Gauger and Mercer: Bifunctional Proteins and Protein Sequence Space


(Ann Gauger) #50

Did you read the article? It’s not a snippet. And who said he gave up? Not me.

What, do you think declaring something almost close to a miracle means you can’t study it? I suppose you also think ID is a science stopper. It’s not. ID gives us more motivation to study the natural world, not less.

(Arthur Hunt) #51

So, he can distinguish “basal” from “delta”. But what about low activity variants of “basal” that come about in his mutational study? In the 2004 paper, how many active variants does he toss out because the assay cannot tell the difference?

I know you have no way to answer this. I asked about an independent calibration of the plating assay so that we might have some idea of the magnitude of this problem, that becomes quite apparent when we look at the 2015 Biocomplexity paper and the considerations raised here. If there is no such calibration, so be it.

One thing we may agree on - when it comes to problems with Axe’s work, the matter of the possible unreliability of the plating assay is a rather minor issue. It probably changes the numbers, but this only adds to the more general uncertainty.

Way back in 2007, I summarized my estimation of this subject thusly:

" 10^-10 -10^-63 (or thereabout): this is the range of estimates of the density of functional sequences in sequence space that can be found in the scientific literature. The caveats given in Section 2 notwithstanding, Axe’s work does not extend or narrow the range.".

I don’t think recovering more low-activity variants in studies such as Axe described is going to extend or narrow this range.

(Arthur Hunt) #52

So, now enzyme activity is not really enzyme activity.

Seriously, Ann. It does no one any good to keep digging this hole. Peptides can be catalysts. Enzymes are catalysts. There really is no question about this. I really cannot see the logic in denying this simple and obvious fact.

(Arthur Hunt) #53

To the rest of the participants of this discussion - sorry for jumping in and monopolizing things, without waiting for more participation. I will sign out for now. Please, please feel free to add comments, questions, and suggestions to this back-and-forth.

(Ann Gauger) #54


I disagree with your assessment, Art. It’s not my work, and I don’t need to defend it, but your arguments are missing the mark. You are applying standards that don’t fit. He was not interesting in measuring absolute enzyme activity. He was comparing the activity of a weakened enzyme before and after a controlled mutagenesis (designed to preserve the fold). The “basal” enzyme had activity at the threshold of beta lactamase activity. He knew its MIC. Then after creating a mutant library, he determined what proportion of mutants retained that level of activity. That’s it. What proportion could still function at the “basal” level. How many mutants inactivated the enzyme below the threshold level he set.

What’s the point of an independent calibration of the plating assay? What do you hope to establish? It sounds fine and fancy, but what’s the point? Do you think we were measuring something other than beta lactamase activity? Do you want a dose response curve?

I assume it is because you want to be sure that every last possible functional fold was detected. Oh. No. You say it wouldn’t matter.

One last thing, and then I am done. Oops, I see you have left me another comment. Well.

This conflates a number of different experimental designs, which do not all ask the same question. As I pointed out before, binding to a column is not an enzymatic reaction. The numbers should not be taken as a range of the same property. And 10^-63 is not to be ignored.

(Ann Gauger) #55


Good night.

(S. Joshua Swamidass) #56

This is an important point. This is really Doug Axe’s work, and he is the one who should be responding to critique. I’m thankful @Agauger has responded some, but this is really Axe’s problem to deal with here.

(John Mercer) #57

We’re not off to a good start. You seem to be misreading again. There was a lot more than one experiment. That would be pretty amazing if we got 2 Cell papers (one cover), 1 Neuron paper (cover), 1 PNAS paper (cover), and several others from a single experiment!

Do you think that casually denigrating the scope of someone’s work like that suggests that you are interested in a serious discussion?

I’m well aware of that, and I would hope that you would know that I am having truly read the paper you apologized for misreading.

You don’t seem to be addressing my point, which is that the mutant could hydrolyze a different substrate while not losing wild-type activity. Acquiring new functions is not what you tested for at all, correct? If so, why are you trying to compare and contrast?

That remark makes no sense.

In what way do you think I’m claiming that they are comparable, exactly? Were you or Axe looking for additional function without losing wild-type function? Are you even aware that there’s more than one myosin, and that we did this to more than one myosin, very easily?

That is completely false, Dr. Gauger. Doug Axe was not measuring the beta-lactamase activity. To measure activity, one measures activity, not survival. Why would you write something so blatantly false to someone who does biochemistry?

I’m familiar with it. It’s horrible.

Actually, you misquoted the first sentence by truncating it in a very interesting way.

The actual sentence is:
Two experimental studies provide reliable data for estimating the proportion of protein sequences that perform specified functions.

That single sentence, or really just the first six words of it, is all one needs to read to see that this paper is absurd, because there are literally thousands of relevant (and far more thorough) experimental studies, but neither you nor Axe cite them. Instead, Axe only cites two studies. That ain’t science.

There you go with the gratuitous denigration again. We didn’t merely measure binding, we actually measured enzymatic activity, something you don’t seem to want to see as relevant.

