Identification of disease specific dependency genes through comparative analysis of b-cell derived malignancies using a novel bioinformatic approach

Abstract

Despite recent improvements in cancer treatment, many patients still relapse after initial treatment and survivors often exhibit serious long-term health effects. Therefore, more cancer specific treatments, leading to increased survival and reduced long-term toxicities are required. Cancer cells exhibit consistent and dramatic changes in global DNA methylation patterns. Our group has recently developed a new bioinformatic approach which integrates these changes in global DNA methylation with genome-wide expression data to identify candidate subtype specific synthetic lethal genes in specific cancer subtypes. A limitation to this approach, is that any potential therapeutic targets identified are only relevant in the specific subset and not all patients with that cancer type. However, as highly similar methylation changes are shared across all B-cell derived malignancies, we hypothesised that this approach could be expanded to allow comparison between different types of B-cell malignancies to allow the identification of Disease Specific Dependency Genes (DSDG) and disease specific tumour suppressor genes (TSG), which would be functionally important in all subtypes of a particular cancer type. This analysis was able to provide proof-of-principle that the bioinformatic approach could successfully identify candidate functionally relevant genes at the whole disease level. In total, 13 candidate genes (seven DSDG and six TSG), were identified. Functional assessment of DSDG candidates did not identify any clear functional impacts of targeting DSDG by siRNA-mediated knockdown, however, this approach resulted in only partial, temporary reductions in expression and alternative approaches to allow stable knockdown will be required. Functional assessment of TSG candidates in ALL identified SLC22A15, as a novel negative regulator of ALL cell growth across all tested ALL genetic subtypes. These results provide initial proof-ofprinciple that our original approach for identifying functionally relevant genes at a subtype specific can be expanded to the whole disease level and can uncover previously unknown functionally relevant genes.