DOI:
10.1039/B612492E
(Method)
Mol. BioSyst., 2007,
3, 36-42
A novel antibody microarray format using non-covalent antibody immobilization with chemiluminescent detection
Received
30th August 2006
, Accepted 31st October 2006
First published on 28th November 2006
Abstract
To date, protein and antibody microarrays have been used in reverse-phase and sandwich-based methods in order to detect known proteins such as biomarkers in samples. Our group developed “libraries” of antibodies against unknown proteins, referred to as mKIAA proteins, and we attempted to discover candidate novel biomarkers by protein expression profiling.
To profile mKIAA protein expression using these antibodies, we established an antibody microarray system using chemiluminescent detection. A number of techniques for protein–antibody microarrays have been reported; however, no entirely suitable protocol for crude protein samples has been established. To address this issue, we immobilized purified antibodies on hydrophilic surface polymer slides (Maxisorp™, Nunc). Although our system is based on the direct labeling of crude protein samples, we achieved sufficient sensitivity (detection limit: 50 pg mL–1) and low backgrounds. This sensitivity is on a level with the sandwich immunoassay -based antibody array system. Using our protocol, we developed an antibody microarray spotted with 960 anti-mKIAA antibodies (total: 3888 spots for quadruplicate assessments), and we carried out protein expression profiling of mKIAA proteins. In this study, we generated an expression profile of 960 mKIAA proteins and compared the present results with those obtained viacDNA microarray.
Introduction
Antibody microarrays hold great promise as a valuable tool for both basic and applied biological research. The continued development of this technology is providing methods that are more practical for routine applications. Several aspects of antibody microarray technology render it well suited for the discovery of disease markers. Antibody microarrays make use of a solid-phase technology that can be used to screen the expression of multiple proteins concurrently.1 Several antibody-based techniques have been developed and used to profile protein expression.2–9 The present platform for antibody microarrays utilizes technologies developed for either DNA microarrays2,6 or sandwich immunoassay techniques.10,11 The use of two differently labeled extracts is similar to conventional DNA microarray analysis and allows for pair-wise comparisons. This type of dual-label system has been described in the literature and is now commercially available.2,6 The micro-sandwich immunoassay technique, another representative approach to antibody microarrays, requires two antibodies to capture and detect a target protein, but its sensitivity and specificity are superior to those of other techniques.11
We previously established a surface plasmon resonance (SPR)-based antibody microarray system.12–14 For the SPR biosensors, target molecules are immobilized on a gold-coated chip, unlabeled biological samples are loaded, and the change in the angle of reflected light is measured.15–18 SPR biosensors exhibit insufficient sensitivity, especially when compared with that obtained using the sandwich immunoassay technique; as a result, SPR biosensors require more protein for the loading of samples.12,13 To overcome these defects, we developed a non-covalently immobilized antibody microarray system with chemiluminescent detection.19,20 Here, we demonstrate the effectiveness of our antibody microarray system for protein-expression profiling. By using this system, we developed 3888 spot antibody microarrays, and we present the results of the protein-expression profiling of 960 mKIAA proteins in five mouse tissues: brain, thymus, kidney, testis, and spleen.
Experimental
Dot-blot analysis
Microarray support materials were gratefully obtained from commercial suppliers and are listed in Table 1. The exact data on the membranes are not shown in Fig. 1, because all positive signals could be distinguished from the background signal. Antibodies and bovine serum albumin (BSA, 100 µg mL–1) were spotted at 0.5 µL per slide. Anti-tubulin and anti-glutathione-S-transferase (GST) monoclonal antibodies were purchased from Chemicon International (Temecula, CA, USA). Anti-maltose binding protein (MBP) antibody was purchased from New England Biolabs (Beverly, MA, USA). Anti-mKIAA antibodies were purified as follows. BSA was purchased from Jackson ImmunoResearch Laboratories (Baltimore, MD, USA). After the slides were air-dried, they were incubated in 10 µg mL–1biotin-labeled mouse brain extract mixed with 10 ng mL–1 recombinant MBP proteins. After being washed three times with PBS-T (0.05% Tween20 in phosphate-buffered saline (PBS, pH 7.4)), the slides were incubated in horseradish peroxidase (HRP)-labeled anti-biotin antibodies (1 : 2000, Roche, Penzberg, Germany). After being washed three times with PBS-T, chemiluminescent reagents (Femtoglow Plus, Michigan Diagnostics, Troy, MI, USA) were added to the slides, and the signal was detected by a modified LAS3000 system (Fujifilm, Tokyo, Japan).
