A self-organized compression network arrests epithelial proliferation

As epithelial development or wound closure approaches completion, cell proliferation progressively slows via contact inhibition of proliferation (CIP) - a mechanism understood as being strictly local. Here we report the discovery of inhibition of proliferation through an unanticipated mechanism that is non-local. As a confluent epithelial layer becomes progressively more jammed, two interpenetrating networks emerge: islands of mechanically compressed non-cycling cells percolating within an ocean of mechanically tensed cycling cells. The evolution of the compression network was found to be susceptible to both specific molecular stimulus and to injury-induced unjamming. Yet, in all circumstances, the size of compressed islands followed a power-law distribution that was well-captured by preferential network theory. Together, these findings demonstrate the existence of a network-based inhibition of proliferation (NIP) that is self-organizing and poised in proximity to criticality.

💡 Research Summary

This paper presents a groundbreaking discovery that challenges the conventional, locally-focused understanding of how epithelial cell proliferation is halted. While the established model is Contact Inhibition of Proliferation (CIP), this work unveils a non-local, network-based inhibitory mechanism termed Network Inhibition of Proliferation (NIP).

The researchers used Madin-Darby canine kidney (MDCK) cells to form a confluent epithelial layer. To probe the mechanical state of individual cells, they developed a “restrained trypsinization” assay that temporarily breaks all cell-cell and cell-substrate adhesions, allowing them to track whether each cell subsequently expands or contracts. Surprisingly, two distinct phenotypes emerged: “contracting” cells and “expanding” cells. Expanding cells were typically smaller, had a higher nucleus-to-cell area ratio, and displayed actin distributed throughout their interior (Phenotype B), whereas contracting cells had cortex-localized actin (Phenotype A).

As the epithelial layer matured and became more densely packed (“jammed”), the proportion of expanding cells increased. Crucially, these expanding cells were not randomly scattered but formed spatially clustered islands and chains embedded within a sea of contracting cells. This pattern resembled force chains in jammed granular materials. When a wound was introduced to create a free boundary, a wave of increased cellular mobility (“unjamming”) propagated from the wound edge. In regions affected by this wave, expanding cells reverted to a contracting phenotype, demonstrating that the compression network is dynamic and reversible.

Functional analysis revealed that expanding cells were far less likely to be in the proliferative phases (S/G2/M) of the cell cycle compared to contracting cells. Conversely, pharmacologically inhibiting cell division arrested the growth of the expanding cell network. This established a clear link: contracting cells are tensed and cycling, while expanding cells are compressed and non-cycling. The expansion of the compressed network depended on cell division events, which primarily occur in tensed regions, but continued even after global tissue density plateaued, suggesting a local growth mechanism.

The key insight came from analyzing the size distribution of the compressed cell islands. Across all experimental conditions—during natural maturation, after wounding, or under drug treatment—the distribution followed a scale-free power law. This statistical signature pointed towards a self-organizing process driven by “preferential attachment,” where a new compressed cell (born from a division event) is more likely to join a large existing island than to start a new, small one. A minimal computational model implementing this biased growth rule on a hexagonal lattice successfully reproduced the qualitative patterns and quantitative power-law distributions observed experimentally.

Further analysis of the network’s evolution showed signs of criticality. As the fraction of compressed cells approached approximately 65%, the average island size increased sharply, and the data collapsed onto a universal scaling function independent of system size—a hallmark of a continuous phase transition near a critical point. This critical point aligns with the theoretical percolation threshold for an infinite hexagonal lattice. However, experimentally, the compressed cell fraction never significantly exceeded 65%, and the power-law exponent stagnated around -1.5. This indicates that the biological system does not cross the full critical threshold but instead self-organizes towards and is maintained in a “quasi-critical” state, likely due to biological constraints like cell extrusion or apoptosis.

In summary, this study redefines the epithelial monolayer as a system composed of two interpenetrating, mechanically defined networks: a percolating network of compressed, non-cycling cells that inhibits proliferation, and a background of tensed, cycling cells. The compression network grows via a self-organizing, preferential attachment mechanism and is poised near a critical point, providing a novel, non-local framework (NIP) for understanding how tissues regulate proliferation during homeostasis and regeneration.