A High-throughput Cell Microarray Platform for Correlative Analysis of Cell Differentiation and Traction Forces

1. Fabrication of Polyacrylamide Substrates

1. Clean glass substrates — either standard microscope slides for endpoint immunofluorescence or glass-bottom 35 mm Petri dishes for TFM — in order to ensure optimal polyacrylamide hydrogel fabrication and integrity during cell culture. Alternatively, use pre-cleaned glass substrates.
   1. Immerse glass substrates in 0.25% v/v Triton X-100 in distilled water (dH₂O). Place substrates on an orbital shaker for 30 min.
   2. Remove Triton X-100 solution and rinse substrates 5 times with dH₂O. Leave substrates immersed in the final rinse and place on an orbital shaker for 30 min.
   3. Remove dH₂O and immerse substrates in acetone. Place substrates on an orbital shaker for 30 min.
   4. Remove acetone and immerse substrates in methanol. Place substrates on an orbital shaker for 30 min.
   5. Remove methanol and rinse substrates 5 times with dH₂O. Immerse substrates in 0.05 N NaOH and place on an orbital shaker for 1 h.
      CAUTION: NaOH is highly caustic and can cause severe skin burns and eye damage. Wear protective gloves, clothing, and eye protection.
   6. Remove NaOH solution and rinse substrates 5 times with dH₂O. Use filtered compressed air to dry substrates and bake at 110 °C on a hot plate until dry (5 – 15 min). Cleaned substrates can be stored at room temperature indefinitely.
2. Silanize clean glass substrates in order to ensure attachment of the polyacrylamide hydrogel.
   1. Immerse clean glass substrates in freshly prepared 2% v/v 3-(trimethoxysilyl)propyl methacrylate (3-TPM) in ethanol. Place substrates on an orbital shaker for 30 min.
      CAUTION: 3-TPM is a combustible liquid. Keep away from heat, sparks, open flames, and hot surfaces and use only in a chemical fume hood.
   2. Remove 3-TPM solution and immerse substrates in ethanol. Place substrates on an orbital shaker for 5 min.
   3. Use filtered compressed air to dry the substrates and bake at 110 °C on a hot plate until dry (5 – 15 min). Silanized substrates can be stored at room temperature for up to 1 month.
3. Option 1: Fabricate polyacrylamide hydrogels on silanized microscope slides for endpoint immunofluorescence.
   1. Prepare a pre-polymer solution in dH₂O with the desired acrylamide/bisacrylamide percentage (w/v) ratio to fabricate substrates with Young's moduli of 4 kPa (4% acrylamide, 0.4% bisacrylamide), 13 kPa (6% acrylamide, 0.45% bisacrylamide), or 30 kPa (8% acrylamide, 0.55% bisacrylamide) and similar porosity, per Wen et al.⁴¹. Vortex solution until clear and filter with a 0.2 µm syringe. Pre-polymer solutions can be stored at 4 °C for 3 months.
      CAUTION: Exposure to acrylamide or bisacrylamide can result in acute toxicity, neurotoxicity, and irritation. Wear protective gloves, clothing, and eye protection.
   2. Prepare a photoinitiator solution of 20% w/v Irgacure 2959 in methanol. This photoinitiator solution cannot be stored and must be prepared fresh each time.
   3. Mix the pre-polymer and photoinitiator solutions in a 9:1 (pre-polymer:photoinitiator) ratio. Optionally degas with a vacuum chamber for 15 min to remove bubbles.
   4. Place silanized slides into a glass drying tray and pipet 100 µL of 9:1 pre-polymer:photoinitiator solution onto each slide. Gently cover each slide with a 22 × 60 mm coverslip while avoiding the creation of bubbles. Note that the coverslip prevents inhibition of the polymerization reaction by oxygen.
   5. Place drying tray in a UV crosslinker and expose slides to 365 nm UV A for 10 min (4 W/m²). Optimize polymerization time as needed. Longer exposures risk difficulty removing the coverslip due to overpolymerization. Shorter exposures risk underpolymerization and low hydrogel stability.
   6. Immerse hydrogels in dH₂O for 5 min. Remove coverslips with a razor, taking care not to damage the polymerized hydrogels.
   7. Leave hydrogels in dH₂O at room temperature for 1 – 3 d, changing dH₂O daily. Dehydrate hydrogels at 50 °C on a hot plate until dry (15 – 30 min) and store at room temperature for up to 3 months.
