Cadherin-based intercellular adhesions organize epithelial cell-matrix traction forces

Abstract

Cell-cell and cell-matrix adhesions play essential roles in the function of tissues. There is growing evidence for the importance of cross talk between these two adhesion types, yet little is known about the impact of these interactions on the mechanical coupling of cells to the extracellular matrix (ECM). Here, we combine experiment and theory to reveal how intercellular adhesions modulate forces transmitted to the ECM. In the absence of cadherin-based adhesions, primary mouse keratinocytes within a colony appear to act independently, with significant traction forces extending throughout the colony. In contrast, with strong cadherin-based adhesions, keratinocytes in a cohesive colony localize traction forces to the colony periphery. Through genetic or antibody-mediated loss of cadherin expression or function, we show that cadherin-based adhesions are essential for this mechanical cooperativity. A minimal physical model in which cell-cell adhesions modulate the physical cohesion between contractile cells is sufficient to recreate the spatial rearrangement of traction forces observed experimentally with varying strength of cadherin-based adhesions. This work defines the importance of cadherin-based cell-cell adhesions in coordinating mechanical activity of epithelial cells and has implications for the mechanical regulation of epithelial tissues during development, homeostasis, and disease.

Fig. 1.

Traction stresses dynamically reorganize in high-calcium medium. (A–C) Differential interference contrast (DIC) images of a three-cell colony at 45 min (A), 6 h (B), and 12 h (C) after calcium elevation. (D–F) Distribution of in-plane traction stresses (red arrows) for cell colony at time points in A–C overlaid on DIC images. For clarity, one-quarter of calculated traction stresses are shown. (G–I) Distribution of strain energy density, w, for cell colony at time points in A–C. The blue lines mark individual cell boundaries. (J) Schematic for calculating azimuthal-like averages for strain energy. Colony outline is eroded inward by distance, Δ, in discrete steps, δ, until entire colony area has been covered. Average strain energy density is then calculated for each concentric, annular-like region. (K) Strain energy profiles for three-cell colony at six time points after calcium elevation. The solid colored lines represent colony’s average strain energy density as a function of distance, Δ, from colony edge. Each profile is mirrored about Δ ∼ R, the effective colony radius. Colony periphery (Δ = 0) is indicated by dashed vertical black lines. Strain energy at Δ < 0 corresponds to regions outside colony periphery. (L) Average strain energy density for entire colony at 15-min intervals from 30 min to 12 h after calcium elevation. Plot colors in K and L are scaled according to time, t, after calcium elevation, from cyan at t = 0 to magenta at t = 12 h. (Scale bars: A–I, 50 μm.)

Fig. 2.

Traction stresses systematically reorganize in high-calcium medium. (A) Distribution of in-plane traction stresses (red arrows) of an eight-cell wild-type colony in low-calcium medium overlaid on DIC image of colony. For clarity, one-ninth of calculated traction stresses are shown. (B) Strain energy distribution, w, of low-calcium colony in C with individual cell outlines in blue. (C) Distribution of traction stresses (red arrows) of a six-cell wild-type colony in high-calcium medium for 24 h overlaid on DIC image of colony. For clarity, one-ninth of calculated traction stresses are shown. (D) Strain energy distribution, w, of high-calcium colony in E, with individual cell outlines in blue. (E) Strain energy profiles for n = 32 low-calcium colonies. (F) Strain energy profiles for n = 29 high-calcium colonies. In E and F, each solid curve represents a colony’s average strain energy density as a function of distance, Δ, from colony edge. Each profile terminates where inward erosion covers entire colony area, at Δ ∼ R, the effective colony radius, indicated by dashed line. The erosion is defined in legend of Fig. 1J. Average strain energy is normalized to value at colony periphery, , giving each colony the same height on the graphs, indicated by the vertical scale bar. For clarity, profiles are spaced vertically according to colony size, with profiles for larger colonies (terminating at larger values of Δ) appearing higher up the y axis. Profile colors correspond to colony cell number given in the legend. (G) Quantification of relative distance from colony periphery (Δ/R) corresponding to 75% of total strain energy, 3W/4, in colonies in low- or high-calcium medium. Small colonies (R < 50 μm, below hash marks in E and F), in low- (n = 8) or high-calcium (n = 8) medium showed no significant difference, whereas large (R > 50 μm) low-calcium colonies (n = 24) had significantly more strain energy closer to colony center than large high-calcium colonies (n = 21). Statistical significance between low- and high-calcium populations is indicated by asterisks (P < 0.001). Error bars indicate 1 SD. (H) Relationship between total strain energy, W, and area, A, of colonies in low- and high-calcium medium. Open symbols correspond to low-calcium colonies, closed symbols to high-calcium colonies. Symbol colors indicate colony cell number, given in the legend. (I and J) Keratinocytes in low-calcium medium (I) or after 24 h in high-calcium medium (J) labeled with anti–E-cadherin and anti-paxillin antibodies and stained with phalloidin to mark F-actin. (Scale bars: A–D, I, and J, 50 μm.) Data for high-calcium colonies in F–H are adapted from ref. .

