Anal Bioanal Chem (2014) 406:7867–7873 DOI 10.1007/s00216-014-8244-3

RESEARCH PAPER

Microfluidic antibody arrays for simultaneous cell separation and stimulus Yan Liu & Todd Germain & Dimitri Pappas

Received: 28 August 2014 / Revised: 29 September 2014 / Accepted: 6 October 2014 / Published online: 30 October 2014 # Springer-Verlag Berlin Heidelberg 2014

Abstract A microfluidic chip containing stamped antibody arrays was developed for simultaneous cell separation and drug testing. Poly(dimethyl siloxane) (PDMS) stamping was used to deposit antibodies in a microfluidic channel, forming discrete cell-capture regions on the surface. Cell mixtures were then introduced, resulting in the separation of cells when specific antibodies were used. Anti-CD19 antibody regions resulted in 94 % capture purity for CD19+ Ramos cells. An antibody that captures multiple cell types, for example antiCD71, can also be used to capture several cell types simultaneously. Cells could also be loaded onto the arrays with spatial control using laminar streams. Both Ramos B cells and HuT 78 T cells were isolated in the chip and exposed to staurosporine in the same channel. Both cell lines had similar responses to the drug, with 2–10 % of cells remaining viable after 20 h of drug treatment, depending on cell type. The chip can also be used to analyze the efficacy of antibody therapy against cancer cells. Anti-CD95 was deposited on the surface and used for simultaneous cell capture and apoptosis induction via the extrinsic pathway. Cells captured on anti-CD95 surfaces had significant viability loss (15 % viability after 24 h) when compared with a control anti-CD71 antibody (81 % viability after 24 h). This chip can be used for a variety of cell separation and/or drug testing studies, enabling researchers to isolate cells and test them against different anticancer compounds and to follow cell response using fluorescence or other readout methods. Keywords Bioanalytical methods . Biotechnological products . Cell systems . Single cell analysis . Microfluidics . Microfabrication . Separations . Instrumentation Y. Liu : T. Germain : D. Pappas (*) Department of Chemistry and Biochemistry, Texas Tech University, Lubbock, TX 79409-1061, USA e-mail: [email protected]

Introduction Microfluidic systems have been widely used in spatially patterned bioassays for high-throughput cellular analysis. Microfluidics enable rapid and reproducible experiments to be performed on small sample volumes [1–5], and enable control of fluid flow and reagent delivery [6–8]. In addition to the fluidic control available, the surface chemistry of labon-a-chip devices can provide additional functionality [9, 10]. For example, the coating and/or patterning of capture molecules in microfluidics has enabled many forms of cell separation [3, 11–15]. It is possible to pattern capture chemistry to capture different cell types in different regions of a microchannel. This type of patterning enables parallel cell analysis without using multiple channels, which minimizes differences in flow effects [12, 16]. In addition to multiplexed cell detection, it is possible to pattern different biomarkers to assess cell response to external stimuli including extracellular matrix or membrane receptor ligands [17–20]. However, patterning different surface chemistry in microfluidic channels remains a complex task [21–23]. Robotic printing is the most commonly used method to deliver different biomarkers, but these systems are costly [24]. Other complex methods based on stepwise surface change through electrochemical [25], photochemical [26, 27], thermally responsive polymer [28], and laminar streams [29] have also been reported. Laminar-flow methods are less complex than the aforementioned methods, but require larger reagent volumes for coating. Microcontact printing is a simple and rapid method to create region-specific patterns by stamping functional groups on substrates, which would enable further binding to DNA or proteins through adsorption, covalent immobilization, or avidin–biotin interactions [30]. Biotin–avidin chemistry has been widely used for microfluidic surface functionalization [31], and streptavidin can be stamped directly onto glass without

7868

loss of activity [32]. In this work, we directly printed neutravidin on glass, followed by enclosure of the surface by PDMS microfluidic channels. The remaining surface chemistry was completed by flowing reagents over the stamped channel surface. The approach was simple and can be adapted to a wide range of microfluidic channels. A stamped-channel PDMS device was used to achieve cell capture and receptor-induced apoptosis. We obtained highpurity capture of cells from mixtures using stamped capture regions (Fig. 1). We then used an antibody to induce apoptosis, to establish the ability to capture cells and to induce biological responses from them. We reveal a powerful cellular bioanalysis technique that can perform cell sorting, drug screening, and cell pathway studies.