![Comparison of various microarray support materials. Panels A to D show the dot-blot analysis. The following antibodies were used for the spot analyses: anti-tubulin (1), anti-MBP (2), anti-GST (3), anti-mKIAA0202 (4), anti-mKIAA1091 (5), anti-mKIAA0641 (6), and BSA (7). The numbers on the slides indicate the position of the antibodies used for spotting. The support materials were aldehyde-coated (A, Matsunami, Tokyo, Japan), poly-l-lysine-coated (B, Matsunami), epoxysilane-coated (C, Nunc), and polymer (D, Maxisorp™, Nunc) slides. The slides were incubated in 10 µg mL–1 mouse brain extract mixed with 10 ng mL–1 MBP recombinant proteins. The signals were detected by chemiluminescence.](/image/article/2007/MB/b612492e/b612492e-f1.gif) |
| Fig. 1 Comparison of various microarray support materials. Panels A to D show the dot-blot analysis. The following antibodies were used for the spot analyses: anti-tubulin (1), anti-MBP (2), anti-GST (3), anti-mKIAA0202 (4), anti-mKIAA1091 (5), anti-mKIAA0641 (6), and BSA (7). The numbers on the slides indicate the position of the antibodies used for spotting. The support materials were aldehyde-coated (A, Matsunami, Tokyo, Japan), poly-L-lysine-coated (B, Matsunami), epoxysilane-coated (C, Nunc), and polymer (D, Maxisorp™, Nunc) slides. The slides were incubated in 10 µg mL–1 mouse brain extract mixed with 10 ng mL–1 MBP recombinant proteins. The signals were detected by chemiluminescence . | |
Table 1 Various products used and respective manufacturers
Product |
Material |
Binding |
Supplier |
Membrane
|
Immobilon-P |
PVDF |
Hydrophobic |
Millipore |
PALL BioTraceTM NT |
Nitrocellulose |
Hydrophobic |
Nippon-Genetics |
Glass
|
DNA microarray slide for amino-modified oligo DNA |
Aldehyde-coated |
Covalent |
Matsunami |
DNA microarray slide TYPE1 |
Poly-L-lysine-coated |
Ionic |
Matsunami |
Microarray slide |
Epoxysilane-coated |
Covalent |
Nunc |
Polymer
|
Maxisorp™ |
Polymer |
Hydrophilic |
Nunc |
Antibody preparation
Anti-mKIAA antibodies were raised by the subcutaneous immunization of rabbits with GST-fused mKIAA protein fragments.21 The antibodies were purified on a Protein A sepharose column (GE, Piscataway, NJ, USA). After Protein A sepharose column purification was carried out, we performed further purifications in order to exclude the IgG fraction with affinity for GST and/or intrinsic proteins derived from Escherichia coli with AN Empore™ affinity azlactone plate (3 M, St. Paul, MN, USA), as recommended by the manufacturer (Fig. 2A). The anti-MBP and anti-GST antibodies used were those mentioned above. An anti-tubulin antibody of polyclonal antiserum was purchased from Chemicon International. The antisera were purified on an Affi-Gel protein A MAPS II Kit (Bio-Rad Laboratories, Hercules, CA, USA). Biotin-labeled anti-rabbit IgG was purchased from Jackson ImmunoResearch Laboratories. Anti-tubulin antibodies (monoclonal, polyclonal) and biotin-conjugated rabbit IgG were used as a positive control. As a negative control, BSA was spotted onto the membranes or slides at the same concentration as that of the antibodies. The concentrations of these antibodies and BSA were adjusted to 10 µg mL–1 ∼ 1 mg mL–1 with PBS.
![High-throughput antibody purification. Schematic diagram of antibody preparations, A. Target IgG fractions were collected by two-step purification. Panels B and C represent Western blot analyses to assess the purity of anti-mKIAA0445. In panel B, an E. coli lysate that expressed empty GST fusion protein was loaded, and then the bands were detected with each protein A-purified antibody and two-step-purified antibody. In the case of two-step purification, the bands detecting GST or other protein impurities disappeared. In panel C, the input protein was E. coli lysate that expressed recombinant MBP-fusion mKIAA0445 protein. By two-step purification, the extra bands were eliminated as indicated by the arrowheads.](/image/article/2007/MB/b612492e/b612492e-f2.gif) |
| Fig. 2 High-throughput antibody purification. Schematic diagram of antibody preparations, A. Target IgG fractions were collected by two-step purification. Panels B and C represent Western blot analyses to assess the purity of anti-mKIAA0445. In panel B, an E. coli lysate that expressed empty GST fusion protein was loaded, and then the bands were detected with each protein A-purified antibody and two-step-purified antibody. In the case of two-step purification, the bands detecting GST or other protein impurities disappeared. In panel C, the input protein was E. coli lysate that expressed recombinant MBP-fusion mKIAA0445 protein. By two-step purification, the extra bands were eliminated as indicated by the arrowheads. | |
Sample preparation: Mouse tissue extract
Adult mice (ICR strain, 8 weeks) were sacrificed, and the tissue was homogenized with CelLytic M (Sigma, St. Louis, MO, USA) containing 0.5% Protease Inhibitor Cocktail (Sigma). The protein content of the lysates was determined by the Bradford protein assay;22 subsequently, the lysates were dialyzed with coupling buffer (0.2 M Na2HCO3, 0.5 M NaCl, pH 8.3) at 4 °C overnight. The lysates were labeled with N-hydroxy-succinimide ester (NHS)-biotin-OSu (Dojindo, Kumamoto, Japan) as recommended by the manufacturer. Streptavidin (Wako, Osaka, Japan) was labeled with NHS near-infrared dye (IR-dye, 800CW; LI-COR, Lincoln, NE, USA) as recommended by the manufacturer. Unconjugated dye was removed by dialysis with PBS 4 °C overnight. The protein content of IR-dye-labeled streptavidin was adjusted to 1 mg mL–1 with PBS. For long-term storage, we added glycerol in the same volume as the lysate (final volume: 50% v/v) and 0.5% Protease Inhibitor Cocktail (Sigma) to the samples, which were stored at –20 °C.
