By Monica Heger
A new sequencing technique allows for the direct comparison of methylation status to other epigenetic markers, like histone modifications, enabling researchers to get a comprehensive picture of the complex relationship between epigenetic events.
Researchers from the Radboud University Nijmegen Medical Center in the Netherlands and the Broad Institute developed a method dubbed ChIP-BS-seq that combines chromatin immunoprecipitation with bisulfite sequencing. The method's details were published last week in Genome Research.
In the study, the researchers demonstrated that the method could examine both histone modifications and methylation. It first uses ChIP to pull down regions of the genome with specific histone modifications and then bisulfite-converts those regions and sequences them on the Illumina Genome Analyzer.
Previous techniques have studied these modifications separately, so piecing together how different epigenetic modifications interact could only be done by correlation. But in this study, the team demonstrated they could make direct measurements of the relationship between the events.
"We can study both modifications at once," Arjen Brinkman, a researcher at Radboud University and lead author of the paper, told In Sequence. "This is particularly important because you are sure that instead of making correlations between the two datasets, you have a direct link between the modifications."
The team used the technique to study H3K27me3, a histone modification involved in gene silencing, in combination with methylation status.
The two epigenetic marks have been widely studied, with some studies suggesting that the two co-occur and others suggesting that they are antagonistic.
"The cross-talk between k27 and DNA methylation is a longstanding issue," Brinkman said.
The team tested its method on a colon cancer cell line and mouse embryonic stem cells. They found that in most cases, the two epigenetic modifications co-occur, but in CpG islands and other regions of high methylation, they are antagonistic to each other, and methylation appears to prevent H3K27me3.
First, the team used chromatin immunoprecipitation to pull down the regions of histone modification. Then, they bisulfite-converted that portion of the genome to achieve single-base methylation resolution, and sequenced it.
To make sure the bisulfite conversion did not affect the histone modification patterns, the team tested both ChIP-seq and ChIP-BS-seq on the colon cancer cell line. Their results showed good correlation between the read densities.
Within the colon cancer cell line there were broad regions of H3K27me3 enrichment, covering genes and intergenic regions. Genes located within an H3K27me3 enrichment block tended to be silent, while genes outside those regions had higher than average expression levels.
Methylation of the H3K27me3-enriched areas was determined in 300-base pair windows and categorized based on function — intergenic, intron, exon, CpG island, or non-CpG island promoter.
In general, the researchers found that H3K27me3 enrichment occurred along with methylation of CpG islands. However, CpG-rich promoter regions with H3K27me3 contained mostly unmethylated CpGs.
For example, the genes SERTAD4, SMAD7, and OVOL2 were located within an H3K27me3 block and were mostly methylated. However, the CpG islands that encompassed the gene promoters were mostly unmethylated.
This pattern held true on a genome-wide scale, with H3K27me3 enrichment and methylation co-occurring except in CpG island promoters. Additionally, all regions with a high density of CpG islands coincided with hypomethylation, while low-density CpG islands were methylated.
The team confirmed these results by testing H3K27me3 enrichment and methylation separately performing ChIP-seq and a MethylCap-seq experiment, which uses methyl binding domain proteins to capture methylated portions of the genome.
Next, they tested the method on mouse embryonic stem cells and found similar results. Areas with the histone modification and with dense populations of CpG islands were unmethylated, while in the rest of the genome, CpG islands were methylated.
Finally, they tested the technique on mouse embryonic stem cells that had been bred to contain no methylation anywhere in their genomes and found that the level of H3K27me3 modification increased. "That's what you'd expect if you have two modifications that are mutually exclusive," said Brinkman.
Other researchers also see the value in being able to interrogate both methylation and other epigenetic markers. A group from the Garvan Institute of Medical Research in Sydney developed a very similar method, dubbed BisChIP-seq, which also combines ChIP with bisulfite sequencing.
In a study describing their method, also published last week in Genome Research, the researchers looked at the relationship between methylation and H3K27me3 enrichment in a prostate cancer cell line.
Their study found nearly opposite results as the Dutch researchers with methylation and H3K27me3 enrichment co-occurring at CpG islands and transcription start sites of silent genes, and "a loss of DNA methylation in intergenic H3K27me3-marked regions," the Garvan Institute authors wrote.
Brinkman said he wasn't sure why the two studies came to different conclusions, but said it could be due to underlying biological differences in the tissues that they studied.
Moving forward, he said that improvements to the method would focus on enabling lower amounts of input DNA. Currently, the protocol has more steps than either ChIP-seq or bisulfite sequencing alone, and at each step some of the sample is lost, so more starting material is needed. Typically, ChIP experiments require at a minimum about half a million cells. Following ChIP, bisulfite sequencing can be done with as little as 100 nanograms of DNA, but Brinkman said that amount was already pushing the lower limits. Going lower is difficult, he said, because bisulfite treatment causes some DNA damage.
"It would be nice to do this in small quantities, in patient samples for instance, or other sources that are a bit more limited," he said. But that's currently "not really possible."
Brinkman said that the team is now looking to use the method to study different epigenetic modifications in conjunction with methylation, including other histone modifications and transcription factor binding sites.
"It would be interesting to use ChIP to capture all the sites where these [transcription factor proteins] bind and look at the methylation states at these fragments, so we can learn the preference for the binding site — methylated or non-methylated," Brinkman said.
Additionally, the technique could be used to get a more in-depth picture of imprinting and X-inactivation, he said.
Have topics you'd like to see covered by In Sequence? Contact the editor at mheger [at] genomeweb [.] com.