Once seen as purely structural, a sort of gift wrap for DNA, histones (and their modifications) have come into their own as important carriers of epigenetic information - information not encoded in DNA sequence.
Histones serve to compact DNA, and can make it either more or less accessible to the cell's transcription machinery. Eight histones and the roughly 150 base pairs of DNA wrapped around them are known as a nucleosome, and make up the basic unit of compacted DNA or chromatin.
Histones can be chemically marked (modified) by either acetylation or methylation. From a drug development perspective, acetyl marks have received more attention to date; histone deacetylase inhibitors are being tested in a number of clinical indications, including a MethylGene Inc. and EnVivo Pharmaceuticals Inc. collaboration. (See BioWorld Today, July 14, 2004).
However, methylation is another way to modify histones. A methyl tag was long thought of as a fairly permanent modification on a histone, so because it is less dynamic, methylation has received less interest from the drug discovery side. But from a basic science standpoint, methylation is as important as acetylation. The recent identification of the first histone demethylase, published in the Dec. 29, 2004, issue of Cell, showed that methylation patterns, too, probably are more dynamic than conventional wisdom holds.
In studies published in the Jan. 28, 2005, issue of Cell, researchers from Harvard University, the Broad Institute of Harvard and MIT, Brigham and Women's Hospital, Harvard Medical School, the Whitehead Institute for Biomedical Research and Affymetrix Inc. used whole-genome tiling arrays for analyzing certain methyl, as well as acetyl, marks in both human cancer cell lines and human fibroblasts.
The researchers studied histone H3 methylation of one specific lysine, lysine 4, across a large region of the genome including human chromosomes 21 and 22 and gene clusters for cytokines and Hox genes. They studied the acetylation of two other lysines, 9 and 14, in the same regions. They also analyzed a smaller region of the mouse genome and compared it to human results.
Lead author Bradley Bernstein, instructor in pathology at Brigham and Women's Hospital and a research associate at the Broad Institute, told BioWorld Today that "the tiling arrays allow the analysis to proceed in a very unbiased fashion. We didn't look at just the regions previously identified as important," such as promoters or known genes. "This technique opens up another avenue for identifying functionally important regions of the genome."
The methylation mark on lysine 4 is associated with active genes; additionally, "many of the proteins that have a methylated lysine 4 are associated with cancer in some way," Bernstein said. The scientists used antibodies to probe for dimethylation of lysine 4, trimethylation of lysine 4 and simultaneous acetylation at both lysine 9 and 14.
Mapping revealed that di-and trimethylation of lysine 4 tended to coincide. Since each nucleosome has two histone subunits, di- and trimethylation of lysine 4 can occur on the same histone, though Bernstein also noted that the resolution of the technique is four or five histones, so perhaps it is adjacent histones that are di- and trimethylated. Trimethylation sites corresponded strongly with underlying gene starts, whereas dimethylation did so to a lesser degree. Acetylation of lysine 9/14 correlated strongly with trimethylation.
The scientists also compared the relative degree of conservation of methylation patterns and underlying gene sequence between the cytokine and IL-4 receptor gene clusters in human and mouse fibroblasts. They found that methylation patterns were evolutionarily more conserved than the underlying gene sequences, and that methylated regions did not show more sequence conservation than unmethylated ones. Asked whether that was a surprise to the scientists, Bernstein replied that within the research team, "it surprised the sequence people. The chromatin people were less surprised," because they have a greater awareness that reflect DNA sequence, but it also could be reflective of higher-order structure and can exert more long-range effects.
Interestingly, the scientists found striking differences in the overall patterns of methylation of Hox genes vs. other gene types.
"Throughout almost all of the regions we examined, we see sharp peaks that often coincide with gene starts. In the Hox regions, we see very broad regions - instead of one mountain peak; it looks like the Himalayas," Bernstein said. The implication is that "this particular cluster is not regulated in a gene-by-gene fashion, like most genes are. Instead, it is regulated together."
He suggested that joint regulation might lead to more robust regulation of the Hox genes, which exist only in multicellular organisms and play a critical and unique role in development. He also noted that his group had done previous research in yeast, a unicellular organism, and the gene methylation there is similar to the gene-by-gene regulation patterns seen in other genomic regions of the human and mouse.
In the Hox clusters, methylation also increased the transcription of intergenic regions; it was seen in a third of methylated but only 10 percent of unmethylated intergenic regions.
Asked what makes a "transcribed intergenic region" different from a gene, Bernstein replied that it is a region that is transcribed but not translated. Whether such intergenic transcription has a function is unclear; it could have a function that is as yet unknown, but also could be a spurious consequence of having DNA unwrapped and accessible.