Sample preparation: Production of GST and MBP proteins
Recombinant GST protein was expressed according to a previously described method.23 To express MBP protein in E. coli, we used the modified pMAL-p2X vector (New England Biolabs).23 Recombinant MBP protein was also expressed according to a previously described method,23 except that 25 µg mL–1 of kanamicin was used instead of ampicillin. The cell pellets were re-suspended in column buffer (20 mM Tris–HCl, 200 mM NaCl, 1 mM EDTA, pH 7.4) and lysed with a probe sonicator (four times for 30 s each, 30 s intervals) in an icewater bath. The samples were then centrifuged at 9000 g for 10 min at 4 °C. The supernatants were loaded onto a 0.5 mL amylose resin column (New England Biolabs) and washed with 10 column volumes of column buffer. The fusion protein was eluted with column buffer plus 20 mM of maltose, and then the samples were dialyzed with PBS.
Construction of the antibody microarray
MicroGrid II (Genomic Solutions, Ann Arbor, MI, USA) was used to construct the microarrays. For the glass and polymer slides, the samples were spotted with 1 nL of each antibody solution. Then the microarray was incubated at 4 °C overnight in a humidity-saturated chamber, prepared by the addition of as much solid NaCl to water as needed to form a 1 cm deep slurry in the bottom of a plastic container with an airtight lid. This forms a chamber with a relative humidity of approximately 75%. For the membranes, the spotted microarrays were air-dried at room temperature. The microarrays were stored in a desiccated chamber at 4 °C until use for the following microarray experiments.
Microarray experiments and statistical analysis
The spotted slides and membranes were blocked in 2% (w/v) IgG-free, protease-free grade BSA (Jackson ImmunoResearch Laboratories) solution at room temperature (rt) for 2 h. The slides were incubated with biotin-labeled mouse tissue extracts at the appropriate concentration for 2 h, and were then washed in PBS-T for 10 min each. For fluorescence detection, the slides were incubated with streptavidin-conjugated Cy3 (1 : 400, GE) for 1 h at rt in 2% BSA, and the signals were detected by FLA8000 (Fujifilm). For IR dye detection, the slides were incubated with IR dye-labeled streptavidin (1 : 400) for 1 h at rt in 2% BSA. After the samples were washed three times with PBS-T, the signals were detected with FLA8000 (Fujifilm). For chemiluminescent detection, the slides were subsequently incubated with HRP-labeled anti-biotin antibody (1 : 2000, Roche) for 1 h at rt in 2% BSA. After the samples were washed three times with PBS-T, chemiluminescent reagents (Femtoglow Plus) were added, and the signals were detected by modified LAS3000 (Fujifilm). To quantify the results, the data analysis was performed using ArrayGauge software (Fujifilm). The strength of the linear relationship between replicate experiments was calculated using Pearson's correlation coefficient.
Results
Optimization of microarray products
Various materials are available for microarray slides on which surface chemical processing is carried out. Here, we investigated the optimal products in such a system for the assessment of protein expression within a crude protein sample. We first evaluated the appropriate support materials for antibody microarrays using crude samples. The microarray slides examined here are listed in Table 1. We then carried out a dot-blot analysis (Fig. 1). A comparison of these microarray products revealed that the Maxisorp™ polymer microarray slides (Nunc, Roskilde, Denmark) immobilized the most antibodies and gave the highest signal-to-noise ratio (Fig. 1). We decided to use Maxisorp™ polymer microarray slides for further experiments.
