Anal. Chem. 2006, 78, 135-140
Use of Porous Membranes Modified with Polyelectrolyte Multilayers as Substrates for Protein Arrays with Low Nonspecific Adsorption Jinhua Dai, Gregory L. Baker, and Merlin L. Bruening*
Department of Chemistry, Michigan State University, East Lansing, Michigan 48824
Coating of substrates with polyelectrolyte multilayers terminated with poly(acrylic acid) (PAA) followed by activation of the free -COOH groups of PAA provides a surface that readily reacts with amine groups to allow covalent immobilization of antibodies. The use of this procedure to prepare arrays of antibodies in porous alumina supports facilitates construction of a flow-through system for analysis of fluorescently labeled antigens. Detection limits in the analysis of Cy5-labeled IgG are 0.02 ng/mL because of the high surface area of the alumina membrane, and the minimal diameter of the substrate pores results in binding limited by kinetics, not mass transport. Moreover, PAA-terminated films resist nonspecific protein adsorption, so blocking of antibody arrays with bovine serum albumin is not necessary. These microarrays are capable of effective analysis in 10% fetal bovine serum. Selective binding of proteins to microarrays of antibodies or other biospecific molecules has the potential to allow fast and parallel analyses of a plethora of proteins in applications such as proteomics, diagnostics and monitoring, medical research, and drug screening and testing.1-8 Despite recent successes and the promise of protein-binding microarrays, implementation of quantitative analyses using these materials still faces several challenges.9,10 Probably the biggest impediment to development of microarray-based protein analysis is production of the required number of biospecific protein binders.2,11 A second important challenge is the creation of microarray supports that resist * To whom correspondence should be addressed. Phone: (517) 355-9715, ext. 237. Fax: (517) 353-1793. E-mail:
[email protected]. (1) Phizicky, E.; Bastiaens, P. I. H.; Zhu, H.; Snyder, M.; Fields, S. Nature 2003, 422, 208-215. (2) Sage, L. Anal. Chem. 2004, 76, 137A-142A. (3) MacBeath, G.; Schreiber, S. L. Science 2000, 289, 1760-1763. (4) Zhu, H.; Klemic, J. F.; Chang, S.; Bertone, P.; Casamayor, A.; Klemic, K. G.; Smith, D.; Gerstein, M.; Reed, M. A.; Snyder, M. Nat. Genet. 2000, 26, 283-289. (5) Zhu, H.; Bilgin, M.; Bangham, R.; Hall, D.; Casamayor, A.; Bertone, P.; Lan, N.; Jansen, R.; Bidlingmaier, S.; Houfek, T.; Mitchell, T.; Miller, P.; Dean, R. A.; Gerstein, M.; Snyder, M. Science 2001, 293, 2101-2105. (6) Schweitzer, B.; Predki, P.; Snyder, M. Proteomics 2003, 3, 2190-2199. (7) Liotta, L. A.; Espina, V.; Mehta, A. I.; Calvert, V.; Rosenblatt, K.; Geho, D.; Munson, P. J.; Young, L.; Wulfkuhle, J.; Petricoin, E. F. Cancer Cell 2003, 3, 317-325. (8) Ng, J. H.; Ilag, L. L. J. Cell. Mol. Med. 2002, 6, 329-340. (9) Lee, Y.-S.; Mrksich, M. Trends Biotechnol. 2002, 20, S14-S18. (10) Wu, P.; Grainger, D. W. Biomed. Sci. Instrum. 2004, 40, 243-248. 10.1021/ac0513966 CCC: $33.50 Published on Web 11/23/2005
