RESEARCH REPORT
Cell adhesion profiling using extracellular matrix protein microarrays Cornelia Kuschel, Heiko Steuer, Andreas N. Maurer, Britta Kanzok, Reinout Stoop, and Brigitte Angres BioTechniques 40:523-531 (April 2006) doi 10.2144/000112134
We have developed a microarray-based system for cell adhesion profiling of large panels of cell-adhesive proteins to increase the throughput of in vitro cell adhesion assays, which are currently primarily performed in multiwell plates. Miniaturizing cell adhesion assays to an array format required the development of protocols for the reproducible microspotting of extracellular matrix (ECM) protein solutions and for the handling of cell suspensions during the assay. We generated ECM protein microarrays with high reproducibility in microspot protein content using nitrocellulose-coated glass microslides, combined with piezoelectric microspotting of protein solutions. Protocols were developed that allowed us to use 5000 cells or fewer on an array of 4 × 4 mm consisting of 64 microspots. Using this microarray system, we identified differences of adhesive properties of three cell lines to 14 different ECM proteins. Furthermore, the sensitivity and accuracy of the assays were increased using microarrays with ranges of ECM protein amounts. This microarray system will be particularly useful for extensive comparative cell adhesion profiling studies when only low amounts of adhesive substrate and cells, such as stem cells or cells from biopsies, are available.
INTRODUCTION Controlled cell adhesion to the extracellular matrix (ECM) is essential for a coordinated morphogenesis and growth of functional tissue during embryonic development, tissue differentiation, and regeneration (1–3). Deregulation of cell adhesion occurs in many diseases and can contribute to the malfunction of tissues. For example, the abundance of integrins, a major family of ECM receptors, changes during tumor transformation. This is believed to contribute to the progression from a noninvasive to an invasive tumor stage (4). Therefore, binding specificities between cells and ECM components as well as changes in binding strengths and molecular mechanisms underlying these changes have been studied for many years. At the same time, the continuing discovery of novel ECM proteins, receptors, and related signal transduction pathways results in increasing numbers of cell-substrate interactions being investigated (5–8). In addition, there is growing interest
in characterizing cell attachment to synthetic materials, such as peptides, polymers, or recombinant ECM protein fragments for the purpose of designing matrices for tissue engineering (9–12). Modulation of cell-substrate interactions by environmental influences, such as growth factors, or intrinsic effects, such as altered gene expression, add additional complexity to the function of cell-substrate interactions being investigated (13). Given the multitude of cell-substrate interactions, systematic screenings of cell attachment to large panels of substrates accelerate a comprehensive characterization and understanding of cell adhesion. Most in vitro studies on cellECM interactions are performed by coating multiwell plates with ECM proteins or other adhesion molecules. This approach requires considerable quantities of substrate and cells. Because most ECM proteins are purified from tissues, they can be expensive and limited in quantity, especially when extracted from human sources. Another limiting factor is cell
availability, especially when stem cells or cells from biopsied material are to be tested against a whole panel of ECM proteins or other substrates. The miniaturization of cell adhesion assays helps to facilitate higher throughput for such studies, and large panels of substrates can be examined rapidly while only small amounts of cells are needed. For this reason, microarrays of diverse cell binding substrates have been developed (14–21). These studies were mainly focused on the preparation of arrays containing various types of substrates and their effects on cells, but little attention has been given to the development of methods for accurate comparative measurements of cell adhesion in this miniaturized format. Here we describe an advanced cell microarray system that provides both reproducible immobilization of protein and a better control of cell culture conditions for differential cell adhesion measurements while using small amounts of cells.
