The following is taken from
a recent post of mine to the ISCID Forum. Since Peter has not responded yet on that forum, I wondered if perhaps he had forgotten the discussion and thought that posting the message here might remind him of it (assuming he still posts here). Some of the information in this post may also be interesting for other forum members.
As a brief introduction: I am a researcher in the lab that has carried out most of the study of the so-called "redundant" human skeletal muscle protein alpha-actinin-3. This protein is absent in about 20% of the normal population due to a polymorphic premature stop codon in the
ACTN3 gene. We have recently submitted a paper (not yet accepted) showing that this protein is not in fact redundant, but actually provides an advantage in terms of sprinting performance. We have also obtained data suggesting that the
absence of this protein provides an advantage in terms of endurance performance. I am currently performing research to elucidate the function of alpha-actinin-3 by generating a mouse model of alpha-actinin-3 deficiency, and learn more about its evolution in humans by studying the patterns of genetic variation which surround the site of the polymorphic stop codon.
The following post is a response to Peter's claims that the apparent redundancy of alpha-actinin-3 poses a problem for modern evolutionary theory. Interested readers may also wish to read other posts to the ISCID thread on this topic (linked above).
=====================================================================
Peter’s claims that the alpha-actinins violate evolutionary theory appear to be based on the following logic (quote below obtained from
a post of Peter’s on the EvC forum). I hope the Administrator will indulge me in a little quote-by-quote responding — I do not wish to be omit any aspect of Peter’s arguments or quote them out of context, and I would like to respond to Peter’s quote in considerable detail. This is, after all, my livelihood that Peter is attacking, since I will be spending at least the next three years specifically researching the function and evolutionary history of the human
ACTN3 gene.
quote:
What exactly is the problem with the redundant actinins? Let's have a careful look.
According to NDT, duplication gave rise to two alpha-actinin.
A close look at these redundant genes reveals that the differences are the result of point-mutations.
This is not entirely true — the two
ACTN genes have also accumulated indels in the highly divergent region at the 5’ end of their coding region (corresponding to the N-terminal region of the two actinin proteins). But for most of their length, the divergence between the two genes is due to point mutations, yes.
quote:
Neutral evolution rate is about 10(exp)-9/nucleotide/year, and recent genome wide studies present evidence that purifying selection worked upon duplicated genes (A. Wagner, mini-review in Genome Biology 2002).
For the sake of argument I’ll accept Peter’s figure for the neutral evolution rate, although it is worth pointing out that this figure can vary widely throughout the genome, over time, and in different populations (it is particularly affected by effective population size, for instance). Of course, it goes without saying that this rate is also utterly inapplicable to regions of the genome which are acted on by natural selection.
I also accept the statement that purifying selection acts on duplicated genes — in fact, to go one better, I will specify that purifying selection has certainly acted on the
ACTN3 gene. Analysis of substitution rates during the divergence of mouse and human
Actn3 genes revealed evidence of very strong purifying selection, as manifested by very low levels of nonsynonymous (amino acid-changing) mutations compared to synonymous (so-called silent mutations).
quote:
The ACTN2 and ACTN3 genes are approximately 3000 bp, and share 85% sequence homology. Moreover, the ACTN3 gene is highly conserved within mammals.
Peter’s sequence homology figures are slightly too high here. Despite relatively high levels of similarity along most of the two genes, the most 5’ regions (corresponding to the N-terminals of the proteins) are so divergent that they cannot be aligned in any meaningful sense. Even when these regions are removed from the calculation, however, the nucleotide sequence similarity between the coding regions of the two human genes is only 76%. Because many of the mutations are silent the encoded proteins are 81% identical and 91% similar or identical.
The
Actn3 gene is certainly highly conserved in mammals — human and mouse
Actn3 coding regions are 89% identical, while the encoded proteins are 97% identical and 98% similar or identical, so Peter is quite correct there.
