Identification and characterisation of novel driver genes in human gastric cancers

Abstract

Next-generation sequencing (NGS) technologies have been increasingly employed to mine mutational data from human normal and disease genomes due to lower costs and better bioinformatics data analysis pipelines and computational infrastructure. Careful downstream validation in independent patient cohorts is essential for confirmation of the significance of the discoveries. Besides, there has yet to be a comprehensive genetic mutational profiling of a large cohort of gastric cancer, a heterogeneous and deadly disease that affects Eastern Asia more than other places in the world. This in-depth study covers a few putative gastric cancer driver genes discovered by our team’s recent whole-genome sequencing study of a large cohort of gastric cancers. The mutational data of these putative gastric cancer driver genes, including MUC6, RHOA, CTNNA2, FMN2, ELF3, PTEN and PREX2, were validated by Sanger sequencing, extended to examine independent patient cohorts and analysed in relation to clinico-pathological features. A stomach-specific mucin, MUC6, was found to be mutated in 10% of gastric cancers, especially in tumours with microsatellite instability (MSI) and older patients. MUC6 expression was significantly downregulated in gastric cancer and its mutation spectrum suggested it to be a mucoprotective tumour suppressor. Besides, I discovered that RHOA mutations and CLDN18-ARHGAP6 and CLDN18-ARHGAP26 gene fusions were both enriched in diffuse-type gastric tumours, identified in 14% and 21% of the samples respectively, and were mutually exclusive in their occurrence. This finding was compared with another dataset derived from The Cancer Genome Atlas (TCGA). Follow-up characterisation of their functional significance revealed defective RHOA signalling: several hotspot mutations in RHOA were shown to inactivate RHOA and its downstream signalling in remodelling the actin cytoskeleton, and enabled small intestinal organoids to evade apoptosis upon dissociation; the CLDN18-ARHGAP26 fusion proteins were found to localise to the plasma membrane and inactivate RHOA. These results correlated with the diffuse-type tumours’ clinical behaviour, and suggest that defective RHOA signalling plays an essential role in the pathogenesis of diffuse-type gastric cancer. Since current bioinformatics tools are not optimised to detect insertions and deletions along microsatellite repeat sequences, exome sequencing data were re-analysed based on a new algorithm optimised to detect InDels in mononucleotide tracts in an attempt to identify coding loci targeted for inactivation by MSI. Candidate MSI target genes were selected for validation. The true mutation rates were compared with the coding and non-coding background as well as published mutation frequencies. MSI target genes that may have true biological significance are those that display significantly elevated mutation rates above the background. Validation of mutations in selected gene loci also enabled the refinement of variant filtering criteria to identify gene sets that may be targeted by MSI. In conclusion, my study provides a comprehensive genomic mutational signature of different subtypes of gastric cancer. The genomic landscape enhances our understanding of the molecular basis for gastric cancer pathogenesis, thence enables the identification of novel biomarkers for diagnostics, prognostics and therapeutic interventions, with the hope of developing a cure for cancer in the near future.