Sessile serrated adenomas/polyps (SSA/Ps) are the putative precursors of the ~20% of colon cancers with the CpG island methylator phenotype (CIMP). To investigate the epigenetic phenotype of these precancers, we prospectively collected fresh-tissue samples of 17 SSA/Ps and 15 conventional adenomas (cADNs), each with a matched sample of normal mucosa. Their DNA was subjected to bisulfite next-generation sequencing to assess methylation levels at ~2.7 million CpGs located predominantly in gene regulatory regions and spanning 80.5Mb; RNA was sequenced to define the samples’ transcriptomes. Compared with normal mucosa, SSA/Ps and cADNs exhibited markedly remodeled methylomes. In cADNs, hypomethylated regions were far more numerous (18,417 vs 4288 in SSA/Ps) and rarely affected CpG islands/shores. SSA/Ps seemed to have escaped this wave of demethylation. Cytosine hypermethylation in SSA/Ps was more pervasive (hypermethylated regions: 22,147 vs 15,965 in cADNs; hypermethylated genes: 4938 vs 3443 in cADNs) and more extensive (region for region), and it occurred mainly within CpG islands and shores. Given its resemblance to the CIMP typical of SSA/Ps' putative descendant colon cancers, we refer to the SSA/P methylation phenotype as proto-CIMP. Verification studies of six hypermethylated regions in an independent series of precancers demonstrated DNA methylation markers’ high potential for predicting the diagnosis of SSA/Ps and cADNs. Surprisingly, proto-CIMP in SSA/Ps was associated with upregulated gene expression; downregulation was more common in cADNs. In conclusion, the epigenetic landscape of SSA/Ps differs markedly from that of cADNs. These differences are a potentially rich source of novel tissue-based and noninvasive biomarkers.

Sessile serrated adenomas/polyps (SSA/Ps) are the putative precursors of the ˜20% of colon cancers with the CpG island methylator phenotype (CIMP), but their molecular features are poorly understood. We used high-throughput analysis of DNA methylation and gene expression to investigate the epigenetic phenotype of SSA/Ps. Fresh-tissue samples of 17 SSA/Ps and (for comparison purposes) 15 conventional adenomas (cADNs)—each with a matched sample of normal mucosa— were prospectively collected during colonoscopy (total no. samples analyzed: 64). DNA and RNA were extracted from each sample. DNA was subjected to bisulfite next-generation sequencing to assess methylation levels at ˜2.7 million CpG sites located predominantly in gene regulatory regions and spanning 80.5Mb (˜2.5% of the genome); RNA was sequenced to define the samples’ transcriptomes. An independent series of 61 archival lesions was used for targeted verification of DNA methylation findings. Compared with normal mucosa samples, SSA/Ps and cADNs exhibited markedly remodeled methylomes. In cADNs, hypomethylated regions were far more numerous (18,417 vs 4288 in SSA/Ps) and rarely affected CpG islands/shores. SSA/Ps seemed to have escaped this wave of demethylation. Cytosine hypermethylation in SSA/Ps was more pervasive (hypermethylated regions: 22,147 vs 15,965 in cADNs; hypermethylated genes: 4938 vs 3443 in cADNs) and more extensive (region for region), and it occurred mainly within CpG islands and shores. Given its resemblance to the CIMP typical of SSA/Ps’ putative descendant colon cancers, we refer to the SSA/P methylation phenotype as proto-CIMP. Verification studies of six hypermethylated regions (3 SSA/P-specific and 3 common) demonstrated the high potential of DNA methylation markers for predicting the diagnosis of SSA/Ps and cADNs. Surprisingly, proto-CIMP in SSA/Ps was associated with upregulated gene expression (n=618 genes vs 349 that were downregulated); downregulation was more common in cADNs (n=712 vs 516 upregulated genes). The epigenetic landscape of SSA/Ps differs markedly from that of cADNs. These differences are a potentially rich source of novel tissue-based and noninvasive biomarkers that can add precision to the clinical management of the two most frequent colon-cancer precursors.

Germline mutations in the mismatch repair ( MM R) genes MSH2 , MSH6 , MLH1 and PMS2 are linked to cancer of the colon and other organs, characterised by microsatellite instability and a large increase in mutation frequency. Unexpectedly, mutations in EXO1 , encoding the only exonuclease genetically implicated in MMR, are not linked to familial cancer and cause a substantially weaker mutator phenotype. This difference could be explained if eukaryotic cells possessed additional exonucleases redundant with EXO1. Analysis of the MLH1 interactome identified FANCD2-associated nuclease 1 (FAN1), a novel enzyme with biochemical properties resembling EXO1. We now show that FAN1 efficiently substitutes for EXO1 in MMR assays and that this functional complementation is modulated by its interaction with MLH1. FAN1 also contributes towards MMR in vivo : cells lacking both EXO1 and FAN1 have a MMR defect and display resistance to N -methyl- N -nitrosourea (MNU) and 6-thioguanine (TG). Moreover, FAN1 loss amplifies the mutational profile of EXO1-deficient cells, implying that the two nucleases act redundantly in the same antimutagenic pathway. However, the increased drug resistance and mutator phenotype of FAN1/EXO1-deficient cells are less prominent than those seen in cells lacking MSH6 or MLH1. Eukaryotic cells thus apparently possess additional mechanisms that compensate for the loss of EXO1.