Combined analysis of chromatin accessibility, promoter interactions and whole genome sequencing solved the missing heritability in gastric cancer

Abstract

Missing heritability in hereditary diffuse gastric cancer (HDGC) ranges from 60 to 90%. These HDGC-like families, despite complying with HDGC clinical criteria, lack CDH1 and CTNNA1 actionable germline variants, and are not offered HDGC-targeted life-saving disease prevention measures. Herein, we explored novel HDGC predisposition mechanisms affecting the CDH1-regulatory network. We called single-nucleotide (SNV) and copy-number variants (CNV) from 19 HDGC-like probands from whole-genome sequencing data and performed gene-ontology analysis. Chromatin enhancer marks and CDH1 promoter interactions were evaluated in normal stomach by ChIP-seq, ATAC-seq and 4C-seq, variant causality was assessed by RT-PCR, immunohistochemistry and microsatellite instability (MSI) analysis in tumours. Functional analysis was performed using CRISPR-Cas9, RT-PCR and flow cytometry in cell lines, and enhancer assays using mouse embryos. Within the CDH1 topologically associating domain (TAD), we found two deletions in Family F4 and F9. F4 carried a heterozygous CDH3 20kb-CNV triggering CDH1 mRNA/protein loss in homozygosity by CRISPR-Cas9 editing, similarly to a CDH1 coding deletion. This 20kb sequence encloses two hypomorphic tissue-specific regulatory elements (REs), each contributing 50% to CDH1 expression regulation. F9 carried a heterozygous 39bp-intergenic CNV downstream of CDH1, triggering CDH1 mRNA/protein loss by CRISPR-Cas9. F15, presenting gastric but not colorectal cancer, carried an MLH1 heterozygous 2.7Kb germline CNV overlapping a stomach-specific RE found by ChIP-seq. The gastric tumour of mixed histology displayed Microsatellite instability (MSI), reduced MLH1 mRNA and protein, and reduced CDH1 and E-cadherin protein. CRISPR-Cas9 clones mimicking the MLH1 heterozygous CNV, triggered loss of MLH1 and CDH1/E-cadherin mRNA and protein, similar to a coding deletion. Beyond the CDH1 TAD and tumour risk syndrome genes, multiple deletions of stomach accessible chromatin sequences were found in particularly young-affected individuals from additional 6 families. This oligogenic pattern impaired specifically mucin genes and multiple immune-related pathways. Herein, we pinpointed novel mechanisms behind HDGC predisposition. One involves deletions of CDH1-REs in the TAD or stomach-specific CDH1-REs in the MLH1 locus. The second involves multiple deletions of stomach REs affecting mucin and immune-related genes, favouring a gastric immune-deficient phenotype. Altogether, by combining stomach-specific chromatin accessibility and promoter interactions with whole genome sequencing, we solved the missing heritability in 47% of HDGC-like families within our cohort.