Ever have a conversation with someone where, try as you might, you just can’t seem to get through? Over the past few weeks University of Toronto professor of biochemistry Laurence Moran and I have been responding to each other’s posts concerning my book The Edge of Evolution and a recent paper by Summers et al. on chloroquine resistance in malaria (a major topic of the book). For previous entries in this running dialogue, see here, here, here, here, here, here, and here.
Unfortunately, I think we have reached the point of diminishing returns, where arguments are repeated to little effect. So I will address his latest (entitled, ironically it seems to me, "Understanding Michael Behe") and let him have the last word if he wishes. Readers can decide for themselves who had the better of the exchange.
He starts off well enough, agreeing with me on two simple points. To my contention that an effect that requires two mutations will be much rarer than an effect that requires one Moran writes, "This is correct." And to my statement that that means a species would either require a much larger population size or a much longer time to acquire the double mutation he also responds, "This is correct." Great! Progress.
But then he goes off the rails. To make my response as clear as possible, I’m going to go through his post in order and in some detail.
I wrote in my last post that I had cited chloroquine resistance in Edge as a likely example of the two-mutation phenomenon, and that Summers et al. recently "confirmed" that it did need two mutations to pump chloroquine. Moran responds, "This is a little bit misleading and possibly a little bit disingenuous. Everyone understood that chloroquine resistance was rare and that it almost certainly required multiple mutations."
I’m afraid it is he who is playing the ing�nue. There’s a big difference between simply requiring serial additive mutations for some maximal effect and requiring multiple mutations before you get an effect at all. The first is a run-of-the-mill, gradualist Darwinian scenario: one mutation comes along, helps a bit, spreads in the population by selection, which increases the base from which the second mutation may arise; the second appears, helps a bit more, spreads, and so forth. Lather, rinse, repeat.
But if the first required mutation (or second, or third) doesn’t help, or positively hurts, then the gradualist scenario is interrupted. The first mutation does not spread in the population (in fact it’s actively kept in check by negative selection), so the number of organisms with the mutation does not increase and can’t provide a larger base within which the second mutation can arise. The Darwinian magic is turned off.
How Much Does that Hurt?
Enormously. For most species, missing even one such baby step increases the required population size/waiting time by a factor of millions to billions. If even one step in a long and relentlessly detailed evolutionary pathway is deleterious, then a Darwinian process is woefully impaired. If several steps in a row are deleterious, you can kiss the Darwinian explanation goodbye.
From here on I will first quote Moran, then respond.
Everyone understood that chloroquine resistance was rare and that it almost certainly required multiple mutations.
That’s revisionist history. Other than the malariologists that I cited in my book (Hayton and Su and Nicholas White), no one I’ve come across who wrote before Summers et al. was published thought multiple mutations would be needed before any chloroquine resistance appeared. For example, the 2005 Science article "A Requiem for Chloroquine" specifically proposed a one-beneficial-mutation-at-a-time scenario.
In The Edge of Evolution I wrote:
Suppose that P. falciparum needed several separate mutations just to deal with one antimalarial drug. Suppose that changing one amino acid wasn’t enough. Suppose that two different amino acids had to be changed before a beneficial effect for the parasite showed up. In that case, we would have a situation very much like a combination-drug cocktail, but with just one drug.
Reviewers of the book writing in major periodicals frothed at the mouth over that idea. As I noted earlier:
But some other people weren’t pleased at all, not one bit — mostly Darwinist reviewers of The Edge of Evolution…. So we got the spectacle of Sean Carroll lecturing in Science:
Multiple replacements can accumulate when each single amino acid replacement affects performance, however slightly, because selection can act on each replacement individually and the changes can be made sequentially.
Well, that’s the theory. But what if a necessary single amino acid replacement doesn’t affect performance at all (because it’s neutral), or "affects performance" by making it worse?
As I show in the post, Kenneth Miller writing in Nature, Nicholas Matzke in TREE, and Richard Dawkins in the New York Times all say similar things. No, "everyone" most definitely did not understand "that it almost certainly required multiple mutations" that had to be present before the effect kicks in. On the contrary, reviewers actively denied it.
Behe developed an explanation based on the idea that two mutations were required and one of them, by itself, had to be very deleterious.
And that turned out to be exactly correct. Summers et al. show that at least two mutations are indeed required for chloroquine-pumping activity, and Lakshmanan et al. showed that parasites with the single K76T mutation languish in the lab. So what’s the problem?
