenol promoted by silver oxide. 2 was subjected to hydrogenation followed by reaction with acryloyl chloride and produced compound 4. Target compound 5 was obtained after removal of the acetyl groups of 4. p-Nitrophenyl-2,3,4,6-tetra-O-acetyl-D-mannopyranoside 2 was prepared as follows: A mixture of 1 equiv of tetra-o-acetyl-Dmannopyranosyl bromide (0.8 g, 1.95 mmol), 2 equiv of p(13) (a) Tsuchida, A.; Akimoto, S.; Usui, T.; Kobayashi, K. J. Biol. Chem. 1998, 123, 712. (b) Okada, M. Prog. Polym. Sci. 2001, 26, 67-104. (14) Houseman, B. T.; Mrksich, M. Top. Curr. Chem. 2002, 218, 1-44. (15) Elias, H. G. Macromolecules; Plenum Press: New York, 1977 (translated by J. W. Stafford). (16) (a) Sauerbrey, G. Z. Phys. 1959, 155, 206-222. (b) Janshoff, A.; Galla, H.; Steinem, C. Angew. Chem., Int. Ed. 2000, 39, 4004-4032. (c) Shen, Z.; Stryker, G. A.; Mernaugh, R. L.; Yu, L.; Yan, H.; Zeng, X. Anal. Chem. 2005, 77, 797-805.

Scheme 2. Synthesis Procedures of N-Acryloyl-aminophenyl-r-D-mannopyranosyla

a Reagents, conditions, and transform ratios: (a) HBr, AcOH, room temperature; (b) Ag O, p-nitrophenol, MeCN, 70%; (c) Pd/C, H 2 2 (30 psi), MeOH, 94%; (d) acryloyl chloride, Et3N, CH2Cl2, 0 °C, 50%; (e) NaOMe, MeOH, 90%.

nitrophenol, and 1.3 equiv of silver oxide in acetonitrile was refluxed for 1 h. After filtration, the mixture was concentrated in vacuo. The residue was diluted by ethanol and then refluxed for 1 h in the presence of charcoal. After filtration, the solution was concentrated in vacuo again. The mixture was allowed to stand at 4 °C overnight, filtered, and washed several times with cold ethanol. The expected compound 2 was obtained as a white powder (0.63 g, 70% yield). 1H NMR (200 MHz, CDCl ) δ 8.2 (d, J ) 9.0 Hz, 2H), 7.2 (d, 3 J ) 9 Hz, 2H), 5.6 (s, 1H), 5.5 (m, 3H), 4.2 (m, 1H), 4.1(m, 2H), 2.1 (m, 12H). p-Aminophenyl-2,3,4,6-tetra-O-acetyl-R-D-mannopyrannoside 3 was prepared as follows: 2 (0.30 g, 0.63 mmol) was dissolved in 150 mL of methanol containing 10% Pd/C (0.20 g). The solution was stirred under H2 (30 psi) at room temperature for 4 h. The solvent was filtered through Celite and evaporated under vacuo. The residue was then dissolved in ethyl acetate, washed with water, and dried over anhydrous Na2SO4. Then the ethyl acetate was evaporated under reduced pressure. Crystallization of the residue from hot ethanol gave the product 3 (0.26 g, 94% yield). 1H NMR (400 MHz, CDCl ) δ 6.88 (d, J ) 8.9 Hz, 2H), 6.59 3 (br, 2H), 5.52 (dd, 1H, J ) 3.5 Hz, J ) 10.0 Hz), 5.40 (dd, 1H, J ) 3.5 Hz, J ) 10.0 Hz), 5.40 (d, 1H, J ) 1.8 Hz), 5.34 (br, 1H), 5.32 (d, 1H, J ) 10.1 Hz), 4.26 (dd, 1H, J ) 5.4 Hz, J ) 12.2 Hz), 4.13 (d, J ) 2.3 Hz, 1H), 4.07 (br, 1H), 3.58 (br, 2H), 2.17 (s, 3H), 2.03 (s, 6H), 2.00 (s, 3H). 13C NMR (100 MHz, CDCl3) δ 170.7, 169.9, 168.7, 148.2, 142.2, 117.9, 116.0, 96.8, 69.5, 68.9, 62.2, 20.9, 20.7. p-Acrylamidophenyl-2,3,4,6-tetra-O-acetyl-R-D-mannopyranoside 4 was prepared as follows: To a solution of 3 (0.20 g, 0.45 mmol) and triethylamine (1.5 mL) in dichloromethane (100 mL) that was stirred at 0 °C was added dropwise a solution of acryloyl chloride (220 µL) in CH2Cl2 (8 mL) over 30 min. The reaction mixture was further stirred at room temperature for another 45 min. Then 50 mL of water was added, and the organic phase was successively washed with equal volumes of 0.5 M HCl, saturated NaHCO3, brine, and water by turn. The organic phase was dried over anhydrous Na2SO4 and evaporated under reduced pressure. The residue was purified by column to give the product 4 (0.10 g, 50% yield). 1H NMR (400 MHz, CDCl ) δ 7.50 (d, J ) 9.0 Hz, 2H), 7.36 3 (br, 1H, NH), 7.03 (d, 2H), 6.39 (d, J ) 16.8 Hz, 1H), 6.21 (d, J ) 10.3 Hz, 1H), 5.73 (br, 1H), 5.52 (dd, J ) 3.5 Hz, J ) 10.0 Hz, 1H), 5.52 (dd, J ) 3.5 Hz, J ) 10.0 Hz, 1H), 5.46 (d, J ) 1.8 Hz, 1H), 5.41 (br, 1H), 5.33 (d, J ) 10.1 Hz, 1H), 4.25 (dd, J ) 5.5 Hz, J ) 12.4 Hz, 1H), 4.08 (d, J ) 2.3 Hz, 1H), 4.05 (br, 1H), 2.17(s, 3H), 2.03 (s, 3H), 2.02 (s, 3H), 2.01 (s, 3H). 13C NMR (100