You also left out the very important fact that we didn’t destroy wild-type activity at all, which was really the major basis for the high impact of all those many experiments and papers that followed. Do you realize that we couldn’t find anything wrong with a homozygous knockin mouse with this substitution?

So let’s try to focus: easily acquiring a new activity without losing the old activity has interesting implications for the feasibility of enzyme evolution, no?

You weren’t measuring activity at all. If you want to make grandiose claims about activity, you need to do the biochemistry and measure it. Did you even sample some of your mutants to attempt to show a correlation?

(John Mercer) #58

That would just be relative activity instead of absolute, but he wasn’t interested in measuring relative enzyme activity either. He and you are still making loads of sweeping claims about relative enzyme activity without actually measuring the thing you’re talking about.

Often different experimental designs give us better insight. That aside, are you endorsing Doug Axe’s claim that there are really only two studies that ask the same question that Axe is asking?

(John Mercer) #59

But Doug doesn’t bother to measure the function he speaks of!

And this is really basic biochemistry, but the primary basis of catalysis is simple binding.

But now I realize that that makes your false claim that we only did binding even more insulting:

(John Mercer) #60

I agree. This is utterly fundamental. If one wants to make claims about enzymatic activity, one should measure enzymatic activity.


It’s a proxy for enzyme activity. In this case it’s probably not a bad one but they tend to be nonlinear.

(Ann Gauger) #62

I am actually not sure where to start in dealing with your response. First you accuse me of bad faith in not having read and commented on your whole body of work. But I was given one paper and it was that paper I was commenting on. I was not making an extensive analysis of that paper either. Yes I know that you measured activity of the modified enzyme and that you demonstrated that it was fully functional. The point I was making was that you didn’t change the function. I was know in no way denigrating your work. Why would you think I was?

In our work we took one enzyme with a particular function that was related structurally to another enzyme with a different function and asked if we could convert the first enzyme to function like the second. That’s the change in function that we were seeking. A change in chemistries.

You don’t need to rehearse all the criticisms of that paper. I am well aware of them. But if you accuse me of not engaging what you wrote in later papers, then at least engage what we actually wrote. I can’t recognize our work in your comments. They seem irrelevant. And we seem to be talking past one another.

You are free to disagree with our work. The hostility is not derived from science, however. A scientific disagreement should not be hostile in its tone. This is something I ask of all those who want to discuss with me. Leave the hostility behind. Take me as operating in good faith. You can critique my science and point out where you think my methodology was wrong or what I should have included. But please critique what we actually did and pay attention to why we did it as described in the paper. Some of you have done this. And we can have a legitimate discussion of why something in the methods should’ve been included or why not. But to misread me or to misunderstand our purpose is unfortunate and counterproductive.

We were not trying to study proteins’ ability to interact with modified substrates if those proteins indeed still carried out the same chemical activity. We were not interested in studying enzymes that had activity toward multiple substrates and asking if they could be shifted to favor one of those substrates. We were asking how hard it is to shift a protein to a new function, a new chemical reaction it had not performed before. We were not trying to recreate history, to go back to the ancestral sequence and trace a pathway forward. We were asking how hard it would be to shift a protein to a new function. At least for our enzymes, the PLP-dependent transferases, you can’t do it with one or two mutations. This is not a surprise to just about every scientist I know. The point is that when you take into account the adaptive changes necessary for that transformation you end up requiring multiple genetic changes, more than just one base change is involved. You need increased expression because of weak activities , a gene duplicate to allow modification of one function while preserving the original. Then you need the base changes required for the new function. If the organism needs that new function, that specific function to solve a problem, and there is only one way to do it, then it must wait or around for all of those things to happen: a gene duplication, the mutations, the over expression. Then you would have a selectable new function. The argument isn’t that complex.

I understand that some of you despise me. If it were just a scientific problem, there would be no reason for such despising. The nearest I can come to understanding is that you think I have betrayed science. In reality I am trying to ask questions using science. You may disagree with my message, my results, or my conclusions. That is part of the scientific debate. Please read and understand what we were doing and critique that. Not some strawman or fantasy creation.

A few last point. First, I understand the view that the only way to get the two proteins we were working with was by retracing the ancestral path. The implication is that there is a very narrow set of solutions to the problem of getting these two functions from one original enzyme. And that should give pause. If evolution needs to be fine-tuned to that extent, where there are only a few paths to get to a final product, then how did we arrive at the thousands of different enzymes we have from some small set of original enzymes? Second, there are the people who say evolution does whatever it can, it chooses the next beneficial step whatever it is without respect to the needs of the organism. Is that the way to produce a functional metabolism? Third, @Mercer in no way did I intend to denigrate your work. You are reading into what I wrote something that isn’t there. Because I only discussed one paper and one part of that paper does not mean I think the rest of what you have done (which I didn’t know about because I didn’t go looking—I thought we were only dealing with the one), doesn’t mean I disrespect anything you have done. Far from it.

