This is my second reply to comments that University of Texas at Austin biologist Martin Poenie posted at the web site of the Christian Scientific Society. Writing under the name gandaulf, Poenie critiqued arguments that Stephen Meyer made about proteins in his new book, Darwin’s Doubt. See my first reply for more background.
To add a bit more, I think Meyer could be so much more compelling to examine some wider data than just what Doug Axe says. I do not know Doug and have nothing against him. Furthermore, I could not care less if the Darwinian paradigm falls apart on the Cambrian explosion. But there is what is known as the Ig superfamily of proteins which contain, as the name suggests, the Ig fold. Members of the Ig superfamily are involved in homotypic adhesion (the foundation for making tissues), receptors, signaling proteins (tyrosine kinases) and of course antibodies and T cell receptors. Now here are two points that are interesting. First, one fold is used for many different and relevant functions, and one in particular that lies at the heart of the Cambrian explosion — mulicellularity — which involves homotypic adhesion. Secondly, what is the sequence variability of the Ig fold in members of the superfamily. If it is as constrained as Meyer portends, then we should see it in the sequence data.
This continues the thought of his first comment, which is that Meyer has no basis to think that the striking variety of animal forms that appeared in the Cambrian explosion would have required new protein folds. The gist of my response was that the distribution of unique genes and proteins (orphans, as they are often called) among extant animal kinds shows that each of the different kinds, right down to the level of species, carries many genes and proteins that are unique — found nowhere else. So even if it is conceivable that animal life in all its diversity could have been formed without lots of new protein folds, that idea is purely hypothetical in that life as we see it doesn’t seem to have been produced that way. Economizing on protein folds doesn’t seem to have been a priority.
In defense of his assertion that new protein folds are not needed for new animal forms, Poenie points out in the above comment that one fold can perform many functions, citing the immunoglobulin fold as a key example. The reasoning here is that once life has a basic set of protein folds, it should be easy for evolution to produce any number of new protein functions by reusing those folds.
Aside from the fact that life doesn’t seem to have limited itself to a common set of basic folds (which was my first point) there is another problem with Poenie’s reasoning here, one that is prevalent in evolutionary thinking. The root of the problem lies in a curious difference between the way biologists think of their science and the way chemists or physicists think of theirs. Biologists, unlike the others, tend to think they are doing science when they name things.
I remember Glenn Seaborg from my undergraduate days at Berkeley, a man who had the honor of naming several elements in the periodic table, and who had the even higher honor of an element being named after him (seaborgium, atomic number 106). But in each of these instances, the science was done well before the naming, and no one thinks the naming of an element establishes anything about its properties.
In biology on the other hand, names are loaded with interpretations to the point where the boundaries between nomenclature and scientific facts are badly blurred. Yuri Lazebnik expressed this very well in a hilarious essay on “the common fundamental flaw of how biologists approach problems,” which he titled "Can a Biologist Fix a Radio?"
Like most biologists, Poenie takes a grouping convention, namely the grouping of proteins into sets called families or superfamilies or folds, to be significant in itself. My question is, how significant can a convention of that kind really be? If you knew nothing about two protein domains other than that they are said by convention to have the same fold, what would you be able to infer from that?
For example, these two protein domains, colored from red to blue along their chains, are classified as having the immunoglobulin fold. On the left is the N-terminal domain from sweet potato purple acid phosphatase (PDB accession 1xzw), as classified by the Structural Classification of Proteins. On the right is a domain from the extracellular region of human tissue factor (PDB accession 2c4f), from the same classification source.
Both of them are portions of larger structures that perform very different functions. These portions have both similarities and differences. In fact, the differences are large enough that, apart from some of the arrow shapes (beta strands) on the left sides of the pictures, there is no clear correspondence of parts even at the level of gross structure, much less the finer level of the genetically encoded amino acids from which the structures are made (not shown).
To be clear, I’m not suggesting that there is no basis for classifying these domains as belonging to a fold group, or that protein classification schemes are unimportant. Indeed, classification is a key step toward bringing conceptual order to what would otherwise be a bewildering assortment of individual protein structures and functions. What I am saying is that the real scientific question of where the many distinct protein structures and functions came from is not reduced or collapsed simply by placing them into groups. If it were, then we could all but eliminate the problem by grouping them under the single heading: proteins.
As it is, the fact that the above domains are conventionally grouped together merely provides a convenient way of grouping the key questions, all of which remain unanswered. Can either of these replace the other without loss of function? Do they have parallel histories, or common histories? Did they evolve in Darwinian fashion from a common starting point? None of these questions is eliminated by the convention of referring to these domains as examples of the immunoglobulin fold. Indeed, the fact that their similarities are lost within the whole protein structures that contain them (below) makes it hard to guess whether those similarities have anything to do with the answers to these hard questions. Meyer is right, then, to refer to protein folds in explaining the problems they pose for Darwinian evolution, and Poenie is wrong to think that the existence of these groupings somehow eliminates the problems that Meyer raises.
In the end, the only way to find out whether structural similarity has any connection to material equivalence or evolutionary relatedness is to perform experiments. My colleague Ann Gauger and I have been doing this for some time now to address the specific question of whether similarity implies that evolutionary transitions are feasible. As skeptics, we decided to look at a pair of enzymes (proteins that do chemistry) with much more striking structural similarity than biologists require in order to infer evolutionary relatedness.
We chose to study this pair of enzymes, called Kbl and BioF:
The structural similarity here is so clear that part-for-part correspondence is completely unambiguous, right down to the level of the individual amino acids from which the chains are built (not shown here). Despite the fact that they are structural twins in this sense, the two enzymes catalyze different chemical reactions with no functional overlap.
We asked whether Darwinian evolution is capable of transforming one of these enzymes to perform the function of the other. Think of this as a much more modest version of the kind of functional transition that Poenie thinks we can safely infer from much less striking similarities among proteins classified as having the immunoglobulin fold. If the logical leap from vague similarity to evolutionary relatedness is really justified, as biologists commonly assume, then transitions between enzymes with striking similarity ought to be a snap. Conversely, if transitions between enzymes with striking similarity are found not to be a snap, then biologists ought to start questioning those logical leaps.
As Ann and I reported, the transition from Kbl function to BioF function appears to be an evolutionary impossibility. Furthermore, we haven’t seen a convincing case that any evolutionary transition from one enzyme function to a genuinely different one is feasible. Even if compelling examples are eventually found, the general difficulty of functional transitions is now well established, and that in itself makes the uncritical inference of evolutionary relatedness from similarity alone bad science. Of course, when you consider the central role this uncritical inference plays in evolutionary reasoning, you’ll understand why evolutionary biologists are loath to rethink it.
Like most biologists, Martin Poenie thinks that fold similarity proves evolutionary relatedness, and because of this he thinks that sequence variability among proteins classified as having the immunoglobulin fold should be a good indicator of constraints. I’ve spent many years examining this kind of reasoning, and I’ve found it to be unsound. If Poenie is willing to examine the evidence, he might find himself agreeing with me.