MHz, CDCl3) δ 170.5, 170.0, 169.9, 169.7, 163.4, 152.3, 133.0, 131.9, 127.7, 121.5, 117.1, 96.1, 69.4, 69.2, 68.9, 66.0, 62.1, 20.8, 20.7. N-Acryloyl-aminophenyl-R-D-mannopyranosyl 5 was prepared as follows: NaOMe (0.5 M) was added into a solution of 4 (0.10 g) in anhydrous methanol (200 mL) at 0 °C until pH ) 10. The solution was stirred for 3 h. Dowex cation exchange resin (H form) was added to adjust the pH to 7, and the mixture was filtered. The filtrate was evaporated to give the product (0.32 g, 90%). 1H NMR (500 MHz, CD OD) δ 7.70 (d, J ) 10.5 Hz, 2H), 7.29 3 (m, 2H), 6.27 (m, 1H), 6.19 (m, 1H), 5.73 (d, J ) 10 Hz, 1H), 5.45 (d, J ) 20 Hz, 1H), 4.04 (d, J ) 5 Hz, 1H), 3.89 (m, 1H), 3.64 (m, 4H). 13C NMR (125 MHz, CD3OD) δ 166.5, 131.6, 131.1, 123.2, 117.4, 98.0, 73.3, 70.3, 68.0, 66.5. FT-IR. The specular reflectance absorption FT-IR spectrum was recorded with a Varian Exculibur 3100 FT-IR spectrometer with a Harrick reflectance kit. Gold thin layer on smooth glass slides were used as substrates. After polymerization on substrate, the polymer thin film was thoroughly washed with water and methanol. Quartz Crystal Microbalance Measurement. AT-cut nonpolished ∼10 MHz resonance frequency QCMs (International Crystal Co., Oklahoma) were used for all the experiments. A QCM is mounted on a homemade Kel-F cell. One side of the gold electrode covered by the c-PAAm is immersed in 1 mL of phosphate-buffered saline (PBS) solution containing saturated Ca2+ and Mn2+ overnight before QCM experiments to obtain a stable baseline. The other side of the gold electrode is exposed to air. Aliquots of Con A samples with various concentrations were added into the PBS buffer with stirring. The resonance frequency and the damping resistance are recorded simultaneously by an Agilent 4395A network impedance analyzer. Atomic Force Microscopy Measurement. A thin layer of gold (100 nm) coated by vapor deposition on the polished quartz plate had a smooth surface and was used as substrate. The gold surface was pretreated in a solution of 1-thiol-2-propene. Polymerization with and without cross-linker was carried out on the substrates. The modified substrate was soaked in a Con A solution containing Ca2+ and Mn2+ for 6 h followed by thorough rinsing with water and PBS buffer. Atomic force microscopy (AFM) images were obtained with a contact mode Molecular Imaging PicoScan atomic force microscope and a MicroMasch (Estonia) ultrasharp (Rc < 10 nm) and ultrasoft (force constant, 0.03 N/m) silicon cantilever. The scan rate was 1 line/s, and the resolution was 512 data points/line. Both the cantilever and the sample were in a PBS buffer filled liquid cell during the experiments. Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

