Mechanical Stiffness Grades Metastatic Potential in Patient Tumor Cells and in Cancer Cell Lines

Cancer cells are defined by their ability to invade through the basement membrane, a critical step during metastasis. While increased secretion of proteases, which facilitates degradation of the basement membrane, and alterations in the cytoskeletal architecture of cancer cells have been previously studied, the contribution of the mechanical properties of cells in invasion is unclear. Here, we applied a magnetic tweezer system to establish that stiffness of patient tumor cells and cancer cell lines inversely correlates with migration and invasion through three-dimensional basement membranes, a correlation known as a power law. We found that cancer cells with the highest migratory and invasive potential are five times less stiff than cells with the lowest migration and invasion potential. Moreover, decreasing cell stiffness by pharmacologic inhibition of myosin II increases invasiveness, whereas increasing cell stiffness by restoring expression of the metastasis suppressor TβRIII/betaglycan decreases invasiveness. These findings are the first demonstration of the power-law relation between the stiffness and the invasiveness of cancer cells and show that mechanical phenotypes can be used to grade the metastatic potential of cell populations with the potential for single cell grading. The measurement of a mechanical phenotype, taking minutes rather than hours needed for invasion assays, is promising as a quantitative diagnostic method and as a discovery tool for therapeutics. By showing that altering stiffness predictably alters invasiveness, our results indicate that pathways regulating these mechanical phenotypes are novel targets for molecular therapy of cancer. Cancer Res; 71(15); 5075–80. ©2011 AACR.

The spread of cancer from its primary site to distant organs, the “invasion-metastasis cascade,” is the main cause of cancer death and invasion of cells into the lymphatics and blood vessels is a crucial step in metastasis, correlating with a poorer patient prognosis (1). Hallmarks of invasion include secretion of proteases, alterations in adhesion receptors, and changes in cell morphologic and migratory properties (2). Drugs targeting the metastatic cascade, including the matrix metalloproteinases (MMP), which degrade the extracellular matrix, or the migratory machinery, are being evaluated in clinical trials (3) but results have been disappointing potentially because of the complexity and redundancy of the metastatic cascade.

Cell stiffness has been postulated to play roles in transmigration of cancer cells through a basement membrane (4). A variety of biophysical techniques including membrane stretching, atomic force microscopy, optical traps, and micropipette aspiration have been used to probe the mechanical properties of cells (5). These techniques use ferromagnetic or superparamagnetic beads to attach to membrane receptors and are followed by application of either a twisting or a pulling motion to the bead and thus to the cell via an electromagnet (6). Magnetic tweezers, like the one described here, provides for a wide range of force magnitudes (10 pN–10 nN), the ability to probe individual cells and to conduct measurements in minutes to understand the time-dependent development mechanical state of a cell (7). While cancer tissue has been found to be generally stiffer than normal tissue, recent studies have shown that cancer cells themselves are more compliant than normal cells (8). However, the extent of the correlation between mechanical properties and specific aspects of cancer progression has not been determined. Specifically, whether modification of cell stiffness might alter key aspects of metastasis such as invasion is unclear. The measurement of metastatic potential through 2-dimensional substrates (migration; movement through membrane pores) and 3-dimensional (3D) motion through tissue matrix material (invasion) are common assays used in cancer cell biology and applied as population studies over longer time durations. Therefore, it is of broad interest from an analytical, diagnostic, and cancer biology context to understand how a mechanical phenotype measure could be utilized relative to current metastatic potential assays. To address these questions, we have used a magnetic tweezer system in parallel with Matrigel invasion assays to examine the specific correlation between mechanical properties of cancer cells and their metastatic potential, specifically their invasive potential.

