Landscape of BRAF transcript variants in human cancer.

Abstract

The BRAFV600E mutant kinase is widely studied as a cancer driver and therapeutic target. Here, we investigated how the annotation of the BRAF-reference (ref) and BRAF-X1 variants has evolved in public databases and addressed challenges posed by their discrimination and quantification from short-read sequencing. We built IsoWorm, a bioinformatic pipeline tailored to discriminate and quantify BRAF variants, and employed it to analyze > 600 cancer cell lines and > 1000 cancer tissue samples. Using FLIBase, we reanalyzed TCGA data from > 9000 cancer tissue samples. We consistently found that BRAF-X1 (now BRAF-204) is very abundant in human cancer and its expression is 1.5-75 times greater than that of BRAF-ref (now BRAF-220). Crucially, we identified KIRP-kidney renal papillary cell carcinoma as a cancer subtype in which a high BRAF-204/BRAF-220 ratio is an independent prognostic factor of poor outcome. Our in silico analyses establish BRAF as a mix of two protein-coding transcript variants, with BRAF-204 being more highly expressed than BRAF-220. These findings prompt us to undertake the systematic benchmarking of BRAF-204 against BRAF-220 in terms of molecular mechanisms, biological activities, druggability, and clinical relevance.