The Gene Genies, Part 1: The Squids of Lamarck.
You know the drill. DNA holds the source code; RNA carries it to the ribosomes; ribosomes build stuff for the cell. Of course, the details of cellular operation are a million times more intricate than this— some RNA acts not to courier code but to switch genes on and off, for example— but it’s this venerable three-step that puts the tinkertoys together.
Now. If a sufficiently unscrupulous RNA molecule had an agenda at odds with the wishes of Daddy DNA, it could do a fair bit of damage. Change an instruction or two while on the road, enlist some hitchhiking enzyme into provoking a frame-shift or a faux-point-mutation. The nucleus mails off an order for Game of Thrones and the ribosome receives one for Spongebob Squarepants.
Who needs gamma rays? This guy hacks his own DNA. (Photo Brandi Noble, NOAA)
The term is RNA editing and it occupies center stage in this recent paper on cephalopod genetics. RNA editing is generally a very rare event. This makes it all the more remarkable that Alon et al report over 57,000 recoding sites for the Longfin Inshore Squid— an order of magnitude higher than reported for any other species. Even cooller, all these hijacked codes seem to be involved in building the nervous system. (“Synaptic vesicle cycle”, “axon guidance”, “actin cytoskeleton”, and “Circadian rhythm” are all processes listed as massively rewritten downstream of the DNA.)
This is part of a squid synapse. Red and yellow bits are subject to change without notice. (from Alon et al.)
It’s right there in the title: The Majority of Transcripts in the Squid Nervous System are Extensively Recoded. As the authors point out, this necessitates a major rethink of the whole squidly evolutionary process. But there are applications beyond such obvious intrinsic biological interest.
If I was interested in rebuilding a cephalopod to my own ends— perhaps adding organic tasers, or extra eye-sockets repurposed as oceanographic sensors (imagine luciferin fluorescence as an indicator of dissolved O2, which trigger photopigments in a modified retina, which in turn send that data back to a central nervous system via an extra optic nerve!)—
Well, let’s just say that a squid who comes pre-equipped with its own set of downstream editing enzymes, targeted to major CNS functions, might come in really handy.
(Coming up in Part 2: Selection-resistant genes. What could possibly go wrong?)
So then the obvious question is what mechanism controls the RNA editing process? And do these changes only happen at embryonic stages of development, or can they be invoked in adults?
See, independent of any tweaking you might want to do to squidly creatures, it would be nice to give my progressively decrepit body a mechanism for making long chain carbon fibres, and then another mechanism for physiogically merging those fibres with my bones. And for that matter, expand the back of my cranium enough to allow my total number of brain cells to expand by 50%.
Occasionally we get a reminder we don’t actually know shit.
What about that reactionless thruster thing? It doesn’t seem to stay dead like all decent magic discoveries should, might there be something to it? Without the Koch brothers being involved?
Also let’s hope this guy turns out to be a crank because we might end up suffering from optimism if it pans out.
More proof, as if we needed it, that the squids are really just biding their time until they decide to eat all our babies and make our more attractive adults into whatever the squid version of a hat-rack is.
This is pretty crazy.
Do you reckon this system allows for rapid tweaking of the phenotype of the nervous system to fit the environment, from a sort of “template” (or set of templates) in the DNA? The authors seem to lean in this direction.
Or is this some apparently unusual (in metazoans) evolutionary system with two parallel systems of information encoding?
Or both? Or something else entirely?
Oh, biological systems and your endless variety. 😀
Fascinating stuff. It feels so satisfying when something confirms my completely irrational intuition that things are far more nuanced and complex than we imagine.
Actually, in all seriousness: are those recoding RNAs produced by a DNA sequence somewhere else in the genome? Maybe somewhere more prone to bursts of rapid evolution? Is this the genome hacking bits of itself that are fairly canalised in the normal course of events? Maybe I should read the paper, eh?
Thinking about this more, why does a squid need extra nervous-system flexibility? Assuming we are seeing a fact about this species that has been proofed or caused by some kind of evolutionary pressure, what kind of pressure makes you better fit if you can radically change your synaptic interfaces? What does it let you do, or prevent you from doing or having happen to you?
Enzymes, editosomes— these things are all proteins. Ultimately, their construction has to start with the genes. Unless of course those genes got edited post-hoc en route to the ribosome, in which case… oh wait…
As to whether the process can be invoked in adults, I’m guessing it depends on the gene. A lot of genes get shut off after they do their developmental thing, but obviously lots of other keep chugging away throughout the life of the cell, doing cellular housekeeping. So if you held a gun to my head, I’d say the first kind are set in stone but the second kind might be tweakable in adults.
I dunno. That’s three independent confirmations of an effect which, some say, violates the laws of physics (the rest say it’s virtual-particle pairs, which confuses me because I thought the whole point on those things was they lived and died below Planck time, and the whole reason for that was that it’s okay if you break the rules so long as you rebalance the books before the universe ticks over another clock cycle. But how can you do that if you’re having an impact at larger scales? And what about Hawking radiation?)
Anyhow, if our model says it doesn’t work and your data say it does, and your data replicate, chances are it’s your model that needs changing. So I’m excited, at least.
