Ratcheting up some splice leaders: a note on directionality

In the sea of eukaryotic genetic diversity also lurk different manners of doing day-to-day genome work itself. Ciliates run two nuclear genomes, trypanosome kinetoplasts contain a chainmail suit of RNA editing circles and dinoflagellates are just weird in every genome compartment they have. Their plastids contain tiny minicircles often containing but a single gene, capable of "rolling" transcription where the minicircle is much like a Mesopotamian cylindrical seal, leaving a concatenated repeated string of genes on the transcript. The mitochondria have linear genomes with short fragmented repeated chunks of important genes all over them. But the nuclear genome is the most fucked up: for one thing, dinoflagellates lack a few histones, and have enormous genomes stored in absolutely bizarre chromosomes. More importantly for our story: every single gene must be trans-spliced with a 'splice leader', a short sequence that attaches at the beginning of the mRNA transcript and brings to it the 3' cap necessary for transcription to work. Oddly enough, Euglenozoans like the trypanosomes and euglenids seem to have a very similar system, evolved entirely by chance* convergence (Lukes et al. 2009 PNAS goes over this remarkable convergence in more detail).

*Or perhaps something happened to both that made them prone to evolve this bizarre system.

Genomic quirks are not just interesting in their own right as some arcane oddities, but can reveal a great deal about the dynamics of genomes in general. The dinoflagellate splice leader system turns out to yield a very crisp illustration of the power of ratchets and the toll of reverse transcription on genomes.

To reiterate, every single nuclear gene transcript in a dinoflagellate must be spliced with the 3'cap-bearing 'splice leader', or else it simply won't work. This means that the dino is full of mature transcripts with splice leaders attached to the transcribed genes. Enter reverse transcriptases, which are prevalent in probably most, if not all, eukaryotic genomes, thanks to viruses and their partners in genomic parasitism crimes, transposons. When they're not busy moving transposons around and helping viruses move in, they reverse transcribe random gene transcripts for fun, that may then, on occasion, be successfully recombined back into the genome. This process probably doesn't happen [successfully] every day, but over thousands or millions of years (and countless individuals) is rampant enough to leave a noticeable trace in the genome.

So we have a load of transcripts floating around with an extra sequence stitched onto them from the splice leader. Do the reverse transcriptases care in the slightest? Of course not: to them, a ribonucleotide is a ribonucleotide, give or take some trace biophysical stuff that might make a couple people cringe at what I just said. (meaning, I wouldn't be surprised if there could be some slight but ultimately detectable biases there too) This means that splice leader, on occasion, actually makes its way back into the nuclear genome attached to the beginning of the gene.

However, this splice leader does not substitute for the usual splice leader trans-splicing, since the 3' cap must be added again, or else the transcript will not be translated. That now-nuclear gene-attached splice leader ends up being completely useless, and is able to gradually degrade into benign junk, provided it doesn't mess with the translation of the gene. What is really cool is that one can actually see this gradual degradation, as shown in Slamovits and Keeling 2008 Current Biol:

Mmmm, actual data! Note how the oldest SL piece closest to the gene (on the right) is the most degraded. (Slamovits & Keeling 2008 Curr Biol)

Once the unnecessary splice leader chunk becomes part of the gene, the gene gets transcribed and trans-spliced like any other – meaning it is once again susceptible to replaying that same process of reverse transcription, except this time it already has a relict sequence. It can acquire a second one on top of that. This explains how there can be several concatenated splice leader relics tagging along, like in the above figure.

Splice leader trans-splicing not necessarily promoting reverse transcription – only makes it easier to detect. In other words, it inadvertently makes for a wonderfully convenient system where you can actually track what happens to a gene after it gets reverse transcribed. Once the gene makes its new home, the old gene copy is still present and they generally would be functionally redundant, so the dual-copy state is extremely unstable as ultimately the loss of one of the copies will be tolerated. If the newly transcribed copy is lost, we never see it and thus don't talk about it in the first place. However, once the clean original is lost, only the gene with the crap from the splice leader remains, and reversal to the original state is so improbable it's practically impossible. In other words, this process is a wonderful example of an evolutionary ratchet.

Ratchets are interesting because they confer intrinsic directionality to a system, even in the absense of external pressures (like selection). The accumulation of splice leader junk in the dinoflagellate's genes isn't particularly healthy, nor is it particularly deleterious – it's effectively neutral. However, one can argue that we do have an example of bloated complexity here. Since you can't go back and lose chunks of splice leaders, this ratchet essentially ensures that left to its own devices, this aspect of genome complexity will increase on its own. At a certain point, there will probably be ever-increasing selection against accumulating further splice leaders, and those lineages that go too far will simply die off – the central tendency doesn't care, and the ratchet will keep on going regardless of what selection 'wants'.

This ratchet example is therefore an elegant case of evolutionary direction that's not particularly well explained by the central dogmas of Modern Synthesis or (neo)Darwinism, where selection is the force that crafts order and directionality, with mutation a mere passive provider of material to be molded. I will go into a deeper discussion of this in another post (there's a cool paper coming out soon), but I think it's worth briefly mentioning here too while we're at it. The "mutation" step (to which, I guess, this trans-splicing and reverse-transcription process can be awkwardly attached) here is what provides a drive, a push in a certain direction, and towards increasing complexity, no less (although that last detail is irrelevant). While selection is present and provides constraints (if both genes are lost, for example, the organism dies), it does not do the 'driving' or 'forcing' in this system. Very crudely put, selection here is the passive phenomenon, and mutation is at the wheel.

Another case of intrinsic directionality, but where reversal is allowed, is your garden variety directional bias – where proceeding in one direction is more probable than going backwards. A very basic example of that is if the replication machinery favours a certain type of nucleic acid – left to its own devices, the genome base composition would be skewed in that direction. Boundaries can also induce an apparent directionality, but in this case it's no longer intrinsic... that's, again, a topic for another day.

This idea was a part of the Mutationism theories in the early 20th century, which were a little extreme and perhaps premature, since mutation was far from being even marginally understood at the time. In the usual melodramatic manner characteristic of academia and the scientific community, the pendulum swung far to the opposite extreme, and Modern Synthesis was born. It became heresy to think that mutation itself can actively contribute to direction and order. The field became engulfed in a false dichotomy, where either selection or mutation can actively provide direction, with the modern folk siding with the former. That is a serious mistake and an unnecessary waste of great explanatory potential – you can go so much farther with selection, drift, mutation and recombination all at the wheel, each pulling with different magnitudes in various directions. Well, technically, you wouldn't if you were the thing being pulled – which resonates so well with the absense of 'ascension' or general active directionality in the evolutionary system as a whole. Evolution is a slow, painful, inefficient and rather stochastic process, partly because the cart is being pulled in so many ways.

(The latter part, concerning directional biases and Mutationism, is based on various publications and conversations with Arlin Stoltzfus and Dan McShea, whom I gratefully acknowledge. =D)

References:
McShea, D. (2001). The minor transitions in hierarchical evolution and the question of a directional bias Journal of Evolutionary Biology, 14 (3), 502-518 DOI: 10.1046/j.1420-9101.2001.00283.x

Slamovits, C., & Keeling, P. (2008). Widespread recycling of processed cDNAs in dinoflagellates Current Biology, 18 (13) DOI: 10.1016/j.cub.2008.04.054

Stoltzfus A (2006). Mutationism and the dual causation of evolutionary change. Evolution & development, 8 (3), 304-17 PMID: 16686641