NMI at the University of Tübingen, Reutlingen, Germany Vol. 40, No. 4 (2006)
BioTechniques 523
RESEARCH REPORT MATERIALS AND METHODS Proteins Type I collagen from human placenta, human plasma-derived fibronectin, human tenascin, and laminin from human placenta were purchased from Chemicon (Temecula, CA, USA). Type III collagen from human placenta, type IV collagen from basement membrane of Engelbreth-Holm-Swarm (EHS) murine Sarcoma, type V collagen from human placenta, human vitronectin, human thrombospondin, EHS laminin, EHS heparan sulfate proteoglycan (HSPG), and poly-L-lysine were purchased from Sigma-Aldrich (Taufkirchen, Germany). Human cellular fibronectin was obtained from Upstate Biotechnology (Lake Placid, NY, USA), type IV collagen from human placenta was purchased from BD Biosciences (Bedford, MA, USA), human placenta-derived type VI collagen was purchased from Acris (Hiddenhausen, Germany), and bovine serum albumin (BSA) was acquired from Roth (Karlsruhe, Germany). BSA was labeled with TAMRA (Fluka Chemie AG, Deisenhofen, Germany), and the labeled protein was purified using a PD10 column (Pharmacia AB, Uppsala, Sweden) according to the manufacturer’s instructions. Preparation of Microarrays Proteins were microspotted on nitrocellulose-coated glass microslides (NMI Technologietransfer GmbH, Reutlingen, Germany). Protein solutions were prepared in 0.5% (w/v) trehalose in phosphate-buffered saline (PBS) containing 0.0003% (v/v) of the nonionic detergent IGEPAL® CA-630 [(octylphenoxy)polyethoxye thanol; Sigma-Aldrich] for collagens or 0.005% (v/v) IGEPAL CA-630 for all other proteins. For the chondrocyte adhesion assays, type III collagen was spotted in 0.4% (w/v) trehalose, 0.001% IGEPAL CA-630, and 5% glycerol. In general, we used the lowest protein coating concentration of a given protein at which the highest number of cells attached to a microspot in an adhesion assay. Furthermore, economic 524 BioTechniques
considerations were taken into account when expensive proteins were used as well as technical feasibility of printing these protein solutions with the ink-jet arrayer. Unless otherwise indicated, proteins were spotted at the following concentrations: 200 μg/mL poly-L-lysine, type III collagen, type V collagen, type VI collagen, fibronectin (human cellular), laminin (human placenta), and laminin (EHS); 400 μg/mL BSA; 50 μg/mL type I collagen (human placenta) and tenascin; 100 μg/ mL type IV collagen (human placenta), type IV collagen (EHS), fibronectin (human plasma), thrombospondin, and HSPG; 250 μg/mL vitronectin. Except for microarrays prepared for chondrocyte adhesion assays, 1 μg/mL BSA-TAMRA was added to each protein solution for the visualization of microspots on slides with a laser scanner (GMS 418 Array Scanner™; Affymetrix, MWG, Ebersberg, Germany). Three nanoliters of protein solution per microspot were spotted with a Packard BioChip Arrayer™ (PerkinElmer Life Sciences, Meriden, CT, USA) onto the nitrocellulose-coated glass slides. Eight by eight microspots were arranged in rectangular arrays, and 12 arrays were placed on each slide. Arrays were printed in regular distances that fit cell cultivation chambers (ProPlat™; Grace Biolabs, OR, USA) such that each array comes to lie in one chamber. After microspotting, the slides were stored dry (without dessicant) at 4°C in a microslide storage box. All arrays in this study were used within threeand-a-half months after manufacturing. Stability tests performed with two cell lines (HEK 293, NIH 3T3) indicated that cells adhered with similar avidity to protein microspots of arrays stored up to 6 months (data not shown). Each experiment in the present study was performed with arrays of one printing batch and thus of the same age. Cell Culture The human embryonic kidney cell line HEK 293 and murine fibroblast cell line NIH 3T3 were propagated in Dulbecco’s modified Eagle’s medium (DMEM; Cambrex, East Rutherford,
NJ, USA) with 10% heat-inactivated fetal bovine serum (FBS) and 2 mM Lglutamine (PAA Laboratories GmbH, Cölbe, Germany) in a humidified 5% CO2 atmosphere at 37°C. Rat adrenal pheochromocytoma cells (PC12) were cultured in DMEM supplemented with 10% heat-inactivated horse serum (PAA Laboratories GmbH), 5% heat-inactivated FBS, and 2 mM L-glutamine in a humidified 10% CO2 atmosphere at 37°C. The chondrocyte cell line C20A4 (22) and human primary chondrocytes isolated from donor knee cartilage (23) were cultured in monolayer in DMEM/ HAM F12 (2:1; Invitrogen, Carlsbad, CA, USA) supplemented with 10% heat-inactivated FBS (PAA Laboratories GmbH) and 54 μg/mL L-ascorbic acid 2-phosphate (Sigma-Aldrich) in a 5% CO2 atmosphere at 37°C. All cultures were supplemented with 100 U/mL penicillin and 100 μg/mL streptomycin (both from Cambrex). Cell Adhesion Assays For HEK 293, NIH 3T3, and PC12 cell adhesion assays, the microarray slides were adjusted to room temperature for 1 h. Cell adhesion assays were carried out in ProPlate cultivation chambers mounted on microarray slides such that each array was positioned in one cultivation chamber that measured 7 × 7 mm. The wells were filled with 200 μL StabilGuard® (SurModics, Eden Prairie, MN, USA), incubated at room temperature for 20–60 min to block unspecific binding sites, and subsequently washed twice with PBS. Cells that had been grown on tissue culture dishes were removed by trypsinization (0.5 mg/mL trypsin, 0.2 mg/mL EDTA in PBS; PAA Laboratories GmbH). After the cells had detached from the culture plate, 10 mL medium containing 10% serum were added to the cell suspension, and the cells were subsequently washed once in serumfree medium. Cells (2 × 104) in 200 μL serum-free medium were then seeded onto each array. The seeded microslides were kept for 4 h at 37°C and CO2 concentrations as indicated above. During the first 2 h of incubation, the slides were agitated for 4 s counterclockwise at 750 rpm every 10 min Vol. 40, No. 4 (2006)