A grad school buddy of mine back at UBC did his thesis on “metavariation”— basically, that certain bits of code can affect overall mutation rate. Don’t know what’s happened more recently (I think it panned out), but you can see the benefits: organisms in a tight spot could up their mutation rates when a wider range of options would be beneficial, then dial it back again when things calmed down (and a high mutation rate would cost more than it returned). I’m guessing this is something like that— except that rather than generate completely novel mutations, these squids just pick and choose from a pre-existing repertoire.
All this is just smoke out my ass, though. The paper itself doesn’t address any of this stuff; it just compares the code with the product and tallies up the mismatches. They didn’t look at the impact of those mismatches in wild systems.
Good question. The authors speculate a bit, and cite previous work—
—which suggests that yeah, this has adaptive consequences. But again: speculation.
The size differential alone–though I’m by no means proficient in squid species and those ranges–might explain some of it. If you grow to 20x your previous size, you might need a few extra sensory nerves to manage the mass increase.
I really wonder what will happen when people look for this stuff in other cephalopods. Squid and octopus in particular seem to have invented a highly-cephalised and very sophisticated nervous and visual system pretty much from scratch, with no real evolutionary background (I could be wrong, but I don’t think any other members of the phylum Mollusca even have proper camera eyes, and they’re certainly nowhere near as cephalised). It might be a lot easier to do that by tweaking the expression of developmental genes responsible for the nervous system than the slow old-fashioned way that us vertebrates did it.
Okay, maybe I’m dumb, but I don’t see anything lamarckian about it.
Cool and awesome, yes, but unless the squid can selectively re-engineer its RNA towards a specific external goal, it isn’t very lamarckian (it’s essentially the same deal as posttranslational cleavage of proteins, only instead of cleaving your proteins, you cleave the RNA, which, come to think of it, might be a slightly more efficient than cleaving enzymes after they already have been “manufactured”)
[…] Wow. […]
[…] Wow. […]
Peter Watts,
[…] Wow. […]
A lot of people are all excited about this without knowing the first thing about how genes actually work, and why this is likely to be a somewhat interesting complication of the basic story without being a breathtaking revelation.
Genes are basically computational rules for a “production system” (rule-based) computer.
The genome is a computer program, and the mechanisms for gene expression are a computer—but NOT the kind of computer that many people are familiar with. It isn’t a sequential von Neumann machine or a Turing machine, but a “production system” computer. (Most people who try to explain genes in basic computational terms get this more or less wrong.)
Production systems were invented by Emil Post even before Turing invented Turing machines, and way before von Neumann machines, which von Neumann didn’t invent. (Or you could say they were invented many millions of years ago by nature.) IMO production systems are usually a better of way of thinking about computation anyhow, and it turns out they’re an excellent fit to what genes do. And IMO that’s VERY interesting.
A typical gene is an if-then rule like
IF (A and B and not C) THEN (D and E and F)
This isn’t an IF-THEN statement like in a sequential programming language like C or Pascal. It’s more like a logical rule, saying that if you know A and B are true but C isn’t, then you can conclude that D and E and F are true.
(About 90 percent of genes are purely computational rules like this, used just to turn other genes on or off, or more generally to make them fire more or less often.)
The left-hand side of the rule is a set of conditions for the rule to fire, and the right hand side is a set of “propositions” that the gene “asserts” if its preconditions are satisfied and the rule fires.
The left-hand side is implemented by the “control region” of the gene, which has binding sites for signaling molecules—if the right molecules are floating around, they will dock to those binding sites and either promote or inhibit the gene from being expressed—i.e., make it more or less likely to fire.
The right-hand side of the rule is implemented by the “coding region” of the gene, which is mainly a sequence of DNA bases that are transcribed to an eqivalent RNA sequence. That RNA sequence is usually transcribe to a sequence of amino acids making up a particular protein. Most such proteins are just signaling molecules that affect other whether other genes are expressed (i.e., whether other rules’ preconditions are satisfied so that they can fire).
The protein typically folds up into a relatively compact lump, with bumps of various shapes poking out. The shapes of the bumps are what’s really important—they implement the “propositions” that the rule “asserts”. So each “proposition” (A, B, etc.) in a rule is implemented by a particular bump shape. There are many thousands of distinct bump shapes.
One complicating factor in this story is that when a gene is transcribed—i.e., its coding region is basically copied to make corresponding RNA sequence—the resulting RNA may not be translated to a protein. The RNA itself may serve as a signaling molecule, because it may itself have bumps of the right shape, so that it can dock to genes’ promoting or inhibiting sites.
Given all this—a pretty powerful production system “programming language”—it’s unclear what RNA editing is mostly used for, or whether it adds much to the power or expressiveness of the basic rule-firing programming language, which is a pretty interesting computational system already, and one which few people have even the most basic understanding of AS a computational system.
(Most molecular biologists do not know about Post’s production systems, much less “massively parallel stochastic fuzzy systems”—they think of a “computer” as something that executes programs sequentially by default, and do not realize they are looking at a computer at all.)
RNA editing may turn out to be pretty dull in those terms—never doing anything that couldn’t be done about as well with propositional rules as described above—and its prevalence in squid may be an artifact of something wrong with squids’ basic genetic program. It may be an ugly hack that evolved to fix something that other organisms fixed in a better way, within the basic rule-firing “programming paradigm.”
Or it could turn out to be really interesting, adding something basic and really useful to the basic programming language, so that evolution can solve programming problems more generally and elegantly. THAT would be very cool, but as I understand it, nobody has any clue at this point whether that’s true.