Fig. 2

Identification of pan-cancer proteomic regulators. A The top pan-cancer protein network drivers. The top bar shows the frequencies of up- and down-regulations of each pan-cancer protein driver in the seven cancers while the 2nd bar from the top shows the frequency of the hub status of each protein driver in the seven cancers. The first and second heat maps from the top represent the enrichment of the up- and down-regulated cancer-type-wise DEP signatures in the neighborhoods of the protein drivers, respectively. The color intensity is proportional to –log10(FDR corrected FET p value). The bottom heatmap summarizes the percentage of significant hits for each protein driver in the CRISPRi screening of cancer cell lines from Archilles database with FDR < 0.05. B–D Pan-cancer neighborhood networks of the top-ranked novel regulators, DDX21 (B), RSL1D1 (C), and SMC2 (D). The links are color-coded by the cancer types. The piechart of each node shows the proportions of links from different cancer types. E–G Anti-tumor activities by silencing the predicted pan-cancer proteome regulators, RSL1D1, DDX21 and SMC2. We conducted shRNA knock-down of the predicted regulators in lung cancer (H847), colon cancer (HCT116), fetal kidney (HEK293T) and breast cancer (MDA-MB-231), with the scrambled shRNAs as controls. E Confluence of different cancer cells transfected by shRSL1D1 (light blue), shDDX21 (brown) and shSMC2 (green), compared to the scrambled control (Scrambled, black). The confluences (y-axis) were measured from day 1 to day 4 (x-axis). F Rate of confluence change in subsequent days. Cases showing significantly lower rate of change, compared to the scrambled control, are marked by red asterisks with different levels of significance shown at the top legend. G Relative cell viability change in comparison to the scrambled control by CTG luminescent cell viability assay