Multiwell stiffness assay for the study of cell responsiveness to cytotoxic drugs

Gel preparation and multiwell stiffness assay assembly

Polyacrylamide (PA) gels of desired stiffness were prepared by mixing acrylamide and bisacrylamide crosslinker to a desired final concentration (Table 1) and assembled in a 96-well plate format (Figure 1). To prepare the gels, Irgacure 2959 at a final concentration of 0.1% w/v was added to all gel solutions. A 500 μl aliquot of a given gel solution was pipetted onto the hydrophilic side of a GelBond sheet. A second GelBond sheet, hydrophobic side down, was placed over the gel solution. Parafilm strips were used as separators to define the thickness of the gel. A glass plate was positioned on top of the GelBond/gel “sandwich”, to assure uniform distribution of the gel solution onto the GelBond sheet. The gels were polymerized under UV (365 nm, Black-Ray UV Bench Lamp, UVP, Upland, CA) for 10 min, after which the glass was removed and the top GelBond sheet was peeled off. The GelBond/gel constructs were left to dry overnight at room temperature and then cut into squares having dimensions equal to those of the well area of square 96-well plates (6.4 × 6.4 mm). To assemble the assay, the GelBond/gel squares were placed, gel side up, into the wells of the 96-well plate on top of a PDMS droplet. The PDMS acted as a biocompatible glue attaching the gels firmly to the bottom of the wells. The multiwell plate was then left in the incubator for 4 h to allow the PDMS to solidify.

Open in a separate window

Figure 1

Schematic of polyacrylamide gel preparation and incorporation into a multiwell plate. The gels were formed between two GelBond flexible sheet supports, where the gel is attached to the bottom support. The top sheet is then peeled off; the gel is dried and cut into pieces that conform to the wells of a multiwell plate. The gels are glued to the bottom of the well with a drop of PDMS, then collagen-coated, and sterilized via UV irradiation.

AFM measurements

The Young’s Modulus (E) of the polyacrylamide gels was measured by AFM, using a Bioscope Catalyst® NanoScope® V device (Bruker, Santa Barbara, CA) attached to an inverted optical microscope (IX71, Olympus, Japan). Polyacrylamide gels were indented with a V-shaped cantilever (MSCT, pyramidal tipped, nominal k = 0.01-0.03 N/m; Bruker) whose spring constant was calibrated by the thermal fluctuations method (Butt and Jaschke 1995; Hutter 1993). The relationship between the photodiode signal and the cantilever deflection was computed from the slope of the force displacement curve obtained at a bare region of the coverslip (without gel sample). The force (F) on the cantilever was computed using Hooke’s law (F = kd, where d is the cantilever deflection).

We probed five different regions for every gel. At every region, we acquired five force–displacement curves (F–z, z being the displacement of the piezotranslator) while the piezotranslator was ramped forward and backward at constant speed (1 Hz, 5 μm amplitude, ~1 μm maximum indentation). Force-indentation data were analysed with the four-sided pyramidal indenter model (Rico et al. 2005):

where E is the Young’s modulus, υ is the Poisson’s ratio, θ is the semi-included angle of the pyramidal indenter, and β is the indentation depth. The parameter ν is assumed to be 0.5 (the water-filled hydrogel essentially is incompressible), and the indentation depth is calculated as δ = z–z₀–d, where z₀ is the tip-gel contact point. E and z₀ were estimated by least-squares fit of this equation to the F-z curve recorded on each gel point (Alcaraz et al. 2003).

Collagen coating

Collagen coating was applied to all gel types to provide cell attachment sites to the otherwise inert polyacrylamide gels. To apply the collagen coating, the gels were first derivatized with Sulfo-SANPAH dissolved in DMSO:PBS at a ratio of 4:96. Briefly, to prepare the sulfo-SANPAH solution, the reagent was first dissolved in DMSO at 10% w/v and stored at −80°C in 20 μl aliquots until further use. Each aliquot was thawed and diluted to 0.5% w/v in deionized water immediately before use. This was then placed on top of the gels at 50 μl/well and activated by exposure to high intensity UV light for 5 min. The unreacted crosslinker was removed by a double wash with PBS, after which 50 μl of 0.2 mg/ml collagen Type I was added to each well and allowed to react for at least 2 h at room temperature. The gels were sterilized under UV for 2 h prior to cell seeding.

Fluorescent collagen was used to confirm collagen coating uniformity of the PA gel surface. For fluorescent labeling, collagen was diluted in 0.1 M Na₂CO₃/0.5 M NaCl buffer of pH 9.3 to a final concentration of 1 mg/ml. The collagen was then reacted with Cy5 for 30 min at 4°C under constant stirring, following the manufacturer’s procedures. Unreacted dye was removed by dialysis in 2 mM HCl with three changes of dialysis buffer. Final collagen concentration was measured by circular dichroism (J810, Jasco, Easton, MD). Fluorescent collagen was stored in 2 mM HCl at 4°C until further use. To attain fluorescent images of the gels as well as the collagen coating, fluorescent latex beads were added to the PA gel solution prior to gelation. Cross-sectional fluorescent images of the resulting gels were obtained with Zeiss LSM 510 META microscope.

