No, not really — my title is a bit of a sensationalistic exploitation of the thesis of a paper by Peterson, Dietrich, and McPeek, but I can buy into their idea that microRNAs (miRNAs) may have contributed to the pattern of metazoan phylogenies we see now. It’s actually a thought-provoking concept, especially to someone who favors the evo-devo view of animal evolution. And actually, the question it answers is why we haven’t had thousands of Cambrian explosions.
In case you haven’t been keeping up, miRNAs are a hot topic in molecular genetics: they are short (21-23 nucleotides) pieces of single stranded RNA that are not translated into protein, but have their effect by binding to other strands of messenger RNA (mRNA) to which they complement, effectively down-regulating expression of that messenger. They play an important role in regulating the levels of expression of other genes.
One role for miRNAs seems to be to act as a kind of biological buffer, working to limit the range of effective message that can be operating in the cell at any one time. Some experiments that have knocked out specific miRNAs have had a very interesting effect: the range of expressed phenotypes for the targeted message gene increases. The presence or absence of miRNA doesn’t actually generate a novel phenotype, it simply fine-tunes what other genes do — and without miRNA, some genes become sloppy in their expression.
This talk of buffering expression immediately swivels a developmental biologist’s mind to another term: canalization. Canalization is a process that leads organisms to produce similar phenotypes despite variations in genotype or the environment (within limits, of course). Development is a fairly robust process that overcomes genetic variations and external events to yield a moderately consistent outcome — I can raise fish embryos at 20°C or at 30°C, and despite differences in the overall rate of growth, the resultant adult fish are indistinguishable. This is also true of populations in evolution: stasis is the norm, morphologies don’t swing too widely generation after generation, but still, we can get some rapid (geologically speaking) shifts, as if forms are switching between a couple of stable nodes of attraction.
Where the Cambrian comes into this is that it is the greatest example of a flowering of new forms, which then all began diverging down different evolutionary tracks. The curious thing isn’t their appearance — there is evidence of a diversity of forms before the Cambrian, bacteria had been flourishing for a few billion years, etc., and what happened 500 million years ago is that the forms became visible in the fossil record with the evolution of hard body parts — but that these phyla established body plans that they were then locked into, to varying degrees, right up to the modern day. What the authors are proposing is that miRNAs might be part of the explanation for why these lineages were subsequently channeled into discrete morphological pathways, each distinct from the other as chordates and arthropods and echinoderms and molluscs.
The authors offer up three lines of argument in support of their idea. The first is the weakest, I think, because it relies on interpretation of some somewhat fuzzy fossil evidence. What they suggest is that if increasing canalization would be expected of phyla over evolutionary time, then early groups ought to show more variation within a species than later groups. We don’t have a record of the actual miRNAs, you see, but a less canalized phenotype would be a visible effect of less tightly buffered developmental processes.
In one study measuring intra-specific variation in trilobites, that’s what was found: the Cambrian trilobites were ‘sloppier’, with a wider array of different body forms within a single species, than Devonian trilobites.
The second line of evidence is the one I thought most interesting. It’s a mathematical argument built on the premise that a major function of miRNAs is to reduce noise in development and therefore reduce the amount of variation in the phenotype, when then has the superficially contradictory effect of increasing the ability of selection to promote specific variations.
The argument goes like this. We’re interested in the amount of change in a particular trait over time, Δz. Δz is the product of the selection differential, S, and the narrow-sense heritability of the trait, h2. That last parameter is the key one: the phenotype is the result of additive genetic variation (VA), which is what evolution is selecting for, and the total phenotypic variance (VP), which is all the environmental pressures and internal random noise and various factors, including non-heritable elements, that contribute to the phenotype.
The heritability, h2, is measured as VA / VP…so reducing VP increases h2, which makes Δz larger for the same selection coefficient.
I know…that’s a smear of symbols. Let’s illustrate it graphically.
Imagine you have a population with 3 alleles, a1, a2, and a3, that affect some quantitative trait — let’s say it is antenna size in an insect. Inheriting allele a1 gives the insect a small antenna, allele a3 produces a large antenna, and a2 is intermediate in size.
However, the alleles exhibit variation in the degree of expression. The a3 allele tends to produce larger antennae, but it isn’t a hard and fixed size: some a3 insects have antennae no larger than the largest a1 insects. This is the situation illustrated in T0 of the diagram below — three alleles, each with a fairly wide phenotypic outcome.
Now look at T1, though: an individual has acquired an miRNA that buffers the expression of its a gene, in this case the a3 allele. It produces an antenna size that is right at the peak of the a3 distribution, and does it more precisely, producing fewer small antennae and fewer very large antennae. It makes the output of the a gene more reliable, but doesn’t change the average size at all.
At T2, the miRNA has begun to spread through the population independently of the a alleles, so other insects with the a1 and a2 alleles have acquired it. Again, it doesn’t change the average size generated by the a gene, but it does fine tune it, so the progeny of those individuals carrying the miRNA will more reliably resemble their parents.
The final panels illustrate what happens if there is selection for the larger phenotype. Without the miRNA, the distributions of the products of the a gene are mushy and broad; selection for large antenna size will act against some carriers of a3 and for some carriers of a1 and a2. The change in the proportions of the carriers, Δz, will be fairly small. With the miRNA, though, there is less overlap, and Δz will be greater even if the strength of selection for that particular phenotype is no greater.
Basically, the presence of a buffering system that reduces the phenotypic variability produced by a given allele makes the property relatively more heritable, and increases the selective evolvability of the organism.
Their first argument is indirect, estimating an increase in miRNA constraints from morphological data; their second is theoretical, suggesting that in principle accumulating more miRNA to canalize animal form should make selection more effective; and the third is an obvious one, to ask whether there is any direct molecular evidence showing that lineages have acquired increasing amounts of miRNA over evolutionary time. To answer that, they dug into the databases to see if they can find a pattern of addition of new miRNA families to various animal lineages, and it turns out there is some convincing evidence that that hasa been going on. In the diagram below, the little boxes contain numbers that indicate the number of new miRNA gene families added since the previous node. It looks like all of us, with the exception of one group of sponges, have been busily stockpiling new miRNAs throughout our histories.
This is a provocative idea, and I rather like it. It relies a bit heavily on the premise that the primary role of miRNAs is to simply stabilize patterns of gene expression, which may not hold up as we learn more about them, and it’s also very much in the hypothesis stage right now, but it does propose an answer to a good question. It doesn’t really say why we had a Cambrian explosion in the first place, but it does give a possible explanation for why we didn’t have explosion after explosion, with each phylum blossoming into a wild riot of new body plans and radical morphological changes at frequent points in their histories. It may be because they gradually evolved mechanisms to increase the reliability of multicellular developmental processes.
Peterson KJ, Dietrich MR, McPeek MA (2009) MicroRNAs and metazoan macroevolution: insights into canalization, complexity, and the Cambrian explosion. BioEssays 31:736-747.