Antibody preparation and antibody microarray development
After the anti-mKIAA rabbit polyclonal antibodies were treated with protein A purification, the IgG fraction still contained anti-GST or antibody against impurities of E. coli bacterial origin (e.g., heat-shock proteins). To exclude such fractions, we performed further high-throughput purification by treating the column with immobilized E. coli lysate expressing empty recombinant GST proteins (Fig. 2A). After purification, extra antibodies against GST protein and other impurities were excluded (Fig. 2B, C). Purified antibodies were spotted onto a Maxisorp™ slide as indicated in Fig. 3A and 4A. The titers of all anti-mKIAA antibodies used here were determined by ELISA, and the respective specificities of these antibodies were checked by several immunological techniques, including Western blot analysis and immunohistochemistry. Some of these findings are freely available through our InGaP (Integrative Gene and Protein-expression database: https://webcreate.kazusa.or.jp/create/)24 a comprehensive database of gene/protein expression profiles of mouse KIAA. Anti-tubulin and biotin-labeled antibodies were used as positive controls, and BSA was used as a negative control (Fig. 3A).
![Antibody microarray layout and detection methods. A: The spatial arrangements of six antibodies (anti-tubulin (monoclonal, 1), anti-tubulin (polyclonal, 2), anti-MBP (3), anti-GST (4), anti-mKIAA0202 (5), anti-mKIAA1239 (6), BSA (7) (used as a negative control), and biotin-labeled anti-rabbit IgG (used as a positive control), spotted onto slides in quadruplicate with total concentrations ranging from 67 to 6700 amol IgG per spot (10 µg mL–1–1 mg mL–1; for BSA, from 150 amol to 15 fmol per spot). The antibodies were spotted at a volume of 1 nL per spot using a MicroGrid II arrayer. B–D: Antibody microarrays incubated with 100 pg mL–1 MBP proteins in 10 µg mL–1 mouse brain extract. The mouse brain extract was labeled with biotin. The microarrays were visualized with streptavidin–Cy3 (B), IR-dye labeled streptavidin (C), and HRP-labeled anti-biotin by chemiluminescent detection (D). In panels A and B, the signals were detected by FLA8000. Panel C was visualized by LAS3000.](/image/article/2007/MB/b612492e/b612492e-f3.gif) |
| Fig. 3 Antibody microarray layout and detection methods. A: The spatial arrangements of six antibodies (anti-tubulin (monoclonal, 1), anti-tubulin (polyclonal, 2), anti-MBP (3), anti-GST (4), anti-mKIAA0202 (5), anti-mKIAA1239 (6), BSA (7) (used as a negative control), and biotin-labeled anti-rabbit IgG (used as a positive control), spotted onto slides in quadruplicate with total concentrations ranging from 67 to 6700 amol IgG per spot (10 µg mL–1–1 mg mL–1; for BSA, from 150 amol to 15 fmol per spot). The antibodies were spotted at a volume of 1 nL per spot using a MicroGrid II arrayer. B–D: Antibody microarrays incubated with 100 pg mL–1 MBP proteins in 10 µg mL–1 mouse brain extract. The mouse brain extract was labeled with biotin. The microarrays were visualized with streptavidin–Cy3 (B), IR-dye labeled streptavidin (C), and HRP-labeled anti-biotin by chemiluminescent detection (D). In panels A and B, the signals were detected by FLA8000. Panel C was visualized by LAS3000. | |
![Development of a 3888 spot antibody microarray system. Panel A shows the fabrication of a 3888 spot antibody microarray. Spotted antibodies were visualized using biotin-labeled mouse thymus extract and HRP-labeled anti-biotin with chemiluminescent detection. B: Reproducibility of the signal detection of 960 antibodies. The mean signal intensities for replicated experiments are plotted against each other. Each microarray was incubated with 5 µg mL–1 of mouse thymus extract. Two microarrays were examined independently. C: Representative protein expression profiles obtained using the present antibody microarray system. We prepared samples derived from five mouse tissues, and the samples were analyzed using our antibody microarray system. Reference-corrected data were ranked based on their signal intensities. The graph shows the number of highest (1st) ranked proteins in each tissue calculated and plotted.](/image/article/2007/MB/b612492e/b612492e-f4.gif) |
| Fig. 4 Development of a 3888 spot antibody microarray system. Panel A shows the fabrication of a 3888 spot antibody microarray. Spotted antibodies were visualized using biotin-labeled mouse thymus extract and HRP-labeled anti-biotin with chemiluminescent detection. B: Reproducibility of the signal detection of 960 antibodies. The mean signal intensities for replicated experiments are plotted against each other. Each microarray was incubated with 5 µg mL–1 of mouse thymus extract. Two microarrays were examined independently. C: Representative protein expression profiles obtained using the present antibody microarray system. We prepared samples derived from five mouse tissues, and the samples were analyzed using our antibody microarray system. Reference-corrected data were ranked based on their signal intensities. The graph shows the number of highest (1st) ranked proteins in each tissue calculated and plotted. | |
