Gauger and Mercer: Bifunctional Proteins and Protein Sequence Space


(Ann Gauger) #98


In his 2004 paper? He started with a beta lactamase fold that degraded ampicillin. He asked how many could degrade ampicillin after mutagenesis. It’s a very specific question.

(John Mercer) #99

Dr. Gauger, I haven’t changed topics at all and I don’t want to discuss any of your papers. You seem to be projecting.

For at least the fourth time, let’s discuss the relevance of our results to your claim:
“So unless functional sequences are easy to find (very common), and/or are clustered together (easily reachable from one functional island to another), explaining current protein diversity without design is impossible.”


Human and C. elegans cytochrome c differs by 40%, so I don’t think that is a good estimate.

I would hope that those who claim evolution can’t produce new function would already know this.

(John Mercer) #101

Who did, if not you?

Note: I have edited my posts to remove the question you did not write.

(John Mercer) #102

Yes, and to that question our work is very relevant, as we shifted function by changing only one residue, as radically as possible by size.

We even shifted function without losing the original function.


By that measure, there would be no human proteins that have a functional fold. Surely you can understand why there are major problems making general statements about functional sequence space based on such a limited study.

(John Mercer) #104

And the fact that there are thousands of other studies relevant to those general statements and questions.

(Ann Gauger) #105

The link you provided listed “Evolution News.” I don’t know who.

I am willing to tackle the question you last posted. But it will have to be later. I am already late.

(Ann Gauger) #106

I’m sorry @T_aquaticus. That doesn’t follow, but we will have to pick it up later.

(John Mercer) #107

What have you read that suggests that?

(S. Joshua Swamidass) #108

It seems like frustration is increasing. A word of encouragement and some direction.

  1. This is an important conversation. We may walk away disagreeing in the end, but this exchange between @Agauger, @art, and @Agauger will be enlightening for everyone. Thank you for participating.

  2. Conversations like this are best if they are well scoped and narrow. Several papers and claims have been raised. Rather than dealing with everything at once, it would be best to deal with one thing at a time. This is especially true because @Agauger is fielding questions from all directions. Speaking from experience, it is hard to keep straight all the details when taking critique from all sides. I reccomend honing in on point at a time.

  3. If you are not a scientist with relevant expertise please keep your comments to clarification, so this isn’t more difficult than it needs to be.

  4. Please give @Agauger the benefit of the doubt. In this case she is outnumbered, and responding in detail to many questions. In contrast her co-author is not engaging with scientists. This deserves genuine respect and kindness in return.

(S. Joshua Swamidass) #109

A post was merged into an existing topic: Side Comments on Gauger and Mercer

(John Mercer) #110

Yes. That was a huge misrepresentation of our work.

Here’s what we did:

  1. Made a substitution to a residue not found in nature (Fig. 1).
  2. Showed that it had normal, even maybe slightly greater, ATPase activity (Fig. 3).
  3. Showed that both ATP and ADP analogs compete effectively (not exactly changing preference, as you claimed) (Table II and Figs. 4 & 6)
  4. Most importantly, we showed that the ADP analogs were highly effective in promoting actin binding. In muscle myosins, this is the mechanism underlying rigor mortis, because after death the [ADP]/[ATP] ratio skyrockets. Again, all of the complexity that we could measure was preserved, both with normal and abnormal substrates and products. That was a really big deal. Myosin is not a simple enzyme with a single activity, but you seem to be persistent in trying to portray it as one.

So, yes, given the rigor (pun intended) of our approach, I feel comfortable in describing what you wrote as a gross misrepresentation. Note, however, I am not asserting that your misrepresentation was deliberate. It is ironic either way, given that you are trying to claim that survival is an accurate measure of enzymatic activity in your work without even attempting to correlate the two.

Have you actually read our first paper yet?

(S. Joshua Swamidass) #111

@Mercer, it would be helpful if you would link to the papers you reference this way. It is not apparent to the rest of us which one is the “first.”

I think we can be certain in this case it was not deliberate. @Agauger already acknowledged she misread your paper.