© 2006 American Chemical Society
nonspecific adsorption and protein denaturation while allowing immobilization of stable, fully active binders.12,13 Nonspecific adsorption in complex mixtures lowers the signal-to-background ratio and also generates false positive identifications.14,15 The most common solution to this problem is to block the microarray surface with bovine serum albumin (BSA) prior to exposure to the solution of interest, but such blocking can sometimes hinder access to binding sites.16,17 A final challenge in creating protein microarrays is developing an appropriate detection system. This is difficult because the concentrations of proteins in a sample such as human plasma can vary from 35 to 52 mg/mL for highabundance species, such as albumin, to lower than 10-12 g/mL for other species, such as hormones.18 This report focuses on developing multilayer poly(acrylic acid) (PAA)/protonated poly(allylamine) (PAH) coatings that both resist nonspecific adsorption and allow for covalent immobilization of arrays of active antibodies. Moreover, deposition of these coatings in microporous alumina supports results in a 500-fold increase in surface area relative to two-dimensional supports.19 This increased surface area decreases protein-microarray detection limits by 2 orders of magnitude. Our research builds on several investigations of both nonspecific adsorption and membrane-based microarrays. A number of efforts focused on modifying surfaces with functional groups, such as oligo-ethylene glycol, to reduce nonspecific adsorption,20,21 and mixed monolayers containing both oligo-ethylene glycol and protein-binding molecules allow for antibody immobilization.22,23 (11) Templin, M. F.; Stoll, D.; Schrenk, M.; Traub, P. C.; Vohringer, C. F.; Joos, T. O. Trends Biotechnol. 2002, 20, 160-166. (12) Seong, S.-y.; Choi, C.-y. Proteomics 2003, 3, 2176-2189. (13) Wilson, D. S.; Nock, S. Angew. Chem., Int. Ed. 2003, 42, 494-500. (14) Kusnezow, W.; Jacob, A.; Walijew, A.; Diehl, F.; Hoheisel, J. D. Proteomics 2003, 3, 254-264. (15) Kusnezow, W.; Hoheisel, J. D. J. Mol. Recognit. 2003, 16, 165-176. (16) Frederix, F.; Bonroy, K.; Reekmans, G.; Laureyn, W.; Campitelli, A.; Abramov, M. A.; Dehaen, W.; Maes, G. J. Biochem. Biophys. Methods 2004, 58, 67-74. (17) Sakaki, S.; Iwasaki, Y.; Nakabayashi, N.; Ishihara, K. Polym. J. 2000, 32, 637-641. (18) Mitchell, P. Nat. Biotechnol. 2002, 20, 225-229. See also: http:// en.wikipedia.org/wiki/List_of_human_blood_components. (19) van Beuningen, R.; van Damme, H.; Boender, P.; Bastiaensen, N.; Chan, A.; Kievits, T. Clin. Chem. 2001, 47, 1931-1933. (20) Prime, K. L.; Whitesides, G. M. Science 1991, 252, 1164-1167. (21) Wischerhoff, E. In Protein Arrays, Biochips, and Proteomics; Albala, J. S., Humphery-Smith, I., Eds.; Marcel Dekker: New York, 2003; pp 159-171. (22) Hodneland, C. D.; Lee, Y.-S.; Min, D.-H.; Mrksich, M. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 5048-5052.
