Gauger and Mercer: Bifunctional Proteins and Protein Sequence Space

In my previous experience discussing science with you, you didn’t read my paper (you literally claimed that we found the opposite of what we found), which is very relevant to the protein functional landscape you published on.

But perhaps you were having a bad day. Would you like to revisit the paper, perhaps in a separate thread?

1 Like

Sure. But I don’t remember which paper you are talking about (or the conversation.) Another thread would be fine.

1 Like

@Agauger, that was a very gracious response.

@JAMercer welcome to Peaceful Science. That was a bit of a non sequitur. Which paper was this? Can you help bring the rest of this up to speed? It sounds like you were an author on it? What type of scientist are you? Looking forward to see where this goes.

Thank you for setting this up.

How hard is it to find functional proteins?

Contrary to the claims of Axe and Gauger that enzymatic activity is tough to find in the protein sequence landscape–

Axe DD, Estimating the prevalence of protein sequences adopting functional enzyme folds.
J Mol Biol. 2004 Aug 27;341(5):1295-315.
DOI: 10.1016/j.jmb.2004.06.058 PMID: 15321723

Axe DD, Gauger AK (2015) Model and laboratory demonstrations that evolutionary optimization works well only if preceded by invention—Selection itself is not inventive.
BIO-Complexity 2015 (2):1–13. doi:10.5048/BIO-C.2015.2

(and others)

I’d like to present a summary of our work for discussion.

This has the same general thrust as Art Hunt’s critique–

but is a series of empirical demonstrations to complement Art’s more theoretical criticisms.

It’s also relevant to this other discussion:

as we made bifunctional proteins very easily.

Our goal was not to learn anything about evolution, but to address a problem in biology caused by evolution: there are huge multigene families that according to sequence relationships and everything else, are the result of gene duplications and diversifications of function. However, contrary to the machine-like picture painted by the ID folks, these functional diversifications are rarely complete, leaving us with schmeers of partially overlapping functions, something we simply don’t see in designed machines. It just screams that it could only have been designed by the iterative process of evolution, much louder than the ID refrain that life is like some sort of human-designed machine.

The practical issue for biologists and biochemists is that partially-overlapping functions make it hard to figure out what a single one of these family members really does. It’s the likely reason why many gene knockouts don’t do much. It also causes problems for biochemical inhibition/activation of specific proteins in cell biology and pharmacology (think drug side effects).

In late 1990s pharmacology, the brilliant Kevan Shokat developed a chemical-genetic method for inhibiting individual tyrosine kinases, which has enormous implications for cancer treatment, since so many oncogenes are tyrosine kinases. With the help of Kevan and his postdoc Kavita Shah, Peter Gillespie and I set out to use this method to study a suspect for the adaptation motor myosin-1c in inner-ear hair cells, a interesting biophysical and physiological system that provides an enormous dynamic range for our hearing.

The above three paragraphs are provided just to establish the relevance of this work outside of any evolutionary context, and to show how evolution impacts our ability to assess functions of proteins in complex critters like mice and humans. I will get to the conflict, in terms of what it says about the sequence landscape, in the next post. You can ignore all of the rationale above if you are only interested in the conflict. :wink:


Here’s the first biochemical paper. I didn’t do much, as I didn’t learn biochemistry until 2009:

I offered this example to Dr. Gauger after she challenged me to “Ask protein engineers how easy it is to redesign a protein to a genuinely new function.”

I replied, "My colleagues and I have done so, so I asked myself and I said, “It was surprisingly easy to add a new function based on our knowledge of evolutionary relationships between entire protein families.”

The idea here was spatial and fairly simple: we identified a bulky tyrosine residue found in most but all myosins–not only right in the most sensitive part, the active site, but associated with the substrate ATP–and changed it to the least bulky residue, glycine (Y61G or Tyr61Gly). Without changing the charge, this was the most radical size change possible. We basically turned a bump in the active site into a hole. More than one myosin turned out to be far more resilient than we ever would have predicted.

The engineering goal was to make a mutant that had two properties: 1) normal wild-type function, and 2) the ability to recognize ATP and ADP modified with a bulky N6 addition (fitting where the tyrosine side chain was missing in the mutant) that would have no effect on the wild-type myosin because it wouldn’t fit in the active site. That way we could turn an individual myosin off or on without affecting the others, using 3 controls to confirm specificity.

