Project description:BACKGROUND & AIM: Sessile serrated adenoma/polyps (SSA/Ps) are the likely culprit of ~20% colon cancers but they are molecularly poorly understood. We investigated their epigenetic phenotype using high-throughput analysis of DNA methylation and gene expression. METHODS: 17 SSA/Ps and, as a comparative group, 15 conventional adenomas (all with matched samples of normal mucosa) were prospectively collected during colonoscopy. DNA from the 64 tissues was analyzed, via bisulfite next generation sequencing, for methylation at ~2.7 million CpG sites located prevalently in gene regulatory regions spanning 80.5Mb (~2.5% of the genome). The transcriptome of these samples was also investigated using RNA sequencing. An independent series of 61 archival lesions was used for targeted verification of DNA methylation. RESULTS: Both SSA/Ps and conventional adenomas showed a profound remodelling of their methylome. Cytosine hypermethylation was more pervasive in SSA/Ps than adenomas, in terms of number of hypermethylated regions (22,147 vs 15,965, respectively) and genes (4938 vs 3443, respectively). In addition, the extent of hypermethylation in a given region was higher in SSA/Ps than adenomas. This methylation pattern of SSA/Ps was reminiscent of the CpG island methylator phenotype (CIMP) of their descendant cancers. We have called this phenotype proto-CIMP since most of the hypermethylation occurred in CpG islands and shores. SSA/Ps were also protected from a wave of demethylation that instead occurred in adenomas outside of CpG islands/shores (4288 vs 18,417 hypomethylated regions in SSAPs vs adenomas). Verification studies of six hypermethylated regions demonstrated the high potential of DNA methylation markers for predicting the diagnosis of SSA/Ps and/or adenomas. Proto-CIMP of SSA/Ps was surprisingly associated with a higher number of up- (618) than downregulated (349) genes, while adenomas showed the opposite trend (516 up- vs 712 downregulated genes). CONCLUSIONS: The epigenetic landscape of SSA/Ps differs substantially from that of conventional adenomas. This huge amount of epigenetic variation represents a rich source of promising diagnostic tools, such as novel DNA markers or histologic stainings, for the tailored management of the two most frequent colon cancer precursors.

Project description:We tamoxifen treated 8-12 week old mice that had floxed alleles of the following: 1) both Apc alleles (giving rise to Apc truncation/inactivation); 2) both Cdx2 alleles (giving rise to Cdx2 inactivation; 3) one Braf allele, that upon Cre-mediated recombination gives a Braf V600E mutant allele (details below), and 4) the combination of both the Cdx2 alleles and the BrafV600E allele. All four of those groups also had a CDX2P-CreERT2 transgene that expresses Cre recombinase fused to a tamoxifen-regulated fragment of the estrogen receptor ligand binding domain. CreERT2 expression occurs only in tissues where the Cdx2 gene is expressed, which is almost exclusively in adult mouse cecum and colon epithelium. A fifth group of mice had the floxed Cdx2 alleles, but no CDX2P-CreERT2 gene. Treating the mice having CDX2P-CreERT2 with tamoxifen permits the Cre recombinase to enter the cell nucleus and recombine the Apc, Braf, and/or Cdx2 alleles containing loxP sequence elements. Mice were treated with intraperitoneal injection of tamoxifen dissolved in corn oil. Three mice per group were used. The control mice did not develop tumors or any morphological or histological changes in their epithelium, but their colons were used to create the 3 control samples. To obtain the BrafV600E allele we used a genetically engineered mouse line previously described by Dankort et al. (Genes Dev 2007, 21:379-84) that can express the BrafV600E mutant protein following Cre-mediated recombination. The Braf(CA) (Braf-Cre-activated) allele mice carry a gene-targeted allele of Braf, where Braf sequences from exons 15-18 are present in the normal mouse Braf intron 14, followed by a mutated exon 15 (carrying the V600E mutation). The exon 15-18 sequence element is flanked by loxP sites. In the absence of Cre-mediated recombination, the Braf(CA) allele expresses a wild type Braf protein. Following Cre-mediated recombination, the Braf exon 15-18 element is removed, and the Braf(CA) allele then encodes the Braf V600E protein (from the introduced mutated exon 15). RNA was purified from tumor or normal tissue, and targets for Affymetrix arrays were synthesized from the mRNAs. We used Affymetrix Mouse Gene 2.1 ST arrays, which hold 41345 probe-sets, but we largely analyzed just those 25216 probe-sets that were mapped to Entrez gene IDs. Raw data was processed with the Robust Multi-array Average algorithm (RMA). Data is log2-transformed transcript abundance estimates. We fit a one-way ANOVA model to the five groups of samples. We supply a supplementary excel workbook that holds the same data as the data matrix file, but also holds the probe-set annotation at the time we analyzed the data, and some simple statistical calculations, which selects subsets of the probe-sets as differentially expressed between pairs of groups, as well as significant Cdx2-/- by Braf V600E interactions. It also gives the homologous human gene IDs we used for enrichment testing, which were 1-to-1 best homologs according to build 68 of NCBI's Homologene. A second supplementary sheet shows the data we enrichment tested after collapsing to distinct human homologs, joins of the results of tests with GSE4045 data and of tests with TCGA data to the mouse genes, and the intersections of selected genes in those data set with our gene selections in mouse. Consumers should consider obtaining more up-to-date probe-set annotation for the array platform.

Dataset Information

Transcriptome of sessile serrated adenoma/polyps shows a relation with MSI-high colorectal cancer and decrease of CDX2

Dataset's files

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets