Label-Free Detection of Proteins in Crude Cell Lysate with Antibody

Anal. Chem. 2005, 77, 7115-7121

Label-Free Detection of Proteins in Crude Cell Lysate with Antibody Arrays by a Surface Plasmon Resonance Imaging Technique Motoki Kyo,*,† Kazue Usui-Aoki,‡ and Hisashi Koga‡,§

Biotechnology Frontier Project, Toyobo Co., Ltd., 10-24 Toyo-cho, Tsuruga, Fukui 914-0047, Japan, Chiba Industry Advancement Center, 2-6 Nakase, Mihama-ku, Chiba 261-7126, Japan, and Kazusa DNA Research Institute, 2-6-7 Kazusa-Kamatari, Kisarazu Chiba 292-0818, Japan

We established a label-free method of measuring proteins in crude cell lysate using antibody arrays and surface plasmon resonance (SPR) imaging. The refractivity of the running buffer was adjusted with that of the lysate to overcome the bulk effect. The chemistries of the fabricated arrays were investigated to reduce nonspecific adsorption on the array surface. We found that the hydrophilicity of the poly(ethylene glycol) moiety and lower electrostatic charge on the surface provided a specific measurement of antigen-antibody interaction. We validated the system by measuring the expression of eight proteins in the mouse brain and comparing the results to those by conventional Western blotting. The detection limit of the antibody array was ∼30 ng/mL in crude cell lysate, on the same order as that of previous SPR research. The system enabled quick, label-free, and high-throughput analysis of abundant proteins with minimal sample volume (∼200 µL). It is expected that our SPR antibody array will be applicable for direct protein expression profiling of cell lysate, as well as for cell phenotyping, food analysis, discovery of new biomarkers, and immunological disease diagnostics. Protein expression analyses are widely used to clarify cellular conditions, signal transduction, and the behavior of diseaseassociated proteins. In fact, such analyses have become essential, because no persuasive correlation had been shown between mRNA expression levels and the abundance of the corresponding proteins.1,2 Proteome profiling, which is the systematic study of protein expression patterns in cells or tissues, was originally achieved with two-dimensional gel electrophoresis.3,4 However, this con* To whom correspondence should be addressed. Phone: +81-6-6348-3893. Fax: +81-6-6348-3833. E-mail: [email protected]. † Toyobo Co., Ltd. ‡ Chiba Industry Advancement Center. § Kazusa DNA Research Institute. (1) Anderson, L.; Seilhamer, J. Electrophoresis 1997, 18, 533-537. (2) Gygi, S. P.; Rochon, Y.; Frenza, B. R.; Aebersold, R. Mol. Cell. Biol. 1999, 19, 1720-1730. (3) Patton, W. F. Electrophoresis 2000, 21, 1123-1144. (4) Tsugita, A.; Kawakami, T.; Uchida, T.; Sakai, T.; Kamo, M.; Matsui, T.; Watanabe, Y.; Morimasa, T.; Hosokawa, K.; Toda, T. Electrophoresis 2000, 21, 1853-1871. 10.1021/ac050884a CCC: $30.25 Published on Web 10/18/2005