This was amazingly easy because it worked with the very first substitution we tried, totally contradictory to what Drs. Gauger and Axe claim as a rule. Note that Fig. 3 shows that the enzymatic function of the Y61G mutant was nearly indistinguishable from wild-type. Fig. 1 shows that 5 of the panel of analogs worked as we hoped they would.

Here’s the amazing thing: when I offered this paper to Dr. Gauger, her reply literally denied the data:

“In this paper you introduced a single mutation into the binding pocket of the protein myosin so that it bound an ADP analog ( slightly different substrate) and essentially gummed up the works.”

Dr. Gauger, maybe something about our writing confused you, maybe you were having a bad day, but I just can’t see how anyone who studies the effects of mutations on enzyme activity could look at Fig. 3 and conclude that we “essentially gummed up” anything. We allowed it to recognize a new, artificial product, without affecting its normal function in these assays. This is the evidence, period. This is not an interpretation.

There’s also the fact that ADP is a product. ATP is the substrate. This is very relevant and basic biochemical distinction that may be causing some confusion. It’s not in the paper, but ATP analogs (we checked a smaller set) are hydrolyzed with the same preferences that we saw for inhibition by analogs of the product, ADP.

Even if you had only read the abstract, which said, “Introduction of this mutation, Y61G in rat myosin-Iβ, did not alter the enzyme’s affinity for ATP or actin and actually increased its ATPase activity and actin-translocation rate,” it is impossible for me to see how you could possibly interpret that as “gumming up” anything. It appears that you only saw what you wanted to see, not the evidence that was there.

So, here I propose that we discuss the implications of this and our other work in the context of your claims of the rarity of function in sequence space. Our subsequent papers showed how well this approach worked for understanding the biology of this and another myosin, beautifully consistent with these biochemical results.

1 Like


Note added after original post: JAM’s post above this did not exist in its present form when I wrote this. Perhaps our postings overlapped. I thank @JAMercer for providing context. But it does make my response below sound stupid.

So far I have no disagreement with your description of proteins as being members of large families and superfamilies whose activities quite frequently overlap. And I agree with your description of the difficulty involved in mutagenesis screens for knockout mutations because of protein promiscuity (for those of you watching this it means a protein has several functions); however, the discussion also seems to be intended as a snare. Perhaps at someone’s instigation? Perhaps to get me to answer Art Hunt’s challenge?

In any case the forum is not intended to be used to spring a trap. For someone to sign on a new member and within half an hour seek me out and lay his trap is unfortunate.

So let’s not proceed as in a trial. You have made your opening statement. Now tell me the paper so I know what I am dealing with. (Thanks for doing so above). What is this paper and when did we discuss it?


See the post above yours. It was 5.5 years ago.

It’s in enormous contradiction to your global claims about the accessibility of function, so I think that this would be a good place to seriously discuss the science, not as some sort of game or trap. IIRC that is the purpose of this forum.

Why wouldn’t you want to answer Art’s challenge? Is there something wrong with it? Do you view it as some sort of game, or as a scientific disagreement?

Just to clarify, I would describe the general problem less as a protein having several functions, but more as related but different proteins covering the same function. I’m not denying the former occurs, of course.


@JAMercer and @Agauger thank you for engaging here. @moderators will watch this thread, and most people should just stay out of it. This thread is going to be primarily for your exchanges. If any disruptions arise, and we will handle them.

I agree with this.

I understand there is some history. My suggestion is that we start with a clean slate here. Perhaps when the data is explained, @Agauger might even revise her position. Or, perhaps that won’t be necessary as her position is explained better.

It sounds like @JAMercer wants to explain the study, and @Agauger is ready to listen. Let’s start with that?

@Art’s challenge is to Axe, not to @Agauger. As I understand it, Ann played a support role to Axes work (though I stand to be corrected). Regardless, Axe is the one who has not appeared here. @Agauger has been engaging here with us for a long time. I have seen her adjust her position, as she should, in the past, maybe this conversation can move things forward.

Sounds like he means the one referenced here: A Conversation about functions in protein sequence space.

The quote he offers:

“In this paper you introduced a single mutation into the binding pocket of the protein myosin so that it bound an ADP analog ( slightly different substrate) and essentially gummed up the works.”

…does not appear on Google, so I imagine it is was a private email exchange.



Adding a few comments before sitting back and enjoying the exchange:

  1. I believe the sort of altered function John and his coworkers described may be relevant to the discussion about information, the Cambrian Explosion, and the like. Not in a specific sense, but it seems to be relevant to the issue of the evolution of proteins that might be expected to be important for morphological change.

