BACR 5: Functional evaluation of low-frequency driver mutations in breast cancer
1Institute of Cancer Research, London, UK
International cancer sequencing efforts have unveiled the complexity of the mutational landscape of breast cancer and have revealed that, aside from a few highly recurrent mutations, the vast majority occur at low frequencies. Whilst some low frequency mutations occur in known oncogenes and have been shown to be oncogenic and subsequently targetable, the majority remain uncharacterised.
Our aim is to functionally investigate the biological significance of potentially oncogenic mutations identified in breast cancer in order to identify novel drivers and therapeutic targets.
We compiled a database of mutations identified in publicly available and in-house exome sequencing data from breast cancer. High confidence variants were run through bioinformatic prediction algorithms FATHMM and CHASM to predict which mutations are most likely to have an oncogenic impact, and CANSAR was used to assess which genes are druggable using existing inhibitors. We identified 8 tyrosine kinases and a total of 22 mutations which were predicted to be oncogenic.
Mutations were engineered using site-directed mutagenesis of full length cDNA constructs and mutant and wild-type constructs were transduced into a panel of cell lines to generate stable clones. The phenotypic impact of mutations was assessed using in vitro 2D and 3D assays.
FGFR2 hotspot mutations were found to be constitutively active, with a high global phosphorylation state, consistent with reports in endometrial and ovarian cancer. Mutant cells were highly sensitive to FGFR2 gene silencing and chemical inhibition with AZD4547 and PD173074. Mutations in INSRR were found to have milder phenotypes which were context dependent. Kinase domain mutations conferred a proliferative advantage in 2D culture, which was more pronounced in 3D spheroids.
In conclusion, we found some lower frequency kinase mutations to be oncogenic. However, the effects of moderate-impact driver mutations can only be fully understood in conditions closely resembling in vivo conditions, such as 3D culture.