© 2005 American Chemical Society

ventional method required special skills and was labor intensive. More recently, mass spectrometry has become the dominant method of proteome analysis.5-7 However, this technique has low throughput and requires expensive, sophisticated equipment. Accordingly, there has been need of a simple, rapid, highthroughput, cost-effective approach. In the past decade, antibody array technology has made rapid progress as a possible means of quick and simultaneous detection with minimal sample volume.8-16 In this method, the antibodies are immobilized on a substrate and interact with target antigens in the sample during the antibody array analysis. This technique may provide a high-throughput alternative to conventional proteome profiling methods. Importantly, more than 2000 antibodies against the products of mouse cDNA clones have been produced in our ongoing project.17,18 A large number of antibodies for several species will be available in the near future. The antibody array is thus considered a promising technique for protein expression analysis. Membranes and glass slides have been used as substrates for the antibody array in previous studies, and this method has shown (5) Gygi, S. P.; Rist, B.; Gerber, S. A.; Turecek, F.; Gelb, M. H.; Aebersold, R. Nat. Biotechnol. 1999, 17, 994-999. (6) Gygi, S. P.; Rist, B.; Griffin, T. J.; Eng, J.; Aebersold, R. J. Proteome Res. 2002, 1, 47-54. (7) Lee, H.; Griffin, T. J.; Gygi, S. P.; Rist, B.; Aebersold, R. Anal. Chem. 2002, 74, 4353-4360. (8) de Wildt, R. M.; Mundy, C. R.; Gorick, B. D.; Tomlinson, I. M. Nat. Biotechnol. 2000, 18, 989-994. (9) Sreekumar, A.; Nyati, M. K.; Varambally, S.; Barrette, T. R.; Ghosh, D.; Lawrence, T. S.; Chinnaiyan, A. M. Cancer Res. 2001, 61, 7585-7593. (10) Wiese, R.; Belosludtsev, Y.; Powdrill, T.; Thompson, P.; Hogan, M. Clin. Chem. 2001, 47, 1451-1457. (11) Huang, R. P.; Huang, R.; Fan, Y.; Lin, Y. Anal. Biochem. 2001, 294, 55-62. (12) Knezevic, V.; Leethanakul, C.; Bichsel, V. E.; Worth, J. M.; Prabhu, V. V.; Gutkind, J. S.; Liotta, L. A.; Munson, P. J.; Petricoin, E. F.; Krizman, D. B. Proteomics 2001, 1, 1271-1278. (13) Angenendt, P.; Glokler, J.; Murphy, D.; Lehrach, H.; Cahill, D. J. Anal. Biochem. 2002, 76, 2916-2921. (14) Miller, J. C.; Zhou, H.; Kwekel, J.; Cavallo, R.; Burke, J.; Butler, E. B.; Teh, B. S.; Haab, B. B. Proteomics 2003, 3, 56-63. (15) Anderson, K.; Potter, A.; Baban, D.; Davies, K. E. Brain 2003, 126, 20522064. (16) Angenendt, P.; Wilde, J.; Kijanka, G.; Baars, S.; Cahill, D. J.; Kreutzberger, J.; Lehrach, H.; Konthur, Z.; Glokler, J. Anal. Chem. 2004, 76, 2916-2921. (17) Hara, Y.; Shimada, K.; Kohga, H.; Ohara, O.; Koga, H. DNA Res. 2003, 10, 129-136. (18) Koga, H.; Shimada, K.; Hara, Y.; Nagano, M.; Kohga, H.; Yokoyama, R.; Kimura, Y.; Yuasa, S.; Magae, J.; Inamoto, S.; Okazaki, N.; Ohara, O. Proteomics 2004, 4, 1412-1416.

