Adhesion Changes Epithelial Fluidity Without Reshaping Cells

Epithelial sheets must rearrange during development, repair, and invasion, yet they also need mechanical cohesion. A common vertex-model account ties this fluidity primarily to a cell shape index set by the balance of cortical contractility and cell-cell adhesion. Bera et al. test whether that geometric proxy remains sufficient when adhesion itself is perturbed. Their central result is a decoupling: weakening E-cadherin-mediated adhesion makes MDCK monolayers more fluid even though the measured shape index, junctional line tension, and substrate traction remain effectively unchanged.

Core Contribution

The paper argues that cell-cell adhesion has two separable effects. Its energetic effect changes the effective interfacial tension, which standard vertex models represent and which can alter cell geometry. Its kinetic effect resists relative motion as adhesions and cortical networks must form, deform, and turn over during neighbor exchange. The latter produces dissipative drag. Treating adhesion only as an energetic contribution therefore misses a route by which a confluent tissue can change its rearrangement dynamics without visibly changing its average cell shape.

That distinction is the substantive contribution rather than the observation that adhesion matters. The authors extend a vertex-model framework with a dissipation parameter alongside the preferred-perimeter parameter $p_0$. In this account, a perturbation can follow a trajectory through the $(p_0,\gamma)$ parameter space that raises cell mobility while leaving the shape index approximately fixed. The model is intended to explain a shape-independent fluidization mechanism, not to replace shape-based descriptions in settings where geometry is the dominant control.

Technical Approach

The experiments use confluent MDCK cell islands exposed to increasing concentrations of the anti-E-cadherin antibody DECMA-1. Nuclei trajectories provide mean-square displacement, long-time MSD exponents, self-diffusivity $D_s$, and structural relaxation times extracted with differential dynamic microscopy. A Voronoi construction estimates the cell shape index $q$; a Cellpose boundary-segmentation analysis supplies an independent check. Laser ablation measures early junction recoil as a proxy for line tension, while traction microscopy measures cell-substrate forces. The study also measures density and verifies DECMA-1 localization with E-cadherin.

Figure 1 is the central experimental sequence: trajectories become more mobile as DECMA-1 increases, the long-time MSD behavior shifts toward fluid-like dynamics, and structural relaxation accelerates. At the same time, the average steady-state $q$ and its distribution show little displacement. The supplementary segmentation analysis reaches the same conclusion, reducing the chance that a Voronoi approximation alone created the apparent decoupling.

The authors use EGTA as a second adhesion perturbation. Unlike the DECMA-1 condition, EGTA changes traction while again producing fluidity changes without a detectable average shape-index change. This does not make the two interventions mechanically identical; it supports the narrower claim that shape need not be the sole observable governing rearrangement dynamics.

Results and Analysis

For the untreated DECMA-1 condition, the MSD fits yield a short-time exponent of 1.3 and long-time exponent of 0.6, with a crossover near 1.5 h and length scale near 17 μm. Across the five DECMA-1 concentrations, the long-time MSD exponent changes with a linear-correlation-test p-value of $8.5\times10^{-7}$, self-diffusivity with $p=0.014$, and relaxation time with $p=3.3\times10^{-3}$. In contrast, steady-state shape index changes only weakly despite a reported $p=1.1\times10^{-3}$ trend, while the independently segmented analysis reports $p=0.13$. The authors interpret this as negligible geometric variation relative to the kinetic changes; that interpretation rests more comfortably on the segmentation check than on the Voronoi result alone.

The mechanical controls sharpen the argument. Initial recoil velocity is not correlated with DECMA-1 concentration ($p=0.55$), and steady traction is likewise not correlated ($p=0.56$). Density does vary statistically ($p=5.8\times10^{-5}$), but remains in the narrow 2,500–3,200 mm⁻² range; the paper reports additional analyses finding no density-kinematics correlation. These controls make a simple explanation through altered line tension, substrate traction, or density less persuasive.

Figure 4 maps the measured diffusivity and shape data onto the extended model. The original vertex model produces diffusivity-versus-shape curves inconsistent with the experimental pattern, whereas intersecting diffusivity and shape contours identify a shape-preserving path involving both energetic and dissipative adhesion terms. This is a coherent mechanistic reconciliation, though its quantitative fit is not equivalent to a direct molecular measurement of junctional friction.

Limits in Practice

The evidence is detailed but narrow: it centers on MDCK monolayers, two chemical adhesion perturbations, and short-timescale in vitro dynamics. The inferred dissipative parameter is model-mediated rather than independently measured at individual junctions. The paper establishes that shape index is insufficient in these conditions; it does not establish that the same mechanism dominates in developing tissues, tumors, or other epithelial cell types.