Effect of Cx3cl1 on tumor-infiltrating immune cells. (A and B) Feature plots overlaid on UMAP representation for Cx3cr1 and violin plots showing relative average expression of Cx3cr1 of the tumor analyzed by ZipSeq in Fig. 6 (A) and Fig. S3, D–L (B). (C) Correlation between an average relative Cx3cr1 transcript level of each cluster and the percentage of cells belonging to the cluster. Combined analysis of two tumors analyzed in Fig. 6 and Fig. S3, D–L. ZipSeq analysis was performed on a total of two tumors. (D) MFI of CX3CR1 on major immune cell types in Cx3cl1-overexpressing and empty vector–expressing CIT6-YFP tumors (n = 20–21). (E) Frequency of lymphocytes and DCs in Cx3cl1-overexpressing and empty vector–expressing CIT6-YFP tumors (n = 20–21). Data in D and E are a representation of two independent experiments. (F) Frequency of TIGIT⁺ TIM-3⁺–exhausted CD4 T cells in Cx3cl1-overexpressing and empty vector–expressing CIT6-YFP tumors (n = 9–10). Experiment was done once. (G) CX3CL1 concentration in supernatants of Cx3cl1 KO CIT9-RFP cell line (left) and CIT9 cell line (right). Data are an average of two experiments with a total of eight tumors per arm. (H) Growth curve of CIT9 parental and CIT9 Cx3cl1 KO tumors (n = 10 per group). (I) CX3CL1 concentration in tumor lysate of Cx3cl1 KO CIT9 cells (n = 8–10). (J–L) Frequency of total macrophages (J), CD206^Hi and non-CD206^Hi macrophages (K), and T cells (L) (n = 10). Data in H–L are a representation of two independent experiments. (M) Ratio of macrophage score and monocyte score (top) and CD206^Hi macrophage score (bottom), based on RNA-seq of mononuclear phagocytes, for human tumors in seven hot immune archetypes (n = 5–28). Statistical significance was determined by Pearson’s correlation test (C), one-way ANOVA (E, F, and I–L), two-way ANOVA (H) with Tukey’s correction, or by Mann–Whitney test with the Bonferroni correction (M). *P < 0.05 and ***P < 0.001. MFI, median fluorescence intensity; ILC,  innate lymphoid cell.