Analytical Chemistry, Vol. 77, No. 22, November 15, 2005 7115

an excellent advantage for cytokine analysis.19-23 However, it also requires a direct or indirect labeling process, which may affect protein function, and takes additional time. Furthermore, the detection protocol has not been optimized yet. The possibility of label-free interaction analyses of proteins has been investigated using surface plasmon resonance (SPR) and quartz crystal microbalance methods. The capture molecule was immobilized on the sensor surface, and the interaction between the capture molecule and analyte could be detected directly in both methods. Initially, these label-free analytical techniques had a disadvantage of low throughput, but over the past decade, this problem has been overcome through the development of multispot SPR for observing interactions with an array format.24-27 Most recently, we reported on the use of the SPR technique for antibody arrays on which 400 antibodies were spotted.28 AntimKIAA antibodies,18 here “m” and “KI” stood for mouse and Kazusa DNA Research Institute, respectively, and “AA” was reference characters, were immobilized by physical adsorption on bare gold, and an SPR instrument with grating26 was used in the research. Though we could measure 400 real-time antibodytarget bindings within a single hour, we also noted three important areas for improvement, i.e., the sensitivity, bulk effect, and nonspecific adsorption. With regard to the sensitivity, the detection limit of the research was ∼5 µg/mL and much higher than that of conventional SPR (80 ng/mL).29 With regard to the second issue, the difference of composition and concentration between the sample and running buffer induced a bulk effect.30 Diluting the sample could reduce the bulk effect, but also led to desensitization. Finally, the crucial nonspecific adsorption caused by inappropriate surface chemistry interfered with the sensitivity, because SPR detects not only specific but also nonspecific adsorption; indeed, it can detect refractivity changes in the vicinity of the sensor surface. In this study, we used an SPR imaging instrument25 and overcame the problem of the bulk effect induced by the refractivity difference between the crude cell lysate and running buffer. The surface chemistries for antibody immobilization on self-assembled monolayers (SAMs) of alkanethiol tethers were optimized. The alkanethiol tethers containing poly(ethylene glycol) (PEG) were examined to reduce the nonspecific adsorptions on the surface.31 (19) Tam, S. W.; Wiese, R.; Lee, S.; Gilmore, J.; Kumble, K. D. J. Immunol. Methods 2002, 261, 157-165. (20) Wang, C. C.; Huang, R. P.; Sommer, M.; Lisoukov, H.; Huang, R.; Lin, Y.; Miller, T.; Burke, J. J. Proteome Res. 2002, 1, 337-343. (21) Lin, Y.; Huang, R.; Santanam, N.; Liu, Y. G.; Parthasarathy, S.; Huang, R. P. Cancer Lett. 2002, 187, 17-24. (22) Lin, Y.; Huang, R.; Cao, X.; Wang, S. M.; Shi, Q.; Huang, R. P. Clin. Chem. Lab. Med. 2003, 41, 139-145. (23) Turtinen, L. W.; Prall, D. N.; Bremer, L. A.; Nauss, R. E.; Hartsel, S. C. Antimicrob. Agents Chemother. 2004, 48, 396-403. (24) Jordan, C. E.; Corn, R. M. Anal. Chem. 1997, 69, 1449-1456. (25) Nelson, B. P.; Frutos, A. G.; Brockman, J. M.; Corn, R. M. Anal. Chem. 1999, 71, 3928-3934. (26) Baird, C. L.; Myszka, D. G. J. Mol. Recognit. 2001, 14, 261-268. (27) Shumaker-Parry, J. S.; Zareie, M. H.; Aebersold, R.; Campbell, C. T. Anal. Chem. 2004, 76, 918-929. (28) Usui-Aoki, K.; Shimada, K.; Nagano, M.; Kawai, M.; Koga, H. Proteomics 2005, 5, 2396-2401. (29) Indyk, H. E.; Filonzi, E. L. J. AOAC Int. 2003, 86, 386-393. (30) O’Brien, M. J., II; Brueck, S. R. J.; Perez-Luna, V. H.; Tender, L. M.; Lopez, G. P. Biosens. Bioelectron. 1999, 14, 145-154. (31) Siegel, R. R.; Harder, P.; Dahint, R.; Grunze, M.; Josse, F.; Mrksich, M.; Whitesides, G. M. Anal. Chem. 1997, 69, 3321-3328.

