To characterize the cancer dependency of NR genes (also called cancer cell fitness, defined as genes required for cancer cell growth), we analyzed the genome-wide CRISPR and RNAi screening data in large-scale cancer cell lines from the DepMap project (Dataset 13). A total of 10 (20.8%) NRs were defined as “strongly selective” essential genes by CRISPR or RNAi screening in select cancer lineages or subpopulations of cancer cells, while no NRs functioned as “common essential” genes across cancer cell lines (Figure 6A). The genes defined as strongly selective were significantly enriched in the NR family compared to non-NR genes in CRISPR screening (Fisher's Exact Test, odds ratio=5.4, p=0.0003). This suggests that select NRs play crucial roles in cell growth and survival in specific cancer cell populations, providing a strong rationale for developing NR-targeting drugs to treat cancer. Importantly, 4 strongly selective NRs (ESR1, AR, RXRA and RARG) are currently serving as therapeutic targets in certain cancer types, 2 of 10 strongly selective NRs are targets of approved drugs for other diseases (Tclin), and 3 of them are targeted by known chemical compounds (Tchem), providing novel opportunities for drug repurposing in cancer treatment. Notably, many “strongly selective” NRs, for example NR1H3, HNF4G and NR5A2, were the subject of fewer studies in biology and cancer (low Pubtator scores), although NR1H3 is involved in early- stage oncology clinical development. This suggests that there is an unmet need for exploring these genes’ functions in tumorigenesis. Based on expression and dependency, the “strongly selective” NRs were divided into 3 groups (Figure 6B). Group 1, including ESR1, PPARG, HNF4A, HNF4G, NR5A2 and AR: the NRs were specifically expressed in certain cancer cells which were functionally dependent upon these NRs. Group 2 including RARG, NR3C1 and NR1H3: the NRs were ubiquitously expressed across all cancer types; however, only tumor cells in select cancer types were highly dependent on these NRs. Group 3 including RXRA: this NR was ubiquitously expressed across all cancer types; however, in each cancer type only a subpopulation of tumor cells was dependent on RXRA. Next, we analyzed the correlation between cancer dependency (gene effects) and mRNA expression for each NR at a pan-cancer level. 21 NRs showed significant and negative correlations between their gene effects and mRNA expression levels across all cancer cell lines (Figure 6C). Among them, 8 NRs, including HNF4A, ESR1, PPARG and RXRA, exhibited high levels of recurrent genomic alterations that were defined by our TCGA analyses, and the cancer cell lines carrying certain genomic alterations were more dependent on the given NR genes compared to the whole cancer cell population (Figure 6D).
