Skip to main content
Premium Trial:

Request an Annual Quote

Methylation's Moment


With the monster challenge that was deciphering the human genome sufficiently wrestled into submission, an increasing number of researchers have directed their attention — and their challenge-grappling skills — toward the epigenome and the role of DNA methylation in gene expression. In the last few months alone, numerous studies have yielded discoveries that promise to pave the way forward for truly in-depth characterizations of the human epigenome.

One such project was conducted by an international team of investigators from the Genome Institute of Singapore and the Scripps Research Institute, which recently completed the first-ever direct comparison of DNA methylation patterns present during two different stages of development in the same human embryonic stem cells. The team used this approach in order to map methylation changes and determine how genes are regulated during human development — a question the research community has only just begun to tackle, says co-author Jeanne Loring, director of the Center for Regenerative Medicine at Scripps.

Much of what Loring and her colleagues have observed about methylation thus far has been a bit surprising. "It was sort of like the first time you see the genome sequence — it's pretty overwhelming. But if you focus down on certain genes, you can see that methylation is clearly responsible for turning some of them on, and others off, during differentiation," she says. "We didn't know exactly which genes were involved, and there was some weird things going on — it's not just CpG islands that are affected. There are single cytosine sites and CpA sites that we see are methylated or unmethylated, so there's some really novel things that we've uncovered and we are just at the surface."

Loring says that, in addition to the study's findings, this was also a strong example of the usefulness of high-resolution methylation maps for discovering developmental regulatory mechanisms. Loring's study follows up on — and helps confirm — research by Joseph Ecker, a professor at the Salk Institute for Biological Studies, who published the first complete mapping of the methylome of two human cell lines in Nature late last year. That study presented the first genome-wide, single-base-resolution maps of methylated cytosines from human embryonic stem cells and fetal fibroblasts. By comparing the methylomes, the researchers were able to observe many differences in the make-up and patterning of cytosine methylation between the two genomes.

Bisulfite and beyond

Even though bisulfite sequencing, a robust method that measures levels of DNA methylation at single-nucleotide resolution, is accepted as the only game in town for differentiating between methylated and unmethylated cytosines, there are no standardized technical approaches. While Loring says she and her colleagues will "unfortunately have to live with it" for the time being, there is a real need for a kit-like approach. "The amount of DNA you're going to have is limiting. We can't do 1010cells, we can make 108 cells if necessary, so we don't have a lot of stuff to play with. We need to have more efficient ways of doing the conversion, which has a detrimental effect on the DNA because it causes the DNA to become fragmented, so it's harder to map," Loring says. "Because we're converting from a four-base code to a three-base code, that raises a lot of challenges for people who are trying to map sequences to a standard genome — which is hard enough because the genome isn't exactly standard. … All of these things are improvable. Our recent paper was just a pioneering effort."

[ pagebreak ]

Efforts to improve upon targeted bisulfite sequencing have yielded promising results. Alexander Meissner, assistant professor of biology at Harvard University, recently led a team that streamlined a method called reduced representation bisulfite sequencing on Illumina's Genome Analyzer platform. In a paper published earlier this year in Nature, Meissner's group demonstrated the method on just 30 nanograms of DNA extracted from human formalin-fixed colon cancer tissue. Unlike classic bisulfite sequencing or random shotgun bisulfite sequencing for the whole genome, Meissner's method is much more precise. "We are specifically enriching for CpG dinucleotides, and that's giving you a very relevant representation of methylation in the genome because wherever we have slight enrichment, we get very good coverage, which are again usually promoter regions and regulatory regions," he says. "In our case, we actually do a single lane of Illumina sequencing and get something like 75 or 90 percent of all the promoters and CpG islands. … While we're not getting everything, we're getting a huge majority of the genome that is relevant for biomarker and novel gene discovery at a dramatically increased throughput and cost reduction, so you can just make it more feasible to screen for hundreds of thousands of samples, which is necessary to find the most appropriate markers."