Materials and methods Reagents and chip-fabrication materials Negative photoresist SU8-2015 and its developer were purchased from Microchem Corp. The positive photoresist AZ 9260 and developer AZ400K were purchased from AZ Electronic Materials. Poly(dimethylsiloxane) pre-polymer and crosslinking agent (Dow Sylgard) were purchased from Ellesworth Adhesives. Glass microscope slides and coverslips were purchased from VWR International. Biotinylated bovine serum albumin (BSA) was purchased from Sigma–Aldrich. Neutravidin was obtained from Pierce. Functional mouse antihuman CD95 (Fas) antibody and biotinylated anti-mouse IgM were purchased from eBioscience. Both biotinylated anti-human Fig. 1 A schematic of the experimental technique. (a) Device fabrication, including neutravidin printing, antibody deposition, plasma sealing, and cell loading. A PDMS stamp is used to define antibody regions on the glass surface, and then the PDMS layer containing a microfluidic chip is aligned and bonded to form the device. (b) Microfluidic affinity device with two antibody assays used for capturing target cells from mixture. A cell mixture is separated into two constituent cell lines, with each cell line captured on its respective antibody. (c) Two laminar streams, each one carrsying a specific cell line, were passed through the affinity channel to create multiple cell patterns

Y. Liu et al.

CD71 and biotinylated anti-human CD19 were purchased from Becton–Dickinson. Sterile phosphate-buffered saline (PBS) was obtained from Invitrogen. Biotin-phycoerythrin (biotin-PE), MitoTracker Red, MitoTracker Green, propidium iodide (PI), and Calcein-AM were purchased from Invitrogen. Chip fabrication Microfluidic designs for microcontact printing and microchannels were generated in an illustration program (Adobe Illustrator or Deneba Canvas) and printed on a 20,000 dpi photomask (CAD/Art services). The device consisted of a Y-shaped channel, with two short channels merged into a main channel: 40 μm deep, 1.0 or 2.0 mm wide, and 0.5 or 30 mm long for short channels and main channels, respectively. Soft lithography was used to fabricate the channel mold. Here, negative photoresist (SU-8 2050) was spin coated (at 1,500 rpm for 30 s) on a 4 in silicon wafer to generate a 40 μm-high fluid-delivery channel. The microfluidic network was fabricated by casting PDMS onto those photoresist-based molds. Once the PDMS elastomer with imbedded channels was released, holes were punched and tubes were connected to provide external access to the channels. Positive photoresist (AZ 9260) was spin coated (at 1,500 rpm for 30 s) to generate stamp molds with pillars 40 μm in diameter. Microcontact printing and surface functionalization To pattern uniform capture chemistry on the channel surface, the PDMS stamp was coated with 10 μL 3 μmol L−1