Detection methods
We next assessed the detection methods. According to the development of DNA microarrays or proteomics (e.g. 2-D electrophoresis), the tools for fluorescent detection have been developed and are utilized well. Although protein microarrays and ELISA-based (sandwich) antibody microarrays are now utilized and have been found to achieve high sensitivity and acceptable backgrounds, low sensitivity and high backgrounds remain major difficulties in the direct labeling method of crude protein samples.14 In particular, as regards the direct labeling of crude protein samples such as tissue extracts, the non-specific binding of various proteins onto support materials is difficult to avoid. Here, we undertook a comparison of several detection methods. Fig. 3 shows the signal intensities of Cy3 (Fig. 3B), of an IR dye that was expected to reduce the backgrounds (Fig. 3C), and of chemiluminescent detection (Fig. 3D). These experiments revealed that the signal intensities from Cy3 could not be detected due to low sensitivity and high backgrounds (Fig. 3B). In the case of the IR dye, although the background was rather low, the signals were insufficient for the direct-labeling method (Fig. 3C). On the other hand, we obtained sufficient signal intensities and acceptable backgrounds by chemiluminescent detection (Fig. 3D). For subsequent analysis, we chose chemiluminescent detection.
Sensitivity of the antibody microarray system with chemiluminescent detection
To approximate sensitivity and dynamic range using chemiluminescent detection systems, we spotted anti-GST antibodies onto slides using the amounts of 67, 670, and 6700 amol per spot (Fig. 5A). We then analyzed the samples containing recombinant GST protein under eight different conditions (0 pg mL–1–100 ng mL–1). The signal intensity was found to vary in accord with both the amount of spotted antibody and the amount of GST protein added. The signal intensity plot of GST protein revealed a dynamic range of over three orders of magnitude (50 pg mL–1–100 ng mL–1) at 67–6700 amol per spot of spotted antibody (Fig. 5A). That is, 50 pg mL–1antigen (1.9 fmol mL–1GST protein) in a crude sample could be detected by chemiluminescent detection, whereby 670 amol per spot antibody was used for immobilization.
![Validation of antibody microarrays with chemiluminescent detection. A: The amount of labeled antigen required for chemiluminescent detection. Anti-GST antibody (67–6700 amol per spot) was spotted onto each slide. The microarrays were incubated with 5 µg mL–1 mouse thymus extract containing recombinant GST at different concentrations (0–100 000 pg mL–1). Graphs show the mean signal intensities of quadruplicate data spots. The relative signal intensities indicate the mean signal intensities divided by the mean of the positive controls. Fifty pg mL–1 GST (1.9 fmol mL–1) was sufficient for detection in a crude sample. B: Graph showing the mean of quadruplicate signal intensities in anti-tubulin (open circles), anti-GST (open squares), and BSA (closed triangles) versus the amount of spotted antibodies per spot. The error bars indicate S.D. Signal intensities were quantified from microarrays using ArrayGauge software. These signals were detected in 50 pg mL–1 GST proteins mixed with 5 µg mL–1 mouse thymus extract. The signals were detectable with 50 pg mL–1 (1.9 fmol mL–1) antigen at 670 amol per spot immobilized antibodies. Using anti-tubulin, the chemiluminescent antibody microarray system was able to detect endogenous proteins. The antigen was detected with the use of as little as 100 µg mL–1 of spotted antibodies (670 amol per spot). C: Reproducibility of signal detection. The mean signal intensities for replicated experiments are plotted against each other. The fabrication of spotted antibodies is as shown in Fig. 3A (anti-tubulin (monoclonal), anti-tubulin (polyclonal), anti-MBP, anti-GST, anti-mKIAA0202, anti-mKIAA1239, and BSA as a negative control). The plot shows a high correlation between two independent experiments (R2 = 0.96). The error bars indicate the S.D. of quadruplicate data spots. Each microarray was incubated with 5 µg mL–1 of mouse thymus extract containing recombinant GST protein at different concentrations (0–100 000 pg mL–1). Two microarrays were examined independently.](/image/article/2007/MB/b612492e/b612492e-f5.gif) |
| Fig. 5 Validation of antibody microarrays with chemiluminescent detection. A: The amount of labeled antigen required for chemiluminescent detection. Anti-GST antibody (67–6700 amol per spot) was spotted onto each slide. The microarrays were incubated with 5 µg mL–1 mouse thymus extract containing recombinant GST at different concentrations (0–100 000 pg mL–1). Graphs show the mean signal intensities of quadruplicate data spots. The relative signal intensities indicate the mean signal intensities divided by the mean of the positive controls. Fifty pg mL–1 GST (1.9 fmol mL–1) was sufficient for detection in a crude sample. B: Graph showing the mean of quadruplicate signal intensities in anti-tubulin (open circles), anti-GST (open squares), and BSA (closed triangles) versus the amount of spotted antibodies per