8981

Scheme 3. Polymerization Reaction on a Gold Surface

RESULTS AND DISCUSSION In our previous work, the free radical copolymerization of acrylamide and carbohydrate-conjugated acrylamides was successfully used to prepare bulk cross-linked PAAm hydrogels that “trap” norovirus. The c-PAAM hydrogel can swell up to hundreds of times its original weight by absorbing water.17 Here, the c-PAAm on the QCM surface was prepared through similar free radical polymerization with AAm and N-acryloyl-aminophenyl-R-D-mannopyranosyl as monomers, 1-thiol-2-propene as surface linker, Bis as cross-linker, and TMEDA and (NH4)2S2O8 as initiators at the same temperature as in the literature, Scheme 3.18 1-Thiol-2propene was preimmobilized on the gold surface by soaking the gold surface of the QCM in its ethanol solution overnight. The cross-linker is 0%, 0.5%, and 1.5% of the total amount of monomers. N-acryloylaminophenyl-R-D-mannopyranosyl 5 is 5% of the total amount of monomers. A specular reflectance FT-IR spectrum of the cross-linked polymer so-prepared on a smooth gold substrate is shown in Figure 1. The spectrum shows both the characteristic peaks of polyacrylamide (3300, 1665 cm-1) and mannose (1262, 1109, 1045 cm-1),19 indicating the successful polymerization of the acrylamide monomers on the surface. The amounts of the surface linker and cross-linker are very small. Their peaks overlapped with the peaks of the polymer backbone so that they are difficult to identify. A bulk polymerization with the same components at the same conditions was done in parallel each time. All experiments with cross-linker gave a bulk cross-linked gel. Without cross-linker, the mixture became very viscous, indicating the linear polymerization. After polymerization, the cross-linked film was rinsed with PBS buffer. Then it was soaked in 0.1% bovine serum albumin solution for 30 min to block the inactive areas before QCM measurement. Figure 2 summarizes the sensing results of the PAAm/cPAAm-modified gold surface of the QCM with and without cross(17) Zhang, Y. L.; Yao, Q. J.; Xia, C. F.; Jiang, X.; Wang, P. G. ChemMedChem 2006, 1, 1361. (18) (a) Feng, X. D.; Guo, X. Q.; Qiu, K. Y. Macromol. Phys. Chem. 2003, 189, 77-83. (b) Mohan, Y. M.; Geckeler, K. E. React. Funct. Polym. 2007, 67, 144-155. (19) Pouchert, C. J., Ed. The Aldrich Library of Infrared Spectra Edition III; Aldrich Inc.: WI, 1981.

8982 Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

linker for binding with Con A. At neutral pH, Con A is predominantly tetrameric with optimal activity. Two types of metal ions (Mn2+ and Ca2+) can bind to Con A; both must be present for carbohydrate binding. When 0.5% Bis was used, the c-PAAmmodified QCM gave 6 Hz frequency change at a Con A concentration of 5 × 10-10 mol/L. The response increased almost linearly to about 1050 Hz when the Con A concentration was increased to 5 × 10-7 mol/L. The binding was very weak when there was no cross-linker (i.e., 19 and 126 Hz at a Con A concentration of 50 and 400 nM, respectively). The values are about 1 order of magnitude lower than that of c-PAAm. From the data in Figure 2A, binding avidities (Ka) were calculated according to the reported method.6e,f The Ka of c-PAAm binding with Con A is 3.2 × 106 M-1, which is about 4-fold larger than the Ka from PAAm without Bis, 7.7 × 105 M-1. The monobinding constant between Con A and mannose is reported to be 103-104 M-1 in solution.20 When mannose is immobilized on a solid surface to form monolayers, the value of Ka for Con A binding was determined to be 8.7 ( 2.8 × 105 M-1 by the QCM technique in our previous work.6f This result indicates that the c-PAAm has significantly

Figure 1. FT-IR spectrum of c-PAAM on a gold surface.

