C. Y. Yu, M. L. Liu, I. C. Kuan, et al

Low-Density on Nitrocellulose Membrane for the Detection of Human Cytokines

Chi-Yang Yu, Mao-Lung Liu, I-Ching Kuan and Shiow-Ling Lee

Department of Bioengineering, Tatung University, Taipei, Taiwan, Republic of China

A low-cost, low-density protein microarray on nitrocellulose membrane was developed for the de- tection of human cytokines. The microarray was derived from chemiluminescent sandwich-type . Anti-human IL-2 and anti-human IL-4 polyclonal were arrayed and im- mobilized on the membrane. After the membrane was blocked and then incubation with human cytokines IL-2 and IL-4, biotinylated anti-human IL-2 and anti-human IL-4 monoclonal antibodies were added to form complexes with the cytokines and the immobilized polyclonal antibodies. Streptavidin-horseradish peroxidase conjugate was applied to detect the level of biotin. Finally, substrates for peroxidase were added to initiate chemiluminescence. The printed spots were uni- form and consistent in size. Under optimized conditions, the limits of detection for human IL-2 and IL-4 were 0.5 and 0.125 ng/ml, respectively. The linear range of the calibration curves spanned over two orders of magnitude. The application of 10 ng/ml streptavidin-peroxidase polymer was found to facilitate the analysis process. Key words: Chemiluminescence, cytokine, microarray, nitrocellulose

tions [5]. Human cytokines IL-2 and IL-4 were selected Introduction as model analytes to compare our system with others [6,7]. The microarray was based on chemiluminescent Protein microarray has become an important research sandwich-type immunoassay. Anti-human IL-2 and tool in . While the combination of two- anti-human IL-4 polyclonal antibodies were printed and dimensional and mass spectrometry immobilized on nitrocellulose membranes with a manual is very useful for discovery-oriented proteomics, protein microarrayer. The membrane was then blocked with bo- microarray is well suited for the study of a collection of vine serum albumin (BSA). After the membrane was related [1]. The most important advantage of incubated with human cytokines, biotinylated anti-human protein microarray is the ability to perform multiplexed IL-2 and anti-human IL-4 monoclonal antibodies were analysis. Other advantages include small sample vol- added to form sandwiched complexes with the cytokines umes, low reagent consumption, low limit of detection and the polyclonal antibodies immobilized on the mem- (LOD), and high sensitivity [2]. brane. The biotin label was detected by adding strepta- Although protein microarray is a powerful research vidin-horseradish peroxidase conjugate (SA-HRP). The tool, one major disadvantage is its high cost. The cost of peroxidase emitted chemiluminescence when its sub- a commercial system ranges from 25,000 to 220,000 strate was added. A digital camera originally designed USD [3]. Colorimetric detection has been applied to for fluorescence microscopy was converted to a simple protein microarray as a low-cost alternative [4]. How- detector for the chemiluminescence. ever, the LOD remains high compared to fluorescence or chemiluminescence. Materials and Methods We developed a low-cost, low-density protein mi- croarray system suitable for protein-detecting applica- Substrate Preparation and Printing

Received: June 1, 2006 Revised: July 31, 2006 Accepted: September 11, 2006 Address for correspondence: Chi-Yang Yu, 40 Chung-Shan N. Rd., Sec. 3, Taipei, Taiwan 104, R.O.C., Tel: 886-2-25925252-3315-27 Fax: 886-2-25854735 E-mail: [email protected]

