The main finding of the current study is that both the CDH1 promoter CGI and the group of CpGs located within the 7th enhancer of CDH1 showed increased methylation in the inflamed ileal mucosa of CD patients. Notably CDH1 CGI was found to be methylated in the ileal mucosa of 90% of CD patients, versus only 50% of our « control » group. Also, the increased average methylation of the 4 CpGs in CDH1 enhancer was relatively large (45.2 ± 17% in CD patients vs 35.5 ± 17% in controls). The intestinal epithelium is constantly renewed by intestinal stem cells throughout life. It is interesting to note that the CDH1 locus was also hypermethylated in the inflamed mucosa of patients with UC [41, 49–51] and that this methylation of the CDH1 locus could contribute to predicting the severity of [36, 45, 46]. This observation supports that intestinal inflammation or its local consequences could increase methylation at the CDH1 locus in both CD and UC. While the intestinal epithelium responds to changes in diet, age, microbiome, and immune activation, it is unknown whether these responses occur in mature or progenitor cells and whether they involve epigenetic reprogramming. Indeed, epigenetic mechanisms have recently been recognized as operating at the interface between the microbiome and the intestinal epithelial cell (IEC) genome [60–62].
In PBMC, percentages of methylated CDH1 CGI in our « control » group matched those observed in 1036 healthy controls [63]. We found no difference of CGI methylation between CD and control patients, consistently with previous epigenome-wide association studies (EWAS) performed in blood cells [31–38]. Methylation levels of CDH1 CGI and the 4 enhancer CpGs in PBMC correlated with those found in intestinal mucosa in our controls, as reported in normal people [64]. This correlation suggests that progenitors of PBMC and intestinal epithelium may have shared a common early programming of CDH1 methylation in the endoderm germ layer, with derived cell types retaining these patterns decades later as a stable lineage mark [65]. In contrast, the absence of correlation between intestinal and PBMC methylation in CD patients supports that increased methylation has more recently occurred in the progenitors or mature mucosal cells of CD patients. As discussed in Heijmans et al works [66], high inter-tissue concordance may be present for DNA methylation changes induced early in development (and potentially propagated soma-wide) than for changes occur during aging, which are more likely to remain tissue-specific.
Mucosal tissue from surgical specimens is composed of heterogeneous cell populations that are different in non-inflamed « control » mucosa and inflamed ileal areas of CD. There is a clear concern that our analyses, as those of others [41, 42], were conducted using whole mucosal samples containing mixed epithelial and non-epithelial cell populations, known to have different methylation [41]. Given that methylation signatures are cell-type-specific, the question arises as to whether the epigenetic patterns we observed in CD mucosa arise from the epithelial or non-epithelial cells, and whether they might be confounded by the different cell populations present in inflamed CD versus control mucosa [46]. Among non-epithelial cells, intra-epithelial lymphocytes show lineages diversity and functional states in the intestinal mucosa under both healthy and CD conditions, as well as altered spatial distribution that potentially correlates with transmural inflammation [67]. Since we did not perform mucosal cell purification [67], CDH1 methylation could not be analyzed in infiltrated intra-mucosal lymphocytes. The only indirect information regarding mucosal lymphocytes in our CD patients comes from circulating PBMC, in which CDH1 methylation was comparable to non-inflamed control mucosa or control PBMC. Since enterocyte number largely exceeds non-epithelial cells in inflamed ileal lesions, it is unlikely that the increased methylation levels observed in inflamed CD mucosa could be explained by infiltrated lymphocytes. Indeed, if local inflammation were able to increase CGI methylation in resident lymphocytes or non-epithelial cells, this would not result in the 90% proportion of methylated CGI that we observed in our mucosal samples.
Our findings suggest that mucosal cells of UC and CD may share epigenetic mechanisms at the CDH1 locus despite heterogeneity in location, severity of inflammation and phenotypes [68].
The current observation may also be relevant to the risk of colorectal cancer (CRC) in CD patients. Numerous genes, notably CDH1 have been reported to be hypermethylated and silenced in sporadic CRC [69, 70]. In addition, hypermethylation of CDH1 CGI characterizes UC-associated CRC [56,58] and was supposed to serve as a useful biomarker for detecting UC patients at high risk of developing CRC [71]. Since CD increases the risk of CRC [59], the question arises whether the increased methylation that we found in CD at the CDH1 locus could help predict disease course and associated CRC.
Clearly, methylation marks at the CDH1 locus were not influenced by genomic sequence variation in cis and thus contribute to the epigenetic signature of the inflamed mucosa independently from genetics.
Many questions remain, since the current study is only observational and exploratory and carries the following weaknesses. Although the ileal mucosa studied in the control group was macroscopically normal and taken out of pathological lesions, one cannot consider our control population as truly normal, i.e., non-pathological. Also, our study does not provide extensive information about all CpG residues at the CDH1 locus. It does not provide information about non-inflamed CD mucosa, as could be obtained from intestinal biopsies [41]. It is unfortunate in this respect that the latter study does not provide a comparison between non-penetrating CD mucosa and normal mucosa [41].
More globally, our observation does not elucidate whether increased methylation at the CDH1 locus contributes to the causality of inflammatory lesions of CD or instead is a consequence of the local inflammation induced by CD. The first alternative is that preexisting locally increased methylation at the CDH1 locus predisposes some regions of the intestinal mucosa of future CD patients to inflammation and possibly CRC, as it seems to be the case for UC. Following this hypothesis, methylation marks preexisting in the intestinal mucosa would result from early epigenetic programming and/or later environmental exposures, while such epigenetic marks do not develop in blood cell progenitors. An alternative hypothesis is that the methylation changes that we observed in CD mucosa are merely induced by the local inflammation or other environmental exposures of intestinal cells, such as changes in microbiome [60–62]. In fact, the two hypotheses are not mutually exclusive. They could combine their effects to increase both the pre-existing and secondary methylation at the CDH1 regulatory locus, which may perpetuate local inflammation and trigger CRC risk in the intestinal epithelium of CD patients.