One of my ongoing interests is the ecology the numerous repetitive, "parasitic" DNA sequences which make up almost a third of the human genome. These sequences are descendants of DNA stretches either containing information for moving themselves to a new location, or able to be moved by proteins encoded by other DNA. Movement by these elements causes changes which are almost always harmful to to the host (the human). For example, a novel parasitic DNA insertion can gum up the production of messenger RNAs (mRNAs) necessary for production of human proteins.
In this month's PLoS biology, computational biologists have pulled out evidence that human cells use RNA editing to target Alu repeats which occur in mRNA.
The main thrust of the article is an effort to understand "A-to-I" editing, a process with important consequences especially for certain neurotransmitter receptors in the brain. To learn more about A-to-I editing, the authors searched a database of mRNAs and found 26 examples of the specific mismatch which this editing generates. Interestingly, in all but one of these cases, the A-to-I editing happened within an Alu repeat. They believe this happenes because the machinery which does A-to-I editing seems to recognize special regions of mRNAs called hairpin duplexes. Hairpins can form when multiple Alu repeats integrate in opposite orientation in the same mRNA. A-to-I editing might weaken the base pairing which creates the hairpin.
Assuming that A-to-I editing is a sort of control mechanism (see more below), the evolutionary success of the Alu sequence repeats, which are self-similar by definition, in jumping into mRNAs seems to be the exact trait which is used to target them.
So what exactly is the impact of A-to-I editing? The story gets even stranger in the second half of the paper, in which they give examples in which an Alu repeat present in an mRNA changes that mRNA, and is changed again by this editing. Exons are made and lost, and mRNAs are spliced together into alternative forms under the influence of this process. In all, the authors think about 1500 mRNA transcripts, or about 1.4% of all transcripts, have changes in their exons because of editing of Alu repeats-- this means different protein products! They believe this is a conservative estimate because diversity among Alu sequences might have prevented detection of some editing events.
Finally, when the authors looked at which tissues are doing this the most, the answer came back that most transcripts are brain- or thymus- derived. (However, the trachea was also active. Go figure.) The reliability of this finding is a bit less robust because it relies on text annotations in the database.
So why? and why the nervous system? The activity of A-to-I editing follows the expression of (separate) adenosine deamidase enzymes within the nervous system and the immune system. These, in turn, may be active because of special features of those two organs. In particular, I still have not blogged about the recent Nature paper showing increased LINE element mobility during neurogenesis. Activity of the LINE family of jumping DNA sequences can also mobilize the far more numerous Alu repeats, so it may be that individual neurons in general have to cope with novel insertions in their transcriptome. The A-to-I editing might have originally arisen to break up cellular "indigestion" resulting from hairpins, and become successively adapted as an Alu-monitoring mechanism and finally a gene editing device. Thus relaxation of DNA "quality control" during neurogenesis and in the immune system (indicated e.g. by LINE mobility and tolerance of aneuploidy) seems to enable greater activity of parasitic repetitive elements. The host organism responds not by reinstating checkpoint controls, but by apparent ad-hoc countermeasures, and even by adapting to (more precisely, exapting) the changes introduced by the jumping genes.
If you can't beat 'em, enjoin 'em. But I guess the question is, why is DNA damage control relaxed in these specialized cells in the first place?
Update: there is a pretty considerable older literature about A-to-I editing, which I completely missed in this blog. See for example here .