Side Comments on Gauger and Mercer
(Blogging Graduate Student) #63

I don’t see this as a problem. It’s like saying that since innumerable very specific life events had to happen to shape the personalities of everyone on earth to be precisely they way they are today, how could that have happened? It’s a version of anthropic reasoning. The answer is that if those specific events didn’t happen, then other events would have, giving rise to a different set of circumstances today.

In terms of “replaying the tape of life”, I think at the molecular level, Gould’s position is more correct, while at the whole-organism (e.g. morphological) level, Conway-Morris’ position is pretty well substantiated.

(Ann Gauger) #64


Which paper are we discussing now?

MIC is a genuine reflection of relative enzyme activity, when the assays being compared are
performed under the same in vivo conditions. Insisting on a biochemical assay when we were making no claims about the in vitro activity is pointless. Sweeping claims for the paper I was involved in is a bit of an exaggeration.

No. But what you quote does not say “there are only two studies.”

As long as you are careful not to mix apples with oranges.

(Ann Gauger) #65


Which paper are you discussing? 2004, 2011, 2015? Axe, Gauger and Axe, Axe and Gauger? Are you a biochemist? A protein biochemist? Then you should know that activity varies depending on the environment of the enzyme. Different solutions change activity, and in particular in vitro vs. in vivo changes activity. To determine functional activity of an enzyme you need to measure it in vivo. For the 2004 paper, Axe was interested in what variants could still perform their in vivo function after mutagenesis. Since MIC is a valid proxy for relative activity when comparing two libraries of enzymes, getting an in vitro number is not necessary. The actual number need is the number of variants still functional, not how much activity they retained. He knew they attained at least as much activity as was needed to survive the basal level of ampicillin.

(S. Joshua Swamidass) #66

8 posts were split to a new topic: Side Comments on Gauger and Mercer

(Ann Gauger) #67

Which paper is this discussing? I assume it is Gauger and Axe 2011.

Making much of a copying error. I was probably answering on my phone where I have difficulty getting the touch screen to work. I have answered else where the critique about references. I am always in favor of more citations.

Denigration? The point I was trying to make is that in this paper you weren’t changing the actual chemistry of the reaction, which is different than our goal. Since this seems to be a stumbling block I will try to reproduce a figure from our paper to show what we mean by changing activity. The first shows a nearest neighbor joining tree for thr PLP transferase superfamily. I can’t get the figure to load from the paper, so I copy it here.

and the second, an analysis of the chemical reactions of nearest neighbors, structural distance, and chemical reactions.

For us, changing function means going from an aminotransferse, to an amino transferase with decarboxylase activity. We chose the nearest neighbor pair with the smallest structural distance for our examination, Kbl and bioF, as known by the gene names, or 2-amino-3-ketobutyrate CoA ligase and 8-amino-7-oxononanoate synthase.

“So let’s try to focus” is gratuitous and unnecessary.

It has interesting implications for that enzyme’s function, and for the theory that you have to give up the old in order to favor the new. Does your newly acquired activity meet the standards above? Because that is what needs to be explained. Meeting the requirement for completely new chemistry, without shared substrate or shared reaction type. We have to be able to explain aminotransferases, ligases, decarboxylases, methyl transferases, hydrolases, lyases, oxidoreductases and isomerases from small pool of starting proteins.

(John Mercer) #72

Let’s deal with this bit first.

I gave you one paper, but given that you claim to read the literature and you make grandiose claims about the prevalence of function in sequence space, you should be familiar with, or at least interested in, our work (or at a bare minimum, some of Kevan Shokat’s 300+). You clearly are neither. In other words, you don’t appear to read the relevant literature.

And even if you want to take the unprofessional stance of only looking at the paper I mentioned, it has a lot more than one experiment. There is no other conclusion than that you were trying to portray our work as insignificant.

But you didn’t portray it that way. You portrayed it as mere binding. We’re supposed to be accurate for the benefit of other readers.

That’s not true. The point I was making was that we changed Myo1c to acquire a new function without losing the old function.

Dr. Gauger, you were the one who brought up your work! I did not!

I brought up our work and noted that you completely mischaracterized it when I first pointed it out to you, but more importantly noted that it does not support your global claims:

Homologous proteins from different organisms often show substantial variation yet are still functional. In proteins, contact sites that interact with partners are typically less forgiving than other sites.

“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.”

Can you deny that our work is very relevant to what you wrote?

So I propose that we discuss only our work, only in the context of your sweeping claims, without you bringing up your work at all, OK?

(Ann Gauger) #74

You are not alone in holding that opinion. The difficulty for me is the interconnected flexibility yet stability that is required. Reactions require certain substrates and produce certain products. If the cell can’t find what it needs it has to make it. That making frequently requires a pathway of multiple enzymes.

The cell functions with roughly 20 amino acids. The first cell eventually had to leaarn to make each amino acid. Threonine needs 3 enzymes, tryptophan 5, histidine 8 I think. Could there be shorter ways to make histidine? Perhaps not. The problem still needs to be solved.