with the Quant Array Software (Packard), and curves were fitted with Origin® 6.0 software (Microcal Software, Northampton, MA, USA). RESULTS The Microarray System
Figure 1. Array layout and cell colonization of ECM microarrays. (A) Laser-scanning image of a typical microarray slide. To visualize protein microspots, proteins were supplemented with fluorescently labeled BSA. Two arrays carrying poly-L-lysine only were printed at the upper end of the slide, and 10 microarrays were printed at the other positions. Scale bar, 4 mm. (B) Laser-scanning image of a typical array (top) and micrograph of the same array after colonization with HEK 293 cells stained with Coomassie Brilliant Blue (bottom). Proteins were microspotted in rows of four microspots. Scale bar, 500 μm. (C) Micrograph of a laminin microspot colonized with HEK 293 cells stained with Coomassie (top) and DAPI (bottom). Scale bar, 50 μm. ECM, extracellular matrix; BSA, bovine serum albumin; PLL, poly-L-lysine; HEK, human embryonic kidney.
by a programmable shaker (Variomag Teleshake; H+P Labortechnik AG, Oberschleißheim, Germany). After cultivation, the chambers were removed from the slides, the slides were washed by immersing in PBS containing Ca2+/ Mg2+ (Cambrex) in a 50-mL screwcap tube, and the cells were fixed and stained in a Coomassie™ solution (0.05% w/v Coomassie Brillant Blue, 50% methanol, 10% acetic acid). The cells were washed in PBS three times for 5 min, stained with 4′,6-diamidino2-phenylindole (0.5 μg/mL DAPI in PBS), and the slides were mounted in 0.2 M Tris-HCl, pH 8.5, 25% (w/v) glycerol, and 10% (w/v) Mowiol® (Calbiochem, La Jolla, CA, USA). The colonized microarrays were analyzed with a Axiovert 35M fluorescence microscope (Carl Zeiss, Oberkochen, Germany) using a 10× objective and motorized stage (Prior, Cambridge, UK). Images were acquired with a DC300F camera (Leica, Wetzlar, Germany). Macros for stage control, semiautomated image acquisition with Vol. 40, No. 4 (2006)
predictive autofocus, and subsequent quantitation of cells on each microspot by counting DAPI-stained nuclei were written in Qwin 2.7 (Leica). The JMP™ Version 5.0 software (SAS Institute, Cary, NC, USA) was used for statistical analyses of the data. For chondrocyte adhesion assays, arrays were prepared as described above. Chondrocytes (106) suspended in culture medium were labeled with 5 μL/mL DiD (Molecular Probes™; Invitrogen) for 20 min at 37°C and subsequently washed two times with medium. Unless indicated otherwise, 104 labeled cells in 200 μL medium were seeded on each array and incubated at 37°C as described above. Arrays were washed with PBS to remove unbound cells. Remaining cells were fixed with 4% paraformaldehyde in PBS for 20 min. The slides were washed with deionized water immediately before they were laser-scanned with a confocal fluorescence scanner (GMS 418) to detect the DiD fluorescence signal. Images were analyzed
During the production of the ECM microarrays, we particularly focused on providing optimal conditions for the measurement of differential cell adhesion. We used microslides with a thin nitrocellulose layer to ensure adequate protein adsorption for cell binding and sufficient optical clarity for bright-field microscopy (Figure 1B, bottom, and Figure 1C, top). The nitrocellulose layer can be easily blocked with appropriate reagents, such as StabilGuard, to avoid unspecific cell attachment. Because quantitative microspotting is of particular importance for the reproducible generation of protein microspots of similar protein content, we used a piezoelectric arrayer that dispenses constant volumes of protein solution and showed no carryover of protein from microspot to microspot. An overview of the microarray slide and its application in a typical cell adhesion assay is shown in Figure 1. Twelve arrays each consisting of 64 microspots of protein mixed with low amounts of fluorescently labeled BSA were placed on one microslide (Figure 1A). Microspots measured 300 μm in diameter and were spotted with a 500μm center-to-center distance (Figure 1B, top). Each microarray consisted of 14 different ECM proteins printed in rows of four microspots and a positive (poly-L-lysine) and negative (BSA) control substrate (Figure 1B, top). Laser scanning of the microslides showed that each quadruplet of microspots containing the same substrate had similar fluorescence intensities, confirming the reproducible spotting quality that can be achieved by the piezoelectric-driven dispension. Arrays colonized with cells show different numbers of cells adhering to microspots of different types of substrates (Figure 1B, bottom). An example of a subconfluent colonization of a microspot is shown in Figure 1C.