J Biomed Lab Sci 2006 Vol 18 No 1~4 15

Protein microarray for cytokine detection

Protran® nitrocellulose membrane (Schleicher & Schuell, streptavidin-peroxidase polymer (Sigma, St. Louis, USA). Dassel, Germany) was cut into 2.2×6 cm. Two pieces of 2.2×0.3 cm double-sided tapes were attached to the 2.2 cm sides of the membrane. The membrane was then at- Results and Discussion tached to a standard glass slide using the tapes; the 2.2 cm side of the membrane was flushed with one of the Printing Quality of the Microarray short sides of the slide. The margins from the 6 cm sides of the membrane to the long sides of the slide were kept MicroCaster hand–held microarray system has a maxi- × equal. Pins of the MicroCaster hand–held microarray mal printing density of 768 spots (24 32; x-axis pitch= μ μ × system (Schleicher & Schuell, Dassel, Germany) were 750 m and y-axis pitch=1125 m) in a 1.8 3.6 cm area. coated with the supplier’s pin conditioner according to The capillary action of the porous membrane may their protocol. enlarge the printed spots. To examine this effect, we Enzymes, antibodies, and recombinant human cyto- printed 0.3 % methylene blue on the membrane. Maxi- kines were purchased from Pierce (Rockford, USA). mal printing density was obtained without overlapping PBS buffer was printed as negative control; 0.1 mg /ml spots. The spots were almost circular under microscope × of horseradish peroxidase and 0.5 μg /ml of biotinylated (4 10 magnification). Using ocular micrometer, the av- anti-IL-4 monoclonal were printed as positive erage diameter of the spots was 634.6±12.3 μm (CV= controls. For the detection of cytokines, 0.1 and 1 mg/ml 1.9 %, n=60). The arrayer has eight printing pins. We polyclonal antibody were used. used average diameter from each grid (spots printed by the same pin, n=8) to determine the variation in diameter Detection of Cytokines caused by individual printing pins. The CV between grids was only 0.5 %, suggesting the pins were almost The membrane was detached from the slide after print- identical in terms of printing size. ing and placed in a small zip-locked plastic bag. All in- The radial diffusion of the chemiluminescent reac- cubation steps were performed at room temperature on tion intermediates often results in a larger spot image an orbital shaker. The membrane was blocked in 1 ml of than the actual printed size. This effect may cause larger PBS containing 1 % (w/v) BSA for 1 h. After blocking, variation in spot diameter and decrease the practical ar- the membrane was incubated with 0.5 ml of cytokine raying density. We printed 25 μg/ml horseradish peroxi- standard (mixture of IL-2 and IL-4) prepared in the dase onto the membrane and initiated the chemilumi- blocking solution for 2 h. The membrane was then nescence reaction by incubating the membrane with Su- washed three times with 15 ml of PBS containing 0.1 % perSignal West Femto substrate for 5 min. The diameter (v/v) Tween 20 in a petri dish for 5 min. One milliliter of of the chemiluminescent spot determined in pixel num- 62.5 ng/ml biotinylated antibody (mixture of anti-human bers using Scion Image software (www.scioncorp.com) IL-2 and anti-human IL-4) in PBS was added and incu- was uniform (CV=3 %, n=60). Maximal printing density bated for 2 h, and then the membrane was washed as (768=24×32) could not be achieved due to spot overlap- previously described. After incubation with 1 ml of 10 ping. The highest printing density was decreased to 192 ng/ml SA-HRP in PBS for 1 h, the membrane was (12×16; x-axis pitch=1500 μm and y-axis pitch=2250 washed again, followed by incubation with 0.5 ml of μm). A 10×10 pixel area within the spot was selected to SuperSignal West Femto substrate (Pierce, Rockford, determine the uniformity of chemiluminescence. The CV USA) for 5 min. The membrane was covered in Saran of the intensity ranged from 1.5 to 5 % (n=60), suggest- wrap and placed in the detector. ing the printed spots were uniform. The simple chemiluminescence detector was based The complete system can be built for less than on a Penguin 600CLM digital camera (Pixera, Los Gatos, 10,000 USD. The cost of the printing substrate is under 1 USA) that is originally designed for fluorescence mi- USD, which is at least ten times lower than most com- croscopy imaging. The camera was first attached to a mercial products [3]. macro video zoom lens (Optem, Fairport, USA) and the assembly was mounted on a dark box. The exposure Optimization of Cytokine Detection time was set to 60 s (2 integrated frames, ISO=400). The images were analyzed using ScanAlyze software The specificity of the antibodies was examined by incu- (rana.lbl.gov/EisenSoftware.htm). For signal amplifica- bating the microarrays with 12.5 ng/ml IL-2 or IL-4 in tion, SA-HRP was replaced by various concentrations of PBS containing 1 % (w/v) BSA. No cross-reaction be-

