Functional identification of non-coding regulatory regions that drive breast and skin carcinogenesis.

Whole-genome sequencing has revealed hundreds of cancer driver mutations in protein-coding genes, while our knowledge of how non-coding mutations contribute to tumorigenesis remains limited. Here, we develop and perform large-scale functional screens to identify non-coding regulatory elements whose mutations drive cancer development in non-transformed human breast epithelial cells and mouse models of breast and skin cancer. We identify 12 frequently mutated regulatory elements associated with 16 genes as candidate cancer drivers. Mechanistically, we characterize two elements associated with the microRNA MIR21 and the tumor suppressor gene, ZFP36L2. Mutating these regulatory elements dysregulates the expression of MIR21 and ZFP36L2, thereby leading to oncogenic transformation. We identify ZFP36L2 as a p53 target gene and find that mutating the p53-binding motif in the ZFP36L2 promoter is sufficient to drive tumorigenesis. Conceptually, our data indicate that cancer driver mutations in non-coding regulatory elements are much more prevalent and potent than initially assumed.