Fig. 3: A transcriptomic signature predicts intratumoral clone expansion.

Cells from bilateral tumors excised on days 14 and 21 from 3 mice were processed by a combination of single-cell RNA/TCR-sequencing and bulk TCR-sequencing to enable integration of single-cell transcriptomic data of clones with their expansion dynamics. a Uniform Manifold Approximation and Projection (UMAP) map of 2928 cells from tumors excised on day 14, colored and labeled by cell type. b−f) Precursor exhausted cells from tumors excised on day 14, belonging to expanding and contracting clones, were investigated for transcriptomic signatures that could predict future clone expansion. c UMAP map of precursor exhausted cells colored by their expansion dynamics. d Differential gene expression analysis of genes overexpressed in precursor exhausted cells from expanding clones over contracting clones. Notable genes are labeled. e Scatter plot comparing the expansion signature score with the expansion (the log2 fold-increase in clone frequency from day 14 to 21) of the clone for individual cells. Cells belonging to the same clone have the same expansion rate. f Scatter plot comparing the mean expansion signature score with the expansion of clones. Clones were scored for their expression of the expansion signature in constituent precursor exhausted cells on day 14. Analysis on clones with at least 5 precursor exhausted cells on day 14. Dots represent genes (d), cells (a, c, e) and clones (f). Scatterplots are displayed with a line indicating the linear regression model fit and 95% confidence intervals obtained by bootstrapping (e, f). Statistical testing via two-sided Wilcoxon rank-sum test corrected with the Benjamini–Hochberg procedure (d) and two-sided Wald test (e, f) (****p < 0.0001; ***p < 0.001; **p < 0.01; *p < 0.05; ns p > 0.05). Illustrations created with cartoons from Iraustoya.com (b). Source data are provided as a Source Data file.