Gauger: Answering Art Hunt on Real Time Evolution


(S. Joshua Swamidass) #24

We agree with this. There were two sorts of manipulations: to the environment and to the genetics. You and Behe point to both types of manipulation as examples of “guidance” and intelligent design, which render the experiment irrelevant. I do not think this is a valid conclusion.

What were the manipulations? There were two types of manipulations: genetic and environmental. You and Behe objected to both manipulations.

  1. External Environment. The investigators placed the cells in an environment where they would benefit from evolving new functions. This simulates what is already happening in the wild, in a controlled laboratory setting. In what sense is “placing the bacterial in a environment where it will be rewarded for innovating” grounds for dismissing the capacity of these cells to evolve new functions?

  2. Internal Genetics. the genetic manipulations are merely so that the researchers could better measure and study the evolutionary process, but they do not actually altering the mechanism. A good analogy might be tagging a wild animal with a radio transmitter, and then releasing him back into the wild to track his movements. Yes, there is an immense amount of designed technology in the transmitter, but this just helps us observe what the wild animal does on its own, without directing him. It does not appear the manipulations alter or direct the course of the evolutionary mechanism, so why is this a valid objection?

It does not appear you have even begun to answer these questions.

I disagree. It was not observed without careful experimental work. It seems to be the full truth to say that these new functions evolved by natural evolutionary mechanisms, in a context designed to give researchers visibility into what happens at a genetic level. We see evidence of this mechanism of evolution in the wild too, but we can’t track it as easily. We need the genetic manipulations to directly observe the innovation, but not to direct it.

We have not yet even considered the implications for Axe’s (and your) understanding of enzyme function. This experiment, it would appear, should be impossible if you were both right. You’ve produced calculations that demonstrate this experiment is impossible. I will let others press you on this, but this experiment appears to be just as challenging to Axe’s understanding of sequence space as this one: Antibody Enzymes and Sequence Space.

(Mikkel R.) #25

It seems to me the only kind of “manipulation” the researchers did was to set up the experimental conditions to mirror the kind of situation theorized to have made it possible for the HisA and TrpF genes to evolve from a common ancestral enzyme in the first place, once upon a time.

If once upon a time an organism existed that only had one of these genes, and this original gene only could catalyze one of the two reactions now catalyzed by each, then surely the scientists can’t be accused of “manipulation” (in a way that is unrealistic), in setting up the genetics of the organism to be like that original situation, by deleting one of the extra genes?

In setting up the environment so that only one nutrient is present, and the other is limited or lacking entirely, how is this any more of a piece of “manipulation” than when similar situations happens in the wild?

Some times particular nutrients are limited, or simply nonexistent, and these are then the conditions that favor mutations that give the organism some some ability to make these compounds themselves from other precursors if they are essential for growth. The only necessity in such a situation is that the condition where one essential nutrient is lacking does not persist indefinitely.

Of course, tryptophan (for example) is only conditionally essential because there now exists proteins that incorporate it in their sequence. This indicates that before tryptophan biosynthesis had first evolved in any lineage, no protein in organisms at that time would have had it in their sequence, so it’s absence would not have limited growth.

(S. Joshua Swamidass) split this topic #26

3 posts were split to a new topic: Comments on Gaugers Response to Hunt on RTE

(Arthur Hunt) #29

@Agauger, what, specifically, did the authors do to duplicate codons 13 to 15 of the HisA coding region, and to substitute residue 10 from D to G? Which methods did they use? How did they arrive at these changes? Which predictive programs might they have used to know that these changes in HisA would give a bifunctional enzyme?


Just to be clear, here are the specific genetic manipulations:

All they did was remove a section of DNA upstream of the gene so that it would be expressed even in the presence of histidine. If the gene is not expressed then there can’t be selection for beneficial mutations in that gene. In other words, they simply made it possible to select for and detect mutations that changed the activity from his+ to trp+. The genetic manipulations themselves did not cause those mutations. The mutations that did occur in the hisA gene are entirely natural.

Let’s compare the methods used by Nasvall et al. to the methods used by Gauger and Axe:

What did Gauger and Axe conclude from this work?