7116

Analytical Chemistry, Vol. 77, No. 22, November 15, 2005

The sensitivity and reproducibility were evaluated using reference antibodies, and finally, we compared the SPR results for eight proteins in crude cell lysate of the mouse brain with those by Western blotting, a conventional immunological method. EXPERIMENTAL SECTION Materials. PEG6COOH undecanethiol (HS-PEG-COOH; SensoPath, Bozeman, MT), PEG3OH undecanethiol (HS-PEG-OH; SensoPath), 11-carboxy-1-decanethiol (11-CDT; Dojindo, Kumamoto, Japan), 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC; Pierce, Rockford, IL), N-hydroxysuccinimide (NHS; Pierce), bovine serum albumin (BSA; Nacalai, Kyoto, Japan), amine-PEG4-alcohol (H2N-PEG-OH; Quanta Biodesign, Powell, OH), anti-actin monoclonal antibody (anti-actin; Chemicon, Temecula, CA), anti-tubulin monoclonal antibody (anti-tubulin; Chemicon), anti-streptavidin polyclonal antibody (anti-SA; Vector, Burlingame, CA), and streptavidin (SA; Vector) were used without further purification. BSA and gelatin were not included in the antibody solutions. The phosphate buffer (phosphate 10 mM, pH 7.4, NaCl 150 mM) was used for antibody immobilization reactions. HEPES buffer (HEPES 20 mM, pH 7.2, NaCl 150 mM) or HEPES buffer supplemented with BSA was used in the SPR analysis as a running buffer. Anti-mKIAA Antibody Preparation. Six anti-mKIAA polyclonal antibodies were generated by the subcutaneous immunization of rabbits with mKIAA protein fragments, which were expressed as GST-fusion proteins. The GST-mKIAA proteins were produced in Escherichia coli using an in vitro recombinationassisted method according to the procedure previously described.17 Subsequently, we validated the expressed antigens by a peptide mass fingerprint method using MALDI-TOF-MS (AXIMACFR, Shimadzu, Kyoto, Japan). After the validation, 0.25 mg of each antigen mixed with Freund’s complete adjuvant was subcutaneously injected into rabbits biweekly. Under achievements of antibody values, the rabbits were sacrificed 7 days after the last boost, and then the sera were collected. The antibodies were purified from the sera with a 1-mL column volume of protein A Sepharose (Amersham; Piscataway, NJ). The information for corresponding mKIAA proteins was obtained from the ROUGE Protein Database (http://www.kazusa.or.jp/rouge/).32 Fabrication of Antibody Arrays. The surface chemistries of the fabricated arrays are shown in Figure 1. A gold-coated chip (Toyobo, Osaka, Japan) was immersed in the 1 mM ethanol solution of HS-PEG-OH. The PEG-covered surface was photopatterned by a UV light source (Ushio, Tokyo, Japan) at 40 mW/ cm2 for 2 h with a chromium quartz mask, which had 96 square holes of 500 µm.33-35 The COOH group was introduced via the three schemes shown in Figure 1a, i.e., immersion in an ethanol solution of 1 mM 11-CDT (scheme A), 1 mM HS-PEG-COOH (scheme B) or a mixture of 0.1 mM HS-PEG-COOH and 0.9 mM HS-PEG-OH (scheme C). The carboxyl groups introduced on the chip were reacted for 1 h with 0.2 M EDC and 0.05 M NHS in (32) Kikuno, R.; Nagase, T.; Nakayama, M.; Koga, H.; Okazaki, N.; Nakajima, D.; Ohara, O. Nucleic Acids Res. 2004, 32, D502-504. (33) Brockman, J. M.; Frutos, A. G.; Corn, R. M. J. Am. Chem. Soc. 1999, 121, 8044-8051. (34) Kyo, M.; Yamamoto, T.; Motohashi, H.; Kamiya, T.; Kuroita, T.; Tanaka, T.; Engel, J. D.; Kawakami, B.; Yamamoto, M. Genes Cells 2004, 9, 153-164. (35) Inamori, K.; Kyo, M.; Nishiya, Y.; Inoue, Y.; Sonoda, T.; Kinoshita, E.; Koike, T.; Katayama, Y. Anal. Chem. 2005, 77, 3979-3985.

Figure 1. Surface chemistries of the fabricated antibody array. (a) Carboxyl groups were introduced on a gold surface via three procedures. The SAMs of 11-CDT (scheme A), HS-PEG-COOH (scheme B), and a mixture of HS-PEG-COOH and HS-PEG-OH (scheme C) were formed on the UV-irradiated surface. (b) The introduced carboxyl groups were activated by EDC and NHS and then reacted with antibodies. The unreacted NHS ester groups were blocked by H2N-PEG-OH.

the phosphate buffer.36 The chip was washed with water, dried by airflow, and placed into an automated spotter (Toyobo). The spotter consisted of a contact-type cylindrical pin with an i.d. of 450 µm and a humidifier, which could maintain the humidity inside the spotter. It delivered ∼10-nL drops of 100-1000 µg/mL antibody solutions automatically on the NHS ester-modified surface. The spotted droplets of the solutions could not be dried out during spotting, because the relative humidity inside of the spotter were maintained at 82%. The immobilization reaction was carried out for 2 h in a moist chamber. The NH2 groups on antibody molecules could be reacted on the surface; consequently, the antibodies were immobilized at random orientations. In a final step, 200 µL of 2 mg/mL H2N-PEG-OH was poured on the surface for 1 h to block the unreacted NHS ester group (shown in Figure 1b). Then the surface was rinsed with a buffer. A phosphate buffer (10 mM phosphate, pH 7.2, 150 mM NaCl) was used for all reactions in the array fabrication. Cell Lysate Preparation. Brains of sacrificed adult male mice (ICR strain, 8 weeks, male) were homogenized with mammalian cell lysis/extraction reagent (Sigma, St. Louis, MO) containing 0.5% protease inhibitor cocktail (Sigma). The cell lysate was centrifuged at 10 000 rpm for 3 min at 4 °C, and the supernatant was used for SPR analysis. The protein content of lysate was determined by Bradford protein assay, in which a calibration curve was obtained with using BSA. The refractive index (nD) was measured by a refractometer (Atago, Tokyo, Japan) at room temperature. (36) Johnsson, B.; Lofas, S.; Lindquist, G. Anal. Biochem. 1991, 198, 268-277.

