The co-evolution of the genome and epigenome in colorectal cancer

Abstract

Colorectal malignancies are a leading cause of cancer-related death1 and have undergone extensive genomic study2,3. However, DNA mutations alone do not fully explain malignant transformation4–7. Here we investigate the co-evolution of the genome and epigenome of colorectal tumours at single-clone resolution using spatial multi-omic profiling of individual glands. We collected 1,370 samples from 30 primary cancers and 8 concomitant adenomas and generated 1,207 chromatin accessibility profiles, 527 whole genomes and 297 whole transcriptomes. We found positive selection for DNA mutations in chromatin modifier genes and recurrent somatic chromatin accessibility alterations, including in regulatory regions of cancer driver genes that were otherwise devoid of genetic mutations. Genome-wide alterations in accessibility for transcription factor binding involved CTCF, downregulation of interferon and increased accessibility for SOX and HOX transcription factor families, suggesting the involvement of developmental genes during tumourigenesis. Somatic chromatin accessibility alterations were heritable and distinguished adenomas from cancers. Mutational signature analysis showed that the epigenome in turn influences the accumulation of DNA mutations. This study provides a map of genetic and epigenetic tumour heterogeneity, with fundamental implications for understanding colorectal cancer biology.

An Epigenetic LINE-1-Based Mechanism in Cancer

“…We would like to note that the model proposed here, by and large, is consistent with experimental evidence published in two very recent studies of a large number of colorectal cancers [ 114 , 115 ]. We expect that the novel view emerging from these studies might indeed apply to cancers of many different origins.…”

Phenotypic plasticity and genetic control in colorectal cancer evolution

“…We used our extensive single-gland, multi-region WGS data (deep WGS, median depth 35×, between 3 and 15 samples per patient (median, 8) and low-pass WGS (median depth 1.2×, between 1 and 22 samples per patient, median 8) for accurate identification of clonal and subclonal somatic variants (https://doi.org/10.6084/ m9.figshare.19849138 from ref. 23 ) and to call somatic copy number alterations in each tumour (note that this included additional tumours lacking RNA-seq data). We specifically examined the clonality of 69 genes (excluding PARP4, LRP1B and KMT2C, which we excluded due to a high number of false-positive low-frequency variants in these genes) on the IntOGen list 41 of putative CRC driver genes (Methods and Fig.…”

“…The method of sample collection and processing is described in a companion article (ref. 23 ). Sequencing and basic bioinformatic processing of DNA-, RNA-and ATAC-seq data are included there as well.…”

“…To concentrate on tumour epithelial cell gene expression, genes were further filtered out if they negatively correlated with purity as estimated from matched DNA-seq data (see associated ref. 23 for methodology of purity estimation). Specifically, for the 157 tumour samples that had matched DNA-seq and therefore accurate purity estimates, a linear mixed-effects model of 'expression (VST) ~ purity + (1 | patient)' (where '~' represents 'is distributed as') was compared via a chi-squared test to 'expression ~ (1 | patient)'.…”

“…For each tumour with at least five tumour samples (n = 17 tumours; note that, except for the large advanced C516 adenoma, adenomas used in ref. 23 did not undergo RNA-seq), mean expression and s.d. of expression were calculated for every filtered expressed gene (n = 11,401) using DESeq2 VST normalized counts (inspired by ref.…”