Jean-Pierre Issa, a professor at the University of Texas MD Anderson Cancer Center who focuses on the role of methylation in aging and cancer, says that sequencing technologies are not yet advanced enough to completely abandon microarrays. "I would love nothing more than to get rid of microarrays, and in fact, we rarely do them anymore," he says. "But the problem with deep sequencing is the cost and throughput, so improvements there would be essentially transformative. And in a few years this will be the main way that we would do DNA methylation, at least from a whole-genome way."

This February, Issa and his colleagues published a study that looked at the role that DNA methylation plays in aging mammals. He and his team performed DNA methylation profiling of promoter regions in aging mouse intestine using methylated CpG island amplification along with microarray analysis. Their findings showed a surprisingly high rate of both hyper-methylation and hypo-methylation as a result of age in normal mouse small intestine tissues, leading them to conclude that a common feature of the mammalian aging cycle is epigenetic deregulation.

Instead of using whole genome shotgun bisulfite sequencing, which has disadvantages related to the number of tags and amount of analysis required, Issa's lab uses a variation on a restriction enzyme-based approach in which distinguishing methylation site is done through differential sensitivity to restriction enzymes. "The way we are doing it right now, we get anywhere from 50,000 to a couple hundred thousand CpG sites measured in one sequencing run at a depth of coverage that is quantitative enough for clinical applications," Issa says. "I think that a variation on this is probably going to be more helpful than simply doing whole genome shotgun bisulfite sequencing, at least in the next few years."

Researchers such as Yuan Gao, an assistant professor at the Center for the Study of Biological Complexity at Virginia Commonwealth University, are also looking for ways to improve bisulfite sequencing. In April 2009, Gao and his colleagues published two papers side by side in Nature Biotechnology. One describes a targeted methylation-profiling method called a bisulfite sequencing padlock probe that uses padlock probes to capture tens of thousands of bisulfite-converted short genomic targets that contain CpG sites for sequencing. The other paper presents a profiling method called methyl-sensitive cut counting, which profiles methylation across whole genomes by using two restriction enzymes: one that cuts unmethylated CCGG sites and one that cuts both methylated and un- methylated regions. "The basic idea for estimating CpG methylation in the CCGG site is to use sequencing read counts matched to the HpaII-cutting sites and MspI-cutting sites," Gao says. "By using a spike-in standard with known methylation level, the unknown methylation level can be estimated from the read counts derived from HpaII and MspI libraries."

[ pagebreak ]

Methylation as a biomarker

Issa and his team have also recently demonstrated the potential of DNA methylation as a predictive biomarker in the clinic. Because an epigenetic therapy that uses the DNA hypo- methylating agent decitabine has proven to be effective in the treatment of myelodysplastic syndromes, they investigated the link between DNA methylation and the clinical outcome of MDS. Their research showed a significant correlation between reduced methylation over time and clinical responses, indicating that DNA methylation can be useful for predicting survival and treatment response in patients with MDS.

Earlier this year, Peter Laird, an associate professor at the Keck School of Medicine, completed a study that looked at DNA methylation as a potential biomarker for cardiovascular disease. He and his colleagues found that elevated peripheral blood leukocyte DNA methylation is associated with the prevalence of cardio-vascular disease, predisposing conditions, and obesity in Singaporean Chinese. However, Laird remains cautiously optimistic about what these findings mean for the clinic, as huge studies are needed over many years in order to demonstrate that a marker is really worth something. "The cardiovascular paper we published hints at having predictive value, but since it's a cross-sectional study, it can't really make that claim — although it suggests that it might be the case because we have some blood samples that were collected before people were diagnosed," Laird says.

When it comes to DNA methylation in the clinic, the interest seems to primarily center on early detection as opposed to monitoring disease progression, because the potential is larger and it's easier to follow up on risk assessment. "There's a lot of interest in early detection because DNA methylation changes and epigenetic changes are early in many cases and they can certainly be detected in biopsies. There is also interest in developing it for bodily fluids, whether it's blood for DNA that is leaked into the bloodstream from the lymphatic [system], from the growing tumor, or in the luminal contents," Laird says. However, he says that academic labs and research institutes likely don't have deep enough pockets and the wherewithal to push things through the FDA approval process for clinical implementation of DNA methylation biomarkers.