Human ovarian cancer cell lines, OVCA429, IGROV, SKOV3, HEY, DOV13, OV2008, and Ovca420 and ovarian cancer stable cells lines, Ovca429Neo, Ovca429TβRIII were cultured, and TβRIII were cultured, derived, and characterized as previously described (9). Antibody to phosphorylated myosin light chain (pMLC) was obtained from Cell Signaling Technologies (catalogue no. 3671) and pan-cytokeratin antibody was obtained from Santa Cruz (catalogue no. 81714).

Primary short-term epithelial ovarian cancer cell cultures were established from the ascites of patients with stage III or IV epithelial ovarian cancer as described previously (10). Cells were seeded and grown on 10 μg/mL fibronectin-coated culture dishes in Roswell Park Memorial Institute media containing 20% FBS and 1% penicillin/streptomycin solution at 37°C in 5% CO₂. Adhered cells were subject to limited dispase digestion for the first passage to remove fibroblasts and stained with a pan-cytokeratin antibody to confirm epithelial origin.

Immunofluorescence was conducted essentially as described previously (9) and images were obtained using a Nikon inverted microscope.

Cancer cells were seeded at a cell density of 25,000 to 70,000 on either Matrigel-coated or uncoated filters and allowed to invade for 18 to 24 hours toward 10% FBS in the lower chamber. Cells invading and migrating through the Matrigel layer were visualized and counted as described (9). Percent cell migration or invasion was determined as the fraction of total cells that invaded through the filter. Blebbistatin (100 μM; ref. 10) was added to the top chamber of the transwell and migration and invasion allowed to proceed. Each assay was set up in duplicate, and each experiment was conducted at least 3 times with 4 random fields from a 10× magnification analyzed for each membrane.

The 3D force microscope (3DFM; ref. 11) was used for applying controlled and precise 60 to 100 pN local force (Supplementary Fig. S1) on 2.8 μm magnetic beads (DYNAL Biotech) coated with fibronectin (Sigma Aldrich). Briefly, cells were plated on coverslips followed by addition of the beads. Cells and beads were incubated for 30 minutes followed by force application and resultant bead displacements were recorded and analyzed. The displacement of the beads was recorded with high-speed video camera (JAI Pulnix) and tracked using Video Spot Tracker (http://cismm.cs.unc.edu). The mean creep compliance was calculated from the tracked displacements as described. Spring constants were derived by fitting the compliance curves to a Jeffrey's model for viscoelastic liquids (Supplementary Fig. S1). For pharmacologic experiments, blebbistatin (100 μM) was added to the cells for 30 minutes, before addition of the beads and the reagent left in for the remainder of the experiment.

The invasiveness and migratory capacity of a panel of ovarian cancer cell lines and primary cells derived from ascites of patients with advanced stage ovarian cancer was determined by using transwell assays in the presence or absence of reconstituted Matrigel (Supplementary Methods, Supplementary Fig. S2). Although both ovarian cancer cell lines and primary cancer cells were able to invade through Matrigel, the degree of invasiveness varied widely among individual lines, with the most invasive and migratory cell line, HEY, being 2 orders of magnitude more invasive (I_HEY = 0.85%, I_IGROV = 0.006%) than the least migratory and invasive cell line, IGROV (Fig. 1A). Similarly, while primary cells were obtained from patients at either stage III or stage IV disease (Supplementary Table S1), the most invasive primary cell line, OV207, was an order of magnitude more invasive than the least invasive primary cell line, OV445 (Fig. 1B; I_OV207 = 0.193%, I_OV445 = 0.006%). Mechanical properties of the cancer cells from the same passage as used for invasion studies were determined in parallel by using a 3DFM-based magnetic tweezer system (11). The creep compliance (deformability) was calculated as the average time-dependent deformation normalized by the constant stress applied (

