Chromatin accessibility shapes tumor architecture and microenvironment in clear cell renal cell carcinoma

Abstract The molecular landscape of clear cell renal cell carcinoma (ccRCC) has been annotated by a number of large-scale sequencing studies, complemented by recent proteomics efforts. However, these bulk analyses are limited by the diverse cell populations that compose these tumors. Here, we obtained single-nucleus RNA-seq (snRNA-seq) and ATAC-seq (snATAC-seq) data from 31 and 13 samples from ccRCC patients, respectively, with matching bulk proteogenomics data collected through the National Cancer Institute Clinical Proteomic Tumor Analysis Consortium (CPTAC). We found an average of 4 distinct populations of tumor cells with substantial transcriptional differences per sample and 7.5% of clusters with epithelial-to-mesenchymal transition (EMT) features, characterized by high expression of known mesenchymal genes, and Wnt pathway genes and enriched TWIST and AP-1 complex (JUN, ATF2) motifs. We also observed substantial tumor genomic heterogeneity, including cluster-specific chromosome 3p loss and 5q gain from the same patient, suggesting both events can drive ccRCC, possibly through tumor cell interactions. We identified growth factor-receptor interactions between IGF1 (macrophages) and IGF1R (tumor cells), HBEGF (macrophages) and EGFR (tumor cells), in addition to the well-known interaction between VEGFA (tumor cells) and FLT1 (endothelial cells). Using paired snATAC-seq and snRNA-seq, we found ccRCC cells were most similar to proximal tubule cells among the six nephron epithelial cell types identified. We also identified previously unknown tumor cell-specific transcription factor (TF) motifs, including those for NFIC, RORC, and RBPJ, regulating genes involved in glycolysis (PFKP), and others potentially in collaboration with HIF1A. Finally, tumors harboring genetic alterations in chromatin remodeling genes (BAP1/PBRM1) showed distinct genome-wide and fine-scale chromatin accessibility profiles compared to other tumors, with enriched motifs for the nuclear factor-κB TFs. In summary, we find that various differential transcriptional network activities are likely key drivers in ccRCC pathogenesis and progression.