4. Option 2: Fabricate fluorescent bead-containing polyacrylamide hydrogels on silanized 35 mm glass-bottom Petri dishes for live evaluation of cell–substrate interactions using TFM.
   1. Sonicate a stock solution of 1 µm fluorescent beads for 15 min to disperse aggregates.
   2. Prepare a pre-polymer solution in dH₂O with the desired acrylamide/bisacrylamide percentage (w/v) ratio to fabricate substrates with Young's moduli of 4 kPa (4% acrylamide, 0.4% bisacrylamide), 13 kPa (6% acrylamide, 0.45% bisacrylamide), or 30 kPa (8% acrylamide, 0.55% bisacrylamide) and similar porosity, per Wen et al.⁴¹. Vortex solution until clear and filter with a 0.2 µm syringe. Pre-polymer solutions can be stored at 4 °C for 3 months.
      CAUTION: Exposure to acrylamide or bisacrylamide can result in acute toxicity, neurotoxicity, and irritation. Wear protective gloves, clothing, and eye protection.
   3. Add fluorescent beads to the pre-polymer solution at a final concentration of 0.2% v/v and vortex to mix.
   4. Prepare a photoinitiator solution of 20% w/v Irgacure 2959 in methanol. This photoinitiator solution cannot be stored and must be prepared fresh each time.
   5. Mix the pre-polymer/bead and photoinitiator solutions in a 9:1 (pre-polymer/bead:photoinitiator) ratio. Optionally degas with a vacuum chamber for 15 min to remove bubbles.
   6. Place silanized 35 mm glass-bottom Petri dishes into a glass drying tray and pipet 20 µL of 9:1 pre-polymer/bead:photoinitiator solution onto the center of each dish. Gently cover each slide with a 12 mm circular coverslip while avoiding the creation of bubbles. Note that the coverslip prevents inhibition of the polymerization reaction by oxygen.
   7. In order to distribute the fluorescent beads to the surface of the hydrogel, invert the dishes and leave at room temperature for 20 min, per Knoll et al.⁴².
   8. While still inverted, expose dishes to 365 nm UV A for 10 min (4 W/m²). Optimize polymerization time as needed. Longer exposures risk difficulty removing the coverslip due to overpolymerization. Shorter exposures risk underpolymerization and low hydrogel stability.
   9. Immerse hydrogels in 0.1 M 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) buffer and leave at room temperature in the dark overnight. Remove coverslips carefully with a razor, taking care not to damage the polymerized hydrogels.
   10. Dehydrate hydrogels at 50 °C on a hot plate until dry (15 – 30 min). Hydrogels can be stored at room temperature in the dark for 3 months.

2. Fabrication of Arrays

1. Prepare buffers to print biomolecules. Use the printing buffer appropriate to the biomolecules of interest. Growth factor (GF) printing buffer is broadly suitable for other classes of molecules, such as cell–cell ligands.
   1. To prepare 2× ECM protein printing buffer, add 164 mg of sodium acetate and 37.2 mg of ethylenediaminetetraacetic acid (EDTA) to 6 mL dH₂O. Vortex and incubate at 37 °C to thoroughly solubilize. After solubilization, add 50 µL of pre-warmed Triton X-100 and 4 mL of glycerol. Vortex and incubate at 37 °C again to solubilize. Add 40 – 80 µL of glacial acetic acid, titrating to adjust the pH to 4.8. 2× ECM protein printing buffer can be stored at 4 °C for 1 month.
      CAUTION: Acetic acid is flammable and corrosive. Wear protective gloves, clothing, and eye protection.
   2. To prepare 2× GF printing buffer, add 105.5 mg sodium acetate and 37.2 mg EDTA to 6 mL phosphate-buffered saline (PBS). Vortex and incubate at 37 °C to thoroughly solubilize. After solubilization, add 100 mg 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS) and 4 mL of glycerol. 2× GF protein printing buffer can be stored at 4 °C for 1 month.
2. Prepare source plate.
   1. In a 384-well V-bottom microplate, combine equal volumes of 2× printing buffer with each biomolecule solution at double the target concentration.
