Tools for Identifying DNA Methylation Targets

When it comes to eukaryotic heredity, the genome, as they say, is only the beginning. There’s also the epigenome, a set of heritable chemical modifications on both DNA and histones that controls how the genome is read.

5-mC and 5-hmc

One such chemical modification is DNA methylation. Typically found in mammalian genomes on the five-position of cytosine residues (5-methylcytosine, or 5-mC) in the context of CpG dinucleotides, methylation is typically concentrated in CpG islands and gene regulatory elements, and is associated with gene silencing.

Researchers have been aware of DNA methylation for decades. 5-methylcytosine is sometimes called the “fifth base” of DNA, and epigenetics researchers have developed an array of molecular tools to find and interrogate it. But in 2009, two teams – one led by the Rockefeller University’s Nathaniel Heintz, and the other by Harvard Medical School professor Anjana Rao (for whom, full disclosure, I did postdoctoral research in the late 1990s) -- threw the epigenetics world a curveball with the discovery of yet another modified form, 5-hydroxymethylcytosine (5-hmC). [1, 2]

“Probably one of the biggest changes that’s occurred since 2009 has been the focus and interest now in these DNA methylation variants,” says Kyle Hondorp, Product Manager at Active Motif, a company whose tagline reads, “Enabling epigenetics research.”

Found in those first two papers in both mouse Purkinje neurons and embryonic stem cells, 5-hmC seems to have a different cellular and genomic distribution than 5-mC, being concentrated in gene bodies, for instance. It is created from 5-mC by the action of enzymes in the so-called TET family. [2] Yet many of the assays used to study DNA methylation cannot distinguish the two forms, says Michael Sturges, Epigenetics Product Manager at EMD Millipore. That means that at least some of the methylation analyses over the past few decades may require a second look, he says.

“The sense is that through all of the data we’ve collected over the past 15 or 20 years, some degree of that data is contaminated with hydroxymethylcytosine events,” says Sturges’ colleagues, John Rosenfeld, Epigenetics Research and Development Manager at EMD Millipore.

5-fC and 5-caC

The genetic alphabet probably has more than six characters. Besides A, C, G, T, 5-mC, and 5-hmC, there are at least two other known variants. Eukaryotic cytosine residues are methylated by the action of DNA methyltransferases. Tet1 can convert 5-mC to 5-hmC, but the enzyme doesn’t stop there; it can further oxidize 5-hmC to 5-formylcytosine (5-fC) and finally to 5-carboxycytosine (5-caC), an enzymatic pathway that seems to be involved in actively erasing 5-mC marks from DNA, says Rosenfeld. (For more on this pathway, see Active Motif’s DNA Methylation Guide. [3])

One strategy for studying these different forms is affinity capture. Traditionally, when studying standard 5-methylcytosine, researchers use anti-5-mC antibodies or methyl DNA-binding domain proteins (MBDs) to capture modified DNA fragments in procedures called MeDIP (methyl-DNA immunoprecipitation) and MIRA (methylated CpG island recovery assay), respectively. Now they can do the same for 5-hmC with antibodies targeting that modification, as well.

Several companies now offer anti-5-hmC antibodies, including EMD Millipore. Active Motif has one, too, which it bundles in an hMeDIP (hydroxymethyl-DNA immunoprecipitation) kit to pull down DNA sequences containing the modification. Alternatively, researchers can enrich 5-hmC-containing fragments using the company’s new Hydroxymethyl Collector™ kit. (For those interested in 5-fC or 5-caC, Active Motif has antibodies available for those modifications as well.)

“Hydroxymethyl Collector is probably one of the most interesting technologies we’ve launched in the past year,” Hondorp says.

Unlike the company’s MethylCollector Ultra™ kit, which purifies methylated DNA using a blend of MBD2b and MBD3L1, the Hydroxymethyl Collector kit leverages an enzyme, beta-glucosyltransferase (also available from Millipore), that specifically transfers a glucose moiety to 5-hmC, Hondorp explains. In the case of Hydroxymethyl Collector, this glucose is modified with an azide group, which is then chemically coupled to biotin via “click” chemistry, enabling a sample’s 5-hmC-containing DNA population to be captured on streptavidin beads.

The result of both hMeDIP and Hydroxymethyl Collector is a pool of DNA fragments, each containing at least one 5-hmC. This pool can be sequenced to provide a genome-wide view of hydroxymethylation or analyzed locus-by-locus via PCR. Another option is to interrogate the DNA pool on microarrays. Microarrays intended for methylation studies include Agilent Technologies’ DNA Methylation Human CpG Island Microarray and Illumina’s Infinium® HumanMethylation450 BeadChip. Roche NimbleGen still offers its 10-chip Human DNA Methylation 2.1M Whole-Genome Tiling array set, but that product will be discontinued by the end of the year as NimbleGen shifts its focus away from microarrays and into sequence capture products, says Roche Applied Science Marketing Manager Clotilde Teiling.

New England Biolabs’ EpiMark® 5-hmC and 5-mC analysis kit also uses beta-glucosyltransferase to glucosylate 5-hmC residues. The assay distinguishes 5-hmC from 5-mC by differential sensitivity of the resulting DNA to restriction endonucleases: untreated 5-hmC containing sequences are sensitive to MspI (5'-C / CGG-3'), whereas glucosylated sequences are not.

Another enzyme that specifically recognizes 5-hmC is the restriction enzyme, PvuRts1I, also available from Active Motif. PvuRts1I binds to 5-hmC-containing DNA, cutting between 9 and 13 nucleotides to either side. [4] The result is a 24-base pair fragment with a centrally located 5-hmC, and treatment of genomic DNA with that enzyme produces a pool of such fragments whose sequencing would reveal 5-hmC locations throughout the genome much as MeDIP sequencing would.