spot. The error bars indicate S.D. Signal intensities were quantified from microarrays using ArrayGauge software. These signals were detected in 50 pg mL–1 GST proteins mixed with 5 µg mL–1 mouse thymus extract. The signals were detectable with 50 pg mL–1 (1.9 fmol mL–1) antigen at 670 amol per spot immobilized antibodies. Using anti-tubulin, the chemiluminescent antibody microarray system was able to detect endogenous proteins. The antigen was detected with the use of as little as 100 µg mL–1 of spotted antibodies (670 amol per spot). C: Reproducibility of signal detection. The mean signal intensities for replicated experiments are plotted against each other. The fabrication of spotted antibodies is as shown in Fig. 3A (anti-tubulin (monoclonal), anti-tubulin (polyclonal), anti-MBP, anti-GST, anti-mKIAA0202, anti-mKIAA1239, and BSA as a negative control). The plot shows a high correlation between two independent experiments (R2 = 0.96). The error bars indicate the S.D. of quadruplicate data spots. Each microarray was incubated with 5 µg mL–1 of mouse thymus extract containing recombinant GST protein at different concentrations (0–100 000 pg mL–1). Two microarrays were examined independently. | |
The signal intensities of 50 pg mL–1 GST protein in 5 µg mL–1 thymus extract are shown in Fig. 5B. At least 670 amol per spot antibody was detected. Furthermore, to confirm the detection of endogenous proteins, the signal intensity of anti-tubulin in 5 µg mL–1 thymus extract is shown in Fig. 5B. Anti-tubulin was used to detect endogenous tubulin proteins immobilized on the slides at 670 amol per spot. Based on these data, we used 670 amol per spot antibody for subsequent experiments (Fig. 5A,B).
To assess the reproducibility of this antibody microarray system, we performed a regression analysis using the results of the independent experiments. Fig. 5C shows a scatter plot of the signal intensities of the corresponding targets on differently spotted slides. The results using thymus samples showed high reproducibility between replicate experiments, with a correlation coefficient of 0.98 (R2 = 0.96), thus demonstrating the high reproducibility of this system.
Development of 3888 spot antibody microarrays
Purified 960 anti-mKIAA antibodies and 12 control spots were spotted in quadruplicate onto Maxisorp™ microarray slides (total: 3888 spots, Fig. 4A). We examined the protein expression profiling of mKIAA proteins in five mouse tissues: whole brain, thymus, kidney, testis, and spleen (Fig. 4C). Fig. 4B shows the reproducibility of 960 antibodies, and the values indicate the means of quadruplicate spots. The targets of our antibody ‘libraries’ were mKIAA proteins, and their cDNAs were isolated mainly from cDNA libraries derived from brain tissue.21,23 We had already verified that nearly half of mKIAA genes are predominantly expressed in brain tissue, as determined by cDNA microarray analysis.23 We therefore used our antibody microarray system to examine the tissue-specific expression pattern of mKIAA at the protein level. We prepared samples derived from five mouse tissues and confirmed the predominant expression, as well as the transcriptional level, of mKIAA protein. Reference-corrected data were ranked based on their signal intensities, and then the number of highest (1st) ranked proteins from each organ were calculated. Approximately 42% of the mKIAA proteins in the brain had first-place rankings (Fig. 4C).
Discussion
In this study we assessed various support materials and concluded that Maxisorp™ polymer slides were the most suitable type of slide for the present system. Maxisorp™ polymer slides were intended primarily for the immobilization of biomolecules by electro-forces and/or ionic bonds. The amount of immobilized antibodies on Maxisorp™ slides exceeded the amounts on glass slides treated with surface chemistries such as aldehyde, poly-L-lysine, epoxy, or other agents. The direct labeling of crude samples is known to be associated with high backgrounds, which can be a serious problem, whereas the signal-to-noise ratio is acceptable with the use of Maxisorp™ slides. Although PVDF and nitrocellulose membranes exhibit a high capacity for the immobilization of antibodies, non-specific hydrophobic adsorption also occurs. Therefore, these membranes are unsuitable for the direct labeling of crude protein samples. Angenendt et al. already examined various antibody microarray conditions using culture-cell extracts.25 In that report, they detected signals with 16 µg mL–1 culture-cell extracts by 1500 amol per spot immobilized antibodies on Maxisorp™ slides using the fluorescent direct-labeling method.25 Recently, Ahn et al. examined the protein expression profiling of angiogenin-induced human umbilical vein endothelial cells (HUVECs) using an antibody-arrayed ProteoChip (Proteogen, Seoul, Korea).7,26 They carried out their antibody microarray analysis with 1 mg mL–1 fluorescence-labeled culture-cell extracts by 100 µg mL–1 immobilized antibodies.7 It should be noted that antibodies spotted on Maxisorp™ slides were found to remain active for about two months under desiccated conditions at 4 °C (data not shown).