Microfluidic antibody arrays for simultaneous cell

neutravidin in Tris–HCl buffer at room temperature. To avoid solvent evaporation, a coverslip was placed on the stamp for 15 min. After rinsing with water and drying under air, the stamp was gently brought into contact with a clean glass substrate for 15 min, during which neutravidin patterns were transferred to the glass substrate. Antibody solutions (10 μg mL−1) in PBS were added to the stamped spots and incubated for another 15 min. For anti-CD71 and anti-CD19 immobilization, the biotinylated antibodies were added to the neutravidin spots directly. For anti-CD95 immobilization, biotinylated antimouse IgM was added to neutravidin and incubated as indicated above. Afterward, mouse anti-human CD95 (IgM) was added and incubated for 15 min to complete the capture surface. After all conjugation reactions had taken place, the chip surface was washed with buffer to remove unreacted antibodies. Once surface functionalization was complete, the microfluidic channels were sealed to the surface to finish device fabrication. An oxygen plasma (1 min, 200 W, 200 mTorr O2) was used to functionalize the PDMS channel layer and glass surface for sealing. An auxiliary PDMS strip trimmed to the channel sizes was used make a reversible seal on top of protein assays during plasma treatment. This reversible sealing strategy enabled antibody molecules covered by this protective PDMS layer to avoid damage or denaturing. After plasma treatment the protective PDMS layer was removed and the channel layer and glass surface were brought into conformal contact and sealed irreversibly. Tubing (30 gauge) was connected to the chip inlets and outlet, and a 10 mL syringe was connected to the inlet tubing for each fluid channel for reagent delivery. Reagent flow was controlled using a syringe pump. After sealing the chip, the channel surface was coated with BSA to reduce nonspecific binding on the glass surface surrounding the stamped capture patterns. Cells and cell culture Ramos B lymphocytes, Jurkat T lymphocytes, and HuT 78 T lymphocytes were purchased from American Type Culture Collection. Each cell line was cultured in RPMI 1640 medium (HyClone, Logan, UT) with 10 % fetal bovine serum and 2 % penicillin–streptomycin (Sigma–Aldrich) at 37 °C, 5 % CO2 atmosphere. Before analysis, cells were centrifuged at 4,500 rpm for 5 min, and resuspended in buffer (3 % BSA in PBS) with concentrations between 105 and 106 cells mL−1. Cells were loaded into the channel using a syringe pump at a flow of 0.05 mL h−1 for 2 min and allowed to settle for another 5 min for cell capture. The channel was then rinsed with buffer at 2.0 mL h−1 to remove unbound cells.

7869

study was to pattern regions of anti-CD19 and anti-CD71 capture, respectively, for tandem cell-affinity separations [12, 33]. A mixture of cells was introduced into chips which had 4×8 spots of anti-CD19 followed by 4×8 spots of anti-CD71. Cells were prestained with MitoTracker Red. PBS with 3 % BSA was used as the carrier fluid for cell loading to further minimize nonspecific binding. After 5 min incubation on the chip surface under stop-flow conditions, the channel was rinsed to remove unbound cells. Cell purity was quantified using fluorescence microscopy. Capture purity was defined as follows [3]: Purity ¼

ntarget  100% ntotal

where ntarget is the number of antigen-matching cells and ntotal is the total number of cells bound to the antibody spot. After cell separation, captured cells were exposed to a solution of 1 μmol L−1 straurosporine in medium for 6 h to induce cell apoptosis. White-light and fluorescence images were acquired at the very beginning and the end of this experiment, with 3 % BSA buffer flowing in the channel to reduce background signal. Control experiments were accomplished by incubating cells in the same affinity device with no drug stimulation. For the second study, stamped regions of anti-CD71 (control) and anti-CD95 were generated. Devices were tested using Jurkat and HuT 78 cells separately. Anti-CD71 captures all proliferating cells, and anti-CD95 also captures these cell lines. The anti-CD95 antibody used in this study is a functional clone that can induce receptor-mediated apoptosis [19, 20]. After cell capture, fluorescence microscopy was used to determine apoptotic and dead cells at the specified time frame. Fluorescence microscopy An inverted fluorescence microscope (IX71, Olympus, Center Valley, PA) was used to obtain all images. All images were captured by a CCD camera (Orca-285, Hamamatsu, Bridgewater, NJ) and analyzed by ImageJ (v. 1.41, National Institute of Health). To evaluate the stamp size features, transfer efficiency, and protein activity, 1 μmol L−1 biotin-PE in PBS was incubated with the printed neutravidin for several minutes. After rinsing, the surface was subjected to plasma treatment with the protective layer in place. Fluorescence microscopy was used to observe pattern size and distribution, and the presence of the proteins before and after plasma treatment.