Circle Reader Service No. 184
RESEARCH REPORT
Because all microspots were located in one incubation chamber, we had to ensure that differences in the numbers of cells attached to microspots are due to specific cell-substrate interactions and not to an uneven distribution of cells in the incubation chamber. To determine conditions required for a uniform cell distribution, we incubated HEK 293 cells on microarrays consisting solely of fibronectin microspots. Because cells bind with the same avidity to all microspots of this substrate, similar cell counts on each microspot would indicate an even cell distribution. We used an electronically controlled shaker to agitate slides and distribute cells within the incubation chamber. Because high cell densities may lead to a saturation of microspots with cells and may not reveal differences in cell distribution in the array, we also determined the appropriate cell number to be seeded on an array for a subconfluent colonization of each microspot. Best results were achieved when microarrays were incubated with 2 × 104 cells in a volume of 200 μL and were agitated in a circular mode at a frequency of 750 rpm for 4 s. This agitation was repeated every 10 min for 2 h to resuspend unattached cells for redistribution on the microarray. A typical array of attached HEK 293 cells obtained under these conditions is shown in Figure 2A. A cone graph in Figure 2B shows similar cell numbers on all 64 microspots. For statistical evaluation of cell distribution, we incubated HEK 293 cells on 12 microarrays, printed on one slide. The number of cells per spot were found to be normally distributed as analyzed by the Shapiro-Wilk normality test (W = 0.98 and P = 0.2091, Figure 526 BioTechniques
cell lines (Figure 3). Comparing cell counts between all three cell lines on each of the 14 ECM substrates (42 pairs), we found 18 significant differences (Tukey-Kramer test; α = 0.05). For example, adhesion of NIH 3T3 cells to human placental type I [type I collagen human placenta (C I hplc)] and type IV collagen (C IV hplc) was significantly lower compared with the adhesion of HEK 293 and PC12 cells
Cell No.
Uniform Cell Distribution on Microarrays
2C). We then tested the reproducibility of a uniform cell distribution by incubating cells on 72 microarrays on a total of six slides. In this experiment, we changed several parameters, such as microslide batch, cell line, experimental day, and operator. The variation coefficients of cell counts per microspot were calculated for each array and ranged from 9.2% to 24.3%, with a median value of 13.4% (Table 1). The median value indicates that the majority of cell counts per microarray were within an acceptable range, and a uniform cell distribution was reproducible among the 72 microarrays. As an additional result of these experiments, we routinely included two microarrays consisting of poly-Llysine microspots only on each slide of microarrays (Figure 1A, top). Cells used in each experiment were incubated in parallel on these poly-L-lysine arrays, and the distribution of cells was judged by microscopic inspection after each experiment. Because all chambers of one slide are subjected to the same incubation conditions in an experiment and cells should bind with similar avidity to each poly-L-lysine microspot, this allowed us to judge whether a uniform cell distribution in all cultivation chambers had been achieved. Comparative Profiling of Cell Adhesion to Different ECM Proteins The multiple substrate array was used to compare the adhesion of different cell types to a panel of ECM proteins. We tested three cell lines, the epithelial cell line HEK 293, NIH 3T3 fibroblasts, and the neuroendocrine cell line PC12. The three cell lines were expected to adhere with different avidities to various ECM proteins due to their specific tissue origin and physiological function. Because differences in cell seeding numbers can occur when using different cell suspensions and because cells of different types can be of different sizes, we normalized cell numbers per spot to cell numbers obtained on poly-L-lysine spots in each array. A comparison of cell adhesion to 14 ECM proteins revealed considerable differences between the three
No. No.of ofMicrospots Microspots
As expected, cells attached to the positive-control substrate poly-L-lysine but not to microspots containing BSA (Figure 1B, bottom panel). Because this binding pattern was reproducible for each cell type and the control microspots were arranged in squares in the upper left corner of each microarray, this pattern was used to recognize the orientation of the array during microscopy and image analysis.