Using the present detection method, we aimed for sufficient sensitivity, similar to that of a sandwich immunoassay -based antibody microarray, as achieved by direct-fluorescence labeling. Although fluorescence labeling is a widely used direct labeling method in the field of proteomics, its signal sensitivity remained insufficient. Moreover, since we expected a reduction in backgrounds, we utilized IR dye.27,28 IR dye was shown to facilitate low background conditions, but the signal intensity was rather insufficient. We then achieved sufficient sensitivity and acceptable backgrounds using chemiluminescent detection with the direct labeling of biotin.
Because the chemiluminescent detection system is based on simple and well-developed procedures such as Western blot analysis, it is not severely affected by viscosity or other sample fluid and buffer conditions. However, the limitations of direct-labeling methods should be addressed in this context. It is important to pay attention to labeling process bias; in other words, it is difficult to label all proteins equally. In this study, we used NHS-activated labeling reagents; NHS substrate reacts rapidly and efficiently with ligands containing amino groups to give a very stable amide linkage. That is, the number of biotin labels per protein depends on the number of amino groups on that particular protein. Furthermore, the process of labeling a protein itself may denature, damage, or mask the epitope . Since several types of labeling reagents are now available, some of which have different active substrates that react with various groups (e.g., –SH, –COOH), it remains necessary to confirm the findings using different labeling reagents when verifying the present results. Moreover, the length of spacer molecules between a protein and biotin sometimes affects antibody–antigen binding. When the spacers are too short, labeled proteins sometimes sterically hinder biotin–anti-biotin binding. On the other hand, overly long spacers in hydrocarbons account for the insolubilization of proteins. In addition, several labeling reagents with various spacer lengths are now available, and therefore it is important to choose the appropriate reagents. We assessed several types of labeling reagents, and chose appropriate ones (data not shown).
It should be noted that with some mKIAA proteins, the results of the cDNA microarray analysis could not be reproduced (Fig. 4C). Protein expression levels depend not only on the levels of mRNA expression, but also on protein stability and/or degeneration; examining protein expression levels using an antibody microarray system will provide additional information about the functional aspects of mKIAA proteins. Furthermore, the solubility of proteins should be addressed in this context. Unlike that of DNA, the solubility of proteins is quite variable. It is impossible to solubilize all protein from tissues, and sometimes labeling a protein leads to insolubilization. Higher expression levels in the thymus may be affected by such conditions. It remains necessary to carefully consider the choice of detergents and labeling reagents during sample preparation.
Conclusions
We developed an antibody microarray system that employs non-covalent immobilization with chemiluminescent detection. We achieved sufficient sensitivity and acceptable backgrounds with this system. By using protein and/or antibody microarray techniques, it is possible not only to examine biomarkers but also to consider a number of important parameters, such as comparmentalization and post-translational modifications (e.g., proteolysis, phosphorylation, glycosylation, steric changes) that regulate protein function.29 It is these parameters, as well as their effects on protein function, that protein and antibody microarrays promise to address as part of the rapidly evolving field of proteomics.
Acknowledgements
The authors gratefully acknowledge all staff members at Kazusa DNA Research Institute. We also would like to thank Mr. T. Takahashi and Mr. S. Ikami (Fujifilm) for the modification of LAS3000. This study was supported by the CREATE Program from JST, a grant from Kazusa DNA Research Institute to H.K., and a Grant-in-Aid for Young Scientists (B) from MEXT KAKENHI (17710173) to K.U.
References
- M. F. Templin, D. Stoll, M. Schrenk, P. C. Traub, C. F. Vohringer and T. O. Joos, Trends Biotechnol., 2002, 20, 160–166 CrossRef CAS
.
- B. B. Haab, M. J. Dunham and P. O. Brown, Genome Biol., 2001, 2, research 0004.1–0004.13 Search PubMed
.
- R. P. Huang, J. Immunol. Methods, 2001, 255, 1–13 CrossRef CAS
.
- M. L. Lesaicherre, R. Y. Lue, G. Y. Chen, S. Q. Zhu and S. Q. Yao, J. Am.Chem. Soc., 2002, 124, 8768–8769 CrossRef CAS
.
- G. MacBeath and S. L. Schreiber, Science, 2000, 289, 1760–1763 CAS
.
