Michael Behe On The Theory of Irreducible Complexity

Michael Behe
Discovery Institute
January 1, 2006
Print ArticleNOTE: Comments from biochemist Dr. Michael Behe, Senior Fellow, Discovery Institute’s Center for Science & Culture in response to Science magazine’s forthcoming article “Evolution of Hormone-Receptor Complexity by Molecular Exploitation,” and regarding irreducible complexity, evolution and intelligent design. Originally published at wwww.idthefuture.com.

The study by Bridgham et al (2006) published in the April 7 issue of Science is the lamest attempt yet — and perhaps the lamest attempt that’s even possible — to deflect the problem that irreducible complexity poses for Darwinism.

The bottom line of the study is this: the authors started with a protein which already had the ability to strongly interact with three kinds of steroid hormones (aldosterone, cortisol, and “DOC” [11-deoxycorticosterone]). After introducing several simple mutations the protein interacted much more weakly with all of those steroids. In other words, a pre-existing ability was decreased.

That’s it! The fact that this extremely modest and substantially irrelevant study is ballyhooed with press releases, a commentary in Science by Christoph Adami, and forthcoming stories in the mainstream media, demonstrates the great anxiety some folks feel about intelligent design.

In the study the authors wished to see if two related modern proteins called the glucocorticoid (GR) receptor and mineralocorticoid receptor (MR) could be derived from a common ancestral protein. Using clever analysis the authors made a protein that they thought represented the ancestral protein. That protein binds several, structurally-similar hormones, as does modern MR. They then introduced two amino acid changes into the protein which are found in modern GR. The two changes caused the ancestral protein to bind the different kinds of hormones anywhere from ten- to a thousand-fold more weakly. That protein bound aldosterone about three-fold more weakly than cortisol. The authors note that modern GR (in tetrapods) also binds aldosterone more weakly than cortisol. So perhaps, the thinking goes, an ancestral gene that could bind both hormones duplicated in the past, one copy accumulated those two mutations to become the modern GR, and the other copy became modern MR.

Click here to read Michael Behe's response to Judge Jones' ruling in the Dover intelligent design trial, "Whether Intelligent Design is Science."

Here are number of comments in response:

1) This continues the venerable Darwinian tradition of making grandiose claims based on piddling results. There is nothing in the paper that an ID proponent would think was beyond random mutation and natural selection. In other words, it is a straw man.

2) The authors (including Christoph Adami in his commentary) are conveniently defining “irreducible complexity” way, way down. I certainly would not classify their system as anywhere near IC. The IC systems I discussed in Darwin’s Black Box contain multiple, active protein factors. Their “system”, on the other hand, consists of just a single protein and its ligand. Although in nature the receptor and ligand are part of a larger system that does have a biological function, the piece of that larger system they pick out does not do anything by itself. In other words, the isolated components they work on are not irreducibly complex.

3) In the experiment just two amino acid residues were changed! No new components were added, no old components were taken away.

4) Nothing new was produced in the experiment; rather, the pre-existing ability of the protein to bind several molecules was simply weakened. The workers begin their experiments with a protein that can strongly bind several, structurally-very-similar steroids, and they end with a protein that at best binds some of the steroids ten-fold more weakly. (Figure 4C)

5) Such results are not different from the development of antibiotic resistance, where single amino acid changes can cause the binding of a toxin to a particular protein to decrease (for example, warfarin resistance in rats, and resistance to various AIDS drugs). Intelligent design proponents happily agree that such tiny changes can be accomplished by random mutation and natural selection.

6) In the “least promising” intermediate (L111Q) the protein has essentially lost its ability to bind any steroid. In the “most promising” intermediate protein (the one that has just the S106P alteration) the protein has lost about 99% of its ability to bind DOC and cortisol, and lost about 99.9% of its ability to bind aldosterone. (Figure 4C)

7) Although the authors imply (and Adami claims directly) that the mutated protein is specific for cortisol, in fact it also binds aldosterone with about half of the affinity. (Compare the red and green curves in the lower right hand graph of Figure 4C.) What’s more, there actually is a much larger difference (about thirty-fold) in binding affinity for aldosterone and cortisol with the beginning, ancestral protein than for the final, mutated protein (about two-fold). So the protein’s ability to discriminate between the two ligands has decreased by ten-fold.

8) One would think that the hundred-fold decrease in the ability to bind a steroid would at least initially be a very detrimental change that would be weeded out by natural selection. The authors do not test for that; they simply assume it wouldn’t be a problem, or that the problem could somehow be easily overcome. Nor do they test their speculation that DOC could somehow act as an intermediate ligand. In other words, in typical Darwinian fashion the authors pass over with their imaginations what in reality would very likely be serious biological difficulties.

9) The fact that such very modest results are ballyhooed owes more, I strongly suspect, to the antipathy that many scientists feel toward ID than to the intrinsic value of the experiment itself.

10) In conclusion, the results (and even the imagined-but-problematic
scenario) are well within what an ID proponent already would think Darwinian processes could do, so they won’t affect our evaluation of the science. But it’s nice to know that Science magazine is thinking about us!