16 J Biomed Lab Sci 2006 Vol 18 No1~4

C. Y. Yu, M. L. Liu, I. C. Kuan, et al tween the antibody pairs was observed when compared antibody pairs. The faint spots from 0.1 mg/ml poly- to the blank. clonal antibody also made the analysis difficult since the To optimize the detection, 0.1 or 1 mg/ml polyclonal ScanAlyze software requires the user visually identify antibody, 20, 62.5, or 200 ng/ml biotinylated monoclonal the locations and contours of the spots. We therefore antibody, 2 or 10 ng/ml SA-HRP were tested. Using 0.1 conclude that 1 mg/ml polyclonal antibody was optimal. mg/ml polyclonal antibody resulted in poor response to Similar result was also reported by others [7]. the cytokines (Fig. 1, grid 3 and 4). For IL-2, the chemi- Although the membrane was blocked in PBS con- luminescence increased slightly, compared to the blank taining 1 % (w/v) BSA for 1 h, the background increased when the highest concentration of cytokine standard was significantly as the concentration of the biotinylated an- applied. For IL-4, the spots could not be observed until tibody or SA-HRP increased (Fig. 2). This might result 0.5 ng/ml IL-4 was applied. The result suggests that 0.1 from the dynamic exchange between the adsorbed BSA mg/ml polyclonal antibody could be insufficient for de- molecules and the biotinylated antibodies (or SA-HRP) tection. This observation was more pronounced for IL-2, since protein adsorption is generally considered reversi- which may be due to the lower binding efficiency of the ble. Extended blocking period and/or different blocking

0 0.125 0.5

1.25 5 12.5 Cytokine (ng/ml)

Fig. 1. Chemiluminescence images of the protein microarrays at different cytokine concentrations. Num- bered white boxes indicate the locations of each grid. Grid 1: 0.1 mg horseradish peroxidase/ml; grid 2: PBS; grid 3: 0.1 mg anti-IL-2 polyclonal antibody/ml; gird 4: 0.1 mg anti-IL-4 polyclonal antibody/ml; grid 5: 1 mg anti-IL-2 polyclonal antibody/ml; gird 6: 1 mg anti-IL-4 polyclonal antibody/ml; grid 7 and 8: 0.5 μg biotinylated anti–IL-4 monoclonal antibody/ml. Concentrations for each grid were the printing concentra- tions. Biotinylated antibody and SA-HRP applied to the membranes were 62.5 and 10 ng/ml, respectively.

J Biomed Lab Sci 2006 Vol 18 No 1~4 17

Protein microarray for cytokine detection reagent may reduce the undesirable increase of the (calibration sensitivity) and R2 value of the calibration background. For 200 ng/ml biotinylated antibody and 10 curve were used as criterions. Among the six combina- ng/ml SA-HRP, the background almost reached the tions of biotinylated antibody and SA-HRP we tested, saturation level of the camera (around 65,000 arbitrary 62.5 ng/ml biotinylated antibody and 10 ng/ml SA-HRP unit for a 16-bit image). The high background precludes were optimal for the detection of both IL-2 and IL-4 (Fig. the application of such combination. 3). The linear range spanned over two orders of magni- To select the optimal reaction condition, the slope tude. The LOD under optimal condition was determined

7.E+04

6.E+04

5.E+04

4.E+04

3.E+04 Intensity (AU) 2.E+04

1.E+04

0.E+00 20 62.5 200 Biotinylated antibody (ng/ml)

Fig. 2. Effect of biotinylated antibody and SA-HRP on the background (□: 2 ng SA-HRP/ml; ▒: 10 ng SA-HRP/ml). The background was averaged from the negative control grid (PBS, n=6). AU=arbitrary unit.

