I've written a few times here about the ongoing work of Joe Thornton, a biologist at the University of Oregon and the Howard Hughes Medical Institute. Thornton studies how molecules evolve over hundreds of millions of years. He does so by figuring out what the molecules were like in the distant past and recreating those ancestral forms in his lab to see how they worked. I first wrote about his work looking at how one molecule in our cells evolved from one function to another (here, here, and here). [Update: These links are now fixed.]
Most recently, I wrote in the New York Times about his latest experiment, in which he and his colleagues found that the evolution from the old function to the new one has now made it very difficult for natural selection to drive the molecule back to its old form. Its evolution has moved forward like a ratchet.
Thornton's new work turned up last week on a web site run by the Discovery Institute, a clearinghouse for all things intelligent design (a k a the progeny of creationism). Michael Behe, a fellow at the Institute, wrote three posts (here, here, and here) about the new research, which he pronounced "great."
This is the same Michael Behe who, when Thornton published the first half of this research, declared it "piddling."
Why the change of heart? Because Behe thinks that the new research shows that evolution cannot produce anything more than tiny changes. And if evolution can't do it, intelligent design can. (Don't ask how.)
I pointed out Behe's posts to Thornton and asked him what he thought of them. Thornton sent me back a lengthy, enlightening reply. Since the Discovery Institute doesn't allow people to comment on their site, I asked Thornton if I could reprint his message here.
You may want to revisit my posts I linked to above to get a more detailed description of Thornton's work before delving into Thornton's reply. And while his reply is quite clear, there are a few terms that may be confusing, so let me preface it with a quick and dirty glossary of terms:
Cortisol: A hormone
Genetic drift: the change in the frequency of an allele (a version of a gene) in a population thanks to chance, not natural selection. Genetic drift can spread an allele through an entire population even if it provides no boost to reproductive success. It can even spread some mildly harmful ones under the right conditions
Glucocorticoid receptor: A receptor that binds cortisol. This is the molecule Thornton has studied, documenting the series of mutations that transformed it from an ancestral receptor sensitive to another hormone.
GR: Glucocorticoid receptor
Steroid receptors: A class of receptors that can bind steroids (a group of molecules that includes hormones such as cortisol).
And without further ado, here's Thornton's message--
Thanks for asking for my reaction to Behe’s post on our recent paper in Nature. His interpretation of our work is incorrect. He confuses "contingent" or "unlikely" with “impossible.” He ignores the key role of genetic drift in evolution. And he erroneously concludes that because the probability is low that some specific biological form will evolve, it must be impossible for ANY form to evolve.
Behe contends that our findings support his argument that adaptations requiring more than one mutation cannot evolve by Darwinian processes. The many errors in Behe's Edge of Evolution -- the book in which he makes this argument -- have been discussed in numerous publications.
In his posts about our paper, Behe’s first error is to ignore the fact that adaptive combinations of mutations can and do evolve by pathways involving neutral intermediates. Behe says that if it takes more than one mutation to produce even a crude version of the new protein function, then selection cannot drive acquisition of the adaptive combination.
This does not mean, however, that the evolutionary path to the new function is blocked or that evolution runs into a “brick wall,” as Behe alleges. If the initial mutations have no negative effect on the ancestral function, they can arise and hang around in populations for substantial periods of time due to genetic drift, creating the background in which an additional mutation can then yield the new function and be subject to selection. This is precisely what we observed in our studies of the evolution of the glucocorticoid receptor (GR).
In our 2007 paper in Science, we showed that multiple mutations were indeed required for the GR to evolve its specificity for the hormone cortisol; some of the mutations that trigger the change in function are deleterious if introduced in isolation, but others are “permissive”: they have no apparent effect on the function of the protein, but once they are in place the protein can tolerate the other mutations that shift and then optimize the new function.
By experimentally characterizing the functional effect of the key historical mutations in various combinations, we showed that there were indeed pathways from the ancestral protein to the new function that passed only through permissive and beneficial intermediate states.
A path to a new function that involves neutral intermediates is entirely accessible to the evolutionary processes of mutation, drift, and selection. Our work showed that these classic neodarwinian processes are entirely adequate to explain the evolution of GR’s new function. (I should mention that pathways involving mildly deleterious intermediates are also accessible in reasonable time under some population genetic conditions; it's just that their relative probability is lower than those involving neutral or beneficial intermediates.)
Behe’s discussion of our 2009 paper in Nature is a gross misreading because it ignores the importance of neutral pathways in protein evolution. We studied whether the key mutations that drove the “forward” evolution of GR’s new function could be reversed in a later version of the GR, restoring the ancestral conformation and function. We found that the later version of the protein could no longer tolerate the ancestral amino acids at these key sites, despite the fact that they had been present in the protein at an earlier stage of evolution.