Analytical Chemistry, Vol. 78, No. 1, January 1, 2006 135
Scheme 1. Procedure for Protein Immobilization in Porous Alumina Membranes
Most relevant to this work, recent studies demonstrate that PAA is highly resistant to protein adsorption from solutions with ionic strengths above 0.2 M.24,25 Dainiak et al. also observed that a PAA coating on an affinity chromatography stationary phase can function as a shielding layer that almost completely prevents the adsorption of yeast cells.26 Several other studies examined protein and cell adsorption on polyelectrolyte films that were deposited by alternating adsorption of polycations and polyanions.27-32 The layer-by-layer deposition of these coatings is intriguing for its simplicity, but in most cases, some amount of protein adsorbs to the surfaces of such films. Schlenoff and co-workers recently examined protein adsorption on a series of multilayer polyelectrolyte films and found that PAA-terminated PAH/PAA films are (23) Houseman, B. T.; Huh, J. H.; Kron, S. J.; Mrksich, M. Nat. Biotechnol. 2002, 20, 270-274. (24) Czeslik, C.; Jackler, G.; Steitz, R.; von Gru ¨ nberg, H.-H. J. Phys. Chem. B 2004, 108, 13395-13402. (25) Czeslik, C.; Jackler, G.; Hazlett, T.; Gratton, E.; Steitz, R.; Wittemann, A.; Ballauff, M. Phys. Chem. Chem. Phys. 2004, 6, 5557-5563. (26) Dainiak, M. B.; Galaev, I. Y.; Mattiasson, B. Biotechnol. Prog. 2002, 18, 815-820. (27) Heuberger, R.; Sukhorukov, G.; Vo ¨ro ¨s, J.; Textor, M.; Mo ¨hwald, H. Adv. Funct. Mater. 2005, 15, 357-366. (28) Ladam, G.; Schaaf, P.; Decher, G.; Voegel, J.-C.; Cuisinier, F. J. G. Biomol. Eng. 2002, 19, 273-280. (29) Salloum, D. S.; Schlenoff, J. B. Biomacromolecules 2004, 5, 1089-1096. (30) Mendelsohn, J. D.; Yang, S. Y.; Hiller, J. A.; Hochbaum, A. I.; Rubner, M. F. Biomacromolecules 2003, 4, 96-106. (31) Yang, S. Y.; Mendelsohn, J. D.; Rubner, M. F. Biomacromolecules 2003, 4, 987-994. (32) Salloum, D. S.; Olenych, S. G.; Keller, T. C. S.; Schlenoff, J. B. Biomacromolecules 2005, 6, 161-167.
136
Analytical Chemistry, Vol. 78, No. 1, January 1, 2006
particularly effective at suppressing protein adsorption.29 The combination of these studies led us to investigate whether PAH/ PAA films terminated with PAA could serve as coatings that resist nonspecific adsorption and allow covalent immobilization of proteins. Activation of the surface -COOH groups of PAA with N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide (EDC) and Nhydroxysuccinimide (NHS) allows covalent attachment of antibodies, and succinimidyl esters that do not react with protein are hydrolyzed by the buffer to reform -COOH groups (Scheme 1). The substrate modification and protein immobilization procedures are simple and straightforward, and the [PAA/PAH]xPAA films show minimal nonspecific adsorption of fluorescent proteins, even without blocking by BSA. Layer-by-layer deposition of polyelectrolytes also allows the modification of three-dimensional substrates, such as porous membranes,33,34 which are appealing for protein microarray analyses because of both their high surface area and the possibility of overcoming mass-transport limitations by flowing solutions through small pores. The concept of accelerating binding between immobilized antibodies and antigens by flowing analyte solution through porous absorbents was clearly demonstrated in enzymelinked immunofiltration assays,35-37 and a few studies reported the use of porous substrates for protein and gene arrays. Xu and (33) Ai, S.; Lu, G.; He, Q.; Li, J. J. Am. Chem. Soc. 2003, 125, 11140-11141. (34) Hollman, A. M.; Bhattacharyya, D. Langmuir 2004, 20, 5418-5424. (35) Clark, C. R.; Hines, K. K.; Mallia, A. K. Biotechnol. Techn. 1993, 7, 461466. (36) Paffard, S. M.; Miles, R. J.; Clark, C. R.; Price, R. G. J. Immunol. Methods 1996, 192, 133-136.