Cell No. Cell No. Figure 2. Distribution and attachment of cells to fibronectin microarrays. (A) Fibronectin array colonized with HEK 293 cells stained with Coomassie Brilliant Blue. (B) Three-dimensional micrograph representing relative cell numbers on microspots of one array. Full-size cones represent maximum numbers of cells. Relative reduction of cell numbers are indicated by cut tips of cones. (C) The number of cells per spot of one slide (12 arrays; 768 microspots) was determined and analyzed by the Shapiro-Wilk normality test. HEK, human embryonic kidney.
Vol. 40, No. 4 (2006)
to the same substrates. On the other hand, adhesion of NIH 3T3 cells to human plasma fibronectin (Figure 3, FN hupls) or vitronectin (VN hupls) was comparable to that of PC12 and HEK 293 cells. Differential Cell Adhesion on Microarrays with Different Substrate Densities
Attached Cells (PLL=100%)
The ECM microarray described here was fabricated by microspotting each protein at a high concentration, resulting in microspots of protein densities at which cells reached
maximum binding. However, changes in cell adhesion must not necessarily be revealed when high densities of matrix proteins are used. Therefore, many researchers test cell adhesion to a range of different amounts of protein (24–26). To adapt this approach to the miniaturized format, we prepared arrays consisting of type II collagen microspotted at different concentrations. Similar to enzyme-substrate EC50 (effective concentration, 50%) calculations, the adhesion to different substrate densities obtained from different coating concentrations can be plotted on a logarithmic scale, and the coating
Figure 3. Comparison of cell adhesion profiles of three cell lines. HEK 293, NIH 3T3, and PC12 cells were incubated on ECM microarrays for 4 h. Cell counts were normalized against average cell counts on poly-L-lysine microspots. Each bar represents the mean value of four microspots. Error bar = sd. PLL, poly-L-lysine; BSA, bovine serum albumin; C I, type I collagen; C III, type III collagen; C IV, type IV collagen; C V, type V collagen; C VI, type VI collagen; FN, fibronectin; LN, laminin; TSP, thrombospondin; HSPG, heparan sulfate proteoglycan; TN, tenascin; VN, vitronectin; hplc, human placenta; hpls, human plasma; hu, human; EHS, Engelbreth-Holm-Swarm murine Sarcoma; HEK, human embryonic kidney; ECM, extracellular matrix.
Table 1. Statistical Analysis of Distribution of Cells on Fibronectin Arrays Slidea
VC of Cell Counts Per Microspotb (%)
Median VC of All Arrays (%)
1B1aI
9.8–12.4
11.0
2A1aI
9.2–18.5
11.8
3A1bII
11.8–15.2
13.3
4B1bI
10.1–24.3
12.1
5B2bIII
13.4–20.2
16.1
6B2aIII
12.9–20.3
15.8
All slides
9.2–24.3
13.4
HEK, human embryonic kidney; VC, variation coefficient. aSlide batch, operator, cell line, and experimental day were varied: A, B: slide batches; 1, 2: operator; a: HEK 293 cells, b: NIH 3T3 cells; I, II, III: experimental day. Twelve microarrays per slide were included. bThis column represents the minimum and maximum values of the 12 microarrays of each slide.
Vol. 40, No. 4 (2006)
concentration at which half-maximal cell binding occurs can be calculated. We incubated different numbers of cells, ranging from 2500 to 20,000 cells on such arrays. Results show that values of half-maximal binding do not significantly differ and are independent of the number of cells seeded (Figure 4). Thus, measuring cell adhesion to a range of different substrate densities is a robust method to compare cell adhesion between different cell populations even if experimental errors in cell counting occur and cell numbers of the used cell populations are not quite comparable. We used such arrays to determine different binding properties between a chondrocyte cell line (27) and primary chondrocytes on type III collagen. For both cell lines, up to a protein coating concentration of 100 μg/mL, the number of cells increased with increasing coating concentrations (Figure 5). At a coating concentration of 100 μg/mL and above, both cell lines reached a plateau in cell numbers binding to the microspots. A reason for this could be that the nitrocellulose surface was saturated with protein at these high coating concentrations or cells could not increase their avidity to the protein when the protein was present above a certain density. The number of attached cells did not significantly differ between the two cell types at a coating concentration of 100 μg/mL and above. This indicates that a difference in adhesion cannot be detected at a concentration of 200 μg/ mL, which is the coating concentration used for most proteins in our ECM microarray. At lower concentrations, however, the primary cells reached half-maximal binding at approximately six times lower coating concentrations than the cell line. Thus, the chondrocyte cell line adheres with lower avidity to type III collagen compared with primary chondrocytes. DISCUSSION With the increasing interest in investigating interactions between cells and large numbers of substrates, there is a concurrent need to increase BioTechniques 527
Printing Concentration at Half-Maximal Adhesion (�g/mL)
Signal Intensity (Arbitrary Units)
RESEARCH REPORT
Concentration in Printing Solution (�g/mL)
Seeded Cells per Array
Figure 4. Adhesion of human primary chondrocytes to increasing amounts of type II collagen. DiD-labeled primary human chondrocytes were cultivated on arrays microspotted with increasing concentrations of type II collagen (0.13–270 μg/mL). The experiment was performed with 2500 (square), 5000 (circle), 10,000 (triangle), and 20,000 (upside-down triangle) cells. (A) Each data point represents the mean value of eight microspots from two arrays. Error bar = sd. For each cell number seeded, half-maximal cell numbers were determined and correlated with the concentration of type II collagen coating concentrations. Half-maximal cell attachment was obtained at 4.5 μg/mL (2500 cells/array), 2.9 μg/mL (5000 cells/array), 3.1 μg/mL (10,000 cells/array), and 2.5 μg/mL (20,000 cells/array) coating concentrations. (B) Distribution of values for coating concentrations of half-maximal cell attachment from six arrays is presented in a box-and-whisker plot for each cell seeding number.