(John Mercer) #112

And I acknowledged her gracious acknowledgement. The problem is that despite that, she is still misrepresenting it.

I do acknowledge significant progress between “essentially gummed up the works” (we couldn’t find anything wrong with it) and "allowed a modified substrate to bind.”

(John Mercer) #113

So as I originally proposed let’s deal with our data and its relevance to Dr. Gauger’s claim:
“So unless functional sequences are easy to find (very common), and/or are clustered together (easily reachable from one functional island to another), explaining current protein diversity without design is impossible.”

This is what I originally proposed in the OP and immediately after:

There’s no reason to Dr. Gauger to conflate her sweeping claims with her or Axe’s individual papers.

(Ann Gauger) #114


So as I originally proposed let’s deal with our data and its relevance to Dr. Gauger’s claim:
“So unless functional sequences are easy to find (very common), and/or are clustered together (easily reachable from one functional island to another), explaining current protein diversity without design is impossible.”

We seem to be using terms differently here. When I say functional island and current protein diversity I mean explaining the existence of amino transferase and lyase and oxidoreductases and decarboxylases etc.

Your paper is a fine paper. It laid the ground work for some important experiments. Myosin is a complex protein. But it does not have the range of activities that need explaining, and as for what you did with your experiments, you yourself say that all the complexity you could measure was preserved. No change in function. A change in the strength of binding to actin by the mutant myosin in the presence of the ADP analog was observed. I don’t know what terminology to use that you will understand, but that’s not a change in function.

When I said “In your example, you made a single change to the binding pocket that allowed a modified substrate to bind.” I was not trying to describe everything you did in your paper. Neither was I downplaying what you did. I was pointing out that you did not change myosin’s function.

In an earlier post I showed figures to illustrate the range of functions that need to be explained.
That is what I am referring to in the quote you selected, and that I have reproduced at the head of this post.

(S. Joshua Swamidass) #115

How are we defining function in this context? That is a squish word without some thing more rigorous behind it, even if it is merely for the context of this “conversation” on Axe’s work.

(Ann Gauger) #116

This is in the context of Axe and my work. By function of an enzyme I mean the particular catalytic activity of that enzyme. Here are comparisons between enzymes of the PLP-dependent transferase superfamily, over a broad range of structural distance, defined in the paper.

(Arthur Hunt) #117

Returning to this discussion, to see where some of the issues raised above lead.

When last I was here, we had Ann ducking and avoiding the problems with Axe’s “assay” for enzyme function. (I had assumed that this was also Ann’s work, since she is a coauthor on one of the papers that use the assay; apparently, she contributed other parts of this work, and I apologize for not picking up on this.) The problem, in a nutshell, is that Axe very likely underestimates the numbers of functional variants in his 2004 paper, that is widely (almost religiously, as if it were holy writ) cited by the ID community as applying to all proteins, and all function.

We didn’t get a chance to explore just what the consequences of this underestimation might be. To look into this, I plugged some differing values into Axe’s equation (first column of p. 1308) to see what the consequences would be. Specifically, I asked what would happen if Axe was missing 10%, 50% and 90% of all functional sequences. The results are that, for these underestimates, the fraction of functional 153-mers in sequence space changes from 5x10^-65 to 4x10^-64, 4x10^-62, and 2x10^-49, respectively. As I stated above, this doesn’t do anything to narrow the range of estimates obtained by different methods. However, it does give us a sense of the uncertainty in Axe’s assay.

Another idea that got left behind by Ann’s curious reaction to a pretty obvious fact has to do with the scaling of Axe’s estimates to short peptides. Specifically, if we use Axe’s numbers to estimate the fraction of functional 20-mers in sequence space, we get a value of 4x10^-9. (For the scenarios discussed in the preceding paragraph, the numbers are 5x10^-9, 10^-8, and 4x10^-7.) What is interesting is that these numbers are actually not too different from values obtained by directly counting active isoforms in random combinatorial libraries. The question this raises is – is there a biochemical or structural basis for this similarity, or is this just coincidence? The answer may tell us something important (and, I will grant, unexpected) about Axe’s approach.

As always, submitted for further discussion …