Extended Data Fig. 7: Isolating EpiSen-high and EpiSen-low cells in model cell lines.

a, Gating scheme for isolation of EpiSen-high and EpiSen-low subpopulations in the JHU006 cell line including doublet and dead cell exclusion. For sorting, the top 10% high and bottom 10% low cells were sorted. Validation of AXL+CLDN4- (EpiSen-high)-sorted cells enriching for the full EpiSen RHP was performed by bulk RNA-Seq of sorted cells (Fig. 5c, Methods). b, FACS analysis of cell cycle by the DNA binding dye propidium iodide (PI) on sorted EpiSen-high and EpiSen-low cells in SCC47, as shown for JHU006 in Fig. 5d. The table summarizes the results. c, Cell cycle scores of the G1/S (X-axis) and G2/M (Y-axis) programs, shown for in vivo HNSCC cells from six tumors (panels). Top 10% of cells scoring highly for the EpiSen-related program (EpiDif1) are shown in black, demonstrating their enrichment among non-cycling cells in each of the tumors (P<0.001, hypergeometric test). d, Three subpopulations were isolated by FACS (EpiSen-high: AXL^-CLDN4⁺, EpiSen-low population: AXL⁺CLDN4^-, control: unsorted) from SCC47 (left) and JHU006 (right), and analyzed immediately after sorting (day 0, top) and at two additional time points in culture (day 7 and 14, middle and bottom, respectively). Density plots correspond to the pie charts in Fig. 5e.