Are these two mechanisms wholly independent or is there a linkage?
A paper published in 1994 by Wassenegger et al. suggests that there is, indeed, a feed-back loop between transcript abundance and methylation of the cognate gene. The system they used is wholly artificial: these investigators introduced cDNA copies of a potato spindle tuber viroid genome into tobacco plants. Viroids do not normally give rise to DNA copies and integrate into the genome and viroids do not encode proteins. However, transcription of an artificial cDNA can give rise to an internal viroid infection.
1. The authors found that the viroid cDNA sequence was methylated if the integrated cDNA encoded a replication-competent form of the viroid, but was not methylated if the integrated cDNA encoded a replication-incompetent sequence. That is, if the viroid RNA accumulated to high levels as a result of replication, the gene was methylated. If it was just transcribed, the gene did not get methylated. Moreover, the replication incompetent viroid sequence became methylated if the transgenic line was infected with the viroid. Large amounts of viroid RNA were detected only when the viroid could replicate and the implication of these observations is that there is a feedback loop from the production of large amounts of RNA to methylation of the coding sequence.
2. Recently, this group has looked at the methylation patterns in detail, reporting that all of the C residues in both strands are methylated in the region corresponding to the cDNA. They detected increased methylation of adjacent sequences, but at somewhat lower levels.
3. Thus these investigators have tapped into an endogenous mechanism that shuts down genes that are expressed at high levels.