Gauger and Axe used designed site directed mutagenesis to change amino acid residues, and they concluded that this was a fair test for evolution. For those who may not be familiar with this process, it involves the purposeful targeting of a base or several bases to produce a specific change in sequence. Gauger and Axe also had the manipulated gene on an engineered plasmid that guaranteed its expression in all conditions, just as was the result of the manipulations Nasvall et al. introduced. Gauger and Axe also grew the bacteria in an artificial environment where they added or subtracted biotin.

In comparison, Nasvall et al. allowed the gene to be mutated naturally by the mechanisms already present in the bacteria. @Agauger is now saying that allowing the gene to be mutated naturally is not valid for testing evolution, but is already an author on a paper where there were designed and targeted mutations, artificial promoters, and artificial environments. Even with all of these conditions, the experiments in the Gauger and Axe paper were considered a valid test of evolution. These positions don’t square up.

(S. Joshua Swamidass) #34

…set up the operon of the plasmid in the correct way for their specific question. (shrug)

It strikes me that Axe and @Agauger’s work seems to have demonstrated the sharp limits of human designed modification. While the RTE paper shows that natural evolution far exceeds the capacity of human designed modification.


I would also add that the model used by @Agauger and Axe seemed to ignore epistatic effects in the genetic background and possible evolutionary history for the enzyme they were studying. I will need to read the paper much more closely to see if this is the case, and my apologies to @Agauger if I missed something in their paper.


I gotta say the sorts of experimental techniques in the Nasvall, Roth & Andersson paper are quite similar to work I once did, so I’m pretty familiar with the steps and methods used. In fact, I know John Roth (an old fart now, but a sharing, very approachable researcher) and used a number of his strains in my work. The paper described interesting work that yielded completely reasonable and justified conclusions.

I did not get the impression that Art suggested the work was completely natural or was done without any experimental manipulation. I don’t see that he misrepresented the work at all. I’m a bit ticked off that this discussion led to me pulling up & studying various sources to ultimately confirm completely uncontroversial results (even though it was nice to see what Roth has been up to). In the future, I’ll likely be more inclined to discount claims made from some sources.

(S. Joshua Swamidass) split this topic #37

11 posts were merged into an existing topic: Comments on Gaugers Response to Hunt on RTE

(Arthur Hunt) #42

Well, seeing as I have taken the trip back through time, I may as well address this particular insult. My graduate advisor was the first author on this paper, which was one of the ways (maybe the main way) that bacterial geneticists did targeted mutagenesis in the days before recombinant DNA, PCR, etc., etc. Needless to say, I did more than a little of this sort of mutagenesis (as well as a whole lot of other classical bacterial genetics) in graduate school.

(Ann Gauger) #49

I can’t find my way through the thicket of threads to say what I want to say so I will say it here. You seem to have found quite a bit of the relevant stuff on your own, i.e. the Jurgens paper. If you read that paper carefully you will see that they point out that HisA and TrpF carry out the same reaction, but with different R groups. They have the same basic chemistry, As I have said before, enzymes that have the same basic chemistry can frequently be shifted toward the other’s function. This is not a surprise. Likewise, if you have an enzyme that has two catalytic activities, subfunctionization can occur. This is not a surprise. If duplicating a gene is beneficial, that too, will occur. Not a surprise.

So to let you know how not a surprise it all was, I was gathering up references. You have found some already. I will provide others. But before I do, let me lay out the hierarchy of difficulty of evolution of an enzyme to a new function, from easiest to hardest. There must be a selectable level of starting function or function within one or two steps and

  1. Overlapping function, shared chemistry: promiscuous enzyme or
  2. Shared chemistry (types of bond broken and formed, mechanisms shared.)
  3. Substrates in common, but no shared reaction types are unlikely to work.
  4. No shared function, chemistry or substrates won’t work.

The kinds of successful conversions you see in the literature are almost always 1 or 2. If you don’t believe me, look carefully. Read the methods and check for baseline activity before conversion like Rumraket did. I would have pointed it out if he hadn’t.