Bao immobilized proteins through physical adsorption onto a porous cellulose membrane and described the sensitivity and specificity improvements of a filtration-based protein microarray.38 Of particular importance to this work, PamGene recently developed a flow-through platform based on porous alumina, noting the advantages of higher surface area;19 however, the details of the surface modification were not revealed, and the applications published so far focused only on DNA and RNA analysis.39-41 This report shows that the combination of flow through a substrate and a simple protein-immobilization procedure that yields minimal nonspecific adsorption provides a promising system for quantitative, high-sensitivity analyses using microarrays. EXPERIMENTAL SECTION Materials. AffiniPure goat anti-rabbit IgG, anti-rat IgG, and anti-mouse IgG and their corresponding Cy5-labled ChromPure rabbit IgG, rat IgG, and mouse IgG were purchased from Jackson ImmunoResearch Lab (West Grove, PA) and were rehydrated following the manufacturer’s instruction. Fetal bovine serum was purchased from Hyclone (Logan, Utah). Poly(acrylic acid) (Mw ∼ 90 000, 25% solution in water), poly(allylamine hydrochloride) (Mw ∼ 70 000), N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide (EDC), N-hydroxysuccinimide (NHS), and TWEEN-20 surfactant were obtained from Aldrich. Buffers were prepared using analytical grade chemicals and deionized (Milli-Q, 18.2 MΩcm) water. Surface Modification and Protein Immobilization. Prior to deposition of [PAA/PAH]xPAA films, porous alumina membranes (Anodisc, 25-mm disks, 0.2-µm nominal pore size, Whatman) or aluminum-coated silicon wafers (200 nm of Al on Si(100)) were cleaned with UV/O3 for 15 min (Boekel UV clean model 135500). Polyelectrolyte deposition began with a 5-min immersion of the positively charged substrates into a solution containing 0.02 M PAA and 0.5 M NaCl at pH 4.0. (Polymer concentrations are always given with respect to the repeating unit, and pH values were adjusted with NaOH or HCl.) After rinsing with water for 1 min, the substrates were immersed in a 0.02 M PAH, 0.5 M NaCl solution (pH adjusted to 4.0) for 5 min and rinsed with water for 1 min. The procedure was repeated to allow the deposition of [PAA/PAH]3PAA films, and these films were then dried with N2, immersed in a solution containing EDC/NHS (0.1 M of each in pure water) for 30 min, washed with water for ∼30 s, and dried with N2. The EDC/NHS-activated membranes were immediately spotted with 0.2 µL (per spot) of antibodies (1.0 mg/mL) and placed in a sealed Petri dish that was saturated with water vapor. (This procedure results in antibody spots that are ∼2 mm in diameter.) To prepare films for reflectance FT-IR characterization of antibody binding, 200 µL of anti-rabbit IgG (1.0 mg/mL) was spread onto EDC/NHS-activated [PAA/PAH]3PAA films on Al(37) Morais, S.; Gonza´lez-Martı´nez, M. A.; Abad, A.; Montoya, A.; Maquieira, A.; Puchades, R. J. Immunol. Methods 1997, 208, 75-83. (38) Xu, Y.; Bao, G. Anal. Chem. 2003, 75, 5345-5351. (39) Hokaiwado, N.; Asamoto, M.; Tsujimura, K.; Hirota, T.; Ichihara, T.; Satoh, T.; Shirai, T. Cancer Sci. 2004, 95, 123-130. (40) Maekawa, M.; Nagaoka, T.; Taniguchi, T.; Higashi, H.; Sugimura, H.; Sugano, K.; Yonekawa, H.; Satoh, T.; Horii, T.; Shirai, N.; Takeshita, A.; Kanno, T. Clin. Chem. 2004, 50, 1322-1327. (41) Wu, Y.; de Kievit, P.; Vahlkamp, L.; Pijnenburg, D.; Smit, M.; Dankers, M.; Melchers, D.; Stax, M.; Boender, P. J.; Ingham, C.; Bastiaensen, N.; de Wijn, R.; van Alewijk, D.; van Damme, H.; Raap, A. K.; Chan, A. B.; van Beuningen, R. Nucleic Acids Res. 2004, 32, e123.
