While investigating the in vitro activity of DNA methyltransferases on annealed oligonucleotide substrates, Australian researchers recently found that murine DNA methyltransferase I will quite happily methylate non-CpG sites in DNA bubbles.
Jason Ross and Peter Molloy from the CSIRO, in collaboration with Japanese researchers from the Institute for Protein Research in Osaka, observed a high degree of cytosine methylation by Dnmt1 at CpT sites on both strands of the unannealed region in a DNA bubble. No appreciable CpT methylation was observed on single stranded DNA, DNA duplexes, or R-loops – where the DNA bubble has an RNA strand annealed in the single-stranded region. Nor was there any detectable non-CpG methylation on the DNA bubble by the de novo DNA methyltransferase enzymes, Dnmt3a and Dnmt3b.
They speculate this property may be associated with DNA replication and/or RNA transcription and have relevance to mechanisms that underlie the significant non-CpG methylation recently reported in human embryonic stem cells.
In the same paper they also investigate the mechanism of transcriptional gene silencing (TGS) in mammals using model substrates. Currently there are two TGS mechanistic camps. The RNA/DNA model suggests that the antisense strand of an siRNA is guided into an R-loop structure and DNA methyltransferases methylate around that R-loop. Alternatively, the RNA/RNA model proposes that the siRNA binds to a transcript create a dsRNA which in turn directs silencing of neighbouring DNA.
The team found strong evidence against the RNA/DNA model of TGS, which dovetails with a number of recent papers supporting the RNA/RNA model. Nice.
Thanks to Jason Ross for bringing this one to our attention and for contributing the details. Check them out for yourself in the Biochemical Journal (September 2010)