Fig 1.
Experimental colonization patterns.
Snapshots of one field of view at confluence for matrix non-producing strains at low (a) and high (b) initial cell densities, and matrix-producing strains at low (c) and high (d) initial cell densities. The inset of each panel shows the initial distribution of founder cells. Initial densities: a) 0.01 cells/μm2, b) 0.162 cells/μm2, c) 0.012 cells/μm2, d) 0.113 cells/μm2.
Fig 2.
Model validation: Correlation length comparison.
Experimental correlation lengths measured in the matrix-producing (pale green dots) and non-producing (black dots) strain, and their model equivalent σ = 1 (dark-green diamonds), respectively σ = 0 (gray squares). Numerical results are shown for flow intensity f = 1, which gives the best agreement with the experiments, averages taken over 2x106 independent realizations. Error bars represent the standard deviation. The insets show snapshots of colonization patterns obtained in the experiments (right) and the model (left) at initial colonization densities indicated by the gray pointers. Experimental surface occupation pictures used for this figure are available at: https://figshare.com/s/654db6149e509881588686.
Fig 3.
Model validation: Distribution of correlation lengths for matrix-producing strains.
Estimated theoretical (full line) and experimental (dashed line) correlation length distributions using kernel density estimation (KDE) techniques. The symbols represent the experimental distribution prior to smoothing estimations. Each color represents a range of colonizing cell densities: green, 10−1 cells/μm2 for the model and high density experimental data (cluster of data around 10−1 cells/μm2 in Fig 2); blue, 10−2 and 2.15x10-2 cells/μm2 for the model and intermediate density experimental data (7x10-3 < ρ0 < 3x10-2 cells/μm2); and red, 10−3 and 2.15x10-3 cells/μm2 in the model and low density experimental data (ρ0 < 5x10-3 cells/μm2).
Fig 4.
Model output: Mean cluster size and variability.
a) Difference in correlation length resulting from investing in cell adhesion for different flow intensities and initial colonization densities. The dashed line indicates the values of f and ρ0 at which this difference is equal to zero. Averages are taken over 5x104 independent realizations. b, c) The standard deviation of the correlation length is a proxy for lineage segregation variability in highly-adhesive strains, (b; σ = 1) and weakly-adhesive cells (c; σ = 0). Averages are taken over 2x106 independent realizations of the model.