Saturation genome editing of BRCA1 across cell types accurately resolves cancer risk

Germline pathogenic BRCA1 variants predispose women to breast and ovarian cancer. Despite accumulation of functional evidence for variants in BRCA1, over half of reported single-nucleotide variants (SNVs) lack a definitive clinical interpretation. Furthermore, the extent to which variant effects are consistent across cell types remains largely unexplored. Here, we performed saturation genome editing (SGE) of BRCA1 in HAP1 cells to score 4,113 variants not previously assayed. Additionally, we developed a new SGE assay in human mammary epithelial cells (HMECs), allowing effects of variants to be compared across cell lines, drug treatments, and genetic backgrounds. We identify 538 variants impacting function via diverse mechanisms, including impairment of the BRCA1-PALB2 interaction and disruption of splicing, transcription, and translation. Function scores from experiments in HAP1 discriminate known pathogenic and benign variants with near-perfect accuracy. Intriguingly, however, nearly half of variants impacting function in HAP1 were found to be neutral when assayed in HMECs. We show that discordantly scored variants are hypomorphic and confer intermediate cancer risk. These results will be highly valuable for clinical interpretation of BRCA1 variants. Moreover, this work illustrates how revealing context-specific variant effects across cell types can enable more accurate resolution of disease risk.