BioF and Kbl were type #4. That was our purpose, to see if it was possible to convert an enzyme with no chemistry or substrate preference in common to another enzyme’s function. As trpF and HisA have shown, sequence identity doesn’t matter. Apparently shared chemistry does. That and the kind of fold the enzyme has, a beta barrel, with the active site located external to the barrel, on unstructured loops, accessible and easy to modify. BioF and Kbl have a much more complicated reaction and active site.

Nasvall et al was a fine study. It used a great deal of previously gathered knowledge about these genes and their regulation to make sure they could make the double function enzyme (they designed and carried out the selection–they created it, even though they did not pick the mutations), and then get it to express first in one condition and then the other without interference. Could that his regulatory mutant have happened on its own? Sure. But the sequence, starting with the set up in a cassette to guarantee constitutive expression, and the selection to split trpC and trpF and then get the double function enzyme, and then to improve it, and then to allow expression in the presence of histidine, and then the extended culture without interference from competing wildtype cells… It’s a nice proof of concept experiment but it won’t happen on its own. How many steps? 6? This is what I was upset about. Just because you can get something to happen in a lab does not mean it will happen for real. How many of those steps would be beneficial in a mud puddle or intestine? We know some bacteria have dual function genes, so it can happen, but not by this elaborate path.

Maybe this is my complaint in a nutshell. You all seem to assume I think this kind of thing won’t happen, that gene duplication and recruitment never works, or exaptation. It does, but under limits.

And that leads to the other ridiculous thing about the use of Nasvall in the first place. Behe himself doesn’t disagree with Nasvall et al. He said so in his review. Those who tried to say it disproved his work misunderstand his work

I never said or implied that a conversion like Nasvall et al described could not work. That would be ridiculous. These were papers I looked at when we were considering what system to use.

From an early draft of the bioF paper:

While numerous experimental studies have aimed to demonstrate the feasibility of reconfiguration, several complicating factors tend to be overlooked when results are interpreted. First, there is surprisingly little discussion of how evolutionary feasibility falls off as the number of required mutations increases. Schmidt and coworkers pay more attention to this than most, observing that “if natural divergent evolution of mechanistic diversity is to be accomplished, only a limited number of substitutions can be required for selective advantage; otherwise, the probability of accumulating the necessary mutations would be too low” [12]. But while they claim that most experimental demonstrations require too many nucleotide substitutions, they offer no way to draw a line between what is feasible and what is not.

Another point, which they also raise, is that many demonstrations of enzyme reconfiguration merely modify substrate specificities without changing the mechanism of reaction. In most cases the modifications simply shift the relative preference for existing substrates (see for example [13–17 ]), but even when mutations cause enzymes to recognize genuinely new substrates [18–19 ], this restricted category of diversification leaves the most important kind of metabolic innovation—new pathways built on new chemistries—unexplained.

Modify this to mention recent papers describing very low limits for feasible transitions, including your manuscript in review.

Ref 19 belongs here: Jurgens et al 2000 and add Leopoldseder et al 2004

Do you mean same chemistry, new substrates, or do you mean new chemistry here? I have given refs for new chemisrty in the next comment. For a case where the chemistry is similar but the substrate different, and near wild-type activity was achieved, see Claren et al 2009 in HisA Trp F folder( Establishing wild-type levels…)

New chemistry: Cite Toscano et al for a review, plus 18, Xiang et al 1999, Schmidt et al 2003, Ma and Penning 1999, Vick et al 2007 (this paper describes optimization of OSBS with four mutations). Delete 19 here.

And additional reference on trpF and HisA

Enzymes are known to be amazingly specific and efficient catalysts. However, many enzymes also have so-called promiscuous functions, i.e., they are able to catalyze other reactions than their main one. The promiscuous activities are often low, serendipitous, and under neutral selection but if conditions arise that make them beneficial, they can play an important role in the evolution of new enzymes. In this thesis, I present three studies where we have characterized different enzyme families by structural and biochemical methods. The studies demonstrate the occurrence of enzyme promiscuity and its potential role in evolution and organismal adaptation. In the first study, I describe the characterization of wild type and mutant HisA enzymes from Salmonella enterica. In the first part of this study, we could clarify the mechanistic cycle of HisA by solving crystal structures that showed different conformations of wild type HisA in complex with its labile substrate ProFAR (N´-[(5´-phosphoribosyl)formimino]-5- aminoimidazole-4-carboxamide ribonucleotide). In the second part of this study, structures of mutant enzymes from a real-time evolution study provided us with an atomic-level description of how HisA had evolved a new function. The HisA mutants had acquired TrpF activity, either in addition to (bifunctional generalists) or instead of (TrpF specialists) their HisA activity.

