The danger arises because nascent RNA can anneal to the template DNA strand, thus forming an R loop. The nontranscribed strand is left as a single strand, potentially susceptible to attack by nucleases.
Bacteria combat this tendency by tightly coupling translation to transcription. In eukaryotes, this is not an option as transcription and translation are nuclear and cytoplasmic, respectively.
Li and Manley did not set out to discover a genome-protective mechanism, but chanced upon it while studying ASF/SF2. The Columbia team put this splicing protein under the control of a tetracycline-responsive promoter. Shutting it off led to cell death, but many revertants became resistant to tetracycline repression.
The ASF transgene in the revertants had, along with many other genes, been shuffled into new areas via DNA rearrangements. Sure enough, markers of DNA double strand breaks appeared within 12 h of turning off ASF, and fragmented DNA appeared within 24 h. These changes were not seen with the tetracycline-mediated shut-off of other essential genes.
R loop structures appeared during the shut-off, and all these shut-off symptoms could be abolished by overexpression of RNase H, which can degrade the RNA in RNA–DNA hybrids. In vitro, ASF suppressed R loop formation during transcription as long as the transcribing RNA polymerase II was phosphorylated on its COOH-terminal domain, as it is in vivo. This phosphorylated domain was already shown to recruit ASF to the transcription reaction.
R loops are not always a bad thing. In B cells they are essential for initiating the DNA rearrangements that mediate class switching of antibody heavy chains. Keeping this machinery away from nascent RNA in other cell types may be one important, presplicing function of ASF and related proteins.