coated Si (2 × 1 cm) and allowed to incubate for 2 h in a sealed Petri dish that was saturated with water vapor. These films were then rised for 5 min with washing buffer, vortexed for 3 min in washing buffer, and rinsed with water. Microarray Protein Assays. After an overnight incubation, the antibody-spotted membranes were washed intensively with washing buffer (20 mM phosphate buffer containing 0.5 M NaCl and 0.1% v/v TWEEN-20, pH 7.2). The washing included a 5-min rinse from a wash bottle followed by a 30-min flowing wash (4 mL/min using a peristaltic pump and a syringe filter holder, Millipore, Catalogue no. SX0002500) and three 5-min washes in a Millipore stirring cell (model 8010). The washed membranes were then exposed to antigen solutions by a simple immersion (quiescent mode), by immersion and rotational agitation (shaking mode), or by flowing antigen solution through the pores (flow mode). For the flow-mode interrogation, the membranes were sandwiched in a syringe filter holder, and antigen solutions were forced through the membrane with a peristaltic pump. After exposure to the antigens, the membranes were again washed intensively with buffer as described above, rinsed briefly with water, dried under a gentle N2 stream, and subjected to fluorescence scanning. Imaging and Data Analysis. The membrane protein microarrays were scanned using an Affymetrix 428 fluorescence scanner. Images were obtained using excitation at 635 nm and a detector gain of 45. Fluorescence intensities were quantified using GenePix 3.0 software provided by the scanner manufacturer. Reflectance FT-IR spectra were obtained using a Nicolet MagnaIR 560 spectrometer with a Pike grazing angle (80°) accessory, and the cross-sectional fluorescence images were acquired using a Zeiss LSM 210 confocal laser scanning microscope. RESULTS AND DISCUSSION Covalent Attachment of Proteins to [PAA/PAH]3PAA Films. Scheme 1 shows our strategy for attachment of antibodies to [PAA/PAH]3PAA films using activation of PAA with EDC and NHS and covalent coupling via amide formation. Intensive rinsing with pH 7.2 washing buffer results in hydrolysis of succinimide esters surrounding the antibody spots and recovery of the carboxylate groups of PAA. Reflectance FT-IR spectra of [PAA/ PAH]3PAA films on Al-coated Si (Figure 1) confirm the steps of this procedure. (To prepare films for reflectance FT-IR spectra, the entire film was exposed to protein solution, not just a few spots.) After activation, new peaks due to the symmetric and asymmetric stretches of succinimide and the carbonyl stretch of the succinimide ester appeared at 1810, 1780, and 1740 cm-1, respectively (Figure 1b). Disappearance of the acid carbonyl peak at 1710 cm-1 (compare spectra 1a and 1b) also suggests the formation of the activated ester. After exposure to a 1.0 mg/mL solution of anti-IgG in pH 7.2 buffer for 2 h, peaks due to the active ester vanished, and an amide I absorbance at 1640 cm-1 appeared (Figure 1c), but the amide absorbance could result from protein, amide cross-links formed between PAA and PAH,42,43 or the amide linkage between PAA and the protein. Comparison of (42) Richert, L.; Boulmedais, F.; Lavalle, P.; Mutterer, J.; Ferreux, E.; Decher, G.; Schaaf, P.; Voegel, J.-C.; Picart, C. Biomacromolecules 2004, 5, 284294. (43) Yang, S. Y.; Lee, D.; Cohen, R. E.; Rubner, M. F. Langmuir 2004, 20, 59785981.