- A. Sreekumar, M. K. Nyati, S. Varambally, T. R. Barrette, D. Ghosh, T. S. Lawrence and A. M. Chinnaiyan, Cancer Res., 2001, 61, 7585–7593 CAS
.
- E. H. Ahn, D. K. Kang, S. I. Chang, C. S. Kang, M. H. Han and I. C. Kang, Proteomics, 2006, 6, 1104–1109 CrossRef CAS
.
- H. Y. Cai, L. Lu, C. A. Muckle, J. F. Prescott and S. Chen, J. Clin. Microbiol., 2005, 43, 3427–3430 CrossRef CAS
.
- E. Kopf, D. Shnitzer and D. Zharhary, Proteomics, 2005, 5, 2412–2416 CrossRef CAS
.
- Y. Lin, R. Huang, N. Santanam, Y. Liu, S. Parthasarathy and R. P. Huang, Cancer Lett., 2002, 187, 17–24 CrossRef CAS
.
- C. C. Wang, R. P. Huang, M. Sommer, H. Lisoukov, R. Huang, Y. Lin, T. Miller and J. Burke, J. Proteome Res., 2002, 1, 337–343 CrossRef CAS
.
- K. Usui-Aoki, K. Shimada, M. Nagano, M. Kawai and H. Koga, Proteomics, 2005, 5, 2396–2401 CrossRef CAS
.
- M. Kyo, K. Usui-Aoki and H. Koga, Anal. Chem., 2005, 77, 7115–7121 CrossRef CAS
.
-
K. Usui-Aoki, M. Kyo, M. Kawai, M. Murakami, K. Imai, K. Shimada and H. Koga, in Frontiers in drug design & discovery, ed. G. W. Caldwell, Atta-ur-Rahman, M. R. D'Andrea and M. I. Choudhary, Bentham Science Publishers Ltd., 2006, vol. 2, pp. 23–33 Search PubMed
.
- Y. S. Day, C. L. Baird, R. L. Rich and D. G. Myszka, Protein Sci., 2002, 11, 1017–1025 CrossRef CAS
.
- Y. S. Day and D. G. Myszka, J. Pharm. Sci., 2003, 92, 333–343 CrossRef CAS
.
- D. G. Myszka and R. L. Rich, Pharm. Sci. Technol. Today., 2000, 3, 310–317 CrossRef CAS
.
- R. L. Rich, L. R. Hoth, K. F. Geoghegan, T.A. Brown, P. K. LeMotte, S. P. Simons, P. Hensley and D. G. Myszka, Proc. Natl. Acad. Sci. U. S. A., 2002, 99, 8562–8567 CrossRef CAS
.
- U. Isacsson and G. Watermark, Anal. Chim. Acta, 1974, 68, 339–362 CrossRef CAS
.
- L. W. Turtinen, D. N. Prall, L. A. Bremer, R. E. Nauss and S. C. Hartsel, Antimicrob. Agents Chemotherapy, 2004, 48, 396–403 Search PubMed
.
- N. Okazaki, R. Kikuno, R. Ohara, S. Inamoto, H. Aizawa, S. Yuasa, D. Nakajima, T. Nagase, O. Ohara and H. Koga, DNA Res., 2003, 10, 167–180 Search PubMed
.
- T. P. Whitehead, L. J. Kricka, T. J. Carter and G. H. Thorpe, Clin. Chem., 1979, 25, 1531–1546 CAS
.
- H. Koga, S. Yuasa, T. Nagase, K. Shimada, M. Nagano, K. Imai, R. Ohara, D. Nakajima, M. Murakami, M. Kawai, F. Miki, J. Magae, S. Inamoto, N. Okazaki and O. Ohara, DNA Res., 2004, 11, 293–304 Search PubMed
.
- M. Murakami, K. Shimada, M. Kawai and H. Koga, DNA Res., 2005, 12, 379–87 Search PubMed
.
- P. Angenendt, J. Glokler, D. Murphy, H. Lehrach and D. J. Cahill, Anal. Biochem., 2002, 309, 253–260 CrossRef CAS
.
- Y. Lee, E. K. Lee, Y. W. Cho, T. Matsui, I. C. Kang, T. S. Kim and M. H. Han, Proteomics, 2003, 3, 2289–2304 CrossRef CAS
.
-
R. Raghavachari, Near-infrared applications in biotechnology, 2001, Marcel Dekker, New York, p3 Search PubMed
.
- J. Sowell, L. Strekowski and G. Patonany, J. Biomed. Opt., 2002, 7, 571–575 CrossRef CAS
.
- R. B. Jones, A. Gordus, J. A. Krall and G. MacBeath, Nature, 2006, 439, 168–174 CrossRef CAS
.
|
This journal is © The Royal Society of Chemistry 2007 |