Fig. 3.

Cadherin-based adhesions are required for organization of traction stresses in high-calcium medium. (A) Localization of E-cadherin, phalloidin (F-actin), and paxillin in colony of three wild-type keratinocytes in high-calcium medium for 24 h with DECMA-1. (B) Distribution of traction stresses (red arrows) of five-cell colony in high-calcium medium for 24 h with DECMA-1 overlaid on DIC image of colony. For clarity, 1/16th of calculated traction stresses are shown. (C) Strain energy of colony in B with individual cell outlines in blue. (D) Strain energy profiles for n = 15 DECMA-1–treated colonies. Each solid curve represents colony’s average strain energy density as a function of distance, Δ, from colony the edge, as defined in Fig. 1J. For clarity, profiles are spaced vertically according to colony size. Each profile terminates where inward erosion covers entire colony area, at Δ ∼ R. (E) Localization of E-cadherin, phalloidin (F-actin), and paxillin in a colony of three E-cadherin–knockout/P-cadherin–knockdown (KO/KD) keratinocytes after 24 h in high-calcium medium. (F) Distribution of traction stresses (red arrows) of a colony of three KO/KD keratinocytes in high-calcium medium for 24 h overlaid on DIC image of colony. For clarity, 1/16th of calculated traction stresses are shown. (G) Strain energy distribution of colony in F with individual cell outlines in blue. (H) Strain energy profiles for n = 14 KO/KD colonies after 24 h in high-calcium medium. As in D, profiles were calculated as defined in Fig. 1J. Profile colors in D and H correspond colony cell number given by the legend between Fig. 2 E and F. (I) Comparison of the strain energy distribution for large low-calcium (n = 24), large high-calcium (n = 21), DECMA-1 (n = 15), and KO/KD (n = 14) colonies. Values represent proportion of the colony from periphery inward, Δ/R, necessary to capture 75% of total colony strain energy. Error bars indicate 1 SD. Higher proportions indicate higher strain energy nearer colony center. Statistical significance between pairs of colony conditions is indicated as follows: *P < 0.05, **P < 0.01, or ***P < 0.001. (Scale bars: A–C and E–G, 50 μm.)

Fig. 4.

Minimal physical model captures cadherin-dependent organization of traction stresses. (A) Schematic of planar colony of three hexagonal cells. (B–D) Strain energy distributions for colony of three hexagonal cells with different spring stiffness, k, expressed in units of E/L, where E is the Young’s modulus of the cell and L the side length of each hexagon. (E–G) Spatial profiles of average strain energy as a function of distance, Δ, from colony edge for different values of k corresponding to data in B–D. Other parameters were as follows: ℓ_p/L = 0.2, E = 1 kPa, ν = 0.4, σ_a = 4 kPa, h = 0.2 μm, and Y = 2 × 10⁶ N/m³ (SI Text). (Scale bars: B–D, 50 μm.)