In Chapter 5 of my 1999 book, Finding Darwin's God, I presented a series of specific experimental studies showing that Michael Behe is incorrect in his assertion that Darwinian evolution cannot "account for the molecular structure of life." One of these was a 2-page description of Barry Hall's studies on the evolved beta-galactosidase (ebg) system under the heading "Parts is Parts."

Michael Behe was so concerned about my discussion of this system that he posted a critique entitled "A True Acid Test" on the web site of the Discovery Institute, of which he is a Fellow.

Why did this section of my book get such attention? Quite possibly because I took on one of Dr. Behe's most cherished and oft-repeated claims - namely, that evolution cannot produce a complex, multipart, biochemical system. And I showed, in a rather matter of fact way, that it could.

The Experiments in Question

In 1982, Barry Hall of the University of Rochester began a series of experiments in which he deleted the bacterial gene for the enzyme beta-galactosidase. The loss of this gene makes it impossible for the bacteria to metabolize the sugar lactose. What happened next? Under appropriate selection conditions Hall found that the bacteria evolved not only the gene for a new beta-galactosidase enzyme (called the evolved beta-galactosidase gene, or ebg), but also a control sequence that switched the new gene on when glucose was present. Finally, a new chemical reaction evolved as well, producing allolactose, the chemical signal that normally switches on the lac permease gene, allowing lactose to flow into the cell.

In my book I quoted evolutionary biologist Douglas Futuyma's description of these experiments:

"Thus an entire system of lactose utilization had evolved, consisting of changes in enzyme structure enabling hydrolysis of the substrate; alteration of a regulatory gene so that the enzyme can be synthesized in response to the substrate; and the evolution of an enzyme reaction that induces the permease needed for the entry of the substrate. One could not wish for a batter demonstration of the neoDarwinian principle that mutation and natural selection in concert are the source of complex adaptations." [ DJ Futuyma , Evolution, ©1986, Sinauer Associates, Sunderland, MA. pp. 477-478.]

What's Wrong with this Description?

Behe, predictably, was unimpressed with these experiments, and went so far as to call my claims "extravagant." Were they? Let's see:

First, Behe says that Hall did not wipe out a "multipart system" as I claimed; he deleted just one gene. Well, that's what I wrote, too. My description clearly and correctly states that Hall started his experiments "by deleting the structural gene for galactosidase," a single gene. However, I did indeed write that this deletion had knocked out a "multipart system." Why? Because once the gene was deleted, three components had to evolve to replace its function: First, a new galactosidase enzyme, second, a new lactose-sensitive control region, and third, a new way to switch on the lac permease gene. And, just as Futuyma and I pointed out, that's exactly what happened - all three parts eventually evolved.

Second, Behe is particularly scornful of the fact that the "new" galactosidase enzyme didn't evolve from scratch, but was produced by a small number of mutations in an existing gene, albeit in an operon far distant from the deleted galactosidase gene. In a similar way, the gene for the repressor of this newly-evolved galactosidase, a protein that controls its expression, was rendered lactose-sensitive by a simple mutation in its sequence. In other words, the "new" 2-part system was produced by a couple of rather minor mutations in two pre-existing genes.

Professor Behe may be unimpressed by these mutations, but he's missing the point. This is how evolution generally works - by minor modifications of pre-existing genes to serve new purposes. He emphasizes that the ebg gene is "homologous " to the lac proteins and overlaps them in "activity," but these statements are quite misleading. The pre-existing enzyme activity of the ebg gene is not enough to support the metabolic needs of the cell, and the ebg gene is actually only 34% homologous to the gene whose activity it replaces (meaning that about 2/3 of the protein is quite different from the galactosidase gene whose function it replaces). The repressor (control) gene is even more different, showing just 25% homology to the lac repressor.

Therefore, my original descriptions of these two genes and the mutations that produce galactosidase activity in them were accurate and correct.

Third, Behe points out that the lac permease which was eventually activated by yet another mutation is a pre-existing protein. It has to be there before the experiments are carried out, and it is not produced from scratch. That's true, of course, but that's also exactly what I wrote. What is new and different, of course, is that the ebg gene product, which originally could not catalyze the chemical reaction needed to switch on the permease gene, acquired that ability by means of another mutation. This chemical reaction, not the permease protein itself, is the third part of the system, even if Dr. Behe is unimpressed by the mundane way in which it evolved.

Fourth, he notes that I did not mention that the bacteria in the experiment were supplied with the artificial inducer IPTG. The use of this inducer, he charges, amounts to "intelligent intervention" in the process, thereby invalidating the results as an example of Darwinian evolution. His criticism, once again, misses the point. We cannot even begin such an experiment without deleting the beta-galactosidase gene, and that is necessarily an act of "intelligent intervention." He does not object to that, of course.

