Nature has recently published an interesting paper which places severe limits on Darwinian evolution. The manuscript, from the laboratory of Joseph Thornton at the University of Oregon, is titled, “An epistatic ratchet constrains the direction of glucocorticoid receptor evolution.” The work is interpreted by its authors within a standard Darwinian framework, but the results line up very well with arguments I made in The Edge of Evolution. This is the second of several posts discussing it.
Using clever synthetic and analytical techniques, Bridgham et al (2009) show that the more recent hormone receptor protein that they synthesized, a GR-like protein, can’t easily revert to the ancestral structure and activity of an MR-like protein because its structure has been adjusted by selection to its present evolutionary task, and multiple amino acid changes would be needed to switch it back. That is a very general, extremely important point that deserves much more emphasis. In all cases — not just this one — natural selection is expected to hone a protein to suit its current activity, not to suit some future, alternate function. And that is a very strong reason why we should not expect a protein performing one function in a cell to easily be able to evolve another, different function by Darwinian means. In fact, the great work of Bridgham et al (2009) shows that it may not be do-able for Darwinian processes even to produce a protein performing a function very similar to that of a homologous protein.
Before reading their paper, even I would have happily conceded for the sake of argument that random mutation plus selection could convert an MR-like protein to a GR-like protein and back again, as many times as necessary. Now, thanks to the work of Bridgham et al (2009), even such apparently minor switches in structure and function are shown to be quite problematic. It seems Darwinian processes can’t manage to do even as much as I had thought.
(As an aside into the circus world of popular-level debates on Darwinism, the work of Bridgham et al (2009) nicely shows the fallacy of the anti-ID retort that, say, a mousetrap is not irreducibly complex because, if the catch and holding bar are removed, parts of it can still be used as a tie clip. The same principle that holds for cellular machinery would hold for all machinery. Something that was shaped by selection to work as a tie clip would not look like an ancestor of a mousetrap, or easily be converted to one by random changes plus selection. If you look at images of tie clips on the internet, none of them resemble mousetraps — except for those purposely designed by folks who were arguing against irreducible complexity.)
Another point worth driving home in this post concerns the frequently encountered argument that, well, just because one kind of protein can’t develop a useful binding site or selectable property easily doesn’t mean that some other kind of cellular protein can’t. (In keeping with their Darwinian framework, Bridgham et al (2009) seem to allude to this.) After all, there are thousands to tens of thousands of kinds of proteins in a typical cell. If one of them is ruled out, the reasoning goes, many more possibilities remain.
This argument, however, is specious. For any given evolutionary task, the number of proteins in the cell which are candidates for helpful mutations is almost always very limited. For example, as I discussed in EOE, out of thousands of malaria proteins, mutations in only a handful are helpful to the parasite in its fight against chloroquine, and only one is really effective — the mutations in the PfCRT protein. Ditto for the human proteins that can mutate to help resist malaria — there’s just a handful. In the case of the hormone receptors discussed by Bridgham et al (2006), one can note that, out of ten thousand vertebrate proteins, the one that gave rise to a new steroid hormone receptor was an already-existing steroid hormone receptor. This should be quite surprising to folks who believe the many-proteins argument, because the steroid receptor was outnumbered 10,000 to 1 by other protein genes, yet it won the race to duplicate and form a new functional receptor. If all things were equal, we should be very surprised by that. But of course not all things are equal. The reason the receptor duplicated to give rise to a closely-related receptor is because no other protein in the cell is likely to be able to do so in a reasonable amount of evolutionary time.
The bottom line is that, for a given evolutionary task, at best only a handful of proteins will likely be helpful to evolve, at worst none may help. To calculate the probability of, say, a helpful protein-protein interaction developing in response to any particular selective pressure, it’s mistaken to gratuitously multiply odds by the total number of proteins in a cell. Combined with the point made by Bridgham et al (2009), that even tiny structural/functional changes may not be achievable by random mutation/ selection, these considerations pretty much squelch the likelihood of Darwinian processes doing much of significance during evolution.
Bridgham,J.T., Ortlund,E.A., and Thornton,J.W. 2009. An epistatic ratchet constrains the direction of glucocorticoid receptor evolution. Nature 461:515-519.
Bridgham,J.T., Carroll,S.M., and Thornton,J.W. 2006. Evolution of hormone-receptor complexity by molecular exploitation. Science 312:97-101.