Analytical Chemistry, Vol. 78, No. 1, January 1, 2006
137
Figure 1. Reflectance FT-IR spectra of (a) a [PAA/PAH]3PAA film; (b) an EDC/NHS-activated [PAA/PAH]3PAA film, (c) film b after reaction with anti-rabbit IgG (1.0 mg/mL for 2 h) and rinsing, and (d) an EDC/NHS-activated [PAA/PAH]3PAA film after immersion in pH 7.2 buffer (without protein) for 2 h.
the spectrum of the film exposed to protein (spectrum 1c) with that of a [PAA/PAH]3PAA film activated with EDC and NHS and exposed to only buffer (spectrum 1d) shows that more than half of the band at 1640 cm-1 is due to protein or the amide formed in its attachment. Because of overlap with a carboxylate stretch, the amide II absorbance cannot be quantified. The similarities between the spectrum of a [PAA/PAH]3PAA film before (spectrum 1a) and after activation and exposure to buffer (spectrum 1d) also suggest that activated esters undergo hydrolysis in buffer. Control experiments that probed the binding of antigens to antibodies on membranes provided further evidence for covalent attachment of proteins. We spotted anti-rabbit IgG onto alumina membranes coated with either [PAA/PAH]3PAA or EDC/NHSactivated [PAA/PAH]3PAA and washed with buffer and subsequently flowed (flow rate of 4 mL/min for 30 min) a solution of 0.5 µg/mL Cy5-labeled rabbit IgG in 0.5 M NaCl and buffer through these membranes. After extensive rinsing, fluorescence intensities were >65 000 (saturation of the detector) counts per second (cps) at antibody spots on the EDC/NHS-activated coating but only 400 cps (marginally detectable) at spots on the unactivated [PAA/PAH]3PAA. This observation implies that physically adsorbed protein was essentially completely removed from the [PAA/PAH]3PAA film and that covalent attachment is required to keep the protein bound to the surface. Similar experiments with an uncoated alumina membrane yielded signals of ∼4200 cps, showing that physisorption is less on [PAA/PAH]3PAA than on alumina. Blocking of [PAA/PAH]3PAA films with BSA is not necessary to eliminate nonspecific adsorption, and this should avoid any difficulties relating to masking of some binding sites by BSA. Effect of Flow Rate on IgG Binding to Anti-IgG Covalently Immobilized on [PAA/PAH]3PAA-Coated Porous Alumina. One asset of membrane-based antibody arrays is that flow of sample through small pores should result in kinetic, rather than mass-transport, control of antigen-antibody interactions because of the minimal diffusion distances required to reach binding sites. Consistent with this supposition, binding of Cy5-labeled IgG to anti-IgG immobilized in alumina was essentially constant over a flow rate range of 1-8 mL/min (Figure 2). The linear velocity at 138 Analytical Chemistry, Vol. 78, No. 1, January 1, 2006
Figure 2. Fluorescence intensity due to bound Cy5-labeled IgG as a function of flow rate through [PAA/PAH]3PAA-coated porous alumina membranes that were triple-spotted with anti-rabbit IgG. The sample contained 0.5 ng/mL Cy5-labeled rabbit IgG in buffer (20 mM phosphate + 0.5 M NaCl, pH 7.2), and the flow rates from a to e were 1, 2, 4, 6, and 8 mL/min. The blue and pink open diamonds represent two sets of independent data with a total of 10 membranes. Antibody spots are ∼2 mm in diameter.
a flow rate of 1 mL/min is only 0.64 cm/min (effective membrane diameter of 2.0 cm, porosity of 50%), which is considerably smaller than the ∼100 cm/min cross-flow velocity that Ligler and coworkers suggested is necessary to achieve kinetic control with two-dimensional substrates.44 Thus, the elimination of diffusion limitations at low flow rates through the membrane probably stems from small diffusion distances within the 150-nm-diameter pores and not from drastic reductions in boundary layer thickness due to flow. (The nominal diameter of the pores in these alumina membranes is 200 nm, but the presence of the [PAA/PAH]3PAA films should decrease the accessible diameter by 50 nm.) Figure 2 also demonstrates the reproducibility of these membrane-based assays and the resistance of [PAA/PAH]3PAA films to nonspecific adsorption. The fluorescence intensities with five different flow rates and 10 different membranes have a standard deviation of 6%, suggesting that these arrays should be capable of quantitative analyses. The dark background in the images in Figure 2 and in all of our other results is representative of