Results and discussion

Cell isolation and anti-cancer-compound testing

Characterization of functionalized surfaces

Two studies were conducted using this affinity microdevice for studying cell response to anti-cancer compounds. The first

The PDMS stamp used to pattern the capture surface had features 40 μm in diameter with 150 μm spacing.

7870

Y. Liu et al.

Each stamped antibody spot can capture cells from a mixture by use of antigen–antibody binding. It is possible to control the coating process to enable different antibody regions to be stamped in the chip, enabling multiple immunophenotyping measurements [35]. The number of antibody spots can be controlled by changing chip geometry or antibody-spot pitch. To determine the cell-separation capabilities of this approach, anti-CD71 and anti-CD19 antibody spots were generated in the microfluidic channel using the stamping process (Fig. 3). When a 1:1 mixture of Ramos B cells and HuT 78 T cells was passed through the chip, the anti-CD19 antibody regions

had a capture purity of 94 %. Anti-CD71 captures both cell types, although the cell capture depends on the interaction time between the cell and the capture surface, the number of bonds formed, and the cell-surface contact area. Both the area and duration of cell-surface contact are expected to remain the same. Therefore, the critical difference between Ramos and HuT 78 cell capture will be the number of bonds formed, which is determined largely by the antigen density of each cell line. Therefore, capturing a cell mixture with a common antibody is a method of gauging relative antigen density, because we have already revealed that cell capture scales with antigen density [13]. When the 1:1 cell mixture was passed through the chip, 60 % of the cells were Ramos B cells and the remaining 40 % were HuT 78 T cells. For the antibody spot sizes used, an average of 28±8 cells were captured on each spot (mean ± standard deviation). This variation between spots is caused in part by flow effects and the random distribution of cells in the channel during cell loading. Using more than one spot of the same kind minimizes this source of error, and also serves as an internal control to ensure the capture chemistry is functioning. In this chip design, cells are exposed to flow continuously after capture. Therefore, shear effects have a function in cell capture and also affect the spatial orientation of the cells. It is possible to reduce shear forces at the expense of increased nonspecific binding. The shear stress can also be reduced by increasing the height of the channels, because shear stress scales with the square of channel height. However, increased channel heights will result in reduced capture efficiency, because cells will interact less with the surface because of laminar flow. It is also possible to use laminar-flow effects to deliver cells with high precision to a specific location on the chip. In Fig. 3c, Ramos B cells were directed to the “north” half of the channel whereas HuT 78 cells were directed to the “south” half. As expected, Ramos cells were captured in the antiCD71 stamped regions on the north half of the channel but not the south. The opposite result was obtained for HuT 78 cells. Cells were retained on the spots for at least 24 h under continuous flow. This approach is useful for patterning multiple cell types for parallel analysis of cells in the same chip.

Fig. 3 (a, b) Cell patterning onto the channel with two antibody arrays. (a) Anti-CD19 captured B lymphocyte cells (grey cells, 94 % capture purity). (b) Anti-CD71 captured both B cells (green cells, 60 % capture

purity) and T cells (grey cells, 40 % capture purity). (c) Cell loading through laminar flow with Hut 78 T cells on “north” side and Ramos B cells on the “south” side

Fig. 2 Microcontact printing of neutravidin and conjugation of PElabeled biotin to the printed area: (a) proteins patterned on a clean glass substrate; (b) PDMS stamp after neutravidin transfer. Fluorescence intensities of each spot differed by 16 % across the array