Cell culture and imaging

Cancer cell lines of different origin were used (Table 2). All cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum and incubated in a humidified incubator at 37°C and 5% CO₂. Cells were harvested by a 5 min exposure to trypsin/EDTA and plated onto a 96-well stiffness assay plate at 10⁴ cells/well. Cells were imaged under a phase contrast on an EVOS microscope (AMG, Mill Creek, WA), using a 10× or 20× objective. Cell spreading area was analyzed with the Micron EVOS proprietary software after manually tracing the boundary of each cell from phase contrast images.

Cell proliferation and adherence

To perform cell proliferation and adherence experiments, MTS assays were conducted 3-5 h post cell seeding and 72 h post seeding for each cell type. The comparative cell proliferation for each cell type was then calculated using the MTS results at 72 h post seeding, and comparative cell adherence was calculated using the MTS results at 3-5 h post seeding. To aid comparison between cell types, the MTS results were normalized as follows: [OD₄₉₀ for 1 kPa gel] / [OD₄₉₀ for 100 kPa gel]. MTS was used as recommended by the manufacturer. Briefly, 20 μl of the MTS solution for every 100 μl of medium was added directly to the wells and the cells were then incubated for up to 1 h in a humidified incubator at 37°C and 5% CO₂. Prior to absorbance measurements, 80 μl of the MTS:medium solution was transferred from each well into a well of a new 96-well plate in order to avoid background absorbance from the gels. Absorbance in the no-gel 96-well plate was measured at 490 nm (Spectra Max Plus, Molecular Devices, Sunnyvale, CA).

Drug screening

Paclitaxel, a microtubule-stabilizing agent, was the drug of choice. Cells were grown for 4 – 24 h prior to paclitaxel administration. Paclitaxel was administered at various concentrations between 0 and 200 nM. No-cell negative control and a no-growth control, where cell growth was halted by addition of sodium azide at 0.2% final concentration just prior to drug administration were also used. Viable cell numbers were evaluated after 72 hours of paclitaxel exposure using an MTS assay. Dose response curves and drug IC₅₀ (50% inhibition concentration), were analyzed with GraphPad Prism software by fitting a four-parameter Hill slope to the normalized dose-response data. Percent cell viability was calculated as (OD₄₉₀ of cells at X nM Paclitaxel - OD₄₉₀ of cells at 200 nM Paclitaxel)/(OD₄₉₀ of cells at 0 nM Paclitaxel - OD₄₉₀ of cells at 200 nM Paclitaxel) ×100.

Cell growth and proliferation on multiwell stiffness assay

Various cell types have been reported to proliferate more rapidly on stiffer substrates (Klein et al. 2009; Tilghman et al. 2010; Wells 2008) where matrix stiffening has also been implicated to play a role in tumorogenesis (Levental et al. 2009; Paszek et al. 2005). The majority of the studies on the stiffness dependence of cell proliferation have been focused on a single cell type (Subramanian and Lin 2005; Yoshikawa et al. 2011). The true contribution of our assay is allowing a rapid, low-cost, and systematic testing of multiple parameters and conditions. Thus, to explore cell proliferation dependency on stiffness in a systematic manner, while validating the utility of the developed assay, we examined the proliferation of several cancer cell lines 72 h post-seeding as a function of matrix stiffness (Figure 3). Viable cell number was measured at 72 h post-seeding with an MTS assay, which measures the metabolic activity of live cells. To aid comparison between cell types, the results were normalized by the average proliferation (OD₄₉₀) on the 100 kPa PA gels for each cell type. The chosen cell lines represent soft solid tumors for which the native healthy tissues have stiffness comparable to the stiffness range covered by our assay (Dall et al. 1993; Levental et al. 2009; Miller et al. 2000; Myers et al. 2008; Yeh et al. 2002). Since the developed assay ultimately depends on cell-matrix interactions, all cell types were also adherent cancer cells for which the ECM microenvironment plays an important role in cell growth and proliferation (Sounni and Noel 2013).

Open in a separate window

Figure 3

Cell proliferation is affected by gel stiffness in a cell-dependent manner. Cell proliferation was measured by an MTS assay on 1 kPa gels (gray bars) or 100 kPa gels (dashed line). To aid comparison between different cell lines, cell proliferation was normalized by the cell proliferation on the 100 kPa gels for each cell type (signified by the dashed line); n ≥ 3. Asterisks designate statistical differences from cell proliferation on the 100 kPa gels (p < 0.05).