This is because he used an incorrect value for the mutation rate that was several orders of magnitude too low.
Wrong. I worked with a mutation rate of 10-8. The best literature value for malaria is 2.5 � 10-9. That’s a difference of only a factor of four from the rate I used, not "several orders of magnitude." What’s more, it’s a factor of four greater than the literature value, not "too low" — the opposite direction from what Moran thought. (Those negative exponentials can be confusing….) I prefer rounding the value up to 10-8 because using several significant figures gives a misleading impression of undue precision.
He based his calculations on the assumption that the two mutations had to arise in a single infected patient.
Those are Moran’s words not mine, but it’s probably correct anyway and certainly makes little difference to the math. The two required mutations do have to be present, not only in a single patient but in the very same pfcrt gene of a single malaria cell. That’s why they’re called "required."
The recent paper by Summers et al. (2014) shows that seven of the chloroquine resistant strains that have been observed have at least four mutations and some of them are relatively neutral. This refutes and discredits the scenario that Michael Behe put forth in his book.
It’s hard to see why Professor Moran thinks the second sentence in this quote belongs after the first. Close your eyes and envision a pathway to a malaria parasite that has four mutations. The first mutation is deleterious, the second rescues the first and makes the parasite marginally chloroquine resistant. Subsequent steps are all beneficial by dint of either improving chloroquine resistance or of stabilizing the structure of the mutated PfCRT, which is required for malaria survival. Once a parasite can survive at least marginally in the presence of chloroquine, further mutations can be added one at a time (no longer two at a time) in each cycle of infection because the population size (1012) greatly exceeds the inverse of the mutation rate.
In the argot of chemical kinetics, getting beyond the deleterious mutation is the "rate-limiting step." After that hurdle is passed further mutations can be added singly — the way Darwinists like — and comparatively rapidly. Since they would be added rapidly, they would be difficult to detect in the wild. Hence the pattern described by Summers et al. fits the scenario I described perfectly.
[H]is sycophants are promoting the idea…
Name-calling is puerile.
As we all know by now, the "guesstimate" by Nicholas White refers to the possible frequency that a chloroquine resistant strain will be detected someplace in the world. This is a far cry from the probability that the correct mutations will actually occur.
It takes chutzpah to deride a value calculated by one of the world’s foremost malariologists — a Fellow of the Royal Society, who lives and works in the middle of malarious regions of the world, who reasons deeply and quantitatively about the development of drug resistance — as a "guesstimate." White’s calculation was based on intimate knowledge of many details of the biology and epidemiology of the parasite.
Malaria is a horrendous scourge, and so is followed closely by international health organizations. In this age of easy and rapid sequencing, one would expect independent strains of malaria to be detected quickly by the many researchers in the area. If Professor Moran has any actual evidence that a large number of de novo origins of malaria have gone undetected, we would all love to see it. Otherwise it’s reasonable to assume that the number so far identified is close to the number that have in fact arisen.
But, according to Michael Behe, there’s a 10,000 (104) fold difference between the probability of the mutations occurring and their actual frequency. This is important because he attributes that difference to the fact that both of the two mutations are deleterious on their own.
For brevity I didn’t quote Moran’s preceding paragraph, which refers to this. Professor Moran seems terribly confused here. The mutation rate he himself favors, about 10-10, would also require a mutational step to be deleterious to get to the observed frequency of 1 in 1020. Otherwise, accumulation of neutral mutations would raise it. I myself do not "attribute that difference" of a factor of 104 between the (very reasonable) rate of genetic production I use and Nicholas White’s number of 1 in 1020 to the mutations being deleterious. Rather, the factor of 104 or so is simply the factor remaining of White’s empirical statistic after accounting for the double mutation rate.
What’s more, the factor of 104 accommodates an expected fractional selection coefficient and leaves room for the possibility of an additional required mutation (a third and/or fourth one). Moran’s number does not. Thus he is in the awkward position of advocating a scenario that would seem to predict an even less frequent occurrence of de novo chloroquine resistance than White estimates, all the while arguing (above) that many origins have gone undetected, which would make it more frequent. He seems to really believe the old saw that consistency is the hobgoblin of little minds.
Summer et al. (2014) showed conclusively that this calculation is wrong. They showed that among the seven strains analyzed there were multiple pathways to resistance and that a minimum of four mutations were required for effective resistance. They showed that one mutation (K76T) was essential in all strains. All strains had one of two additional mutations that seemed to be required early on, N75E or N326D.
