Fig 1.
Schematic representation of cell population dynamics model of muscle regeneration.
The diagram delineates the interactions among various cell types involved in the muscle regeneration process following injury, including the infiltration, transformation, proliferation, death, and exfiltration of cells. Boxes indicate cell types Md (dead myonuclei), N (neutrophils), Nd (dead neutrophils), M (monocyte), M1 (M1 macrophages), M2 (M2 macrophages), QSC (quiescent satellite cells), ASC (activated satellite cells), Mc (myocytes). Solid arrows represent transformations/transitions between cellular states, while dashed arrows indicate regulatory influences.
Fig 2.
Cell typing analysis of McKellar et al. [33] (A–D) and De Micheli et al. [34] (E–H) scRNA-seq mouse muscle regeneration datasets.
(A/E) UMAP projections of McKellar/De Micheli cells. (B/F) Dot plots of marker expression used to assign cell types to McKellar/De Micheli cell clusters. (C/D) UMAP projections of McKellar/De Micheli muscle satellite cell subsets. (D/H) Dot plots of marker expression used to assign cell types to McKellar/De Micheli satellite cell subsets.
Fig 3.
Cell-type population fractions from McKellar’s [33] and De Micheli’s [34] datasets.
Line graphs show the mean cell proportions across replicates of each cell type, as a fraction of all cell types included in our model. Each subplot corresponds to a different cell population in the model: N (neutrophils), M (monocytes), M1 (M1 macrophages), M2 (M2 macrophages), QSC (quiescent satellite cells), ASC (activated satellite cells), and Mc (myocytes). The empirical data do not include numbers for the dead cell types, Md and Nd.
Table 1.
Initial parameter values before optimization, search bounds, and optimized parameters.
Fig 4.
Model predictions versus empirical data over a seven-day time course following muscle injury.
The solid black lines represent the trajectories obtained from the ODE model trained on the McKellar dataset. Empirical data points from the McKellar dataset are depicted as blue dots, with vertical bars indicating standard deviation across replicates. The e validation data from the De Micheli dataset are shown as red dots. The alignment of model predictions with the McKellar (training dataset) data points indicates the model’s accuracy, whereas consistency with the De Micheli (validation dataset) data points supports the model’s generalizability.
Fig 5.
Sensitivity of modeled cell trajectories to parameter perturbations.
Each heatmap shows the influence of a small positive perturbation to each model parameter on one of the nine cell types as a function of time. Red represents increase in the state variable, blue decrease. Within each heatmap, values are scaled to fill the full color range, so strengths of derivatives can be compared within a heatmap, but not across heatmaps.