, where a is the radius of the bead and r_max is maximum bead displacement). We find that the most invasive cell line, HEY, was 10 times more deformable than the least invasive cell line, IGROV. In addition, OV207 that exhibited 30-fold greater invasion than OV445, had a J_max = 3.1 Pa⁻¹ in contrast with the J_max = 0.3 Pa⁻¹ observed for OV445 (Fig. 1C and D). Hence, both cell lines and primary cells that exhibited high invasive behavior also presented high J_max values and were more compliant. To further examine the relationship between cancer cell deformability and invasive potential, the effective shear modulus (here on referred to as the stiffness, k) of the cell was calculated by fitting a modified Kelvin–Voigt model (12) to the compliance using a least squares fit (see Supplementary Materials). The cancer cell lines and primary cells were classified both by their stiffness and their invasiveness, with both parameters falling into 3 classes of low, medium, and high stiffness or invasiveness, respectively. Cell lines within a given class did not exhibit statistically significant differences in their stiffness, whereas cell lines between classes, were significantly different (P ≤ 0.05). Consistent with previous findings, the distribution of all stiffness values for the less invasive cell line (IGROV) showed a log normal distribution whereas the highly invasive cell line (SKOV3) showed a normal distribution (Supplementary Fig. S4; ref. 8). Scaling the cell line and primary cell correlations separately with their respective highest cell stiffness values resulted in a single parameter power law (Fig. 2D). A similar correlation and power law was also observed for stiffness and cell migration (Supplementary Fig. S2; P_{cell lines} = −0.95 and P_primaries = −0.96 in log–log scale). Although previous reports have shown alterations in cell stiffness of cancer cells either from body fluids or tumors (13), our data using ovarian cancer cell lines and cells from patient ascites show that cancer cells across a given disease population exhibit a varying degree of stiffness, a phenomenon previously not described. In addition, the variability in stiffness correlates directly with a specific measure of metastatic progression as determined by using in vitro 3D invasion assays.

Stiffness and deformation are strongly regulated by actomyosin contractility (14, 15). Phosphorylation of the 20-kD regulatory MLC subunit on the Ser19 (mono) or on Ser19/Thr18 (di; ref. 16) has been shown to promote cell contractility via changes in the actin myosin network (17). Visualization of actin in the stiffest and least invasive cell line, IGROV, revealed strong cortical actin staining with little to no cell protrusions or lamellipodial structures (Fig. 3A). In contrast, the compliant and invasive cell lines, including SKOV3 and HEY cells, exhibited distinct lamellipodial and protrusive structures with limited cortical actin. In addition, pMLC was found distinctly along the cell periphery in IGROV (Fig. 3B) whereas SKOV3 and HEY cells had little to no peripheral pMLC localization. This phenotypic difference between the stiffest/least invasive and the compliant/most invasive cell lines might reflect differences in epithelial character of the cells. Accordingly, we examined the expression of an epithelial marker (E-cadherin) and a mesenchymal marker (vimentin) in these cell lines. Indeed, the stiffest/least invasive cell lines expressed more E-cadherin and less vimentin, whereas the compliant/most invasive cell lines expressed less E-cadherin and more vimentin (Supplementary Fig. S5). Intriguingly, while the Ovca420 and IGROV cells both exhibited high cortical actin (Fig. 3A and B), high E-cadherin expression and low vimentin expression (Supplementary Fig. S5), they exhibited a 1.7-fold difference in stiffness, which corresponded to a 4-fold difference in invasion (Figs. 1 and 2). These data show that cell stiffness measurements conducted as described may be a more discerning measurement of metastatic potential than examining cell structure or epithelial character.