The results shown in Figure 3 indicate that cell proliferation is affected by matrix stiffness in a cell type-dependent manner. All cell types, except SW620 and PC3, showed higher proliferative capacity on stiffer substrates. In general, we saw no correlation between the stiffness dependence of cell proliferation and the tissue of origin. For example, the two colon cancer cell lines HT29 and SW620 exhibited stiffness-dependent and stiffness-independent growth profiles, respectively. Overall, our current results correlate with trends discussed in the literature, where other researchers have shown that most cell lines are responsive to the stiffness of the substrate: for example, MDA-MB-231 cells exhibit >20 fold increase in cell numbers over 5 days on 9.6 kPa gels as opposed to 1-5 fold increase on 0.15 kPa gels (Tilghman et al. 2010) and HepG2 cells show a 12-fold higher proliferative index on 12 kPa PA gels versus 1 kPa gels (Schrader et al. 2011).

To confirm that the results shown in Figure 3 are indeed due to proliferation rather than variable cell adherence on substrates of different stiffness, we seeded the cells on the multiwell stiffness assay for 3-5 hours and measured cell metabolic activity via an MTS assay (Figure 4) but, in order to measure only the adhered cells, the media was changed prior to using the MTS assay. Our results indicate that at 3-5 hours post-seeding all cell types adhered equally well on PA gels of different stiffness. These results were expected, as cell adherence correlates with the existence of a sufficient number of integrin binding sites, influenced by the density of the collagen coating, which should be comparable for all gel types. Similar cell adhesion efficiency to collagen-coated PA gels irrespective of elastic moduli was also shown by Tilghman et al. for A549 and MDA-MB-231 cancer cells (Tilghman et al. 2010).

Open in a separate window

Figure 4

Cell adherence is not affected by gel stiffness. Cell adherence was measured by an MTS assay 3-5 h post cell seeding on 1 and 100 kPa gels; To aid comparison between different cell lines, cell adherance was normalized by the cell adherance on the 100 kPa gels for each cell type (signified by the dashed line); n = 3. No significant differences were noted between the OD measurements for gels of different stiffness (p < 0.05).

Cell spreading area

Most, but not all, of the cell lines showed greater cell spreading on stiffer than on softer substrates. The cell lines SW620 and PC3, for which proliferation was independent of stiffness, were also irresponsive to stiffness with respect to cell spreading. Cell lines that demonstrated rigidity-dependent growth also showed rigidity-dependent spreading, with the exception of HT29 (Figure 5, Supplemental Figure 4). Both ACHN and MDA-MB-231 cell lines exhibited >2 fold increase in mean spreading area on PA gels of 100 kPa as opposed to gels of 1 kPa, SY5Y, HepG2, HT29, and 786-0 all showed ~1.5 fold increase, while Hela showed only ~1.2 fold increase. Our findings that cell response to stiffness differs between cell lines are in good agreement with the literature (Yeung et al. 2005). For example, Tighman et al., reported no change in cell spreading area for PC3 cells, but >2-fold increase in cell spreading area on 4.8 kPa versus 0.15 kPa collagen-coated PA gels for MDA-MB-231 cells. Further, Lam et al. observed >2 fold increase in average neurite length of SH-SY5Y cells on collagen-coated PA gels of 50 kPa versus 1 kPa (Lam et al. 2010).

Open in a separate window

Figure 5

Mean cell spreading area is affected by PA gel stiffness. All cell lines, with the exception of HT29, PC3, and SW620, exhibited significant increase in cell spreading area on 100 kPa PA gels as opposed to 1 kPa PA gels. Cell spreading area was measured 72 h post cell seeding; n ≥3. Asterisks designate significant differences (p < 0.05).

Compared to the other cell lines, the breast cancer cell line MCF-7 exhibited a unique response. The cells formed 3D cell aggregates on the soft 1 kPa matrices and 2D cell sheets on the stiffer 10 kPa and 100 kPa matrices (Supplemental Figure 4), which prevented us from measuring cell spreading area of individual cells. Similar behavior for breast cancer cells MCF-10A cultured on 400 Pa, where they formed 3D aggregates, and 1250 Pa PA gels, where they formed 2D aggregates, was observed by Shebanova et al. and was attributed to decreased mobility and increased cell-cell interactions on the soft substrates (Shebanova and Hammer 2012). The spreading area of the MCF-7 cell aggregates after 72 h of culture was ~20 times smaller for the 1 kPa gel than for the 10 and 100 kPa gels, possibly due to higher proliferation on the stiffer gels as observed on Figure 3.

We thank Dr. Raimon Sunyer for valuable technical discussion and practical help with performing atomic force microscopy measurements of gel stiffness, Dr. Elena Makareeva and Dr. Sergey Leikin for help with the fluorescent collagen labeling, Danielle Ferguson for technical assistance, and Colleen Cole for image analysis assistance. This work was supported by funds from the Intramural Research Program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health.