NEW YORK (GenomeWeb) – A Salk Institute-led team has generated a map of the human methylome, gaining insight into patterns of DNA methylation of various tissues.
By analyzing 18 tissue types from four individuals, the Salk's Joseph Ecker and his team developed a map of cytosine methylation along the genome. The map, along with matched genomic and transcriptomic data, revealed tissue-specific variations in methylation patterns, the researchers reported today in the online, early edition of Nature.
"What we found is that not all organs we surveyed are equal in terms of their methylation patterns," Ecker said in a statement. "The signatures of methylation are distinct enough between organs that we can look at the methylation patterns of a tissue and know whether the tissue is muscle or thymus or pancreas."
Analysis of the map also found that non-CG methylation was more common than previously thought and enabled the researchers to predict genes that might be escaping from X-chromosome inactivation.
To survey the human epigenome with a particular emphasis on cytosine methylation, the researchers obtained post-mortem tissue samples from 18 different parts of the body from four individuals. In addition to performing MethylC-seq on the samples, they also conducted mRNA-seq and genome sequencing.
From this, they reported that about 15 percent of the CG sites they examined were differentially methylated. According to the researchers, some 60 percent of the differentially methylated regions they uncovered were novel.
Hypomethylation at these differentially methylated regions was linked with tissue-specific functions, Ecker and his colleagues said. For instance, they reported that strongly hypomethylated regions in aortic tissue overlapped with aorta-specific super-enhancers located near MYH10, a gene involved in blood vessel function.
Tissues belonging to the same organ system — such as heart and muscle tissues — tended to cluster together when the researchers performed a hierarchical clustering analysis on hypomethylation at differentially methylated regions. They observed a similar clustering in their transcriptome-level data.
Further, methylation at these differentially methylated regions correlated inversely with gene expression, with that correlation becoming stronger at spots close to transcription start sites. The researchers noted that the strongest negative correlation wasn't actually at the gene promoters, but at locations slightly downstream of promoters.
"In the past, people have really thought the promoter or the upstream regions are where everything is happening," Ecker added. "But we found that methylation changes that are most correlated with gene transcription are often in the downstream regions of the promoter."
He and his colleagues also found that predicted enhancers and putative promoters together only accounted for about half of the intragenic differentially methylated regions.
The remaining undefined intragenic differentially methylated regions (uiDMRs) could represent a previously unrecognized type of functional elements. Methylation of these uiDMRs, the researchers added, correlated strongly with the expression of the genes that contained them.
In addition, Ecker and his team uncovered evidence of non-CG methylation, which had previously been found in human embryonic stem cells and a few other cell types. This mCH methylation, they further reported, could be split into two groups based upon the context of the methylation, whether it fell in a TNCAC or NNCAN motif.
"The only place this had been observed before was in the brain, skeletal muscle, germ cells, and stem cells," first author Matthew Schultz, a former Ecker lab graduate student now at Human Longevity, said in a statement. "So, to see it in a variety of normal adult tissues was really exciting."
They also noted a negative correlation between expression and mCG and that, except for in brain cells, mCH levels dropped during differentiation.
The methylome map also highlighted female-specific promoter mCG hypomethylation and gene body mCH hypermethylation that appeared to mark genes that escape X chromosome inactivation. Gene body mCH, they added, was particularly able to predict bi-allelically expressed genes.
For instance, they found that genes like FUNDC1 exhibited female-specific hypermethylation in a number of tissues, except for in neurons, which suggested to the researchers that there could be tissue-dependent regulation of X inactivation escape.
Ecker and his colleagues also argued that such methylome maps could help researchers understand the interplay between disease and epigenetics.
"You could imagine that eventually, if someone is having a problem, a biopsy might not only look at characterizing the cells or genes, but the epigenome as well," Ecker said.