To investigate the role of stiffness as impacted by the cytoskeleton on migration and invasion, we determined the effect of altering actomyosin contractility on these properties. Because cells with differential invasiveness (IGROV vs. HEY) had distinct pMLC localization and cytoskeletal architecture (Fig. 3), we used blebbistatin, a myosin II inhibitor on the stiffest cell line (IGROV) and examined the effect on cell stiffness, migration, and invasion. Blebbistatin, at a concentration that disrupted cortical pMLC localization but did not affect viability, increased cell invasion by 2.5-fold, cell migration by 4-fold, and decreased cell stiffness by 2-fold (Fig. 4C and D). We also observed concomitant alterations in the actin cytoskeleton of IGROV cells (Supplementary Fig. S3), supporting a relationship between actomyosin contractility, cell stiffness, and invasion of cancer cells. Another factor implicated in regulating migration and invasion either via the cytoskeleton or via secretion of proteases like MMPs is TGF-β (18). Early in carcinogenesis tumors become resistant to the homeostatic effects of TGF-β. One of the proposed mechanisms is loss of expression of the type III TGF-β receptor (TβRIII), which has been shown in a number of human cancers, including cancers of the breast, lung, ovary, pancreas, and prostate (reviewed in ref. 18). TβRIII has been shown to regulate cancer cell motility via alterations in the actin cytoskeleton (9). To further investigate the contribution of the cytoskeleton and stiffness on cancer cell invasion, we examined the effect of TβRIII expression on cancer cell stiffness using stable cell lines Ovca429Neo (no TβRIII expression), and Ovca429TβRIII (TβRIII expression restored; ref. 9). We found that Ovca429TβRIII cells were 2-fold stiffer than Ovca429Neo cells (K_{ovca429TβRIII} = 2.9 Pa, K_Ovca429Neo = 1.29 Pa; Fig. 4B). Furthermore, this increase in stiffness corresponded to a decrease in invasiveness, similar to the correlation observed in the cancer cell lines (Fig. 4A). In addition, treating Ovca429TβRIII cells with blebbistatin increased their invasiveness by 4-fold and decreased the stiffness by 2-fold (K_{ovca429TβRIII-blebbistatin} = 1.48 Pa, I_{ovca429TβRIII-blebbistatin} = 0.04%) similar to effects seen with blebbistatin treatment of the stiffest ovarian cancer cell line, IGROV (Fig. 4C and D). Hence, cytoskeletal stiffness and effects on myosin II function may mediate suppression of migration and invasion by TβRIII.

Our results are the first evidence that metastatic potential measured through cancer cell invasion shows an inverse power-law relationship with cell stiffness. The particular exponent we derive may depend on the methodology employed for mechanical property determination. As cancer cells get progressively more invasive, they display softer mechanical characteristics that result in cell deformation and shape changes suitable for a metastatic population. We also find that cell lines having similar cytomorphology and cells from patients with similar stage disease can have widely different invasive potential that correlates with differences in stiffness. Currently, cell-based diagnoses in cancer rely on histology examination of the removed tissue sample through antibody labeling of specific markers. This complex process is not always reliable and lacks quantified assessment of the disease state. Hence, sensitive biophysical measurements such as those shown here, can be done in short periods of time, on samples obtained from either ascites or circulating cells, providing potentially unique information about the patient's cancer including metastatic potential. Application of more sophisticated models to quantify scale-free cell mechanics will provide further insight into this relationship (19, 20). These insights into biomechanical changes during cancer progression have the potential to lead to novel therapy for treatments. Our observation that the relationship between invasiveness and stiffness is maintained across a series of cancer cell lines, in patient tumor specimens, and under cell biochemical modifications that increase and decrease cell stiffness, suggests that magnetic bead assays for stiffness may be an clinically applicable predictor of invasive potential, and that treatments that affect cellular stiffness, independent of mechanism, may be useful antimetastatic approaches.

V. Swaminathan, K. Mythereye, G.C. Blobe, and R. Superfine are invested in Rheomics Inc., a company developing medical diagnostics informed by the results of this report. All other authors have declared no conflict of interest.

We thank Susan Murphy for sharing cancer cell lines and for useful discussions.

This work was funded in part by NIH grant R01-CA135006 (G.C. Blobe), Department of Defense cancer research grant W81 XWH-09-1-0265 (K. Mythreye), and P41-EB002025-23A1 (R. Superfine) supporting Computer Integrated Systems for Microscopy and Manipulation and Carolina Center for Cancer Nanotechnology Excellence NIH-U54-CA119343 (R. Superfine).