Fig. 2: In vitro evolution of aneuploid HMEC lineages leads to convergent selection of breast cancer-associated arm-level CNAs.
From: Chromosome evolution screens recapitulate tissue-specific tumor aneuploidy patterns
a, Diagram of the in vitro evolution experiments with aneuploid and diploid HMEC clones. b, Heatmap summary of all original (first screen, solid squares) and newly selected (triangles) copy number events in long-term evolution experiments in diploid, 2N-range and 4N-range aneuploid HMECs, plotted by arm. The first (wider) column indicates copy number gain or loss frequencies in the breast cancer TCGA cohort. All subsequent columns represent independent in vitro evolution experiments grouped and labeled by clonal lineage and ploidy. Colors inside triangles indicate final copy number state after the newly acquired event. Diploid clone names are indicated by lowercase letters, and tetraploid clones are indicated by uppercase letters. True arm-level events that probably involve broken chromosomes are highlighted in yellow. Right of the heatmap includes a summary of the evolution experiments in daughter subclones of clone CQ. c, Correlation between true arm-level event frequencies in the HMEC in vitro evolution screen (n = 90 in vitro evolution experiments; gain minus loss frequencies) and breast cancer arm-level event frequencies (TCGA cohort, n = 722). Pearson’s correlation coefficient (r = 0.68) and associated P value (P = 9.25 × 10−7) are shown. Dashed gray line indicates linear regression model of the data. d, Heatmap of Pearson’s correlation P values from comparisons of chromosome arm-level gain minus loss frequencies in evolved HMECs and various solid tumors (see Fig. 1f legend for tumor type abbreviations and numbers of patient samples). e, Transcriptomic GSEA-based immune infiltrate analysis by immune cell type for breast cancers with various CNAs (TCGA database). Gain of 16p (which is not selected in vitro but is frequent in breast cancer) is significantly associated with reduced CD8 T-cell, natural killer (NK) cell and macrophage signatures. P values are calculated by assessing the frequency with which the enrichment score of a gene set in a ranking exceeds that of random ranking permutation (10,000 permutations) and is adjusted for multiple gene sets testing. T.reg, regulatory T cells.