This is referring to a quotation of mine, not cited here for brevity’s sake. I have no idea why Professor Moran thinks any of this shows my "calculation is wrong."
The recent data by Summer et al. refutes three of Behe’s assumptions: (a) that only two mutations are required to account for the appearance of resistant strains, (b) that a resistant parasite had to arise from wild type within a single infected individual, and (c) that one or more of the individual mutations are deleterious on their own. Other than that, his calculations and his "predictions" are perfect!!!!
Summers et al.’s paper does not even address Moran’s (a), (b), or (c), let alone resolve matters the way he claims. The paper measured the ability of a number of mutant PfCRT proteins to pump chloroquine across a membrane in an artificial test system consisting of frog eggs. It also examined the ability of two mutant malaria constructs carrying artificial genes on a transfected plasmid to survive in a medium containing a certain level of chloroquine that usually kills half of sensitive malaria cells. That’s it!
They did not determine that more than "two mutations are required" for resistant strains. They did show that several strains with three mutations that could pump chloroquine nonetheless did not have higher resistance against a certain level of chloroquine in test tubes in the lab. However, this does not show that the strains could not survive at some levels of chloroquine in some humans in the wild. That remains an unanswered question.
Summers et al. did absolutely nothing to determine whether or not "a resistant parasite had to arise from wild type in a single infected individual." Nor did they even try to investigate whether "one or more of the individual mutations are deleterious on their own." Rather, mutant PfCRT proteins were examined in an elegant test system of frog eggs, which can’t tell you what the effect of the protein will be in a malarial cell within a human. The several mutant proteins they did test in malaria cells were coded by genes carried on plasmids — meaning that the malaria cells also expressed their own, wild-type, normal, genomic, unmutated gene, which of course could compensate for any missing required activity of a nonfunctional mutant plasmid gene, so they couldn’t detect if a mutant protein were deleterious by itself. On the other hand, it was shown years ago that malaria cells constrained to express only mutant K76T PfCRT would not grow in the lab, demonstrating the mutation is indeed deleterious.
Other than that, Moran’s comments are perfect.
It is true that any specific set of 4-6 mutations is extremely unlikely. Nobody disputes that any more than they dispute the probability of a specific hand of bridge being dealt in the next deal. That’s only going to happen once in 1028 tries. It’s impossible that you will see it in your lifetime. But you still get to play bridge.
The bridge analogy is inapt (see my previous post). P. falciparum dies without the needed mutations. It will never play bridge again.
For example, let’s say that the difference in needle clusters between red pine and white pine are determined by a single gene. Let’s say that three specific mutations are required to change from a cluster of two needles to a cluster of five needles…. When such a triple mutation arises we recognize that it was only one of millions and millions of possible evolutionary outcomes. There was no a priori requirement that Earth contain red pines and white pines just as there’s no a priori requirement that you get a specific hand in the next deal.
Were there "millions and millions of possible evolutionary outcomes" that could make malaria resistant to chloroquine? I can think of a few possible alternative scenarios. Instead of a chloroquine pump appearing, maybe a chloroquine-degrading enzyme could have arisen, or maybe malaria’s membrane could somehow be altered to stop chloroquine from entering the cell, or maybe the cell could become dormant until the chloroquine passed.
But none of those scenarios happened. Why not? Because any evolutionary pathway leading to those outcomes was even less probable than the pathway that occurred. The odds of 1 in 1020 are not just the probability of PfCRT acquiring the right mutations — it’s the minimum odds of any chloroquine-resistance mechanism arising in P. falciparum. It may very well be the case that there are no other resistance mechanisms that can be reached by malaria.
And just as those alternative chloroquine resistance pathways are imaginary, Professor Moran’s "millions and millions of possible evolutionary outcomes" are imaginary. In the absence of actual evidence that a huge number of relevant unrealized biochemical features could have been built by Darwinian processes, it is illegitimate to arbitrarily multiply probabilistic resources.
Moran is right that there is "no a priori requirement that Earth contain red pines and white pines." But he doesn’t follow his own logic far enough. In fact there’s no a priori requirement that Earth contain any life at all, or that any life that does exist be able to successfully traverse a mutational pathway by Darwinian means to give rise to a form of life significantly different from itself.
In the absence of an a priori requirement, science is obliged to investigate whether or not such pathways exist. Right now the evidence we have in hand militates strongly against it.
This is hard for IDiots to understand.
See my remark above about name-calling.
Photo source: Kathleen Tyler Conklin/Flickr.