We identified the specific “restrictive” historical mutations, which occurred after the shift in function, that either clashed with or failed to support the ancestral conformation. If these mutations are reversed first before the key function-switching mutations, the ancestral structure and function can be restored.
Reversing the restrictive mutations alone does not enhance the ancestral function, but in some orders they have no effect on the GR’s function. These restrictive mutations are simply the flip-side of permissive mutations: reversal of a restrictive mutation is a permissive mutation for reverse evolution. As Fig. 4 in our paper shows, there are several pathways back to the ancestral sequence that pass only through steps that are neutral or beneficial with respect to the protein’s functions.
Thus, all pathways to the ancestral sequence and structure are not blocked, as Behe says they are. The chance effects of genetic drift could allow the protein to “float” along such paths, producing the appropriate background in which the function-shifting mutations could be reversed. However, selection alone would not be sufficient to drive the protein deterministically through the neutral steps. If selection for the ancestral function were imposed, reversal to the same sequence and conformation as the ancestor would be unlikely, though not impossible.
Taken together, the existence of permissive and restrictive mutations indicates that neutral paths to specific adaptive combinations of mutations are opening and closing during evolution. If the clock could be turned back and history allowed to run again, it’s likely that some different path would be followed, and different protein forms would evolve by the natural processes of evolution.
This brings us to Behe’s second error, which is to confuse reversal to the ancestral sequence and structure with re-acquisition of a similar function. We showed that restrictive mutations make selection alone insufficient to drive the protein back to the same form as that found in the ancestor. But nothing in our results implies that, if selection were to favor the ancestral function again, the protein could not adapt by evolving a different, convergent, underlying basis for the function.
Indeed, directed evolution experiments in the laboratory have shown that mutation and selection alone can cause steroid receptor proteins to rapidly evolve sensitivity to new hormones; some of the mutations involved are different from those that occurred during the historical evolution of ancient proteins.
Our paper shows that re-evolution of the underlying ancestral form is unlikely, but it says nothing about the re-evolution of the ancestral function. We found that chance processes play a key role in determining which adaptive forms actually evolve under selection, but this does not mean, as Behe alleges, that no adaptive form can evolve.
Finally, Behe erroneously equates "evolving non-deterministically" with "impossible to evolve." He supposes that if each of a set of specific evolutionary outcomes has a low probability, then none will evolve. This is like saying that, because the probability was vanishingly small that the 1996 Yankees would finish 92-70 with 871 runs scored and 787 allowed and then win the World Series in six games over Atlanta, the fact that all this occurred means it must have been willed by God.
Consider the future: there are countless possible that could emerge from our present state, making the probability of the one that actually does evolve extraordinarily low. Does this mean that the future state that will ultimately emerge is impossible? Obviously not. To say that our present biology did not evolve deterministically means simply that other states could have evolved instead; it does not imply that it did not evolve.
Consider your own life history as an analogy. We can all look back at the road we have traveled and identify chance events that had profound effects on how our lives turned out. "If the movie I wanted to see that night when I was 25 hadn't been sold out, I never would have gone to that party at my friend’s house, where I met my future spouse….” Everyone can tell a story like this. The probability of the life we actually lead is extraordinarily small. That obviously doesn't mean that its historical unfolding was impossible.
That we inhabit an improbably reality requires a divine explanation only if we, like Behe, take the teleological view that this is the only reality that could exist. But if we recognize that the present is one of many possibilities, then there is no difficulty reconciling the nature of evolutionary processes with the complexity of biological forms. As history unfolds, potential pathways to different futures are constantly opening and closing. Darwinian processes are entirely adequate to move living forms along these pathways to a remarkable realization – but just one realization out of many others that could have, but didn't, take place.
I considered hard whether I should address Behe’s argument or ignore it. I am well aware that Behe and his supporters might portray my response as an indication that there is scientific debate over the possibility of adaptive protein evolution: “Look, an evolutionary biologist who actually does scientific research is arguing with me; let’s teach this controversy in public schools!” Because Behe has grossly misinterpreted the results of my research to support his position, however, I feel some responsibility to set the record straight.
Behe’s argument has no scientific merit. It is based on a misunderstanding of the fundamental processes of molecular evolution and a failure to appreciate the nature of probability itself. There is no scientific controversy about whether natural processes can drive the evolution of complex proteins. The work of my research group should not be misintepreted by those who would like to pretend that there is.