We report the occurrence of an isomerase with a putative (βα)8‐barrel structure involved in both histidine and trypto‐phan biosynthesis in Streptomyces coelicolor A3(2) and Mycobacterium tuberculosis HR37Rv. Deletion of a hisA homologue (SCO2050) putatively encoding N ′‐[(5′‐phosphoribosyl)‐formimino]‐5 amino‐imidazole‐4‐carboxamide ribonucleotide isomerase from the chromosome of S. coelicolor A3(2) generated a double auxotrophic mutant for histidine and tryptophan. The bifunctional gene SCO2050 and its orthologue Rv1603 from M. tuberculosis complemented both hisA and trpF mutants of Escherichia coli . Expression of the E. coli trpF gene in the S. coelicolor mutant only complemented the tryptophan auxo‐trophy, and the hisA gene only complemented the histidine auxotrophy. The discovery of this enzyme, which has a broad‐substrate specificity, has implications for the evolution of metabolic pathways and may prove to be important for understanding the evolution of the (βα)8‐barrels.

(Ann Gauger) #50

I object to being plopped down here. But I am glad to know Art know bacterial genetics. So do I.

(S. Joshua Swamidass) #51

@Agauger, this is a different objection. Are you now conceding that your first objection is not valid? Remember, you write:

From your ENV article:

Näsvall et al. Demonstrates the Effectiveness of Intelligent Design

To summarize, Näsvall et al.’s work does not invalidate Behe’s book. On the contrary, the paper shows exaptation occurring under the guidance of an intelligent agent.

This was your initial objection. Are you giving up on it? I would still assert what happened was completely natural. The only manipulations that were done were to change the environment of the cells and make it easier to track the genetic changes.

It appears that you are giving up on your initial set of objections. Am I understanding you correctly?

Not sure why you object. This is where we are discussing this.

(Ann Gauger) #52

OK folks, please do read the BioF Kbl paper carefully. I acknowledge it is not an easy read (guess who wrote it), but just about everything you raise is addressed there.
We chose to use two modern proteins despite their sequence dissimilarities to demonstrate what was possible for two proteins with common structure and active site and even a great deal of chemistry in common, but not all, to accomplish. TrpF and HisA are far apart in sequence space but they have shared chemistry. See what I wrote above. (I had to edit this because it got rearranged, apparently. I wish you would stop moving my stuff.)
Of course epistasis is a problem. Not for His A and TrpF though. The active site sits on the outside of a beta barrel, relatively unconstrained. Our active site is deeply buried and constrained. as we acknowledge in a figure.

(Ann Gauger) #53

No, I do not give up my original objections. They are all the same thing. Read carefully.

(Ann Gauger) #54

My intent was to start a new thread, fresh, like you like to do.

(S. Joshua Swamidass) #55

Okay, you write this:

As I recall, it was THREE steps, and I do not understand why this is thought to an elaborate path. It appears to be exactly the scenario evolutionary theory envisions in the wild, without any human direction.

Actually the complaint is that you say because it happened in a designed experiment it was caused by intelligent design. This seems to be a fallacy. You still have not shown us why it isn’t.

(Ann Gauger) #56

The manipulations do indeed alter and direct the course of the evolutionary path. They don’t choose the bases, but they choose what kinds of change will be rewarded. They set up the selection. That’s what geneticists do. I have highlighted the six distinct selections they did.

(S. Joshua Swamidass) #57

Which is covered here:

(Ann Gauger) #58

They place the cell in six distinct environments, depending on what they were selecting for. (I forget–did they evolve or engineer the his regulatory mutant? If engineered, it’s only 5 environments. Oh, and the set up in a cassette was engineered. Was the trpC and trpF split engineered? That means 3 selection and 3 engineered steps. I don’t think that improves your case.