Sometimes the best defense is a good offense. That appears to be the case for transcription factors (TFs) that resist de novo DNA methylation by binding to their sites in CpG islands. Although this idea has been around for awhile, a team of German and American researchers led by Michael Rehli recently provided the first convincing evidence that TF binding is a general mechanism for methylation protection.
The work builds on early reports by Adrian Bird (Genes Dev. 1994) and Cedar (Nature 1994) that an Sp1-binding site is necessary for protection of the APRT gene from de novo methylation in mice and humans. These findings were intriguing, but nobody knew whether the Sp1/APRT story was an isolated case or a general mechanism.
More recently, groups such as the Vertino Lab have performed genome-wide computational studies to identify sequences associated with methylation-resistant CpG islands in cancer samples. These studies didn’t detect any known TF consensus sites, which seemed to argue against a general protective role for TFs.
The team first generated genome-wide methylation maps using methyl-CpG immunoprecipitated samples on Agilent CpG Island Arrays, then mined the data with a powerful de novo motif finding algorithm called HOMER. Using this approach, Rehli’s team have identified consensus binding sites for several general TFs within CpG islands that remain unmethylated in both normal and malignant cells. When ubiquitous TFs such as Sp1, nuclear respiratory factor 1, and yin-yang 1 are bound to these sites, the region is highly resistant to de novo methylation. But one bound TF is not enough: two or more TFs must bind cooperatively at neighboring sites to get protection.
How do TFs stand up to the methylation machine? Some reports have suggested that RNA polymerase II (Pol II) controls the epigenetic fate of CpG islands. According to Rehli, “There is likely a great deal of overlap between TF binding and Pol II recruitment, since the identified TFs are known to recruit Pol II.” However, Rehli’s team identified some TF-bound, methylation-resistant sites in intergenic regions that likely don’t bind Pol II. “So we think that the association [of methylation protection] with TFs is broader than Pol II,” says Rehli. “I do generally think that cis-acting sequences (and the corresponding factors) play an important role in establishing and maintaining epigenetic states.” Because the identified TFs are known to recruit epigenetic modifiers such as Polycomb group proteins, the TFs could play a direct role in establishing an “active” chromatin environment that impedes CpG island methylation.
As sophisticated techniques for motif discovery continue to be introduced, we’ll likely learn a lot more about how certain DNA sequences can “resist the establishment” of DNA methylation, so stay tuned.
For now, check out the latest details at Cancer Research February 2010