      NOTE: An appropriate target concentration for most common ECM proteins is 250 µg/mL while target concentrations for other types of arrayed factors vary depending on retention in the hydrogel and biological function. The total volume in each well can be as low as 5 µL and need not be more than 15 µL. In addition to the biomolecule combinations of interest, include an arrayed fluorescent marker in order to facilitate downstream image analysis. Use rhodamine-conjugated dextran (2.5 mg/mL).
   2. Mix each well thoroughly by pipetting, taking care not to generate bubbles. Centrifuge the source microplate for 1 min at 100 × g. Fabricate microarrays using source plates prepared on the same day and stored at 4 °C until microarray fabrication.
3. Clean pins according to the instructions of the manufacturer before each microarray fabrication run. Load clean pins directly into the printhead of the microarrayer.
4. Prepare microarrayer and program using the manufacturer's software. Although the steps below are partly specific to the particular microarrayer used here, the operation of most microarrayers is similar.
   1. Turn on the humidifier unit, adjust the set point to 65% RH (non-condensing), and wait until the rheometer matches the set point. Place source plate in the appropriate adaptor.
   2. Dehydrate hydrogel substrates at 50 °C for 15 min and place into the appropriate adaptor. The microarrayer has adaptors for both microscope slides and microplates. For arraying 35 mm glass-bottom Petri dishes, load the dishes into a 6-well microplate and place the microplate into the microplate adaptor on the arrayer.
   3. Adjust the parameters of the program to accurately reflect the layout of the source plate, array design, and desired format (e.g., microscope slide or microplate containing 35 mm Petri dishes). Include wash steps using both water and dimethyl sulfoxide (DMSO) between each condition in order to prevent carry-over and cross-contamination.
   4. Start array fabrication, checking no less frequently than once an hour that the humidity has not dropped below 65% RH (non-condensing) and that the pins are not clogged. If the humidity has dropped unexpectedly, pause arraying to fill the humidifier and clear associated tubes of condensation. If the pins are clogged, pause arraying to clean the pins or otherwise replace with pre-cleaned pins. Note that it is possible to array multiple types of biomolecules sequentially on the same substrates provided sufficient drying time (i.e., 4 h to overnight).
   5. Once the program is complete, place fabricated arrays in a slide box or microplate covered with aluminum foil at room temperature and 65% RH (non-condensing) overnight. Note that it may be necessary to evaluate array quality and retention using general protein stains or immunofluorescence; see Brafman et al. for more details²⁵.

3. Cell Culture and Assay Readout

1. The day after fabrication, place arrayed substrates in 4-chambered dishes (microscope slides) or 6-well microplates (Petri dishes) and immerse in 1% v/v penicillin/streptomycin in PBS; use 4 mL for slides and 3 mL for dishes. Expose to UV C for 30 min. Exchange penicillin/streptomycin solution for cell culture media.
2. Collect and count cells. Seed onto arrays at 500 × 10³ -  2 × 10⁶ cells/array at 4 mL per microscope slide and 3 mL per 35 mm Petri dish. Incubate array cultures at 37 °C and 5% CO₂ for 2 – 24 h or until the formation of well-populated cell islands. Adjust both seeding density and time as needed for your cells and particular application. Underseeding (i.e., low density or seeding time) may result in poor array population and skewed biological outcomes. Overseeding (i.e., high density or seeding time) may result in reduced array integrity due to island detachment.
3. After allowing for formation of cell islands, wash array cultures twice with pre-warmed cell culture media; again use 4 mL for slides and 3 mL for dishes. Optionally add appropriate controls and treatments (e.g., small molecule inhibitors, growth factors, etc.) of interest to the biological system. Change the media of the arrays every 1 – 2 d in order to maintain the concentration of any treatments. Evaluate cell marker expression and cell function by immunofluorescence or cell–substrate interactions by TFM within 1 – 5 d of initiating array cultures — see Option 1 and Option 2 below.
4. Option 1: Perform endpoint immunofluorescence. Note that immunofluorescence of some proteins may require more stringent permeabilization using methanol, ethanol, or HCl. Due to potential damage to arrays, evaluate and optimize each permeabilization protocol before use in larger-scale experiments.
   1. Aspirate cell culture media from array slides in 4-chambered dishes and add 4 mL/slide of freshly prepared 4% v/v paraformaldehyde (PFA) in PBS. Incubate for 15 min at room temperature.
      CAUTION: Exposure to PFA can result in acute toxicity and can also irritate or corrode skin on contact. Wear protective gloves, clothing, and eye protection and use only in a chemical fume hood.