However, when Hall grew the bacteria under selective conditions designed to favor re-evolved galactosidase activity, Behe cried foul. As he should know, and as Futuyma wrote, "... mutation and natural selection in concert are the source of complex adaptations." All that Hall had done was to set up conditions where the bacteria would survive (although just barely), and would prosper only if they evolved a system to replace the one he had deleted. Behe calls this "intervention," implying that the investigator had to intervene directly to produce the new system. He didn't of course. All that Hall did was to use that inducer to set up growth conditions that would ensure that the mutants, if they appeared, could survive to be recovered and analyzed. In short, he screened for mutants, he didn't produce them as Behe implies.

Behe is perfectly free to describe the results of these experiments as "a series of micromutations," but he's missing the key question. That question, of course, is whether or not these "micromutations" assembled a system that would fit Behe's description of "irreducible complexity." As I will show, they do.

Achieving Irreducible Complexity

Does Barry Hall's ebg system fit the definition of irreducible complexity? Absolutely. The three parts of the evolved system are:

(1) A lactose-sensitive ebg repressor protein that controls expression of the galactosidase enzyme
(2) The ebg galactosidase enzyme
(3) The enzyme reaction that induces the lac permease

Unless all three are in place, the system does not function, which is, of course, the key element of an irreducibly complex system. Behe quotes a single sentence from Hall's 1999 Paper (FEMS Microbiology Letters 178: 1-6) to the effect that "reacquisition of lactose utilization requires only the evolution of a new beta-galactosidase function." The quote is accurate, but Hall is describing only the enzymes directly involved in lactose metabolism (number 2 in my list above), not the regulatory parts that make the pathway function (numbers 1 and 3). In the very same paper Hall wrote:

"Genes for enzymes with new or improved catalytic activities do not arise from random DNA sequences; they evolve from existing genes whose products exhibit activities that are more or less related to the "novel" activity."

As I wrote in Chapter 5 of my book, the well-matched parts of the newly evolved system include both the new enzyme and both new regulatory steps:

"Lactose triggers a regulatory sequence that switches on the synthesis of an enzyme that then metabolizes lactose itself. And the products of that successful lactose metabolism then activate the gene for the lac permease, which ensures a steady supply of lactose entering the cell. Irreducible complexity."

The fact that each of these parts were scavenged from pre-existing genes doesn't compromise this example a bit. At the time Hall deleted the true galactosidase gene, not one of these three components existed in its final, functional form. Mutation and selection produced each of them, not from scratch as Behe would demand, but from pre-existing genes. As Meléndez-Hevia and his co-authors paraphrased Jacob in their study of the Krebs cycle "evolution does not produce novelties from scratch: It works on what already exists" [ J Mol Evol 43: 293-303 (1996)].

Are Adaptive Mutations non-Darwinian?

One of the interesting aspects of the ebg system, as Behe notes, is that the several mutations that produce the evolved system seem to appear much more frequently than one might expect for a truly random process. Hall and others refer to the elevated mutation rates observed during prolong non lethal selection as part of a process of "adaptive mutation." Behe says that it is "misleading" to use such a system "to argue for Darwinian evolution," because it violates the most basic assumptions of Darwinism with respect to the randomness of mutations. James Shapiro, one of the leading workers in the field of adaptive mutations, would find this reasoning strange indeed. In a 1997 review on adaptive mutations Shapiro noted that:

"the ability to increase the frequency of potentially useful mutations is beneficial (adaptive) for the bacterial population undergoing selection. It is worth noting that observations on the influence of environmental conditions on the levels of hereditary variation are hardly novel and go back at least as far as the opening chapter of Darwin's Origin of Species." [Shapiro, JA, Trends in Genetics 13: 98-104 (1997)]

In short, Darwin would not have been surprised. And he certainly would not have found the phenomenon to be a violation of basic Darwinian principles, as Behe would have readers believe.

The Bottom Line?

Neither I nor any one else has put forward the ebg system as an example of unlimited evolutionary potential. However, the interesting (and, to Behe, the dangerous) part of Hall's experiments is the fact that they produce an interlocking, functional, three-part system under direct experimental observation.

Michael Behe has invested a great deal in his "Biochemical Challenge to Evolution," and it is hardly surprising that he reacts strongly (even to the point of name-calling) to my assertion that "the evolution of biochemical systems, even complex multi-part ones, is explicable in terms of evolution." Nonetheless, Hall's experiments speak for themselves.

As Barry Hall wrote in a 1999 review, "The genome of each organism contains not only information for functioning in its current environment, but the potential to evolve novel functions that will allow it to thrive in alternative environments" (Hall, BG, FEMS Microbiology Letters 178: 1-6 [1999]). For Michael Behe's Biochemical Argument from Design, the existence of experimental evidence that organisms can evolve novel functions is very bad news. Nonetheless, whether Behe wishes it or not, that evidence is there, and the news is beginning to come out.