Compressive stress drives adhesion-dependent unjamming transitions in breast cancer cell migration

Abstract

Cellular unjamming is the collective fluidization of cell motion and has been linked to many biological processes, including development, wound repair, and tumor growth. In tumor growth, the uncontrolled proliferation of cancer cells in a confined space generates mechanical compressive stress. However, because multiple cellular and molecular mechanisms may be operating simultaneously, the role of compressive stress in unjamming transitions during cancer progression remains unknown. Here, we investigate which mechanism dominates in a dense, mechanically stressed monolayer. We find that long-term mechanical compression triggers cell arrest in benign epithelial cells and enhances cancer cell migration in transitions correlated with cell shape, leading us to examine the contributions of cell-cell adhesion and substrate traction in unjamming transitions. We show that cadherin-mediated cell-cell adhesion regulates differential cellular responses to compressive stress and is an important driver of unjamming in stressed monolayers. Importantly, compressive stress does not induce the epithelial-mesenchymal transition in unjammed cells. Furthermore, traction force microscopy reveals the attenuation of traction stresses in compressed cells within the bulk monolayer regardless of cell type and motility. As traction within the bulk monolayer decreases with compressive pressure, cancer cells at the leading edge of the cell layer exhibit sustained traction under compression. Together, strengthened intercellular adhesion and attenuation of traction forces within the bulk cell sheet under compression lead to fluidization of the cell layer and may impact collective cell motion in tumor development and breast cancer progression.

Keywords: cell shape; cell–cell adhesion; collective migration; traction force microscopy; unjamming transition.

FIGURE 1

Compressive stress inhibits migration in MCF10A cells and enhances migration in 4T1 cells. (A) Representative fluorescence images of MCF10A and 4T1 wound areas at the indicated time points post-wound with and without compression. Cell nuclei are labeled with Hoechst 33342. Lines are scratched on each well using a p-200 pipette tip and cell migration is captured by fluorescence microscopy at 30-min time intervals for 16 h post-wound. Cell edges used to calculate wound area are outlined by white dashed lines. Scale bars, 200 µm. (B) Heat maps show spatiotemporal evolution of the velocity for control and compressed cells. (C) Quantification of wound area (between white dashed cell edges) for each cell type and compressive pressure. Mean wound area at each time point is plotted from three independent replicates with the individual experiments plotted as thin lines.

FIGURE 2

With compression, MCF10A cells and nuclei become more compact, whereas 4T1 cells and nuclei elongate. (A) Representative binary images outlining fixed MCF10A and 4T1 cells labeled with E-cadherin with and without compression. (B) Boxplot of cell shape index for MCF10A and 4T1 cells subjected to 0, 600, or 1,200 Pa for 12 h. (C) Cell aspect ratio (AR), which emphasizes elongation, of control and compressed MCF10A and 4T1 cells is plotted as mean of AR ( $\overline{\mathrm{A}\mathrm{R}}$ ) vs. standard deviation (s.d.) of AR for each cell type and compressive pressure. (D) Representative binary images outlining fixed MCF10A and 4T1 cell nuclei (labeled with DAPI) with and without compression. (E) Boxplot of nuclear shape index for control and compressed MCF10A and 4T1 cells. Boxplots of cell and nuclear shape indices show median and quartiles for three independent replicates. Whiskers are maximum and minimum data points, and data from each replicate is denoted as a different color. Number of MCF10A cells and nuclei analyzed: 0 Pa (n = 234), 600 Pa (n = 229), and 1,200 Pa (n = 222). Number of 4T1 cells and nuclei analyzed: 0 Pa (n = 205), 600 Pa (n = 231), and 1200 Pa (n = 224).

FIGURE 3

E-cadherin is upregulated in unjammed 4T1 cells under mechanical compression. (A) Representative immunofluorescence images of MCF10A cells labeled with DAPI and an E-cadherin antibody. Cell monolayers are subjected to specified compressive pressures for 12 h. Scale bar, 20 µm. (B) Quantification of relative E-cadherin fluorescence at MCF10A cell–cell contacts. Mean fluorescence intensity at the cell membrane ±S.E. is plotted from three independent replicates (n = 12–13). (C) Representative microscopy images of 4T1 cells labeled with DAPI and an E-cadherin antibody in the same experimental conditions as in (A). Scale bar, 20 µm. (D) Quantification of relative E-cadherin fluorescence at 4T1 cell–cell contacts. Mean fluorescence intensity at the cell membrane ±S.E. is plotted from three independent experiments (n = 10). (E) Representative microscopy images of microcontact-printed 4T1 cell islands labeled with DAPI and an E-cadherin antibody. Micropatterned cell islands are exposed to specified stresses for 12 h. Scale bars, 80 µm (top) and 20 µm (bottom). (F) Quantification of relative E-cadherin fluorescence at the cell–cell contacts of 4T1 cell islands. Mean fluorescence intensity at the cell membrane ±S.E. is plotted from three independent replicates (n = 8–15). qPCR analysis of E-cadherin, N-cadherin, and vimentin mRNA levels with and without compression (1,200 Pa) for MCF10A (G) and 4T1 (H) cell monolayers. Transcript levels are calculated using the ΔΔC_t method normalized to GAPDH. Mean mRNA level ±S.E. is plotted from three independent experiments with duplicates per experiment.

