Fig. 3

ICB responders and non-responders show different clonal response patterns. a The size of tumors from the same mouse is positively correlated. b Correlation of barcode distribution of tumors from the same mouse is significantly higher than that from different mice. c Hierarchical clustering of in vivo tumor samples based on barcode distribution. Generally, responders cluster separately from non-responders and control-treated tumors. Each row represents a specific barcode, and each column represents a tumor sample. To assist visualization, rows were ordered by the difference between the responders and the control IgG group. d Clonality of barcode distribution is significantly higher in ICB responders (mean ± SD; **P < 0.01, ***P < 0.001; one-way ANOVA with Benjamini-Hochberg post-test multiple comparison). “res” are responders and “non” are non-responders. e Summary of percentage of intra-tumoral CD4+ and CD8+ cells in control IgG, ICB responders, or non-responders (mean ± SD; ^#P < 0.1, ***P < 0.001; one-way ANOVA with Benjamini-Hochberg post-test multiple comparison). f GSEA analysis of bulk tumor RNA-seq reveals higher expression of genes involved in T cell proliferation or B cell-mediated immunity in responders compared to non-responders. The entire list of enriched gene sets can be found in Additional file 3: Table S2. g Heatmap of relative expression of the top differentially expressed genes. Within the gene sets that were enriched, we picked representative genes to reflect the differential enrichment of gene sets. Responders had higher expression of genes in adaptive immunity. h TIDE reveals higher T cell exclusion scores in non-responders than responders (mean ± SD; *P < 0.05; n.s., not significant; one-way ANOVA with Benjamini-Hochberg post-test multiple comparison). i Correlation of bulk tumor expression in each group with MDSC, cancer-associated fibroblast (CAF), or M2 macrophage. Responders showed lower correlation with MDSC and M2 macrophage than control IgG or non-responders (mean ± SD; *P < 0.05; n.s., not significant; one-way ANOVA with Benjamini-Hochberg post-test multiple comparison). j, k Differential gene expression signature in ICB responder from our study correlates with better survival in the Mariathasan et al. (j) or Van Allen et al. (k) studies. P values were derived using Cox-PH regression treating signature score as a continuous variable. l Cox-PH regression coefficient value (mean ± SEM) and z score of the signature score in its correlation with survival hazard in multiple ICB treatment clinical trials