SPR Imaging Analysis. The antibody array on the gold chip was placed into an SPR imaging instrument (Toyobo). HEPES buffer (20 mM HEPES-NaOH, pH 7.2, 150 mM NaCl) or HEPES buffer supplemented with BSA was used as an SPR running buffer. The SPR measurement was performed at an angle, at which the reflectivity at the antibody spot was ∼10%, when the SPR running buffer flowed. The running buffer and the cell lysate were applied on the array surface at 100 µL/min. The signal data were collected with an analysis program (Toyobo). One SPR signal corresponded to ∆0.15% reflectivity in the program. All SPR experiments were performed at 30 °C. Western Blot Analysis. Adult mice (ICR strain, 8 weeks, male) were sacrificed, and individual organs (heart, lung, muscle, kidney, testis, pancreas, thymus, spleen, brain, prostate) were homogenized with CelLytic M (Sigma) containing 0.5% protease inhibitor cocktail (Sigma). Protein concentrations of lysates were determined by the Bradford protein assay. Subsequently, 20 µg of each sample was resolved by 8% SDS-PAGE, transferred to a PVDF membrane (Immobilon-P; Millipore), and probed with the diluted anti-mKIAA antibodies (1:2000). After washing with PBS containing 0.05% Tween 20, the filter was incubated with horseradish peroxidase-conjugated anti-rabbit IgG (1:2000, Amersham), and specific proteins were detected using an enhanced chemiluminescence system (ECL Plus; Amersham) with optimized exposure times. RESULTS AND DISCUSSION Refractivity Adjustment Overcame a Bulk Effect in SPR Analysis of Cell Lysate. Antibody arrays with anti-actin, antiAnalytical Chemistry, Vol. 77, No. 22, November 15, 2005

7117

Figure 2. SPR signal changes by exposures of mouse brain lysate (nD ) 1.3391) and mouse brain lysate supplemented with 100 ng/ mL SA. (a) HEPES buffer (nD ) 1.3345) was used for the running buffer. (b) HEPES buffer containing 2.5% BSA (nD ) 1.3390) was used. The array was rinsed with HEPES buffer (BSA-free) for 1 min during switching injections.

tubulin, and anti-SA were fabricated for this analysis. The photopatterned chip was immersed in an ethanol solution of a mixture of 0.1 mM HS-PEG-COOH and 0.9 mM HS-PEG-OH (Figure 1a, scheme C). 200 µg/mL antibody solution was prepared and spotted onto the NHS ester-modified surfaces (Figure 1b). The antibody array was placed on an SPR imaging instrument and exposed to brain lysate and 100 ng/mL SA-containing lysate in sequence. The total protein concentration and refractive index (nD) of the brain lysate were 11.3 mg/mL and 1.3391, respectively. The SPR signal changes are shown in Figure 2a, when the HEPES buffer was used for the SPR running buffer. The crucial baseline changes were observed due to the bulk effect, because the nD of HEPES buffer was 1.3345, which is 0.0046 smaller than that of the brain lysate. The signals during the use of the running buffer could be analyzed, and revealed that actin and tubulin were detected in the lysate, and SA was detected in the lysate supplemented with SA. However, the signal changes during lysate injection could not be observed, hence the signal saturation points, when the antigen-antibody interactions reached their equilibriums, were unclear. Additionally, the slight changes and the disassociative interactions may have been missed due to the crucial baseline changes. To reduce the bulk effect, the cell lysate or serum were diluted with running buffer in previous reports.28,37,38 However, the sensitivity became lower according to the dilution ratios. To 7118 Analytical Chemistry, Vol. 77, No. 22, November 15, 2005