FIGURE 4

E-cadherin knockdown inhibited compression-induced upregulation of E-cadherin in 4T1 cells, triggering jamming. (A) Representative microscopy images of 4T1 scramble and E-cad KD cells labeled with DAPI and an E-cadherin antibody. E-cad KD is induced in 4T1 shE-cadherin cells by adding 200 μM IPTG 72 h prior to experiments. Cell monolayers are exposed to specified stresses for 12 h. Scale bar, 20 µm. (B) Quantification of relative E-cadherin fluorescence at the cell–cell contacts of 4T1 scramble and E-cad KD cells. Mean fluorescence intensity at the cell membrane ±S.E. is plotted from three independent replicates (n = 22–36). qPCR analysis of E-cadherin (C), N-cadherin (D), and vimentin (E) mRNA levels in 4T1 scramble and E-cad KD cells. Transcript levels are calculated using the ΔΔC_t method normalized to GAPDH. Mean mRNA level ±S.E. is plotted from three independent experiments with duplicates per experiment. (F) Representative images of 4T1 E-cad KD wound area at the indicated time points post-wound. 4T1 shE-cadherin cells express mNeonGreen. Cell edges used to calculate wound area are outlined by white dashed lines. Scale bar, 200 µm. (G) Heat maps of spatiotemporal evolution of the velocity for 4T1 E-cad KD cells under different levels of mechanical compression. (H) Quantification of wound area (between white dashed cell edges) for each condition. Mean wound area at each time point is plotted from three independent replicates as a representative trace. (I) Summary depicting the effect of compressive stress on collective migration in MCF10A WT, 4T1 WT, and 4T1 E-cad KD cells. Strong cell–cell contacts are denoted by black dashes. Red dashes indicate weak cell–cell adhesion. Number of small black arrows (right) represent relative cell velocity during wound healing.

FIGURE 5

Compressive stress reduced traction stresses within the bulk cell sheet. (A) Representative microscopy images of 4T1 cell islands labeled with DAPI and a vinculin antibody. Vinculin staining is shown at two different imaging planes. Micropatterned cell islands were exposed to specified pressures for 12 h. Scale bars, 80 µm (top) and 20 µm (bottom). (B) Quantification of relative vinculin intensity of individual cells at the basal plane and at intercellular junctions. Mean fluorescence ±S.E. is plotted from three independent replicates (n = 18). (C) Traction stress vector field and traction stress magnitude of micropatterned cell islands with and without 1,200 Pa compression for 3 h. Cell nuclei are labeled with Hoechst 33342. Scale bar, 80 µm. Mean traction stresses (D) and total strain energy (E) with and without compression on micropatterned islands. Number of images analyzed for MCF10A: control (n = 21) and compressed (n = 21). Number of images analyzed for 4T1: control (n = 18), compressed (n = 30). (F) Traction stress vector field and traction stress magnitude of a wounded edge with and without 1,200 Pa compression for 3 h. Cell nuclei are labeled with Hoechst 33342. Scale bar, 80 µm. Mean traction stresses exerted by MCF10A cells (G) and 4T1 cells (H) at the leading edge (within 5–7 cell layers of the wound margin) and within the bulk monolayer for 0, 600, and 1,200 Pa compressive stress for 3 h. Number of images analyzed for MCF10A: 0 Pa (n = 37), 600 Pa (n = 26), and 1,200 Pa (n = 32). Number of images analyzed for 4T1: 0 Pa (n = 28), 600 Pa (n = 33), and 1,200 Pa (n = 24).

FIGURE 6

Mapping the experimentally observed tissue states to the theoretical jamming–unjamming phase diagram. The jamming–unjamming phase diagram is shown in terms of the two pertinent parameters of the SPV model: the single-cell motility v ₀ and the target cell shape index p ₀. By mapping the experimentally observed cell traction forces and cell shapes to theoretical simulation results, the positions of 4T1 and 10A cells are placed on the phase diagram (see Methods).