Fig. 3. Calibration curves for IL-2 (▲) and IL-4 (□) under optimal reaction condition (62.5 ng biotinylated antibody/ml and 10 ng SA-HRP/ml). Cytokine standard (mixture) 0.125, 0.5, 1.25, 5, and 12.5 ng/ml were prepared in PBS containing 1 % (w/v) BSA. The fitted lines have slopes of 1826 (R2=0.92) and 4058 (R2=0.92) for IL-2 and IL-4, respectively. The printing concentration of the polyclonal antibody was 1 mg/ml.

18 J Biomed Lab Sci 2006 Vol 18 No1~4

C. Y. Yu, M. L. Liu, I. C. Kuan, et al by further diluting the standards to 0.005, 0.0125, 0.05, requirement. By immobilizing capture antibody on fluo- 0.125, and 0.5 ng/ml cytokine. The LOD (~ 95 % confi- rescence-coded beads, the analyte being measured on dence level) is defined as the concentration that produces individual bead can be identified and the corresponding a chemiluminescence signal greater than blank+3SD [8]. fluorescence signal from the antibody- sandwich The LOD for IL-2 and IL-4 were 0.5 and 0.125 ng/ml, can be recorded when the bead passes through a flow respectively. cytometer. Hill and Martin were able to analyze ten dif- Compared to a fluorescent protein microarray sys- ferent cytokines by MBAA using only 75 μl patient se- tem, which also utilized nitrocellulose as the printing rum [14]. A recent review has shown that similar quanti- surface, the LOD for IL-4 was similar [7]. However, the tative results can be obtained from both ELISA and LOD for IL-2 in our system was an order of magnitude MBAA if identical antibody pairs, similar diluents, and higher. The difference may be caused by the lower bind- blockers are used [15]. Although MBAA appears to out- ing affinity between the antigen and antibody pairs, the perform protein microarray, it has several disadvantages amount of antibody immobilized, or the nature of the including cumbersome immobilization procedure, lower detectors. The linear range of the fluorescence system multiplexity (up to one hundred analytes for current spanned over two to three orders of magnitude, which commercial assays), and the requirement of an expensive was similar to ours. flow cytometer. In addition to antibodies, have The LOD for both cytokines were similar to early been used as ligands to bind cytokines in various formats immunoassays [9,10]. However, the LOD of recent [16]. Bock et al. developed a fluorescent photoaptamer commercial ELISA can reach several picograms per mil- microarray for seventeen target proteins including sev- liliter when testing standards (see www.piercenet.com or eral cytokines [17]. Since photoaptamers bind to the www.idsltd.com). It has been shown that membrane pro- target analytes covalently, the microarray can withstand tein microarray coupled with radiographic film may lack vigorous washing to improve signal to noise ratio and the sufficient sensitivity to detect cytokine levels in com- limit of quantification. Photoaptamers can also be applied plex biological fluids when compared to ELISA [11]. to the chemiluminescent protein microarray described in One possible explanation for the high LOD ob- this manuscript to improve its performance. The im- served in microarray is the amount of capture antibody munoproteomics approach is perhaps the most recent used. For 1 mg/ml printing concentration, approximately development in cytokine detection [18]. Immunopro- 3 ng capture antibody was printed (calculated from the teomics uses a single reagent (proACTR) for immobili- delivery volume of 3 nl of the arraying pin). The amount zation of antibody and capture of antigen. The immobi- of the capture antibody used per microtiter plate well in lized antibody-antigen complex is then applied to the commercial ELISA is approximately two logs higher, in protein chip and analyzed directly by SELDI-TOF mass the sub-microgram range. However, when fluorescent spectrometry. Comparing to protein microarray, the ma- label and laser microarray scanner were applied to mi- jor advantage is its speed. The assay can be completed in croarrays, the reproducibility and LOD for sixteen dif- less than 1 h. However, the current degree of multiplex- ferent human cytokines were similar to ELISA results ing is low and expensive instrumentation is necessary. [12]. For limited number of analytes, protein microarray has no clear advantage over traditional ELISA in terms Effect of Streptavidin-Peroxidase Polymer of performance and ease of use. Nevertheless, protein microarray would be an appropriate method when a com- The polymer is made by covalently linking multiple plex concentration profile is needed due to its ability to streptavidin and horseradish peroxidase molecules to a perform multiplexed analysis. hydrophilic polymer backbone according to the supplier. In addition to protein microarray, several new tech- The number of enzyme labels per biotin binding site can nologies for cytokine detection have also emerged in be significantly increased and thus enhance the chemi- the past two decades. The enzyme-linked immunospot luminescence signal. Rolling circle amplification and (ELISPOT) uses the same concept from traditional ELISA silver enhancement have been applied to amplify the [13]. Instead of measuring the cytokine concentration in signal from protein microarrays [4,19]. the sample, the ELISPOT assay allows the quantitative We tested the potential signal enhancing effect of measurement of frequency of cytokine secreting cells at 10, 20, and 33 ng/ml polymer. Both 20 and 33 ng/ml the single cell level. Another immunologically based polymer showed no response to IL-4 since the chemilu- method, the multiplex bead array assays (MBAA), has minescence already reached the saturation level of the the advantages of multiplexing and small sample volume digital camera (Fig. 4). When 10 ng/ml polymer was