overcome this problem, we adjusted the nD of the running buffer by adding BSA, which was known as a noninteractive protein toward antibodies. We consider that the refractivity can be coordinated with various water-soluble molecules (e.g., proteins, sugars, salts, synthetic polymers, and alcohols). We chose BSA for nD adjustment, because cell lysate contained 11.3 mg/mL proteins. The nD of the BSA-added HEPES could be determined by the following formula: n ) 0.0018C + 1.3345 (R2 ) 0.994), where n and C are the nD of the buffer and the BSA concentration (%), respectively. Therefore, we prepared a HEPES buffer containing 2.5% BSA, the nD of which was 1.3390, or only 0.0001 different from that of the lysate, as the SPR running buffer. The SPR signal changes resulting from the use of the buffer containing BSA are shown in Figure 2b. The array was rinsed with BSA-free HEPES buffer for 1 min during the switching of injections to avoid signal fluctuations by the interfusion and diffusion of BSA and cell lysate ingredients. The signal changes during lysate injection could be monitored, because the baseline changes were negligible. A gradual signal increase was observed on the anti-SA spot by exposure to the lysate supplemented with SA. By the technique, concentrations near the detection limit could be observed as slight changes without the bulk effect. Nonspecific binding of BSA on an array surface was observed by sequential exposures of BSA-free and BSA-added buffer (data not shown). The nonspecific binding of BSA on each spot was negligible. Therefore, we considered that BSA did not have a blocking effect on the array surface. PEG Moiety and Lower Density of the Carboxyl Groups in SAM Reduced Nonspecific Binding. We introduced carboxyl groups onto the surfaces by the three procedures described in Figure 1a. The SAM of 11-CDT was formed on gold in scheme A. In scheme B, PEG moiety-inserted alkanethiol, HS-PEG-COOH, was used. In scheme C, HS-PEG-COOH and HS-PEG-OH were immobilized and the density of the COOH group was reduced. Then the antibodiessi.e., six anti-mKIAA antibodies, plus antiactin, anti-tubulin, and anti-SAswere immobilized on the array. The 200 µg/mL antibody solutions were prepared and spotted for the reactions. The total protein concentration and refractive index (nD) of the brain lysate were 5.6 mg/mL and 1.3371, respectively. The 1.5% BSA-containing HEPES buffer (nD ) 1.3370) was used as an SPR running buffer. The SPR analysis results for mouse brain lysate using the antibody array prepared with scheme A are shown in Figure 3a. The signals on anti-actin and anti-tubulin were significant. However, the signal on blank spots (gray line), on which 11-CDT was densely packed and no antibody was immobilized, was increased by the nonspecific adsorption. In contrast, the signal on the background (black line), on which HS-PEG-OH was immobilized, indicated a subtle change. We speculated that the nonspecific adsorption on blank spots was raised by hydrophobic and electrostatic effects. The detectible SPR signals on antibody spots were obtained, but it was difficult to distinguish the specific interactions. (37) Ritter, G..; Cohen, L. S.; Williams, C.; Richards, E. C.; Old, L. J.; Welt, S. Cancer Res. 2001, 61, 6851-6859. (38) Campagnolo, C.; Meyers, K. J.; Ryan, T.; Atkinson, R. C.; Chen, Y. T.; Scanlan, M. J.; Ritter, G.; Old, L. J.; Batt, C. A. J. Biochem. Biophys. Methods 2004, 61, 283-298.

Figure 3. SPR signal changes on the antibody array by exposure of mouse brain lysate (nD ) 1.3371) using HEPES buffer supplemented with 1.5% BSA (nD ) 1.3370) as the running buffer. The antibody arrays were prepared via (a) scheme A, (b) scheme B, and (c) scheme C in Figure 1a. The schemes of antibody immobilization are depicted in (a′), (b′), and (c′), which correspond to (a), (b), and (c), respectively.

To overcome the nonspecific adsorption by hydrophobic effect, we used HS-PEG-COOH instead of 11-CDT. The antibody array prepared by scheme B showed an improvement on the blank spots (Figure 3b). We think that the PEG moiety in HS-PEG-COOH reduced the nonspecific adsorption, because the repeat unit of ethylene glycol was known as a hydrophilic and flexible spacer, which could suppress nonspecific adsorption.31,34,39,40 However, the nonspecific adsorption remained on the blank spot, on which HS-PEG-COOH was densely packed. We speculate that this was due to the electrostatic effect of the carboxyl groups. To avoid the electrostatic nonspecific adsorption, we reduced the density of the carboxyl groups on the surface by using a mixture of HS-PEG-COOH and HS-PEG-OH (Figure 1a, scheme C). Figure 3c shows that the SPR signal change on the blank spot, prepared by the mixture solution with 0.1 mM HS-PEG-COOH and 0.9 mM HS-PEG-OH, was negligible. The SPR signals on antibody spots in Figure 3c were smaller than those in Figure 3a and b. We consider that the signals in (39) Uchida, K.; Otsuka, H.; Kaneko, M.; Kataoka, K.; Nagasaki, Y. Anal. Chem. 2005, 77, 1075-1080. (40) Okumura, A.; Sato, Y.; Kyo, M.; Kawaguchi, H. Anal. Biochem. 2005, 339, 328-337.