Fluorescein-labeled biotin was used to visualize neutravidin deposition on the glass surface (Fig. 2). The deposited neutravidin spots matched the pillars on the PDMS stamp. The stamped boundaries were clearly observed, with uniform stamp pitch across the surface. The transfer of neutravidin occurred because the binding strength between the glass and protein was stronger than the force between the proteins and the PDMS stamp. We and others have revealed [33, 34] that PDMS–neutravidin adhesion is weaker than glass– neutravidin adhesion. The fluorescence-intensity difference between the 12 stamped spots shown in Fig. 2 is 16 % across all spots. However, for cell capture the antibody density deposited on the surface greatly exceeded the antigen density [19], so this variation was not expected to have a significant effect on cell capture. Cell capture on antibody spots

Microfluidic antibody arrays for simultaneous cell

7871

Fig. 4 Parallel drug stimulation of Hut 78 T and Ramos B cells in a single channel. HuT 78 (a) and Ramos (b) cells, patterned in the same channel, were exposed to staurosporine for 0 h, 4 h, and 20 h. Control-group HuT 78 cells (c) were cultured in the medium without drug stimulation. The reduction in MitoTracker Red fluorescence intensity reveals the membrane-potential loss during apoptosis. At 4 h, 40 % of HuT 78 cells and 7 % of Ramos cells were apoptotic. At 20 h, 93 % of HuT 78 and 90 % of Ramos cells were apoptotic. No control cells were detected to be apoptotic at 4 or 20 h

Anti-cancer-drug testing of captured cells Apoptosis regulation is one of the mechanistic targets for many chemotherapy drugs [35]. The intrinsic pathway is typically initiated using several cytotoxic compounds, for example straurosporine and camptothecin [36, 37]. After cells were captured on their respective antibody regions, 1 μmol L−1 staurosporine was passed through the channel to induce apoptosis. For this study, HuT 78 T cells were loaded first and stop flow was used to saturate one set of anti-CD71 spots. Afterward, Ramos B cells were loaded to attach to the anti-CD19 regions and anti-CD71 regions not exposed to HuT 78 cells. The shear stress in the channels was controlled by adjusting the flow rate so that shear damage did not induce apoptosis or necrosis. The medium-perfusion steps were performed at 0.05 mL h −1 , resulting in a shear force of 0.26 dyn cm−2. The washing step to remove cells used a flow

Jurkat CD71 0h

CD71 6h

Fig. 6 Cell culture on anti-CD95 and anti-CD71 (control) spots for 24 h. Control cells (a) had similar morphology to drug-treated cells at 0 h. The drug-treated cells (b) underwent significant morphological changes as apoptotic bodies were formed. Cell viability was assessed by calcein-AM staining and propidium iodide to identify dead cells. Viable cells (green fluorescence) and dead cells (red fluorescence) were counted in each array. Control-cell viability (c) was 81 % at 24 h, whereas the antiCD95-captured cells had a viability of 15 % (d)

of 2.0 mL h−1 (10 dyn cm−1). The washing step was brief to minimize shear stress. Control experiments (Fig. 4) indicated that the perfusion and washing steps did not induce apoptosis or necrosis in cells. One defining characteristic of apoptosis is mitochondrial membrane-potential loss, which can be evaluated using fluorescent probes, for example MitoTracker Red (MTR) [38]. Staurosporine was revealed to reduce mitochondrial fluorescence in both Ramos B cells and HuT 78 cells in the same chip. After cell capture, as staurosporine was just introduced (0 h), cells had a bright fluorescence and cell adhesion was primarily limited to the stamped antibody spots. At 4 h, both cell lines had reduced fluorescence as apoptosis was induced in the cells (Fig. 4). At 20 h, most cells lost mitochondrial membrane potential, but were still readily observed in whitelight images (not shown). Using a threshold of μ0hr −3s, where μ0hr was the mean cell fluorescence at 0 h (before drug introduction) and s was the standard deviation of that fluorescence mean, 93 % of HuT 78 cells at 20 h had lost membrane potential and were deemed to be apoptotic. Similarly, 90 % of