Now, Peter’s facts are more or less accurate, with some relatively minor errors. These are the conclusions he draws from them:
quote:
1) These data say that approximately 450 bp changes occurred on neutral positions, i.e. positions not under selective constraint.
2) These data mean that it would take about 10(exp)6 years for 3 random mutations to occur in the duplicated gene. Thus 150 million years for 450 neutral mutations.
3) These data imply that after each point-mutation there was (neutral) purifying selection. So, here we have to introduce neutral purifying selection again. What exactly is selection on neutral genes? Not a single evolution biologist was able to tell me, thus far.
The number of substitutions is actually considerably higher than Peter’s figure — there are 629 differences between the coding regions of two genes even if the divergent region is excluded, and the number if this region is considered is larger and includes a number of indel events as well as point mutations.
Peter’s calculation of the divergence time for
Actn2 and
Actn3 is also off by approximately a factor of two. The best estimate for the divergence time, based on phylogenetic models, is closer to 300 million years (Dixson
et al. 2003. J Mol Evol 56(1):1-10).
But these are quite trivial errors compared to the erroneous claim that these changes must have occurred on neutral positions... not under selective constraint. This is entirely false, and this error is (IMO) largely responsible for Peter’s misconceptions regarding the evolution of the actinin gene family.
This is an extremely important point, and warrants emphasis:
Just because a gene is redundant in one species, it does not follow that it has always been redundant.
In other words, even if
ACTN3 really were redundant in humans (and it is not, as my lab’s study on genotype frequencies in athletes, discussed above, has shown), this does not mean that it has been evolving without selective constraints — i.e. neutrally — since its inception. Not only is Peter’s claim to the contrary illogical, it is also directly contradicted by the available evidence, e.g. the fact that
Actn3 is not redundant in mice, as its expression in skeletal muscle is not overlapped by that of
Actn2 (the presumed compensatory factor for
ACTN3 deficiency in humans); and humans are the only known mammal suffering from null mutations in
Actn3 — all other studied mammals, including chimpanzees, have fully functional actinin-3.
These data suggest that
Actn3 has been subjected to purifying selection because it performs an important function in the skeletal muscle of most mammals. It appears likely that this function has only recently changed in humans — likely due to evolutionary changes in skeletal muscle, including the loss of very fast-type (IIb) muscle fibres which are present in smaller mammals — allowing a null allele to spread through the human population.
But as the athlete research performed at my lab has shown,
ACTN3 is not redundant even in modern humans — in fact, while functional actinin-3 is required for optimal sprint performance, the null allele appears to have spread through the action of natural selection because it provides an advantage for endurance performance. Thus there is literally no support for the notion that actinin-3 has been undergoing purely neutral evolution, even in modern humans, and there is absolutely no need to postulate neutral purifying selection. Normal, run-of-the-mill purifying selection to maintain protein function is sufficient.
Peter’s final conclusions are:
quote:
As you see, it is highly doubtful that the mechanism that keeps genes (unchanged) in the genome is by natural selection. [In fact this example is a falsification of natural selection. Thus, I demonstrated that both random mutation (see previous posting) and natural selection (this is not the only example) can be falsified at the molecular level. Conclusion: NDT RIP].
As I have shown, Peter’s claims rest on a fundamental misunderstanding: namely, that apparent (and in this case illusory) redundancy of a gene in one species means that that gene must have been redundant throughout its entire evolutionary history. Once that misconception has been set aside it is clear that there is no need for mysterious forces such as neutral purifying selection to explain the conservation of the
Actn3 gene throughout mammalian evolution — normal purifying selection maintaining gene function is sufficient. Thus
ACTN3 presents no difficulties for standard evolutionary theory, and indeed provides a fascinating example of evolution in action in modern humans.
I hope that this post has been sufficient to clarify any confusion regarding the consistency of
ACTN3 with modern evolutionary theory. However if Peter (or anyone else) still has doubts regarding any of the points I have raised I would be more than happy to respond to them.
======================================================================
Mesk.