Perhaps you saw this article from The New York Times last week. It describes some significant new findings in protein evolution:
In work published last year, Dr. Thornton reported how his group reconstructed an ancestral protein of two hormone receptors found in humans. The two, once identical, diverged along different evolutionary paths. One is now part of the stress response system; the other is involved in different biological processes, including kidney function in many animals.
In the new study, the researchers determined the exact positions of more than 2,000 atoms in the ancestral hormone receptor. The receptor existed in animals that lived more than 440 million years ago, before the last common ancestor of people and sharks. Then the researchers examined what occurred during the next 20 million years — before another split of the evolutionary tree that led to bony fish. “That’s the ancestor of you and a salmon,” Dr. Thornton said.
Alas, the paper is not currently freely available online.
The article goes on to describe how the researchers were able to pinpoint specific mutations that allowed the protein to develop different functions in different evolutionary lineages.
Of the glucocorticoid receptors that have been looked at in different species, five specific mutations are always present and distinguish them from the ancestral receptor. When the scientists introduced the five changes into the ancestral protein, they expected that it would be transformed into a glucocorticoid receptor.
Instead, the protein broke, unable to bind to any hormone.
On further investigation, the scientists found that two other mutations, which had negligible effects by themselves, strengthened some of the protein’s folds so it could withstand the other five mutations. The researchers were also able to show several sequences in which the seven mutations could have occurred without the protein’s functionality ever deteriorating.
Fascinating stuff! Science Daily offers further details.
The significance of this work to evolution / ID disputes is clear. The fact that proteins can get co-opted to perform novel functions is one of the major fallacies in Michael Behe’s notion of irreducible complexity, a fact that has been pointed out to him many times. Behe’s reply has generally been to mock such scenarios as implausible. For example, in anticipating this objection in Darwin’s Black Box, Behe wrote:
For example, suppose you wanted to make a mousetrap. In your garage you might have a piece of wood from an old Popsicle stick (for the platform), a spring from an old wind-up clock, a piece of metal (for the hammer) in the form of a crowbar, a darning needle for the holding bar, and a bottle cap that you fancy to use as a catch. But these pieces couldn’t form a functioning mousetrap without extensive modification, and while the modification was going on, they would be unable to work as a mousetrap. Their previous functions make them ill- suited for virtually any new role as part of a complex system. (p. 66)
This was already an absurd analogy when Behe first devised it in 1996. The present work makes it simply laughable. It has long been known that proteins can acquire novel functions through evolution, but the level of detail here is simply astounding. Courtesy of this new work, we now have a mutation by mutation account of how an important protein acquired a new function. This shows plainly that one of the major processes offered in refutation of Behe’s arguments about irreducible complexity really is workable in practice.
This is how it always goes in dealing with creationists. They make grand pronouncements, based on nothing, about what is possible and what is not. Then the normal march of scientific progress shows them to be utterly wrong. When this happens, they immediately set about the task of explaining that the new work showing that X is possible in no way undermines their earlier contention that X is not possible.
So here’s Behe trying to explain away this work. Here’s his opening salvo:
A recent New York Times story by Kenneth Chang touted a new paper in Science by the laboratory of Joseph Thornton of the University of Oregon as refuting intelligent design. Thornton’s laboratory has been interested in the evolutionary development of differences between two proteins abbreviated GR and MR. Since the two proteins are very similar, and since they bind very similar small hormone molecules, they likely developed from an ancestral gene by gene duplication and subsequent diversification. Despite Chang’s story, none of that challenges intelligent design, which agrees that minor evolutionary changes can happen by random mutation and natural selection.
Sadly, there are at least two big flaws in this paragraph. The first is that “intelligent design” is not the sort of thing that makes empirical claims about what is and is not achievable by random mutation and natural selection, or any other evolutionary mechanism for that matter. Beyond some bald assertions that certain “irreducibly complex” biochemical systems could not have evolved naturally, ID offers no guidance at all on any empirical question of interest to biologists. His conclusion that the findings of Thornton and his coauthors represent minor evolutionary changes is not based on any rigorous notion of how we distinguish minor from major changes. Rather, it is based solely on Behe’s need to dismiss any and all data contrary to his blinkered biological opinions.
Second, Chang did not say these findings challenged “intelligent design.” He was far more focused than that:
Supporters of intelligent design, who question evolution, have argued that mutations, occurring one by one, could not refold a protein into a new function, because the mutations would first unravel the protein into a useless, unfolded configuration.
The new study refutes that assertion, at least in this instance.
Moving on, Behe offers up the following summary of what the researchers here actually did:
The gist of the new paper is that the workers reconstructed in the lab what they thought would be the ancestral protein, as well as several later evolutionary versions of it. To get to a protein mimicking modern GR, they purposely introduced several mutations to the ancestral form. The first few mutations took the protein’s activity part of the way toward the modern activity. But adding several other mutations that they thought would increase the specific activity to that of modern GR unexpectedly caused the protein to lose all its ability to bind the hormones. So, after thinking awhile, the authors then went back and intentionally introduced other mutations which did not affect hormone binding, but which they hoped would strengthen a particular part of the protein. After deliberately strengthening that part, they found they could add the final mutations, the protein would retain its activity, and its activity would be much more similar to modern GR.