528 BioTechniques
ECM proteins (14–21). In these works, the preparation of substrate was emphasized but little attention was paid to the adaptation of quantitative cell adhesion measurements to the miniaturized
Percent Fluorescence (Normalized to Brightest Microspot)
the throughput of assays while keeping expenditure of material, time, and cells at a reasonable level. We have developed a microarray system and cell adhesion assay protocols that are particularly suited for miniaturized comparative measurement of cell adhesion to large panels of substrates with minimal expenditure of adhesive substrate and cells. Quantitative comparisons of cell adhesion require reproducible substrate preparation and cell culture procedures. Using a nitrocellulose-coated surface in combination with piezoelectric microspotting, we were able to generate arrays of microspots with a highly reproducible protein content. The protein binding capacity of the nitrocellulose coating was well suited to bind a broad range of different amounts of protein, which allowed us to generate arrays of microspots with increasing protein densities. Several cell microarrays have been developed to comparatively measure cell attachment to a variety of substrates, including polymers, glycans, peptides, antibodies, major histocompatibility complex (MHC)-peptide complexes or
format. For example, entire microslides were incubated with cells in cell culture dishes (14,17). Unless the arrays are as large as the microslide, there is a high expenditure of cells relative
Primary cells Cell line C20A4
Primary cells
Cell line C20A4
Concentration in Printing Solution (�g/mL) Figure 5. Differential cell adhesion to increasing amounts of type III collagen. DiD-labeled primary human chondrocytes and C20A4 cells were cultivated for 4 h on arrays with increasing densities of type III collagen (protein concentrations in spotting solution: 0.4–750 μg/mL). (A) Fluorescence signal was normalized to the brightest microspot (= 100%) for each cell type. One data point represents the mean value of eight microspots from two arrays. Error bar = sd. (B) Laser-scanning images of arrays cultured with DiD-labeled cells. Different protein concentrations were printed in rows of four microspots from left to right beginning with the highest concentration in the upper left corner and ending with the lowest concentration in the lower right corner of the array. Vol. 40, No. 4 (2006)
to array area. Approaches that use micropipeting of small volumes of cell suspensions onto individual microspots or arrays without containment can reduce cell expenditure (16,18,19), but these strategies have the disadvantage of being more elaborate. Flaim and coworkers (15) improved this by adding a gasket to the slides to contain cell suspensions to the array area. In addition, they agitated the slides during cell incubations to distribute cells on the array. However, all these approaches required a minimum of 3 × 105 cells to cover arrays with a confluent cell monolayer to ensure that all microspots were in contact with a similar number of cells. Compared with these cell microarray systems, we reduced the expenditure of cells in our system by 10- to 1000-fold. The use of commercially available chambers that can be mounted on regular microslides and provide separate cavities for each array allowed us to use small volumes of cell suspensions. The agitation of the microslides in 10-min intervals led to the resuspension of unattached cells after each incubation period and redistributed the cells on the arrays. Using this protocol, 2 × 104 cells or fewer (as few as 625 cells; data not shown) could be used on our arrays to generate reproducible cell adhesion profiles of 14 ECM proteins. Receptor-mediated interaction of cells with ECM proteins often leads to a strengthening of adhesion during the first 15 min and beyond, after the first contact with the adhesive substrate. During this time, cytoplasmic moieties of the receptor molecules recruit intracellular cytoskeletal structures that lead to the strengthening of the adhesion (28). The 10-min interval in our experiment thus provides sufficient time for this molecular response to be initiated. However, variations of this time interval could possibly reveal differences in the phenotype of cell adhesion behavior, especially when comparing different cell types or cell treatments and densities of substrate molecules. Using the array system described here, it is possible to generate adhesion profiles of 10 different cell populations to multiple substrates on one microslide. Our statistical evaluation showed that this method yields reproducible results
