At the proper place and time, DNA methylation can be a good thing. It’s important in gene regulation, required for normal development and cell physiology. Epigenetic researchers have even shown that methylation patterns are passed on to offspring. But DNA methylation in the wrong location and time can be a very bad thing, erroneously switching genes on or off, and leading to many types of cancer. Yet methylation is only one epigenetic regulator. Today, researchers are trying to understand how different epigenetic factors interact with one another to produce complex phenotypes.
While it has been known for awhile that there are many factors affecting epigenetic regulation, it has been impossible to study these simultaneously. DNA methylation and nucleosome positioning, or the accessibility of DNA to regulatory factors, is one example. DNA methylation can be measured by digesting DNA with two types of enzymes, one sensitive and one insensitive to methylation, and then compare the fragments. But once you cut up the DNA, there is no way to measure nucleosome positioning. Conversely, using nucleases to cut away DNA that is not bound to histone core proteins (in other words, DNA that would be accessible to epigenetic regulatory factors) would preclude DNA methylation analysis. What’s an epigenetic researcher to do? They get creative.
Michael Kladde and his group in the department of biochemistry and molecular biology at the University of Florida Shands Cancer Center, use DNA methyltransferases, rather than nucleases, in a footprinting technique to assess nucleosome positioning (1). Unoccupied regions along DNA not bound to proteins are accessible to methyltransferases, and therefore become methylated at particular residues (cytosines). In contrast, regions of DNA bound to proteins are inaccessible to methyltransferases. The footprint created using this method suggests the positioning of nucleosomes along the DNA.
“DNA methyltransferases, which unlike nucleases do not cut DNA, are more easily expressed in living cells, enabling detection of open versus closed regions of chromatin in their truly native nuclear environment,” explains Kladde. “Conventional footprinting assays cannot detect molecules lacking accessible chromatin nor directly link the degree of accessibility to CpG methylation state.” Nearly always, DNA accessibility is inversely correlated with DNA methylation.
Recently, two groups used methyltransferases to zero in on single strands of DNA. Kladde’s group developed a technique they call MAPit (methyltransferase accessibility protocol for individual templates) to assess DNA methylation and accessibility simultaneously on single DNA molecules (2–3). MAPit takes advantage of bisulfite sequencing to convert unmethylated cytosine to uracil and eventually thymine, while methylated cytosine remained as cytosine. The result is the ability to use the methylation output to assess nucleosome occupancy in yeast, and DNA methylation as well in mammalian studies, on an individual DNA molecule. Such dual epigenetic analyses could never have been achieved using conventional techniques.
Another epigenetics group headed by Peter Jones of the Norris Comprehensive Cancer Center at the University of Southern California, independently began using methyltransferases to study nucleosome occupancy—unbeknownst to Jones, who is quick to credit Kladde with the first results.
“We’ve been interested in a method which would integrate DNA methylation information, and nucleosome positioning, on the same individual DNA molecule,” says Jones, whose technique is termed NOMe-seq. As in MAPit, NOMe-seq initially uses bisulfite sequencing of DNA, and then clones individual DNA molecules into a vector, “capturing a snapshot of an individual promoter, or an individual enhancer, for example,” says Jones, including DNA regions that were accessible or not based on whether the GC’s had been methylated by the commercially available GC methyltransferase discovered by Kladde’s group, known as M.CviPI (1).
The strength of NOMe-seq and MAPit lie in their single-molecule focus; other epigenetics methods give information about a population of DNA molecules, but not specific information about individual sites on a chromosome – which can, of course, differ greatly. “A substantial strength of MAPit is that its single-molecule view avoids averaging methylation at each target site over the population of molecules present in each sample,” says Kladde. “This allows visualization of molecules with distinct classes or subpopulations of chromatin structure.” Jones also stresses the value of looking at single DNA molecules. “If you do it on a molecule-by-molecule basis, you can find out really important information,” he says. “For example, you find that in a female cell, one X chromosome is methylated, and one is not. You can tell that the one that’s methylated is the one that’s completely inaccessible, and the unmethylated one is very accessible.”