This is a very misleading way of putting things. Behe is keen on describing things in a way that puts maximal emphasis on the role of human design in achieving these results. But the researchers didn’t just introduce several mutations. They first reconstructed the most likely form of the gene ancestral to those coding for modern GR and modern MR proteins. They did this by applying standard statistical techniques to a large sample of known variants of these genes in modern organisms. They then used known phylogenies to discover mutations that occurred during the time period in which the evolution of these genes was thought to occur. It was this process that led them to the appropriate mutations to investigate, not just some process of intelligent cogitation.
Likewise, when their initial hypothesis about the effect of the mutations did not pan out, they did not just think awhile and conjure up some new ones to try. Instead, they deduced that certain “permissive mutations” (that is, mutations whose occurrence mitigated the harmful effects of introducing the previously discussed substitutions all at once) must have occurred, and by a series of clever experiments were able to zero in on where those mutations were.
In other words, they weren’t just trying out mutations willy-nilly. They were using a large data set to guide them to where the mutations had to be. This is what elevates their evolutionary hypothesis above the level of a just-so story, and gives it a solid basis in empirical fact.
But Behe is just getting warmed up:
Now, dear reader, can you guess which parts of Darwin’s theory this all neglects? Of course — both random mutation and natural selection. The workers nicely showed it is quite reasonable to think that the one ancestor protein could produce two descendants, but they didn’t even try to address the question of whether it could happen by chance plus selection. Of course, getting a single amino acid mutation by chance is not a problem. But in order to have the mutation be positively selected, it has to benefit an organism. The authors (and news stories) completely ignore that — the authors didn’t measure whether duplicating the ancestral gene, and then modifying it, would benefit an organism that was used to relying on just one protein (admittedly that would be hard to do). What’s more, in order to be confident that a multi-mutation scenario reconstructs a Darwinian process, all subsequent mutations have to be positively selected, too.
Of course, that final sentence is totally false. Mutations do not have to be positively selected to persist in a population. They could be neutral, or even slightly deleterious, and still persist. Such mutations might be of small consequence when they occur, but if they later combine with other mutations to produce a beneficial effect, then selection can act on the whole complex.
It is simply absurd to say that the scenario described in the paper leaves out the part about random mutation and natural selection. The mutations described by the authors are of standard types (mostly just simple substitutions), and all are known to occur randomly in great abundance. Furthermore, the authors were able to describe in great detail the effects of each of the mutations. One need only hypothesize a scenario in which those effects were beneficial (not hard to do) and you have your evolutionary scenario.
But they aren’t. Although they test none of the mutations in actual organisms, the authors themselves feel that the very particular mutations they deliberately introduced, which strengthen the protein but don’t affect hormone binding, would have been neutral. That makes those mutations much, much less likely to spread in a population, to be available later for when the beneficial mutations came along. In other words, the authors themselves think the scenario involves a big stroke of luck. In the New York Times Chang quotes Thornton: “’These very exquisitely adapted bodies we have represent a role of the dice,’ Dr. Thornton said. ‘And they could have turned out very differently.’”
Of course, Thornton said nothing about a big stroke of luck. He said merely that things could have turned out differently. And since the probability that a given neutral mutation will be present in a randomly chosen member of a population depends heavily both on the population size and the amount of time involved, Behe likewise has no basis for asserting that it is “much, much less likely” that such mutations would spread. (Incidentally, less likely than what? A slightly beneficial mutation? A strongly beneficial mutation?)
Behe tells us that this work establishes the plausibility of one ancestral protein giving rise to two descendants. But even though all of the mutations being hypothesized are very small and even though we have a good grasp on the phenotypic effects of the underlying genes, he finds it implausible that standard evolutionary mechanisms provide the explanation. Very well. What explanation does Behe prefer? Directed mutations? Divine intervention? How does an ID proponent make sense of this data? Here’s Behe’s answer:
Big strokes of “luck,” however, point much more to intelligent design than to Darwinism. If evolution were guided or designed to unfold in a particular way, then very improbable events would be expected to be packed into it. The bottom line is that, while the new paper is very clever work, it offers no support at all to Darwinism. If anything, the authors careful work points strongly away from randomness. If even such minor evolutionary differences as those between GR and MR are problems for chance-driven evolution, greater evolutionary difference are almost certainly beyond the edge of random evolution.
What? If the changes from the ancestral gene to the modern GR gene are part of the design of evolution, then they happened with very high probability, perhaps with certainty depending on the extent of the plan. And why, exactly, would we expect seemingly improbable events to be packed into “evolution by design?” A certain amount of chance is what you expect from a process based on randomly occurring mutations and selection in stochastically changing environments. If things are unfolding according to a plan, improbability is precisely what you do not expect to see.
Perhaps one of these days Behe will explain to us precisely what ID does, and does not, predict about nature. I’m not holding my breath.
A while back Michael Behe offered the following description of what sort of evidence he would need before accepting the naturalistic explanation over his preferred version of “God did it.” Not only would he need a step-by-step list of mutations…
…but also a detailed account of the selective pressures that would be operating, the difficulties such changes would cause for the organism, the expected time scale over which the changes would be expected to occur, the likely population sizes available in the relevant ancestral species at each step, other potential ways to solve the problem which might interfere, and much more.
It sure looks to me like the authors addressed all of these points in considerable detail. (With the exception of the population sizes, which is hardly the major sticking point here. Behe only included that one to make his list longer and more impressive.) But — surprise! — it still isn’t enough. In fact, in Behe’s world, when scientists produce a comprehensive, step-by-step, well-documented, evolutionarily feasible sequence of events through which a protein attains a novel function, it somehow ends up being evidence for design nonetheless. What a charming fellow.