As a graduate student, I was very interested in the way one-dimensional sequences of amino acids related to the three-dimensional structure of proteins (which are made of one-dimensional sequences of amino acids). I was especially interested in what consequences would result from changes in the amino acid sequence (naturally-occurring or otherwise). Back then, one of my dream projects that was technically unfeasible was to take a protein sequence and artificially alter every position to every possible amino acid. I wanted to have a comprehensive view of what amino acid substitutions would do to the structure (and subsequently the function) of proteins.
That was then. Now, with so much high throughput technology, such dream projects can actually be accomplished. So my hat's off to McLaughlin et al., authors of a new study published in Nature that managed to accomplish that very project. For the protein, they picked a small one called PSD95pdz3, which binds to other proteins and is involved in cell signaling. McLaughlin et al. mutated every position one at a time to every other possible amino acid, resulting in 1577 variants of the native protein. They then checked these variants for their ability to bind to a short peptide that represents the normal binding partner of PSD95pdz3.
Their results confirm some of what protein biochemists have long believed about protein sequence and function. Substitutions at most sites (about 75%) had little effect on the protein function, and some sites tolerated substitutions of very dissimilar amino acids with little to no functional consequence. More surprisingly, these weren't always surface amino acids. At least one site that has direct contact with the binding partner was quite tolerant of substitutions. Only 20 amino acid sites were sensitive to substitutions, and that's about a fourth of the amino acid positions in the protein.
McLaughlin et al. repeated their experiment with a different protein binding partner that represents a different binding class of the same PDZ protein family. Most positions in PSD95pdz3 behaved exactly as they did with the native binding partner, but nine positions were different. Four of those positions significantly increased the affinity for the non-native binding partner. One substitution, Gly330Thr, converted PSD95pdz3 to a protein that binds both binding partners with high affinity but without specificity. A double substitution of Gly330Thr and His372Ala converted PSD95pdz3 to a protein that specifically binds the nonnative binding partner with high affinity.
Within the creation/evolution debate, these results have quite important consequences. You've probably heard creationists and ID advocates claim that mutations are bad and ruin protein function. As a person with a background in protein biochemistry, I've always thought that claim was exaggerated at best. It certainly didn't remind me of my own experiences with proteins in graduate school, where substitutions were made all the time, and marginally or fully functional proteins resulted. This comprehensive survey of mutations confirms my own experiences: Amino substitutions at most sites (meaning 75%) don't do much of anything, and some of those positions can be altered to any other amino acid without altering the protein's function.
Their second experiment of examining binding to a nonnative binding partner speaks to the ongoing research project of some ID scholars who want to show that a protein really can't evolve even a new function. In the past, this type of research has been accomplished by substitutions targeted to residues that are known to be important to protein function. The results of these studies have shown that multiple substitutions are necessary to convert one protein function to another, which is consistent with McLaughlin et al.'s discovery that two substitutions are necessary to fully change the specificity of PSD95pdz3 to a different binding partner. Whereas ID scholars have concluded based on their research that such conversions are improbable (impossible?) because they involve coordinated mutations, McLaughlin et al. found a plausible intermediate, Gly330Thr, which binds to both binding partners. Since there's a functional intermediate, it's at least plausible that this protein could evolve a new function.
Now we could argue about how general McLaughlin's findings are, and maybe even whether a nonspecific PSD95pdz3 would be beneficial or detrimental. I expect that's exactly how the ID response will go (I have not at this point read any ID blog posts about this paper), and I wouldn't be surprised if some blogger finds a way to spin this paper to support ID claims. My own experience as a protein biochemist tells me that McLaughlin et al.'s discoveries will be a general result for most proteins and that full surveys of possible mutants will reveal various functional intermediates between proteins of slightly different function. I would hope at the very least that McLaughlin et al.'s research would soften the rhetoric about how impossible evolution is or how mutations are always bad. It probably won't happen, but I can always hope.
As for me, I've always thought that the resilience of proteins to mutations spoke well of their original design. Proteins are not incredibly sensitive to mutation, breaking down the minute one mutation happened. That would be a terrible design. Instead, they're built to last, which is what I would expect from a wise designer.
McLaughlin et al. 2012. The spatial architecture of protein function and adaptation. Nature 10.1038/nature11500
Feedback? Email me at toddcharleswood [at] gmail [dot] com.