Figure 4.
Models assuming time-invariant but spatially-variable environment are unable to accurately describe the clustering of T cells of irrelevant specificity in different conditions. (A) Mice were infected with 3 × 105 GFP-expressing Py sporozoites. Twenty hours later 9 × 106 Py-specific activated CD8 T cells (PyTCR), 9 × 106 OT1 T cells (specific to chicken ovalbumin), or mixture of 9 × 106 PyTCR and 9 × 106 OT1 T cells were transferred into infected mice and livers of these mice were imaged using intravital microscopy 6 h later. In total 92 (mice receiving only OT1 cells, B) and 52 (in mice receiving a mix of PyTCR and OT1 cells, C) parasites were randomly chosen and the number of T cells in a 40 μm radius were counted (29). The “two population” mathematical model (Equation 10) was fitted to these two datasets simultaneously assuming two different entry rates θ01 and θ02 and either allowing the fraction of attracting parasites f to vary between the datasets (solid line) or to be fixed between the datasets (dashed line). Fixing the fraction f between the datasets significantly reduced the quality of the model fit of the data as compared to the model in which f could vary (likelihood ratio test, , p < 0.001).