Supplemental Figure S1 shows the identification and characterization of major immune cell populations and states across both timepoints, using UMAP visualization, unsupervised clustering, and marker gene expression profiling. Supplemental Figure S2 shows the distribution of immune cell types across samples, timepoints, and response groups, as well as the lack of association between metastatic site and response to immune checkpoint blockade. Supplemental Figure S3 shows the comparison of immune cell type proportions across timepoints and response groups, and validates the association between NK cells and lack of response using multiple independent melanoma and breast cancer cohorts. Supplemental Figure S4 shows flow cytometry-based characterization of peripheral blood NK (PBNK) cells, including gating strategy, subset distribution, and expression of cytotoxic markers across treatment timepoints and response groups. Supplemental Figure S5 shows the relationship between immune cell composition and tumor-infiltrating lymphocyte (TIL) patterns, CD8⁺ T cell density across regions and response groups, and spatial localization of NK cells in representative lesions from responders and non-responders. Supplemental Figure S6 shows spatial transcriptomic analysis of the tumor microenvironment (TME) using Xenium, including clustering and marker gene expression of TME cells, spatial localization of NK cells in cold and immune-excluded tumors, and neighborhood enrichment patterns of cell types in cold tumors. Supplemental Figure S7 shows the spatial distribution of NK cells in murine melanoma models (YUMM5.2 and NRAS;Ink4a) using immunofluorescence, HInk4a and YUMM5.2 murine tumors across treatment conditions, highlighting immune and malignant cell cell composition, and CD8⁺ T cell abundance in the spleen after depletion of this cell type. Supplemental Figure S10 shows immune profiling of NRAS;Ink4a and MC38 tumors under control and anti–PD-1 conditions using CITE-seq, flow cytometry, and multiplex immunofluorescence, highlighting NK cell dynamics, immune composition, and spatial organization across treatment cohorts.
Poźniak et al. (Thu,) studied this question.