NOMe-seq also helped Jones’ group to settle an epigenetic conundrum – what comes first, the inaccessibility, or the methylation? “It’s an argument that has been raging for years,” says Jones. “Does the DNA becomes inaccessible first, in other words, does the sequence become turned off, and then it become de novo methylated? Or does it become de novo methylated first, and then become inaccessible?” Using NOMe-seq to look at individual DNA strands, Jones’ group found that DNA becomes inaccessible, or switched off, first, and then becomes methylated. “The data is completely clear because you are not looking at the average, you’re looking at an individual molecule,” he says. “So you can see that the molecule becomes switched off first at a particular promoter, and then the methylation occurs.” Jones’ most recent work expands this result genome-wide, sequencing bisulfite-treated DNA using next-generation sequencing (4). “You can then look genome-wide at the structure of all the enhancers, and the structure of all the CTCF sites, and the structure of all the transcription start sites,” he says. “You can see, genome-wide, where the DNA methylation is, and also where the DNA is accessible.”
Cancer and epigenetics
The recent sequencing of hundreds of different cancer cell genomes unearthed a recurring theme of mutations in epigenetic regulators. “But what they didn’t expect were mutations in chromatin remodelers,” notes Jones, citing the example of ARID1A, which works with p53 to regulate transcription and tumor growth in some cancers, and is now thought to be a tumor suppressor (5). It was found to be mutated in over 50% of a particular type of ovarian cancer cells. “What we want to know is, if you have a mutation of this gene, what is the effect of that phenotypically on chromatin structure?” says Jones. “What does it do to the distribution of nucleosomes and DNA methylation in cancer?”
In fact, ARID1A plays an important role in the formation of chromatin remodeling complexes, implying that it may alter nucleosome positioning. But how this leads to cancer remains a mystery. “I think that’s extremely important, because we know that these mutations are what are called ‘driver mutations’,” says Jones. “They’re found in lots of cancers. But we don’t know what the outcome of the mutation is. We know you get cancer, but what does it do to the structure of the chromosomes?” One current line of investigation the Jones’ lab is pursuing is nucleosome positioning genes that were found to be mutated in cancer cells.
Jones’ group is exploring other epigenetic phenomenon as well, including the use NOMe-seq to study cancer cells from real, fresh kidney tumors, examining how the epigenome changes in response to drug treatment in cultured cells. “What we want to know is, if you treat a cell with a DNA demethylating agent, say for example, 5-azacytidine, then globally what happens to DNA methylation and nucleosome positioning at the same time?” says Jones.
Although such epigenetic drug therapies are already being used to treat cancer patients – it’s still unknown why and how the drugs work. “Even in solid tumors, they are getting responses,” Jones says. “I just think it’s a very exciting area of epigenetic therapy, and we’re pleased to have the approach which we think appears to give a much cleaner, clearer view of what’s going on.”
1. Xu, M., M. P. Kladde, J. L. Van Etten, and R. T. Simpson. 1998. Cloning, characterization and expression of the gene coding for a cytosine-5-DNA methyltransferase recognizing GpC. Nucleic acids research 26(17):3961-3966.
2. Jessen, W. J., S. A. Hoose, J. A. Kilgore, and M. P. Kladde. 2006. Active PHO5 chromatin encompasses variable numbers of nucleosomes at individual promoters. Nature Structural & Molecular Biology 13(3):256-263.
3. Pardo, C. E., I. M. Carr, C. J. Hoffman, R. P. Darst, A. F. Markham, D. T. Bonthron, and M. P. Kladde. 2011. MethylViewer: computational analysis and editing for bisulfite sequencing and methyltransferase accessibility protocol for individual templates (MAPit) projects. Nucleic acids research 39(1).
4. Kelly, T. K., Y. Liu, F. D. Lay, G. Liang, B. P. Berman, and P. A. Jones. 2012. Genome-wide mapping of nucleosome positioning and DNA methylation within individual DNA molecules. Genome research 22(12):2497-2506.
5. Wu, J. N., and C. W. Roberts. 2012. ARID1A mutations in cancer: Another epigenetic tumor suppressor? Cancer discovery (December).