NEW YORK (GenomeWeb News) – RNA editing of coding regions is likely to be non-adaptive, the University of Michigan's Guixia Xu and Jianzhi Zhang reported in the Proceedings of the National Academy of Sciences today.
The duo examined nearly 1,800 human coding A-to-G RNA editing sites and focused on how frequently editing occurred at those spots and how evolutionarily conserved or essential those sites were. As they reported in PNAS, Xu and Zhang found that as the importance of a site increases, the less likely it is to undergo RNA editing, leading them to conclude that RNA editing of coding sites is typically not adaptive. Instead, they argue that much of RNA editing is due to promiscuous binding by the editing enzyme.
"RNA editing is a very strange phenomenon from a geneticist's point of view," Zhang told GenomeWeb Daily News.
In RNA editing, after DNA is transcribed to messenger RNA, an additional step changes the sequence of the mRNA, most commonly swapping a G for an A, before it is translated into protein, sometimes also changing the protein sequence in the process.
In some cases when RNA editing is impaired, it leads to severe disorders. This, the researchers said, led many to assume that RNA editing is adaptive and expands transcriptome diversity.
"So, you wonder why the information is not encoded in DNA," Zhang said, adding, "wouldn't it be better if the G was in the genome so it was more secure?"
Recent genomic and transcriptomic studies have uncovered thousands of RNA editing sites up from just a few a couple of years ago, and Zhang and Xu set out to determine what portion of RNA editing sites is advantageous using a comparative genomics approach.
Drawing from six different databases, the researchers pulled together a list of 1,783 A-to-G editing sites located in protein-coding regions.
If editing were advantageous, Zhang and Xu said that it would occur more frequently at sites at which such A-to-G editing would lead to a nonsynonymous change, meaning that editing would affect the protein produced.
However, at a majority of those nearly 2,000 sites, they found that editing led to synonymous change, while nonsynonymous changes occurred at some 600 spots. Nonsynonymous changes, they calculated, were about 38 percent less likely to occur than synonymous ones.
"This phenomenon could be interpreted that the nonsynonymous editing actually is bad, is harmful," Zhang said. The harmful editing, he added, was removed by natural selection.
They also examined the level of RNA editing at the sites. Editing doesn't occur 100 percent of the time at every site, Zhang noted, but if it is important, it should occur more frequently.
The nonsynonymous editing sites, though, have lower editing levels than synonymous editing sites, suggesting that nonsynonymous editing may be harmful, as only sites with low editing levels appear to be tolerated by natural selection.
Xu and Zhang also divvied human genes based on whether or not they were essential. Essential genes, they determined, are ones that cause infertility or death if deleted, based on mouse orthologs.
The fraction of essential sites that are edited is lower than the fraction of non-essential sites that undergo editing, the researchers found. Further, they found that the level of nonsynonymous editing is lower than synonymous editing in essential genes, though not in non-essential genes.
Additionally, sites under functional constraints and ones that are evolutionarily conserved, as determined through phylogenetic variation, are less likely to undergo nonsynonymous RNA editing, the researchers reported.
For instance, the duo examined the phylogenetic variations of 143 editing sites spots, determining whether they were hardwired, conserved, unfound, or diversified sites. They predicted that if nonsynonymous editing is advantageous, then editing at diversified sites would be low, while if it is disadvantageous, then editing at those sites would be high because differences would be tolerated. As they reported in PNAS, they found that the median editing level at diversified sites is higher than at the other types of sites, indicating that it is disadvantageous.
One of the proposed benefits of RNA editing is that it could boost transcriptome diversity, and to test that hypothesis, the researchers examined whether commonly made edits eventually became fixed at the DNA level. They found, though, no difference between edited and unedited As and their conversion to T or G/C.
"Using comparative genomic analysis, we showed that a substantial fraction of nonsynonymous A-to-G editing is too deleterious to survive purifying selection and be detected. Of those nonsynonymous editing sites that are detected, most are slightly deleterious, rather than beneficial," Xu and Zhang wrote. "If coding A-to-G editing is generally harmful, why does it exist?"
The researchers speculated that RNA editing is a byproduct of promiscuous binding by the editing enzyme.
"For a site to be edited, it has to be recognized by a particular enzyme, and so far, we actually don't know exactly how the enzyme finds the position," Zhang said. "Given that there are hundreds of thousands of sites in the genome that are edited and they do not seem to show a consensus sequence, this suggests that this enzyme is not very specific."
The original function of RNA editing, Zhang added, is not clear. He said that it may have been a way to protect against viral infection or perhaps to keep expanding sequences like Alu repeats in check.