   2. Aspirate PFA solution and wash each slide 3 times with 4 mL of PBS. At this point, fixed slides can be stored at 4 °C for 1 week. It is advisable, however, to continue through immunolabeling and mounting on the same day as fixing in order to ensure array integrity.
   3. Aspirate PBS and add 4 mL/slide of 0.25% v/v Triton X-100 in PBS. Incubate for 10 min at room temperature.
   4. Aspirate Triton X-100 solution and wash each slide 3 times with 4 mL of PBS. Add 4 mL/slide of 5% v/v serum matched to the species of the secondary antibody (e.g., donkey serum for donkey secondary antibodies) in PBS and incubate at room temperature for 1 h.
   5. Thoroughly remove the blocking solution from each slide. Add 500 µL/slide of primary antibody diluted in 5% v/v serum in PBS. This volume is sufficient to cover arrays for both 1 h incubations at room temperature as well as overnight incubations at 4 °C.
   6. Wash each array slide 3 times with 4 mL of PBS. Thoroughly remove the final wash and add 500 µL/slide of the appropriate secondary antibody diluted in 5% v/v serum in PBS.
   7. Wash each array slide 3 times with 4 mL of PBS. Wash briefly with dH₂O before carefully removing array slides from solution using forceps. Use a laboratory tissue to wick or dry residual dH₂O.
   8. Pipet 100 µL of mounting solution with DAPI across the slide while visually confirming complete coverage of the entire array.
   9. Place a 22 × 60 mm coverslip over the slide to mount. Seal the edges of the coverslip with clear nail polish. Store in the dark at 4 °C until imaging, no earlier than the next day.
   10. Image entire arrays using either a microarray scanner or inverted fluorescent microscope equipped with a robotic stage. Microarray scanners provide quicker readout but may require Cy3- or Cy5-compatible fluorophores and are often of limited resolution at the order of single cells (i.e., 1 – 10 µm). Fluorescent microscopes provide the option to use a variety of fluorescent channels and higher resolution (<1 µm, ~ 100× total magnification) but provide slower readout depending on the quality of the robotic stage and magnification/objective.
   11. Save captured images of entire arrays from either method as TIFF files in order to prevent data compression or loss associated with other file formats (e.g., JPG).
5. Option 2: Perform live evaluation of cell–substrate interactions using TFM.
   1. Prepare a solution of 1% v/v bovine serum albumin (BSA) and 1% v/v sodium dodecyl sulfate (SDS) in PBS to dissociate cells from substrates during TFM.
   2. Move 35 mm Petri dishes containing array cultures to an incubated (37 °C, 5% CO₂), inverted fluorescent microscope with a robotic stage for TFM measurements.
      1. In one dish, mark the positions (X-coordinate, Y-coordinate) and focus planes (Z-coordinate) of individual cell islands using phase contrast microscopy.
      2. Switch to far red fluorescent microscopy to visualize the beads. Return to each of the positions saved in the previous step and correct the Z-coordinate of the focus plane so that only the first layer of beads below the cell island is in focus. Save the new coordinates and proceed to automated imaging of all cell islands to capture pre-dissociation phase contrast and far red fluorescent images.
   3. Carefully add 150 µL of BSA/SDS solution to the dish and wait 5 min to allow for complete cell dissociation from substrate; monitor cell dissociation using phase contrast microscopy.
   4. After the cell islands have been dissociated from the substrate, return to the marked positions and check that the first layer of beads are still in focus. If these beads are out-of-plane due to deformation induced by cell-generated traction, then correct the Z-coordinate of the focus plane so that they are again in focus. Save the corrected Z-coordinates and repeat automated imaging of all islands to capture post-dissociation far red fluorescent images.
   5. Repeat Steps 3.5.1 – 3.5.4 for the remaining dishes.

4. Analysis of Data

1. Analysis of immunofluorescence data.
   1. Process acquired array images. Split composite array images into files containing individual channels (i.e., red, blue, or green) and convert to 8-bit TIFF images^43,44. Apply binning (e.g., 2×2 or 4×4) to reduce image size to ~ 32 megapixels per channel to reduce memory requirements during downstream single-cell analysis of entire array images. See Supplemental Code File titled "array_processing.ijm" for an ImageJ macro implementation of these array processing steps.
