Dennis Venema is a Fellow of Biology with BioLogos, where he writes a series of posts called “Letters to the Duchess” — an allusion to Galileo’s “Letter to the Grand Duchess Christina.” In the past, I have looked at Professor Venema’s articles on evidences for common descent (see here, here, here, here, here, and here).

In a recent online discussion with Venema, he made the erroneous claim that the mammalian immune system, with its search for, and production of, antibodies, is a good example of why evolving protein-protein binding sequences is not a problem. In fact the mammalian immune system is yet another enormous problem for the theory of evolution.

The mammalian immune system is not a good example because it is designed for this job of creating protein-protein binding sequences. It searches a well-defined design space extremely rapidly, and measures the success of its search experiments accurately and quickly. The fact that our immune system successfully designs antibodies in short order does nothing to address the problem of how random mutations occurring throughout the genome are supposed to have found myriad binding sequences, crucial for life. Venema also referred to another example that he has written about. Unfortunately this example also fails to demonstrate Venema’s claim of “evolution producing a new protein-protein binding event.”

The problem with evolving protein-protein binding is that too much gene sequence complexity is required to achieve the needed binding affinity. You could say it is an “all-or-none” type of problem.

One or two mutations will not generally do the job — you usually need more mutations before the two proteins stick together very well. And stick together they must, on a massive scale, in order to perform their necessary tasks. Even the simplest, unicellular, organisms contain massive protein machines, consisting of dozens of different proteins binding together to perform crucial life functions.

The study Venema referred to did a beautiful job in confirming this “all-or-none” character of protein-protein binding sequences. The study showed that in order for a viral protein to perform a relatively simple switch from one protein to a very similar protein required four types of mutations. Anything less and no dice.

The twist in this study was that subsets of the four mutation types were apparently useful for a different function (strengthening the binding affinity to the original protein). So while in general the evolution of protein-protein binding sequences is astronomically difficult because too many simultaneous mutations are required, in this case the four mutation types could be accumulated, with useful benefit realized at some of the intermediate steps.

This is not a general result. It is not a revolutionary new finding that reverses our understanding of protein-protein binding sequences.

It confirms our knowledge, and adds a fascinating outlier case where the “all-or-none” character is circumvented by intermediate functions that fortuitously “push” the design in the right direction. As the study explains:

The “all-or-none” epistasis among the four canonical phage mutations implies that it would have been unlikely for the new function to evolve on the scale of our experiments, except for the lucky fact that some of the mutations were beneficial to the phage in performing their current function, thereby pushing evolution toward the new function.

The study provides no indication that the untold thousands upon thousands of protein-protein binding problems in molecular biology would enjoy this type of setup. And if they did, oh what a suspicious sign of design that would be.

Venema is mistaken in his attempt to recruit this study as a solution to the evolution of protein-protein binding sequences. Strangely, not only had Michael Behe provided his explanation of the study, but Venema was aware of it at the time of his writing. Venema explained that in his next article he would address Behe’s explanation, but in fact Venema simply rehashed Behe’s original explanation for why protein-protein binding is a problem for evolution. Venema did not address Behe’s explanation but simply concluded that Behe’s original explanation must be false because, after all, this new study demonstrates the evolution of just such protein-protein binding sequences.

This is an unfortunate misuse of a study that most readers will not understand. Venema has force-fitted the study into an evolutionary proof.

Additional Problems

In addition to the basic problem of serendipity, this confirmation of the “all-or-none” character of protein-protein binding sequences was possible only with a very contrived, designed, laboratory experiment. Simply put, a virus population was provided with a willing and well-fed host to live off. In the meantime, many more host targets awaited the virus population. So a few mutations helped the viruses infect the initial hosts, and mere single additional mutation then allowed the viruses to infect the second group of hosts.

It was, again, an entirely artificial, laboratory environment, that wasn’t even intended to replicate a realistic evolutionary environment. Venema nowhere explained this.

Second, the study also discovered even more serendipity. Not only were there “luckily” intermediate fitness benefits, but the finding of the four mutations types also required certain mutations in the host genome. Without them, no dice.

Finally, it is worth noting that across the many different virus populations used in the experiment, the virus protein in question did not incur any synonymous mutations. The study attempted to explain this as a sign of selection:

First, all 248 independent mutations in the 51 sequenced J alleles were nonsynonymous, whereas the expected ratio of nonsynonymous to synonymous changes is 3.19:1 under the null model for the ancestral J sequence. This great excess is evident even if we include only the 82 nonsynonymous mutations in the 24 isolates that did not evolve the new receptor function.

This is suspicious. According to evolutionary theory, a lack of selection will be manifest in relatively few nonsynonymous mutations. So the ratio of normalized nonsynonymous mutations-to-normalized synonymous mutations (the so-called Ka/Ks ratio) will be less than unity. On the other hand, strong selection will be manifest in relatively many nonsynonymous mutations. So the Ka/Ks ratio will be greater than unity.

A high Ka/Ks ratio, and hence an inference of strong selection, should be due to relatively many nonsynonymous mutations. In this study, however, it is in the synonymous mutations where the surprise comes. There were zero. In other words, the Ka/Ks ratio is infinity. To pass this off simply as a sign of strong selection is not good science, even within the normal science of evolutionary theory.

Image: Grand Duchess Christina suppresses a smile at the notion of an implied parallel between Galileo and Dennis Venema, by Tiberio Titi via Wikipedia.

Cross-posted at Darwin’s God.