Towards Data-Driven Modeling of Cell Cycle and Wound Closure Processes

TL;DR

By linking single-cell cycle dynamics with emergent tissue behaviour this work establishes a quantitative approach for exploring how intracellular processes shape repair processes and demonstrates the value of integrating high-resolution data with cell-based mechanical models and provides a foundation for systematic evaluation of therapeutic interventions.

Abstract

Effective wound repair treatments rely on a clear picture of how cell proliferation and migration are coordinated during tissue restoration. Fibroblasts are key contributors to tissue restoration in the dermis, and modern imaging tools allow their cell-cycle progression to be observed directly, enabling comparison between experiments and computational models. Here we investigate how different stages of the cell cycle influence fibroblast-driven wound closure using the Discrete Laplacian Cell Mechanics (DLCM) framework driven by time-lapse microscopy data. \textit{In vitro} assays provide cell positions, migration behaviour, and cycle-stage information, and we show that incorporating proliferation, migration, and cell cycle arrest allows the computational model to reproduce the essential experimental trends. The results reveal that arrest in the G1 phase notably impacts the cell cycle dynamics and that the initial spatial arrangement of cycle states significantly affects wound closure. By linking single-cell cycle dynamics with emergent tissue behaviour this work establishes a quantitative approach for exploring how intracellular processes shape repair processes. More broadly, it demonstrates the value of integrating high-resolution data with cell-based mechanical models and provides a foundation for systematic \textit{in silico} evaluation of therapeutic interventions.

Figures (5)

• Figure 1: From left to right: a) cell data with red representing the G1 phase, green--S/G2/M; b) cell data in black and white and overlayed with a hexagonal mesh and cell center (yellow points) where the image processing finds a cell; c) finally, the DLCM model's representation formulated from the number of cells in each voxel given by the data. Green voxels contain cells in $g$ state; red voxels contain cells in $m$ state (cell states randomized as described in §\ref{['sec:results_setup']}). Darker-shaded voxels indicate that the voxel is doubly occupied (states of the two cells may differ but we do not specifically use another color for the visualization). The white and grey voxels, where there are no cells, represent wounded and unwounded region, $W$ and $V$, respectively.
• Figure 2: Estimated front position of the population along with sample standard deviation of $n = 100$ simulations. Fitting a line through either the data or the simulation results yields a slope of $0.03$, which is also how the Darcy scaling coefficient $D$ in \ref{['eq:cell_migration']} was selected.
• Figure 3: Percentage of cells from a total of $n=100$ simulations at each sub-stage (ticker) for respective cell cycle stage ($g$ and $m$) at the start and end of the simulation for case (iii), where $g$ cells in doubly occupied voxels do not tick. We also show the end of the simulation for case (i) and (ii). Here, green and red bars represent $m$ and $g$ state, respectively.
• Figure 4: Fraction of cells in respective cycle stage, showing all three cases for the cell cycle arrest. The panels represent cases (i)-(iii), respectively. Red and green are the $g$ and $m$ states (or G1 and S/G2/M), respectively; the box plots are the experimental data, and the curves and envelopes represent the simulation mean and 68% confidence interval of $n = 100$ simulations.
• Figure 5: Fraction of cells in respective cycle stage, showing both simulation and data mean and standard deviation of $n = 100$ simulations, with colour scheme as in \ref{['fig:cycle_iii']}. The model is case iii) but using different spatial distribution of the initial cell cycle phases: a) The leftmost cells are in $g$; b) The rightmost cells are in $g$. The most notable difference in behaviour occurs initially.