Figure 2. Isotherms (A) and the sensorgrams (B) upon addition of Con A. PAAm/c-PAAm film prepared on QCM from a mixture of monomer solutions with and without cross-linker, Bis.

higher binding affinity to Con A likely due to the multivalent interactions with Con A. To achieve an intramolecular multivalent binding between a multifunctional receptor and a multifunctional ligand, it has to be thermodynamically favorable and kinetically facile. The affinity constant of an intramolecular polyvalent binding is dictated by the total free energy change of each individual binding (∆Gpoly ) ∑i)1-n ∆Gi). A negative value of ∆Gpoly denotes a thermodynamically favorable binding. The enthalpic component (∆Hpoly) and the entropic component (T∆Spoly) constitute the total free energy change (∆Gpoly ) ∆Hpoly - T∆Spoly). Further, ∆Hpoly includes two major parts: the negative enthalpy change comes from the receptor-ligand binding (∆Hpolybind), and the positive enthalpy change comes from the conformational change of the polymer (∆Hpolyconf). For an n-valent binding with n g 2, the ∆Hpolybind will be larger than that of a monovalent binding. The entropy and enthalpy changes that come from solvation do not depend on the valency and, hence, were neglected in our discussion here. To reduce the ∆Hpolyconf change, an obvious approach is to include a flexible spacer between two carbohydrate ligands. However, bivalent systems joined by flexible linkers are proved to fail for entropic reasons.21 After the first single ligand-receptor binding occurs between two multivalent units, any subsequent intramolecular bindings between any of the rest of the active sites of the ligand and the receptor require additional ligand conformation change to adapt the geometry of the receptor. The more flexible the polymer before the binding, the more entropy cost for the consequent binding. A previous report also showed indirect evidence that steric hindrance is an important factor affecting the polyvalent binding.7a The use of cross-linked polymer decreased the flexibility of the polymer backbone between the carbohydrate binding sites. Therefore, the cost of conformational entropy for multivalent binding was minimized since each of the ligands is already in its best possible location and little change of geometry is needed for the multivalent binding. (20) (a) Ambrosi, M.; Cameron, N. R.; Davis, B. G. Org. Biomol. Chem. 2005, 3, 1593-1608. (b) Lee, Y. C.; Lee, R. T. Acc. Chem. Res. 1995, 28, 321327. (21) (a) Glick, G. D.; Knowles, J. R. J. Am. Chem. Soc. 1991, 113, 4701-4703. (b) Glick, G. D.; Toogood, P. L.; Wiley, D. C.; Skehel, J. J.; Knowles, J. R. J. Biol. Chem. 1991, 266, 23660-23669.

The affinity constants of non-cross-linked polymer binding with lectin obtained in our system are consistent with those reported in the literature.22 There is another reason that may explain the very small response when Bis is absent in the reactant mixture. The surface linker is immobilized, while the free radical polymerization is occurring in the solution. Only a small amount of the surface linker could be involved in the polymerization. When cross-linker was used, the entire product, cross-linked polymer thin layer, could react with the surface linker and be immobilized. However, linear PAAm is water-soluble; most of the PAAm will be washed off from the QCM electrode, and only those polymer chains that terminated at the surface linker will be immobilized on the QCM. The total amount of immobilized c-PAAM is much larger than that of linear PAAm. Elliposometry measurements suggest that the polymer film with 0.5% Bis is thicker than that without Bis. Considering that the film may be heterogeneous, we measured the film thickness at five different sites on each substrate, and the average of the results was used. The precise film thickness is difficult to obtain by ellipsometry due to the uncertainty of the optical parameters of the gold substrate used. 1-Thiol-2-propene has been used as a surface linker to prepare molecularly imprinted polymers on gold substrate and as a connector for gold nanoparticle aggregation.23 Typically, the polymerization was initiated by UV irradiation or by chemical initiators. 1-Thiol-2-propene was also immobilized on a Si(100) surface to form one-dimensional molecular lines through radical chain reaction at room temperature and elevated temperatures in vacuum.24 When it is immobilized on a gold surface to form a compact monolayer, the propagation of free radicals through the 1-thiol-2-propene will be inhibited. Therefore, the immobilized 1-thiol-2-propene will react with free radicals in the solution like (22) (a) Murphy, L. A.; Goldstein, I. J. Biochem. 1979, 18, 4999-5005. (b) Lamb, J. E.; Goldstein, I. J. Arch. Biochem. Biophys. 1984, 229, 15-26. (c) Winter, H. C.; Mostafapour, K.; Goldstein, I. J. J. Biol. Chem. 2002, 277, 1499615001. (23) (a) Wu, A. H.; Syu, M. J. Biosens. Bioelectron. 2006, 21, 2345-2353. (b) Zhao, X.; Ding, X.; Deng, Z.; Zheng, Z.; Peng, Y.; Tian, C.; Long, X. New J. Chem. 2006, 30, 915-920. (c) Zhao, X.; Ding, X.; Deng, Z.; Zheng, Z.; Peng, Y.; Long, X. Macromol. Rapid Commun. 2005, 26, 1784-1787. (24) (a) Hossain, M. Z.; Kato, H. S.; Kawai, M. J. Am. Chem. Soc. 2005, 127, 15030-15031. (b) Hossain, M. Z.; Kato, H. S.; Kawai, M. J. Phys. Chem. B 2005, 109, 23129-23133. (c) Lopinski, G. P.; Wayner, D. D. M.; Wolkow, R. A. Nature 2000, 406, 48-51.

Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

8983

Figure 3. Isotherms upon addition of Con A. PAAm film prepared on bare Au (0) and SAM-modified Au surface (9).

a terminator. The reaction between two neighboring 1-thiol-2propenes would be extremely difficult to occur because the propagation of the free radical within a two-dimensional monolayer would be very difficult due to steric hindrance of the compact self-assembled monolayer (SAM) and the lower reactivity of 1-thiol-2-propene than that of acrylamide monomers in the solution. The active surface linker would prefer to react with the free radicals in the solution, leading to quick termination of the polymerization. Additionally, to reduce the reactivity of the 1-thiol2-propene, instead of setting the polymerization temperature at the typical level, 60 °C, we used 37 °C as the polymerization condition. At this temperature, 1-thiol-2-propene has reasonable activity for our system. For example, 1-thiol-2-propene could be oxidized to diallyl disulfide, which is a natural compound extracted from garlic. Diallyl disulfide is used as an antioxidant or free radical killer in the human body. The activity of the CdC group of diallyl disulfide is similar to that of 1-thiol-propene. Since only a small amount of the 1-thiol-2-propene on the surface reacted with the free radicals in solution, the amount of the new bond was too low to be characterized by spectroscopic methods. Nevertheless, the film was very stable so that it cannot be removed by scotch tape. This result may imply some chemical bonds formed between the surface linker and the polymer film. Calibration curves with and without the surface linker are shown in Figure 3. Without the SAM of the surface linker, there is some immobilized c-PAAm on surface. The value of the frequency change is smaller when there is no surface linker. However, without the surface linker, small amount of c-PAAM could be nonspecifically immobilized on the bare gold surface. The nonspecifically adsorbed polyacrylamide film will not be stable, and the amount will be difficult to control. Therefore, in our work, the 1-thiol-2-propene was used as a surface linker. The surface linker not only increased the total amount of the immobilized polymer but also made the cross-linked polyacrylamide thin film more stable and robust. The binding constant without the surface linker calculated from the data in Figure 3 is 2.5 × 106 M-1. This means the thermodynamic nature of the bindings in these cases is similar. 8984 Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

Figure 4. Isotherms upon addition of Con A. PAAm film prepared on QCM from a mixture of monomer solutions with 0.5% (9) and 1.5% (0) Bis.

The cross-linked mannose polymer made on the gold surface of the QCM has high specificity. No response was observed when the polymer did not contain the mannose-conjugated monomer (figure not shown). ECL is a glycoprotein that has specificity toward galactose residues and appears to have the highest binding activity toward galactosyl-(R-1,4)-N-acetylglucosamine. Shown in Figure 2B, addition of ECL up to 100 nM gave almost no frequency response compared to a frequency change 313 Hz when Con A of similar concentration (80 nM) was added. This suggests higher specificity of cross-linked mannose polymer than the SAMs of mannose.6f The enhancement of the selectivity of cross-linked mannose polymer may be due to two reasons. First, at neutral and alkaline pH, Con A exists as a tetramer of four identical subunits, whereas ECL is a glycoprotein consisting of two different subunits. Multivalent binding of cross-linked mannose polymer with a 4-valency Con A could be achieved with much less difficulty than that of a 2-valency ECL. Second, each subunit of Con A has a molecular weight (MW) of approximately 26 000 Da. Thus, the total MW of the Con A tetramer is about 104 000. The physical size of each Con A subunit obtained by the X-ray crystallographic method is about 6.3-8.9 nm in all dimensions.25 The MW of ECL is about 54 000 Da; it consists of two different subunits of approximately 28 000 and 26 000 Da, respectively. Thus, Con A and ECL are different in size. The use of cross-linked polymer may provide a size selectivity in which the backbone of a crosslinked polymer formed a 3-D network. Figure 4 shows the QCM results when the cross-linker was 0.5% and 1.5%, respectively. Interestingly, the response from 0.5% Bis is larger than that from 1.5% Bis. The binding constant obtained with 1.5% Bis is 2.7 × 106 M-1, which is close to the result from 0.5% Bis, indicating that the mechanisms and valency of 1.5% Bis polymer and 0.5% Bis polymer are similar. The mesh size of the 3-D network of a cross-linked polymer depends on the ratio of the cross-linker to the monomers. The less the cross-linker is, the larger the mesh size is. A 0.5% cross-linker content means that there is one crosslinker in each 200 monomers. In this work, the cross-linker is a (25) Yonzon, C. R.; Jeoung, E.; Zou, S. L.; Schatz, G. C.; Mrksich, M.; Van Duyne, R. P. J. Am. Chem. Soc. 2004, 126, 12699-12676.