   2. Note the coordinates in pixels of the top left, bottom left, and bottom right rhodamine-conjugated dextran markers or arrayed conditions. Use these coordinates to rotate the 8-bit TIFF images to be perfectly vertical and, later, to annotate output from the single-cell analysis with specific arrayed conditions. See Supplemental Code Files titled "rb_array_rotater.ijm", "rg_array_rotater.ijm", "rgb_array_rotater.ijm", and "array_gridding.ijm" for implementations of these array rotating and gridding steps.
   3. Perform single-cell analysis of binned, rotated 8-bit TIFF images in CellProfiler (version 2.1.1)⁴⁵ using the following modules: IdentifyPrimaryObjects, IdentifySecondaryObjects, and MeasureObjectIntensity. IdentifyPrimaryObjects identifies nuclei, IdentifySecondaryObjects identifies immunolabels associated with each cell nucleus, and MeasureObjectIntensity provides quantifications for both nuclear labels and immunolabels.
      1. Output single-cell data from all three modules as a CSV file by channel using the ExportToSpreadsheet module to facilitate later downstream analysis. See Supplemental Code Files titled "b_array_image_analysis.cppipe", "gb_array_image_analysis.cppipe", "rb_array_image_analysis.cppipe", and "rgb_array_image_analysis.cppipe" for CellProfiler pipelines implementing these steps for image sets containing red, green, or blue channels.
   4. To transform data to account for experimental variability and non-Gaussian single-cell distributions, apply quantile normalization by biological replicate⁴⁶. This process generates a shared distribution across replicates and enables unbiased comparisons of changes in immunolabel intensity. Furthermore, unlike Z-scoring and other parametric methods, quantile normalization is non-parametric and does not assume a particular distribution of data, allowing for more representative analyses of single-cell behavior as a function of arrayed condition.
   5. Plot data and interpret. Depending on the biological system and hypothesis, calculate and plot one or more of the following ensemble measures for each arrayed condition:
      1. Calculate and plot cells per island as a combined measure of adhesion and survival over the course of the experiment.
      2. Calculate and plot the quantile-normalized immunolabel intensity as a measure of cell fate or function.
      3. Calculate and plot the percentage of cells positive for an immunolabel as determined by intensity above a consistent threshold, usually 2 s.d. above the mean intensity of a negative control.
      4. Alternatively, plot distributions of immunolabel intensity in order to examine and categorize single-cell behavior as a function of arrayed condition. These distributions can further be characterized using measures of central tendency (mean, median, mode) and variation (variance, coefficient of variation, Fano factor) and hypothesis testing methods such as the Kolmogorov-Smirnov test.
2. Analysis of TFM data. Here is described an approach incorporating a previously developed algorithm by Butler et al. and Wang et al.^40,47.
   1. Use ImageJ to batch convert the images to 8-bit TIFF files. Apply pixel-averaged binning (e.g., 2 × 2) to reduce the computational cost and time of downstream analysis. As algorithms for TFM have focused largely on single-cell analysis, the large cell–substrate interface of the islands (~ 17.5 × 10³ µm²) in comparison to the cell–substrate interface of a single cell (75 µm²) necessitates the binning step.
   2. Input the captured phase contrast and far red fluorescent images (both pre-dissociation and post-dissociation) into a scientific programming environment such as MATLAB and process using the previously developed algorithms of Butler et al. and Wang et al.^40,47.
      1. Select three regions distant from the cell island. These regions are used to account for displacements due to image or sample drift.
      2. Provide the factor to convert from pixels to micrometers (e.g., 0.454 pixels/µm), the Young's modulus of the substrate (e.g., 13 kPa), and the Poisson's ratio (e.g., 0.48 for the polyacrylamide gels described here).
      3. For each island, draw a boundary around the periphery to define the geometrical constraints; all forces outside of this boundary are zeroed. This constrained system is reasonable given the large distance (i.e., 450 µm) between islands.
   3. Calculate the root-mean-square traction stress and contractile moment for each island. The contractile moment is a measure of the residual stress across the cell island and has been shown to reflect the strength of cell–cell interactions⁴⁸. For each arrayed condition, average root-mean-square values over multiple islands and biological replicates and calculate associated variance for hypothesis testing. It is also possible to average the distribution of stresses or moments over many islands to provide a representative map of both measures as a function of geometry, e.g., distance from the center of the island.