Bisulfite sequencing

Still, to get true base pair resolution, researchers have to adopt a different strategy. The “gold standard” for sequencing methylated DNA, all agree, is bisulfite sequencing. Treatment of DNA with sodium bisulfite converts unmodified cytosines to uracil, which shows up as thymine in DNA sequencing reactions. Comparison of that dataset with untreated DNA pinpoints methylated residues.

However, as with other methylation assays, bisulfite cannot distinguish between 5-mC and 5-hmC. Earlier this year, though, researchers described two strategies to circumvent this problem.

The first strategy, called Tet-assisted bisulfite sequencing (TAB-Seq, [5]), relies on the ability of TET proteins to oxidize 5-mC to 5-caC. When treated with sodium bisulfite, 5-caC is converted to uracil just like unmodified cytosine. The method uses beta-glucosyltransferase to first protect 5-hmC residues by glucosylation. Then, the DNA is treated with TET to convert all remaining methylcytosines to 5-caC. Finally, the DNA is bisulfite converted. Upon sequencing, all cytosines and 5-methylcytosines will read as T, while 5-hmC residues appear as C. Comparison of that dataset with one generated by standard bisulfite sequencing allows researchers to determine which bases contained which modification.  Chicago-based startup WiseGene has commercialized the TAB-Seq method.

The second strategy, oxidative bisulfite sequencing (oxBS-Seq, [6]), uses potassium perruthenate to oxidize 5-hmC to 5-fC. After treatment with bisulfite, 5-fC (like cytosine itself) reads as T, such that comparison of standard bisulfite-converted DNA with oxidized/bisulfite-treated DNA can distinguish those bases that were 5-mC from those containing 5-hmC.

Of course, vendors have not neglected 5-methylcytosine. Bisulfite conversion, for instance, is “famously finicky,” says Sturges, requiring researchers to strike the right balance of reagents, time, and temperature, such that all cytosine residues are converted without “overconverting” the methylcytosines, as well. EMD Millipore has released an enhanced bisulfite conversion kit called the CpGenome Turbo Bisulfite Kit.

According to Sturges, the Turbo kit, true to its name, uses new chemicals and optimized reaction temperatures to shorten the typical bisulfite conversion protocol from 16 hours to 90 minutes. In internal tests, he says, the kit has yielded 99.9% conversion efficiency without overconversion.

Targeted selection of methylated DNA

The company has also launched a methyl-DNA enrichment kit called CpG MethylQuest, which uses MBD2 to capture DNA containing 5-methylcytosine, as well as control DNAs containing either no modified bases, 5-mC, or 5-hmC.

Agilent Technologies now offers a target enrichment solution for next-gen sequencing applications, called the SureSelect XT Human Methyl-Seq kit. According to Yong Yi, Associate Director of Marketing for the Genomics Solutions Division, the kit does for methylated DNA what the SureSelect exome kits did for targeted exon sequencing: Reduce the cost and time of sequencing regions of interest.

“Sequencing a whole human genome is cost and time prohibitive, [and] exome sequencing made that much more efficient,” Yi says. “We are leveraging the same idea with [the] Methyl-Seq [kit]. This is a method to do a more targeted Methyl-Seq experiment.”

The kit includes a pool of 120-mer RNA baits to capture methylated regions including CpG islands, as well as “shores” and “shelves” within the human genome. (Shores and shelves are regions 2- and 4-kb away from the edges of CpG islands, respectively, in which many differentially methylated sequences are found, Yi says.) In total, the kit captures about 3.7 million CpGs over 84-Mb of sequence, Yi says. By comparison, Agilent’s methylation-focused DNA microarray covers about 27,800 CpG islands from across the genome.

“It is a very comprehensive design,” Yi says.

Roche NimbleGen doesn’t offer an off-the-shelf target selection product for DNA methylation, but users can create their own custom design, says Teiling, pooling as many as 2.1 million probes in a single tube.

Of course, as epigenetics grows in importance and expands into ever more fields, more and more researchers may find themselves bewildered by the variety of tools and strategies available to them. Service providers can provide the expertise needed to run these experiments, saving the researcher the time and effort required to bring a new technique in-house and optimize it. Active Motif, for instance, recently acquired GenPathway, meaning the company can offer its clients such services as bisulfite sequencing, bioinformatics analysis, hMeDIP, and more.

“Now we can offer that to customers that may not be able to do the methods and data analysis on their own,” Hondorp says.

[1] S. Kriaucionis, N. Heintz, “The nuclear DNA base 5-hydroxymethylcytosine is present in Purkinje neurons in the brain,” Science, 324:929¬–30, 2009.
[2] M. Tahiliani et al., “Conversion of 5-methylcytosine to 5-hydromethylcytosine in mammalian DNA by MLL partner TET1,” Science, 324:930–5, 2009.
[3] http://www.activemotif.com/documents/1654.pdf
[4] H. Wang et al., “Comparative characterization of the PvuRts1I family of restriction enzymes and their application in mapping genomic 5-hydroxymethylcytosine,” Nucleic Acids Res, 39:9294–9305, 2011.
[5] M. Yu et al., “Base-resolution analysis of 5-hydroxymethylcytosine in the mammalian genome,” Cell, 149:1368–80, 2012.
[6] M.J. Booth et al., “Quantitative sequencing of 5-methylcytosine and 5-hydroxymethylcytosine at single-base resolution,” Science, 336:934–7, 2012.

 The image at the top of the page is from Active Motif's HydroxyMethyl Collector™ Kit.