Figure 3a and b may have included the nonspecific adsorption amounts. Even if the signals of the specific interactions were depressed in scheme C, the noise by nonspecific adsorption was negligible. Consequently, the signal-to-noise ratio was much improved. We also examined the concentration of HS-PEG-COOH from 0.01 to 0.5 mM when the total alkanethiol concentration was fixed at 1 mM. However, no significant difference was observed in nonspecific adsorption and SPR signal intensity (data not shown). Serum analyses by SPR with antibodies immobilized on a hydrophilic carboxymethyl dextran surface have been reported previously.37,38 Significant nonspecific adsorptions were observed. We think that molecules in the serums were adsorbed on the surface electrostatically, and thus, the reduction of the carboxymethyl group on surface may have improved the nonspecific bindings. In our surface chemistry, the antibodies were immobilized at random orientations. Nonoriented antibodies may cause nonspecific protein interactions. However, nonspecific adsorptions on antiSA spots were much lower than specific bindings on anti-actin and anti-tubulin spots by exposure of SA-free cell lysate. Therefore, nonspecific bindings on nonoriented antibodies were not significant in our experiment. Analytical Chemistry, Vol. 77, No. 22, November 15, 2005

7119

Figure 4. Relationship between SPR signals and SA concentrations on anti-SA spots.

Sensitivity and Reproductively of Antibody Arrays. We examined the sensitivity of the antibody array by using SA as a model protein. The antiSA-immobilized arrays were fabricated by spotting 1.0, 0.5, 0.2, and 0.1 mg/mL anti-SA solutions on an NHS ester-modified surface, prepared via scheme C in Figure 1a. The mouse brain lysates containing 10, 30, 100, 300, 1000, and 3000 ng/mL SA were applied on the arrays for 600 s sequentially. And the SPR signals were collected at 570-s exposure of each sample, because the SPR signal changes were almost saturated at ∼570 s. The relationships between SA concentrations and SPR signal intensities, with four different anti-SA concentrations, are described in Figure 4. The anti-SA spots with more than 0.2 mg/mL antibody showed approximately the same signal intensities. In the meantime, the inside of anti-SA spots with 1.0 mg/mL antibody showed a lack of uniformity (data not shown). Therefore, we think that the optimum range of antibody concentration for spotting was 0.20.5 mg/mL. The detection limit was ∼30 ng/mL in this method, because no significant differences were observed between 0 and 10 ng/mL signals. The detection limit of the Biacore immunoassay was reported to be 80 ng/mL IgG in bovine milk29 and, thus, was at the same level as that of our system. In our previous report, the detection limit in another array using SPR with diffraction grating was ∼5 µg/mL28 and might be 100-fold higher than that of our novel system, despite differences in the SPR instruments, the surface chemistries, and the model proteins between these two systems. However, we should note that the detection limit of our system was much higher than that of the sandwich immunoassay-based antibody array (2.0-700 pg/mL), in which biotinylated antibodies and fluorescent-labeled SA were used to detect traces of cytokines and chemokines.20 Therefore, these results and those of previous studies indicate that our system could be applied to detect at least moderately expressed proteins (>30 ng/ mL). Two arrays were fabricated via scheme C by spotting 0.2 mg/ mL antibody solutionssi.e., six anti-mKIAA antibodies, plus antiactin, anti-tubulin, and anti-SA. The average and standard deviation (SD) of four spots for each antibody was calculated. The correlation of SPR signals between the two arrays was shown in Figure 5. SDs were indicated as error bars. The data of “array 1” in Figure 5 corresponded to those in Figure 3c. The obtained SPR signals were different between these two array experiments, but there were distinct correlations (R2 ) 0.89). This result implies that the positive and negative control spots are necessary for the prospective quantitative analysis. 7120 Analytical Chemistry, Vol. 77, No. 22, November 15, 2005

Figure 5. Comparison of obtained SPR signals between the two arrays. The signal data and error bars were calculated from four spots on each array. Table 1. Comparison between SPR Signals and Western Blotting Results Western blotting antibody

SPR signal

Mw

mKIAA0202 mKIAA0531 mKIAA0656 mKIAA0675 mKIAA0769 mKIAA0988 actin tubulin SA

2.1 6.6 7.9 1.5 9.2 2.1 8.0 9.5 2.3

57, 53 120 180, 52 140, 160 87 88, 52 42 50 18

a

splicing dominant exposure variants expression time (min) 2 1 2 2 1 2 nda nd nd

brain brain brain brain spleen heart nd nd nd

3.0 1.0 0.5 3.0 0.5 3.0 nd nd

nd, not done.