with standard deviations comparable to those obtained from cell adhesion measurements in multiwell plates (25,29). Subsequent statistical analysis can be applied to determine significant differences in cell adhesion. However, we also showed that at high amounts of immobilized protein, small differences in cell adhesion were not always revealed. This became evident when we probed a chondrocyte cell line against primary cells on an array consisting of microspots with increasing amounts of type III collagen. The chondrocyte cell line had previously been generated by immortalization with simian virus 40 (SV40) large T antigen and is known for its different expression profile of integrins when compared with primary cells (27). Comparison of adhesion profiles of the two different cell populations showed that the cell line has a lower binding capacity to type III collagen than primary cells. However, the difference was only apparent on microspots generated with low coating concentrations. We conclude that arrays of microspots of different protein densities can be used for more sensitive detection of differences in adhesion. Moreover, the abundance of data obtained from a microarray containing an extensive range of protein densities makes the adhesion assay more reliable. In conclusion, we developed a cell microarray system and assay protocols for accurate and sensitive measurements of differential cell adhesion to large panels of substrates while using minimum amounts of adhesive substrate and cells. Many applications that interrogate cell-substrate adhesion can be performed in this system. For example, profiling of cell adhesion to various substrates could be used as a diagnostic tool for diseased cells, effects of therapeutic treatments, or changes of gene expression on cell adhesion could be determined, or the arrays could be used for screening substrates for cell adhesion for tissue engineering. ACKNOWLEDGMENTS
We would like to thank Michael Hartmann for assistance with the
microarrayer equipment and use, Susanne Stumpf for technical assistance in the early development of the microarray system, Markus Templin, Thomas Joos, Dieter Stoll, and Karin Benz for helpful discussions, and Helmut Wurst for critically reading the manuscript. The C20A4 cell line was a kind gift of Mary Goldring. COMPETING INTERESTS STATEMENT
A patent application on the preparation of nitrocellulose-coated glass slides has been submitted on behalf of the NMI and the authors C.K., H.S., B.K., and B.A. The NMI intends to make Multiple Substrate Array™ biochips as described in this manuscript commercially available. The authors C.K., H.S., A.N.M., R.S., and B.A. are currently employed by the NMI. B.K. is a former employee of the NMI. REFERENCES 1. Albelda, S.M. and C.A. Buck. 1990. Integrins and other cell adhesion molecules. FASEB J. 4:2868-2880. 2. Danen, E.H. and A. Sonnenberg. 2003. Integrins in regulation of tissue development and function. J. Pathol. 201:632-641. 3. Watt, F.M. 2002. Role of integrins in regulating epidermal adhesion, growth and differentiation. EMBO J. 21:3919-3926. 4. Mizejewski, G.J. 1999. Role of integrins in cancer: survey of expression patterns. Proc. Soc. Exp. Biol. Med. 222:124-138. 5. Boyd, C.D., R.A. Pierce, J.E. Schwarzbauer, K. Doege, and L.J. Sandell. 1993. Alternate exon usage is a commonly used mechanism for increasing coding diversity within genes coding for extracellular matrix proteins. Matrix 13:457-469. 6. van der Flier, A. and A. Sonnenberg. 2001. Function and interactions of integrins. Cell Tissue Res. 305:285-298. 7. Plow, E.F., T.A. Haas, L. Zhang, J. Loftus, and J.W. Smith. 2000. Ligand binding to integrins. J. Biol. Chem. 275:21785-21788. 8. Schwartz, M.A. and M.H. Ginsberg. 2002. Networks and crosstalk: integrin signalling spreads. Nat. Cell Biol. 4:E65-E68. 9. Rosso, F., A. Giordano, M. Barbarisi, and A. Barbarisi. 2004. From cell-ECM interactions to tissue engineering. J. Cell. Physiol. 199:174-180. 10. LeBaron, R.G. and K.A. Athanasiou. 2000. Extracellular matrix cell adhesion peptides: functional applications in orthopedic materials. Tissue Eng. 6:85-103.