Figure 5. AFM images after absorption of Con A on c-PAAm coved QCM (left panel) and linear PAAm prepared on QCM without cross-linker (right panel).

Figure 6. SPR sensorgram obtained with a c-PAAm film when Con A was added.

tetrafunctional unit and the monomer is a bifunctional unit. Therefore, each cross-linker unit will form four “arms” by the monomers. Each of the four arms has an average of 50 monomers (200 monomers divided by the number of arms, 4). All the crosslinkers join together by the “arms” to form the cross-linked polymer. Therefore, each fragment of the cross-linked polymer (joined by two “arms”, between two neighbor cross-linkers) is an average of 100 repeat units in length (50 plus 50). A hexagonal mesh constituted with six 100-repeat-unit fragments will approximately fit the size of a Con A tetramer. However, when the

cross-linker is 1.5%, the size of many meshes will be too small for Con A to fit. Further evidence of the size selectivity was shown when E. coli W1485 was used as the target analyte. E. coli W1485 (ATCC 12435) carries type 1 fimbriae, which is specific for mannose binding. However, when c-PAAm is used, no binding of E. coli was observed because the E. coli cell is too big to move into the polymer film. Figure 5 shows AFM images obtained at the same conditions as that of Figure 1. After addition of Con A, there are a lot of nanoscaled “particles” adsorbed on the substrate (left image). The height of them is about 25 nm, which is close to the aforementioned size of the Con A tetramer. The diameter of the particles is about 150-250 nm. The average full width at half-maximum is about 50-80 nm, which is in agreement with the AFM image of Con A reported (65 nm).26 The larger observed value than its geometric dimension was rationalized as a result of the tip convolution effect. Without cross-linker, there is no significant binding (right image). Their backgrounds before the addition of Con A are similar (images not shown). The images did not change after multiple scans, indicating a strong binding of Con A on the surface. We further verified the performance in sensitivity and selectivity of the c-PAAm by the SPR technique. A sensorgram obtained with a homemade SPR is shown in Figure 6. Since we did not stir (26) Lebed, K.; Pyka-Fosciak, G.; Raczkowska, J.; Lekka, M.; Styczen, J. J. Phys.: Condens. Matter 2005, 17, S1447-S1458.

Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

8985

the system, it takes more time to reach equilibrium. There is no obvious response when ECL was added, whereas there are significant responses when Con A is added. CONCLUSION The use of carbohydrates as sensing elements incorporated into a cross-linked polymer structure represents a novel approach for molecular recognition. Our results with the c-PAAm membrane showed significant yet selective response to Con A at a concentration as low as 5 × 10-10 M. The response is proportional to the Con A concentration up to 10-7 M in PBS. The binding avidity, at the level of ∼106 M-1, is significantly larger than that of selfassembled carbohydrate monolayers and linear polymers. Furthermore, additional size selectivity could be obtained by control-

8986

Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

ling the amount of the cross-linking reagent used. Using this approach we are developing biosensors for highly sensitive and specific detection of various bacteria toxins such as Cholera toxin or Shiga toxin, to name a few. ACKNOWLEDGMENT X. Zeng thanks Oakland University and the NIH (4R33 EB000672) for support. P. G. Wang acknowledges support from The Ohio State University Ohio Eminent Scholar adornment.

Received for review July 9, 2007. Accepted September 10, 2007. AC071453Q