Antibody Enzymes and Sequence Space



I don’t doubt that Axe’s experimental results are accurate. What we disagree with is the conclusions that he draws from them.

A good counterexample is the emergence of beta-lactamase activity in a limited pool of antibodies that are well, well below the 10^-77 figure given by Axe.

The Extra Face in Mount Rushmore
Darwinism Falsified in Science Long Ago?
(Bill Cole) #2

I agree with you the experiment is one data point. What needs to be sorted out is what is causing the variation in the results of these experiments. Certainly as Art mentioned in his paper it can depend on the methods of the experiment. It also may depend on the nature of the protein that is being tested.


Without reading Art’s paper, I really can’t say. It would be worth asking if some of the mutants had beta-lactamase activity in different conditions (change in temperature, pH, salt, etc.), and that could certainly change the conclusion. It is also interesting that Axe chose a temperature sensitive enzyme which would indicate a lack of protein stability from the very start. However, what I don’t doubt is that if we followed Axe’s exact methodology we would get the same results he got. I always assume published results are truthful and honest until given a reason to think otherwise (i.e. innocent until proven guilty).

What I find more interesting is that we can find beta-lactamase activity in a library of around 1 billion randomly assembled variable regions from antibodies. 10^9 is well below Axe’s 10^77.

(Mikkel R.) #4

Pff, what’s 68 orders of magnitude between friends eh?

(Bill Cole) #5

What does beta-lactamase activity mean?

(Mikkel R.) #6

Catalyze hydrolysis (breaking of a chemical bond with water as one of the substrates) of beta-lactams (a type of organic molecule).

(Timothy Horton) #7

That works out to an error of 1.044 Dembskis. :slightly_smiling_face:

Error Value In Dembskis

(Ann Gauger) #8

@T_aquaticus @swamidass
I just looked at that paper again. They never did get soluble activity. And on phage they never report real enzyme kinetics. In fact they only report pluses and minuses for the relative activities of their 5 clones . The report slopes but no on or off rates, maximal concentrations, nothing! Mercer and Harshman and Hunt should be all over it. They were when it was Axe and they didn’t like the conclusion. They didn’t report if concentration had an effect that I saw.

This is not solid stuff to be making big claims on @Swamidass.

They must have purified them, at least the ones they did structures on. Nothing there either

Whatever they have, unless it exhibits enzyme kinetics, its not an enzyme

(S. Joshua Swamidass) #9

If we find a similar study that does show enzyme kinetics, where would that leave you? I suspect we could contact the authors and find out too. If these antibodies do show enzyme kinetics, where does that leave you? Would that make the case?

We can get more evidence (see above).

@art should probably comment on this, but there is a difference between determining actives and non-actives in this case. A low sensitivity / high specificity test (as did Axe and this study did) does not rule out activity, but it can demonstrate activity. It appears this was exactly this case, and we are justified in rejecting Axes certainty of non-activity (it is indeterminate in his experiment) while having more confidence that these are real positives.

As far as obtaining the enzyme kinetics, that is a good confirmatory experiment. I agree with that, and if it wasn’t done in that study its fine to ask for it. I don’t want to go down that path, though, till you can tell us what it would mean if did get confirmation that had enzymatic activity. Would you agree then that this data demonstrates Axe’s argument fails?

(Ann Gauger) #10

Do you remember what I said when I got so angry? I am on my phone so can’t look it up. If they had been soluble with reasonable kinetics then yes I would reconsider. But don’t expect me to be public about it . I will not be your foil. Fight it out with Doug.

(S. Joshua Swamidass) #11

I’d love to discuss this with Doug. We’ve invited him here many times. Maybe he will be up for it sometime. Who knows?

Alright, let us see what we can find out.

(Ann Gauger) #12

@swamidass do you know how easy those fluors are to degrade? The negative with just phage was ok but what about with random protein. In E. coli just loading up the peroplasmic space with protein provides some resistance to penecillin. Penicillin hydrolyzes easily. You have to store it desicated. It loses half its potency in one day on plates. I haven’t checked about those fluors.

(S. Joshua Swamidass) #13

Okay, I looked at the study. It did in fact measure enzyme kinetics, but was unable to get units for this precise technical reason:

The absence of such activity when scFv P90C2 is expressed in soluble form, suggests that the structure of this antibody is dependent on the experimental conditions, and possibly even on stabilization caused by the link with the PIII phage–protein.