CD95 0h

CD95 6h

Fig. 5 Cell response ligands bound to the antibody spots. Jurkat cell apoptosis was induced during cell capture with an anti-CD95 affinity surface. No control cells were found to be apoptotic at 6 h, whereas 97 % of anti-CD95-captured cells were apoptotic at 6 h

7872

Ramos cells were apoptotic at 20 h. However, the mid-point measurement at 4 h revealed a greater difference between the two cell types. At 4 h, 40 % of HuT 78 cells were apoptotic, whereas only 7 % of Ramos cells were apoptotic. This difference indicated that Ramos-cell response was slower than that of HuT 78 cells. Because both cell types were captured in the same channel, it was possible to directly compare temporal dynamics of the apoptosis process. Stamped ligand arrays for receptor-mediated apoptosis As mentioned previously, it is possible to stamp ligands in the channel that not only result in cell capture, but also induce a biochemical response. Functional anti-CD95 has been revealed to capture T cells by binding to the Fas receptor on the cell surface and inducing cell apoptosis [19, 20]. After 6 h (Fig. 5), 97 % of Jurkat cells were apoptotic, indicating that binding to the anti-CD95 surface induced apoptosis. When anti-CD71 was used as a control, 0 % of cells were apoptotic at 6 h. Mitochondrial-membrane-potential dyes indicate the earliest stages of apoptosis. In addition, MitoTracker Red and similar dyes are susceptible to photobleaching because they are “signal off” detection methods. We also assayed viability at 24 h to determine how many cells were dead at the end of the apoptosis process. Staining with calcein-AM and propidium iodide (Fig. 6) indicated that 15 % of anti-CD95captured Jurkat cells were viable, whereas 81 % of control cells were viable. Although some viability loss is expected in the microchannel over that time span, the large reduction in viability on anti-CD95 spots is caused by the completion of apoptosis.

Conclusion We report an integrated microfluidic affinity device for cell capture and drug and/or ligand testing. This work presents a simple method to pattern several antibody assays in a single channel, which facilitates analysis of cellular response to external stimuli in a parallel fashion. We used multiple antibody patterns to capture different cell types at different defined points of a microchannel, which enabled parallel-cellassay capabilities. At the same time, multiplexed cell analyses were achieved with different intrinsic and extrinsic apoptosis stimuli. The objective of this work was to develop a straightforward method to create an integrated microassay for cell capture and analysis. In future work, we will reduce the spot pitch and increase the complexity of the assay to increasing numbers of spot types in the same channel. For example, we will increase the number of spot types per chip to perform complex immunophenotyping of blood samples, and will also

Y. Liu et al.

investigate the interference effects that coincide with sample complexity. Acknowledgments This project was supported by the National Institutes of Health (Grant RR025782 and GM103550).