Comparison between SPR Signals and Western Blotting. The average SPR signals of two arrays obtained in the experiment depicted in Figure 5 are compared with the results for Western blotting in Table 1. The Western blottings performed with antimKIAA antibodies for the six mKIAA proteins are shown in Figure 6. Among them, three proteins (mKIAA0531, mKIAA0656, mKIAA0769) were thought to be relatively abundant based on the strong intensities of their bands and shorter exposure times on Western blotting. Indeed, endogenous proteins from adult mouse tissues were immunoprecipitated by using the anti-mKIAA antibodies, and the expressions were confirmed on 1D or 2D SDSPAGE (data not shown). These proteins also showed high signals in our SPR system. On the other hand, three other proteins (mKIAA0202, mKIAA0675, mKIAA0988) showed extremely low SPR signals, even though these proteins revealed specific bands but required longer exposure time on Western blotting. The results of Western blotting depend on both the sensitivity and specificity of antibodies against SDS-denatured proteins, while SPR signals depend not only on the sensitivity and specificity against native proteins in solution but also on their molecular weights. In addition, multiple splicing variants are additive for SPR signals. Although we could not explain the underlying cause of the low SPR signals, the concentration of these undetected proteins might be close to the detection limit of our SPR system. It is quite difficult to clarify the exact concentrations of endogenous proteins in crude cell lysate, because the calibration curves for each protein are

Figure 6. Results of Western blot analysis using the samples extracted from adult mouse organs. Each panel shows the results for a different antibody (a, mKIAA0202; b, mKIAA0531; c, mKIAA0656; d, mKIAA0675; e, mKIAA0769; f, mKIAA0988).

not always available. Nevertheless, it is expected that profiling and quantification of the relatively abundant proteins in crude cell lysate by the SPR imaging technique will be achieved soon. CONCLUSIONS In this paper, we established a method for label-free detection of proteins in crude cell lysate by means of an antibody array and SPR imaging. The cell lysate was applied to the array surface without dilution in order to maintain the sensitivity of the assay. The refractivity of the running buffer was adjusted with that of the cell lysate to overcome the bulk effect in the measurement procedure. This made it possible to monitor subtle signal changes during lysate injection. We also investigated the surface chemistries of the fabricated arrays to reduce nonspecific adsorption. The nonspecific adsorption seemed to be contributed by hydrophobic and electrostatic binding. The carboxyl group-terminated alkanethiol with a PEG moiety was used on a gold surface to provide hydrophilicity. The carboxyl compound was blended with the hydroxyl-terminated PEG alkanethiol to reduce the surface charge. The modified surface chemistry showed negligible nonspecific adsorption. (41) Belov, L.; de la Vega, O.; dos Remedios, C. G.; Mulligan, S. P.; Christopherson, R. I. Cancer Res. 2001, 61, 4483-4489. (42) Belov, L.; Huang, P.; Barber, N.; Mulligan, S. P.; Christopherson, R. I. Proteomics 2003, 3, 2147-2154. (43) Ko, I. K.; Kato, K.; Iwata, H. Biomaterials 2005, 26, 687-696.

The detection limit of the antibody array was ∼30 ng/mL in cell lysate and was inferior to the sandwich immunoassay-based antibody array on a glass substrate and the Western blotting. However, the SPR imaging technique enabled quick analysis within 30 min and label-free detection of multiple proteins in 200 µL of cell lysate. Therefore, an antibody array with SPR imaging could be an alternative method for analyzing the expression of at least moderately expressed proteins (>30 ng/mL). Also, this study implied the possibility of quantitative analysis of protein expression. We think our SPR antibody array could be applied for direct protein expression profiling, cell phenotyping,41-43 food analysis,29 discovery of new biomarkers,9,14 and immunological disease diagnostics. ACKNOWLEDGMENT This work was supported by the New Energy and Industrial Technology Development Organization (NEDO). K.U.-A. and H.K. are grateful for a grant from the CREATE program (Collaboration of Regional Entities for Advancement of Technological Excellence) of the Japan Science and Technology Corporation. We also thank Mr. Kazuki Inamori and Ms. Terue Kamiya (Toyobo Co. Ltd.) for their critical input. Received for review May 20, 2005. Accepted August 1, 2005. AC050884A

Analytical Chemistry, Vol. 77, No. 22, November 15, 2005

7121