J Biomed Lab Sci 2006 Vol 18 No 1~4 19

Protein microarray for cytokine detection

Fig. 4. Effect of streptavidin-peroxidase polymer on the detection of IL-4 (□: 10 ng SA-HRP/ml; ▲: 10 ng polymer/ml; ◆: 20 ng polymer/ml; ○: 33 ng polymer/ml). The printing concentration of the polyclonal antibody was 1 mg/ml. Biotinylated antibody was 62.5 ng/ml. Cytokine standard (mixture) 0.005, 0.0125, 0.05, 0.125, and 0.5 ng/ml were prepared in PBS containing 1 % (w/v) BSA. used, chemiluminescence intensity increased about two 5. Kodadek T: Protein microarrays: prospects and prob- fold for all standards, compared to the optimal 10 ng/ml lems. Chem Biol 2001; 8: 105-15. SA-HRP. However, the LOD remained the same sincethere 6. Huang RP, Huang R, Fan Y, et al: Simultaneous detec- tion of multiple cytokines from conditioned media and was a coherent increase in the blank signal (from 16432 ± ± patient’s sera by an antibody-based protein array sys- 1328 to 30595 2071, n=6). Similar result was also ob- tem. Anal Biochem 2001; 294: 55-62. served for IL-2. Although the polymer did not lower the 7. Tonkinson JL, Osborn DS, Stillman BA, et al: Smaller is LOD, the images were easier to analyze due to brighter better. IVD Technology 2003; 9: 29-34. and clearer spots. 8. Skoog DA, Holler FJ and Nieman TA: Principles of in- strumental analysis, 5th ed. Orlando: Harcourt, 1998. 9. Cardenas JM, Marshall P, Henderson B, et al: Human Acknowledgements interleukin 2. Quantitation by a sensitive radioimmuno- assay. J Immunol Methods 1986; 89: 181-9. Financial support by National Science Council (grant 10. Custer MC and Lotze MT: A biologic assay for IL-4. Rapid fluorescence assay for IL-4 detection in supernatants and NSC 92-2213-E-036-003) and Tatung University (grant serum. J Immunol Methods 1990; 128: 109-17. B9208-S05-073) is gratefully acknowledged. 11. Copeland S, Siddiqui J and Remick D: Direct compari- son of traditional ELISAs and membrane protein arrays References for detection and quantification of human cytokines. J Immunol Methods 2004; 284: 99-106. 12. Knight PR, Sreekumar A, Siddiqui J, et al: Development 1. MacBeath G: Protein microarrays and proteomics. Nat of a sensitive microarray immunoassay and comparison Genet Supplement 2002; 32: 526-32. with standard enzyme-linked immunoassay for cytokine 2. Ekins RP: Ligand assays: from electrophoresis to minia- analysis. Shock 2004; 21: 26-30. turized microarrays. Clin Chem 1998; 44: 2015-30. 13. Cox JH, Ferrari G and Janetzki S: Measurement of cy- 3. Sage L: Protein biochips go high tech. Anal Chem 2004; tokine release at the single cell level using the ELISPOT 76: 137A-42A. assay. Methods 2006; 38: 274-82. 4. Liang RQ, Tan CY and Ruan KC: Colorimetric detection 14. Hill HR and Martins TB: The flow cytometric analysis of of protein microarrays based on nanogold probe coupled cytokines using multi-analyte fluorescence microarray with silver enhancement. J Immunol Methods 2004; 285: technology. Methods 2006; 38: 312-6. 157-63.