References 1. El-Ali J, Sorger PK, Jensen KF (2006) Nature 442:403–411 2. Velve-Casquillasa G, Berrea ML, Piela M, Trana PT (2010) Nano Today 5:28–47 3. Gao Y, Li W, Pappas D (2013) Analyst (Cambridge, U K) 138:4714– 4721 4. Gao Y, Li P, Pappas D (2013) Biomed Microdevices 15:907–915 5. Henley WH, Dennis PJ, Ramsey JM (2012) Anal Chem 84:1776– 1780 6. Liu Y, Butler WB, Pappas D (2012) Anal Chimic Acta 743:125–130 7. Liu K, Dang D, Harrington T, Bayer K, Pappas D (2008) Langmuir 24:5955–5960 8. Liu K, Tian Y, Burrows SM, Reif RD, Pappas D (2009) Anal Chim Acta 651:85–90 9. Fiddes LK, Chan HC, Lau B, Kumacheva E, Wheeler AR (2010) Biomaterials 31:315–320 10. Kusi-Appiah AE, Vafai N, Cranfill PJ, Davidson MW, Lenhert S (2012) Biomaterials 33:4187–4194 11. Vickers DAL, Chory EJ, Murthy SK (2012) Lab Chip 12:3399–3407 12. Li P, Gao Y, Pappas D (2012) Anal Chem 84:8140–8148 13. Wang K, Solis-Wever X, Aguas C, Liu Y, Li P, Pappas D (2009) Anal Chem 81:3334–3343 14. Hyun K-A, Lee TY, Jung H-I (2013) Anal Chem 85:4439–4445 15. Choi S, Levy O, Coelho MB, Cabral JMS, Karp JM, Karnik R (2014) Lab Chip 14:161–166 16. Didara TF, Tabrizian M (2010) Lab Chip 10:3043–3053 17. Tang CK, Vazea A, Rusling JF (2012) Lab Chip 12:281–286 18. Flaim CJ, Chien S, Bhatia SN (2005) Nat Method 2:119–225 19. Reif RD, Martinez MM, Wang K, Pappas D (2009) Anal Bioanal Chem 395:787–795 20. Reif RD, Martinez MM, Pappas D (2010) Anal Bioanal Chem 397: 3387–3396 21. Yin Z, Tao S, Cheong R, Zhu H, Levchenko A (2010) Integr Biol 2: 416–423 22. Shi J, Yang T, Cremer PS (2008) Anal Chem 80:6078–6084 23. Rodriguez-Segui SA, Ximenez JIP, Sevilla L, Ruiz A, Clopo P, Rossi F, Martinez E, Samitier J (2011) J Biomed Mater Res 98:245–256 24. Kusi-Appiah AE, Vafai N, Cranfill PJ, Davidson MW, Lenhert S (2008) Lab Chip 8:2197–2205 25. Kaji H, Hashimoto M, Nishizawa M (2006) Anal Chem 78:5469–5473 26. Balakirev MY, Porte S, Vernaz-Gris M, Berger M, Arie J, Fouque B, Chatelain F (2005) Anal Chem 77:5474–5479 27. Custodio CA, San Miguel-Arranz V, Gropeanu RA, Gropeanu M, Wirkner M, Reis RL, Mano JF, del Campo A (2014) Langmuir 30: 10066–10071 28. Huber DL, Manginell RP, Samara MA, Kim B, Bunker BC (2003) Science 301:352–354 29. Didar TF, Foudeh AM, Tabrizian M (2012) Anal Chem 2012(84): 1012–1018 30. Sassolas A, Leca-Bouvier BD, Blum LJ (2008) Chem Rev 108:109– 139 31. Heikkinena JJ, Kivimäkib L, Maattac JAE, Makelab I, Hakalahtib L, Takkinend K, Kulomaac MS, Hytonenc VP, Hormi OEO (2011) Colloids Surf, B 87:409–414

Microfluidic antibody arrays for simultaneous cell 32. Iversen L, Cherouati N, Berthing T, Stamou D, Martinez KL (2008) Langmuir 24:6375–6381 33. Wang K, Marshall MK, Garza G, Pappas D (2008) Anal Chem 80: 2118–2124 34. Sheng W, Chen T, Kamath R, Xiong X, Tan W, Fan ZH (2012) Anal Chem 84:4199–4206 35. Martinez MM, Reif RD, Pappas D (2010) Anal Methods 2:996–1004

7873 36. Yu ZTF, Kamei K, Takahashi H, Shu CJ, Wang X, He W, SILVERMAN R, Radu CG, Witte ON, Lee K, Tseng H (2009) Biomed Microdevices 11:547–555 37. Zeng C, Zhang X, Lin K, Ye H, Feng S, Zhang H, Chen Y (2012) Mol Pharmacol 4:578–586 38. Khanal G, Chung K, Solis-Wever X, Johnson B, Pappas D (2011) Analyst 136:3519–3526