This, to be clear, does not make the antibody any less catalytic. When the antibody is attached to the phage it is an enzyme, but when it separates it is not. This creates a problem in quantitating how much of the protein is antibody versus phage, so you can’t get a well scaled clearance.

  • Note that the concentration of scFv clones is not measurable due to the heterogeneity of the number of scFv presented on the surface of each phage particle.

However, they were able to measure the relative clearance, which scales with Km / Vmax…

Importantly, assessment of catalytic activity was performed with the scFv fragments displayed on the surface of phage, thus preventing us from quantitatively evaluating the catalytic parameters. The initial slope was thus used to compare the relative activity of scFv displayed on phage surface. It is important to note that such a measurement, on phage‐displayed scFv, is indispensable for the validation of the selection strategy and has to be considered as a prerequisite to the soluble expression of the selected scFv fragments.

They present a table that some of the activities, and two of them show higher activity than beta-lactamase itself at 45 nm. That is absolutely evidence of catalytic activity, that was not present before selection. The fact that this is not in soluble form doesn’t change anything. The virus started out without the function, but after selection it had the enzymatic function.

I understand you don’t want to concede anything on Axe’s argument publicly, but I’m not sure why your objections were valid. These are experiments that make very clear that there is enzymatic activity. They are even able to show it is substrate specific, and that it is roughly on par with native beta lactamase.

Right. That is what the negative control is for, and it came out really good. You are suggesting they needed to do more negative controls? Is your claim that it is really easy to get beta lactamase activity that is substrate specific, as they did here?

Okay, when you get a chance to clarify your objection let me know. I do a lot of work analyzing data veery similar to this, and it looks pretty good to me. Maybe you see something I missed though. I’m happy to be corrected.

(S. Joshua Swamidass) #14

The flourophore is based on Nitrocefin. Does Nitrocefin have the same problem? Why would we expect non-funcitonal random protein to hydrolyze Nitrocefin more than the negative control?

(Arthur Hunt) #15

Quite wrong. The phage are soluble.

All irrelevant. The activities that were demonstrated for the phage are authentic. Also, Axe never, ever, EVER measured anything biochemically.

Ann, you are really grasping at straws here. You are not helping your case at all. (I will leave it to others to provide an extensive list of all the qualitative measures of enzyme activity that Ann has just dissed. The instances number in the thousands.)




I agree with the previous posters on the irrelevance of your objections, so I won’t repeat them extensively here. The thing that strikes me is why they were only able to find 5 clones with beta-lactamase activity out of 2.7 billion possible clones? How do you explain that? The only difference between the phage clones was the bit of DNA they inserted into phage genome, so it would seem to me that whatever that bit of DNA resulted in has beta-lactamase activity, perhaps in combination with the other phage proteins.

It is also notable that they found structural similarities between the antibody genes and other beta-lactamases.

I somehow doubt this will convince you.

(John Mercer) #17

Why are you pretending that there’s only one paper on catalytic antibodies, Ann?

The report has enzymatic assays, something Axe didn’t bother with.

More importantly, the authors of this paper do not draw any global evolutionary conclusion, so what’s the problem exactly?

(Ann Gauger) #18


I can only spend a little time here today–major work project. I just want to say why I find this study unconvincing so far.
The antibody has no activity unless displayed on phage. It is missing a stabilizing 3D structure on its own.
The identified clones with activity all have the same triad of amino acids common to serine protease and beta lactamase active sites. These residues are held in the correct orientation by the 3D structure in order to carry out the enzymatic reaction.

If you were to add protein to the soluble phage-free Ab enzyme as you proposed, it would not help produce unless it produce a proper 3d structure. That is not easy to get. I notice you haven’t moved forward with that idea. There are loads of practical difficulties as well.

(S. Joshua Swamidass) #19

Just one thought. Doesn’t that make it irreducibly complex? That makes this enzymatic system even more complex than an antibody.



I also don’t see how this is a problem. Heterodimers are a class of functional proteins.

Points to @swamidass for pointing out the evolution of this irreducibly complex system.

And yet they still got beta-lactamse activity.