  2. Interestingly enough, the research described by John and his coworkers contradicts an assertion Meyer made in our exchange at Biola back in 2010 - namely, that it is not possible to mutationally alter cytoskeleton components without destroying function.

  3. I guess I am not going to find out in this forum why Ann believes I do not understand Axe’s work. I will admit to being more than a bit disappointed. Oh well.

I apologize for the interjection. I am looking forward to learning more about John’s work in this discussion.


Addressing your points respectively,

  1. Exactly. That’s a major reason for bringing it up. My understanding of evolution has played an enormous role in my work, but I’ve never actually studied it.

  2. Meyer’s assertion is truly mind-bogglingly wrong, both from the work I’ve started to present here, and in my more recent work on inherited cardiomyopathies. The cardiomyopathies have the advantage of being natural, but the literature is so granular that it’s harder to get the big picture. The gross generalization is, for all but the most severe single-amino acid residue changes in cardiac cytoskeletal proteins, for every teenager who dies suddenly and tragically from cardiac arrest caused by hypertrophy, the half of her/his family members carrying the same disease allele will not have any problem for decades or will die from something else while never showing any symptoms. I suspect that there’s more tolerance for these mutations in the heart over skeletal muscle because of the plasticity of cardiac physiology. The bottom line is that cardiac cytoskeletal genes are pretty darn polymorphic, we knew that in 2010, and more so now.

  3. I found your analysis to be compelling, so I thought that adding a more empirical approach might stimulate more discussion. We’ll see…


I can verify that @Agauger wants to participate but is occupied with some more pressing concerns. In my experience with @Agauger, she is honest about things like this. This is not a dodge. I expect she will be up for engaging in the coming weeks. In the mean time, I request that we do not impute mal-intent on her.

I’m looking forward to the exchange too.

1 Like

@Agauger, great to see you back. @Mercer sent me a note, hoping to pick up the conversation here if you can.

1 Like

@Mung, @swamidass, @Art, @Mercer

I am about halfway through the boxes at this point. I hope to be able to put some time into this topic this weekend. But before I do that I have to apologize to @Mercer for having misread his paper long ago. I read it quickly and saw that he and his colleagues had change the active site to accommodate a modified substrate, and I then jumped to the conclusion that they were trying to create a protein that could be crystallized with the modified substrate in the active site. That is a standard thing that crystallographers do. The reason for doing it is that the modified substrate does not complete the reaction and remains in the active site, thus “gumming up the works”. But that was not the intent of your work. I did not read carefully enough. Sloppy on my part. But if you go back and read the context of the conversation you might understand.

I will return to the issue of what constitutes a significant change in function as I can this week. Let me just say that I have never disputed that some kinds of changes are possible with a single mutation, especially if it enlarges the binding pocket of the active site. Changing the activities of a promiscuous enzyme to favor one substrate over another is definitely possible. Getting a new chemistry that was not already present in any way is very much harder, and I believe beyond the reach of mutation and selection. Here’s where the debate begins.

My kitchen looks better than this now, after unpacking quite a few boxes today, but I still have no dishes or glasses. We did however find the can opener.


8 posts were split to a new topic: An Intelligently Designed Kitchen

Thanks @Agauger. I really appreciate your humility here. We all make mistakes too. Thanks for this.


Accepted. That is gracious of you.

That’s not really my point, though. It has nothing to do with a single mutation. It has nothing to do with favoring one substrate over another, which we never tested.

Before we did the experiment, I hypothesized that enlarging the active site would likely kill the enzymatic activity. We had a whole set of substitutions lined up to try, but the first, most radical one worked. If you and Doug Axe are right, that never should have happened. The same substitution also worked on a very different myosin.

My point is that function is far more prevalent in sequence space than you argue that it is. The point of my story above is that it was far more prevalent than I ever thought it would be. My experience completely changed my thinking. I would suggest that your misreading of my paper was facilitated by your prejudices.

I should also note that I have never seen the term “promiscuous” applied to any myosin activity, so I am puzzled by your use of it. Besides, subjective terms like “promiscuous” and “new chemistry” are too subjective to be very useful.

1 Like

There are objective meanings to “promiscuous,” in this context, meaning the opposite of specificity.


So are you saying that the myosins are enzymatically promiscuous?

I don’t understand why you think this is contrary to our argument. Did you modify myosin’s function to a new one (a different chemical reaction)?

1 Like

Not that I know of, but you would know more than I in this case.