Fig. 5: Inflammatory basal state of human SCLC.
From: Basal cell of origin resolves neuroendocrine–tuft lineage plasticity in cancer
a, Heat map of bulk RNA sequencing (RNA-seq) expression in n = 944 human SCLC biopsies (Supplementary Table 8) for indicated genes or signatures, grouped by subtype and ordered by basal signature score. b, As in a, with YAP1 subtype included. c, Spearman correlation matrix of genes and gene signatures in n = 944 SCLC tumours, including TIP genes and mouse (M) and human (H) ionocyte (iono) signatures (Supplementary Table 2). Yellow box highlights tuft/iono correlations. d, Expression of human SCLC subtype signatures (from data in a and Supplementary Table 7) in RPM/RPMA basal-derived tumours (from Fig. 2e) by UMAP or in violin plots grouped by SCLC fate. e, Pearson correlation matrix comparing enrichment scores of human SCLC subtype signatures (from four independent datasets: Caris, Liu et al.23, George et al.39 and Lissa et al.61), normal lung cell type signatures (basal, tuft and NE) and ChIP–seq target signatures (A, N or P_targets) in RPM/RPMA tumour data (from Fig. 2e). f, GSEA for ‘antigen presentation’ and ‘T cell inflamed’ signatures per indicated SCLC subtype versus ‘all’ other subtypes (data from a). NES and P values were determined using Kolmogorov–Smirnov and permutation testing. g, Expression of ‘antigen presentation’ and ‘inflamed’ human SCLC signatures8,23 in RPM and RPMA tumours (from Fig. 2e) by UMAP or in violin plots grouped by fate. h, Expression of therapeutic targets in RPM/RPMA tumours (from Fig. 2e) by UMAP or in violin plots grouped by fate. i, Graphical abstract: Waddington landscapes depict SCLC fate trajectories from NE (left) versus basal (right) origins. SCLC-Y from NE origins are transcriptionally similar to A/N cells from basal origins and are therefore omitted. The arrow thickness reflects frequency of trajectories in RPM GEMM. MYC, PTEN and ASCL1 concentrations vary by fate. All violin plots show median and upper/lower quartiles. Statistics are by Kruskall–Wallis tests and post hoc Dunn’s pairwise comparisons with Bonferroni correction. ****P < 0.0001; NS (P > 0.05); other exact P values indicated. NES, normalized enrichment score. Schematic in i was created using BioRender (https://biorender.com).