20 J Biomed Lab Sci 2006 Vol 18 No1~4

C. Y. Yu, M. L. Liu, I. C. Kuan, et al

15. Elshal MF and McCoy JP: Multiplex bead array assays: teomics 2004; 4: 609-18. performance evaluation and comparison of sensitivity to 18. Boyle MD, Hess JL, Nuara AA, et al: Application of im- ELISA. Methods 2006; 38: 317-23. munoproteomics to rapid cytokine detection. Methods 16. Guthrie JW, Hamula CL, Zhang H, et al: Assays for cy- 2006; 38: 342-50. tokines using aptamers. Methods 2006; 38: 324-30. 19. Kingsmore SF and Patel DD: Multiplexed protein profil- 17. Bock C, Coleman M, Collins B, et al: Photoaptamer ing on antibody-based microarrays by rolling circle am- arrays applied to multiplexed proteomic analysis. Pro- plification. Curr Opin Biotechnol 2003; 14: 74-81.

J Biomed Lab Sci 2006 Vol 18 No 1~4 21

Protein microarray for cytokine detection

低密度硝化纖維膜蛋白質微陣列應用於人類細胞激素檢測

游吉陽 劉茂隆 官宜靜 李綉鈴

大同大學生物工程系

本研究發展出低成本、低密度之硝化纖維膜蛋白質微陣列並應用於人類細胞激素檢測。此微陣列 係由冷光三明治式免疫分析法衍生而來。首先將抗人類 IL-2及抗人類IL-4多株抗體打點並固定於薄 膜表面。薄膜經 blocking後與人類IL-2及人類IL-4反應，再加入經生物素標定之抗人類 IL-2及抗人 類IL-4單株抗體與細胞激素及固定於表面之多株抗體形成複合物。以 streptavidin-horseradish peroxidase conjugate偵測生物素，最後加入 peroxidase之基質以產生冷光。微陣列中之點大小一致 且產生均勻冷光。在最適條件下，對人類 IL-2及人類IL-4之偵測極限分別可達 0.5及0.125 ng/ml。標 準檢量線之線性範圍跨越兩個級數。反應中使用 10 ng/ml之streptavidin-peroxidase polymer可使數 據分析更為容易。

關鍵詞：冷光、細胞激素、微陣列、硝化纖維膜

收稿日期：95 年 6 月 1 日 修稿日期：95 年 7 月 31 日 接受日期：95 年 9 月 11 日 通訊作者：游吉陽 台北市中山北路三段 40 號 Tel: 886-2-25925252-3315-27 Fax: 886-2-25854735 E-mail: [email protected]

22 J Biomed Lab Sci 2006 Vol 18 No1~4