RESEARCH REPORT 11. Lu, L., M.J. Yaszemski, and A.G. Mikos. 2001. Retinal pigment epithelium engineering using synthetic biodegradable polymers. Biomaterials 22:3345-3355. 12. Cutler, S.M. and A.J. Garcia. 2003. Engineering cell adhesive surfaces that direct integrin alpha5beta1 binding using a recombinant fragment of fibronectin. Biomaterials 24:1759-1770. 13. Eliceiri, B.P. 2001. Integrin and growth factor receptor crosstalk. Circ. Res. 89:1104-1110. 14. Anderson, D.G., S. Levenberg, and R. Langer. 2004. Nanoliter-scale synthesis of arrayed biomaterials and application to human embryonic stem cells. Nat. Biotechnol. 22:863-866. 15. Flaim, C.J., S. Chien, and S.N. Bhatia. 2005. An extracellular matrix microarray for probing cellular differentiation. Nat. Methods 2:119-125. 16. Belov, L., O. de la Vega, C.G. dos Remedios, S.P. Mulligan, and R.I. Christopherson. 2001. Immunophenotyping of leukemias using a cluster of differentiation antibody microarray. Cancer Res. 61:4483-4489. 17. Falsey, J.R., M. Renil, S. Park, S. Li, and K.S. Lam. 2001. Peptide and small molecule microarray for high throughput cell adhesion and functional assays. Bioconjug. Chem. 12:346-353. 18. Nimrichter, L., A. Gargir, M. Gortler, R.T. Altstock, A. Shtevi, O. Weisshaus, E. Fire, N. Dotan, and R.L. Schnaar. 2004. Intact cell adhesion to glycan microarrays. Glycobiology 14:197-203. 19. Soen, Y., D.S. Chen, D.L. Kraft, M.M. Davis, and P.O. Brown. 2003. Detection and characterization of cellular immune responses using peptide-MHC microarrays. PLoS Biol. 1:E65. 20. Ito, Y. and M. Nogawa. 2003. Preparation of a protein micro-array using a photo-reactive polymer for a cell-adhesion assay. Biomaterials 24:3021-3026. 21. Ito, Y., M. Nogawa, M. Takeda, and T. Shibuya. 2005. Photo-reactive polyvinylalcohol for photo-immobilized microarray. Biomaterials 26:211-216. 22. Goldring, M.B. 2004. Culture of immortalized chondrocytes and their use as models of chondrocyte function. Methods Mol. Med. 100:37-52. 23. Benz, K., S. Breit, M. Lukoschek, H. Mau, and W. Richter. 2002. Molecular analysis of expansion, differentiation, and growth factor treatment of human chondrocytes identifies differentiation markers and growth-related genes. Biochem. Biophys. Res. Commun. 293:284-292. 24. Lelongt, B. and P. Ronco. 2003. Role of extracellular matrix in kidney development and repair. Pediatr. Nephrol. 18:731-742. 25. Kikkawa, Y., N. Sanzen, H. Fujiwara, A. Sonnenberg, and K. Sekiguchi. 2000. Integrin binding specificity of laminin-10/11: laminin-10/11 are recognized by alpha 3 beta 1, alpha 6 beta 1 and alpha 6 beta 4 integrins. J. Cell Sci. 113:869-876.
530 BioTechniques
26. Gout, S.P., M.R. Jacquier-Sarlin, L. Rouard-Talbot, P. Rousselle, and M.R. Block. 2001. RhoA-dependent switch between alpha2beta1 and alpha3beta1 integrins is induced by laminin-5 during early stage of HT-29 cell differentiation. Mol. Biol. Cell 12:3268-3281. 27. Loeser, R.F., S. Sadiev, L. Tan, and M.B. Goldring. 2000. Integrin expression by primary and immortalized human chondrocytes: evidence of a differential role for alpha1beta1 and alpha2beta1 integrins in mediating chondrocyte adhesion to types II and VI collagen. Osteoarthritis Cartilage 8:96-105. 28. Lotz, M.M., C.A. Burdsal, H.P. Erickson, and D.R. McClay. 1989. Cell adhesion to fibronectin and tenascin: quantitative measurements of initial binding and subsequent strengthening response. J. Cell Biol. 109:1795-1805. 29. Schnapp, L.M., N. Hatch, D.M. Ramos, I.V. Klimanskaya, D. Sheppard, and R. Pytela. 1995. The human integrin alpha 8 beta 1 functions as a receptor for tenascin, fibronectin, and vitronectin. J. Biol. Chem. 270:2319623202.
Received 16 September 2005; accepted 3 January 2006. Address correspondence to: Brigitte Angres NMI Natural and Medical Sciences Institute University of Tübingen Markwiesenstr. 55 72770 Reutlingen, Germany e-mail:
[email protected]
Vol. 40, No. 4 (2006)
