Several papers on the topic of “evolvability” have been published relatively recently by the laboratory of Richard Lenski.¹² Most readers of this site will quickly recognize Lenski as the Michigan State microbiologist who has been growing cultures of E. coli for over twenty years in order to see how they would evolve, patiently transferring a portion of each culture to new media every day, until the aggregate experiment has now passed 50,000 generations. I’m a huge fan of Lenski et al‘s work because, rather than telling Just-So stories, they have been doing the hard laboratory work that shows us what Darwinian evolution can and likely cannot do.
The term “evolvability” has been used widely and rather loosely in the literature for the past few decades. It usually means something like the following: a species possesses some biological feature which lends itself to evolving more easily than other species that don’t possess the feature, so that the lucky species will tend to adapt and survive better than its rivals over time. The kind of feature that is most often invoked in this context is “modularity.” That word itself is often used in a vague manner. As I wrote in The Edge of Evolution, “Roughly, a module is a more-or-less self-contained biological feature that can be plugged into a variety of contexts without losing its distinctive properties. A biological module can range from something very small (such as a fragment of a protein), to an entire protein chain (such as one of the subunits of hemoglobin), to a set of genes (such as Hox genes), to a cell, to an organ (such as the eyes or limbs of Drosophila).” ³
Well, Lenski and co-workers don’t use “evolvability” in that sense. They use the term in a much broader sense: “Evolutionary potential, or evolvability, can be operationally defined as the expected degree to which a lineage beginning from a particular genotype will increase in fitness after evolving for a certain time in a particular environment.”¹ To put it another way, in their usage “evolvability” means how much an organism will increase in fitness over a defined time starting from genotype A versus starting from genotype B, no matter whether genotypes A and B have any particular identifiable feature such as modularity or not.
Lenski’s group published a very interesting paper last year showing that the more defective a starting mutant was in a particular gene (rpoB, which encodes a subunit of RNA polymerase), the more “evolvable” it was.² That is, more-crippled cells could gain more in fitness than less-crippled cells. But none of the evolved crippled cells gained enough fitness to match the uncrippled parent strain. Thus it seemed that more-crippled cells could gain more fitness simply because they started from further back than less-crippled ones. Compensatory mutations would pop up somewhere in the genome until the evolving cell was near to its progenitor’s starting point. This matches the results of some viral evolution studies where some defective viruses could accumulate compensatory mutations until they were similar in fitness to the starting strain, whether they began with one-tenth or one-ten-billionth of the original fitness.⁴
In a paper published a few weeks ago the Michigan State group took a somewhat different experimental tack.¹ They isolated a number of cells from relatively early in their long-term evolution experiment. (Every 500th generation during the 50,000-generation experiment Lenski’s group would freeze away the portion of the culture which was left over after they used a part of it to seed a flask to continue the growth. Thus they have a very complete evolutionary record of the whole lineage, and can go back and conduct experiments on any part of it whenever they wish. Neat!) They saw that different mutations had cropped up in different early cells. Interestingly, the mutations which gave the greatest advantage early on had become extinct after another 1,000 generations. So Lenski’s group decided to investigate why the early very-beneficial mutations were nonetheless not as “evolvable” (because they were eventually outcompeted by other lineages) as cells with early less-beneficial mutations.
The workers examined the system thoroughly, performing many careful experiments and controls. (I encourage everyone to read the whole paper.) The bottom line, however, is that they found that changing one particular amino acid residue in one particular protein (called a “topoisomerase,” which helps control the “twistiness” of DNA in the cell), instead of a different amino acid residue in the same protein, interfered with the ability of a subsequent mutation in a gene (called spoT) for a second protein to help the bacterium increase in fitness. In other words, getting the “wrong” mutation in topoisomerase — even though that mutation by itself did help the bacterium — prevented a mutation in spoT from helping. Getting the “right” mutation in topoisomerase allowed a mutation in spoT to substantially increase the fitness of the bacterium.
The authors briefly discuss the results (the paper was published in Science, which doesn’t allow much room for discussion) in terms of “evolvability”, understood in their own sense.¹ They point out that the strain with the right topoisomerase mutation was more “evolvable” than the one with the wrong topoisomerase mutation, because it outcompeted the other strain. That is plainly correct, but does not say anything about “evolvability” in the more common and potentially-much-more-important sense of an organism possessing modular features that help it evolve new systems. “Evolvability” in the more common sense has not been tested experimentally in a Lenski-like fashion.
In my own view, the most interesting aspect of the recent Lenski paper is its highlighting of the pitfalls that Darwinian evolution must dance around, even as it is making an organism somewhat more fit.¹ If the “wrong” advantageous mutation in topoisomerase had become fixed in the population (by perhaps being slightly more advantageous or more common), then the “better” selective pathway would have been shut off completely. And since this phenomenon occurred in the first instance where anyone had looked for it, it is likely to be commonplace. That should not be surprising to anyone who thinks about the topic dispassionately. As the authors note, “Similar cases are expected in any population of asexual organisms that evolve on a rugged fitness landscape with substantial epistasis, as long as the population is large enough that multiple beneficial mutations accumulate in contending lineages before any one mutation can sweep to fixation.” If the population is not large enough, or other factors interfere, then the population will be stuck on a small peak of the rugged landscape.
This fits well with recent work by Lenski’s and others’ laboratories, showing that most beneficial mutations actually break or degrade genes⁴, and also with work by Thornton’s group showing that random mutation and natural selection likely could not transform a steroid hormone receptor back into its homologous ancestor, even though both have very similar structures and functions, because the tortuous evolutionary pathway would be nearly impossible to traverse.⁵⁶ The more that is learned about Darwin’s mechanism at the molecular level, the more ineffectual it is seen to be.
References
¹Woods, R. J., J. E. Barrick, T. F. Cooper, U. Shrestha, M. R. Kauth, and R. E. Lenski. 2011 “Second-order selection for evolvability in a large Escherichia coli population.” Science 331: 1433-1436.
²Barrick, J. E., M. R. Kauth, C. C. Strelioff, and R. E. Lenski, 2010 “Escherichia coli rpoB mutants have increased evolvability in proportion to their fitness defects.” Molecular Biology and Evolution 27: 1338-1347.
³Behe M. J., 2007 The Edge of Evolution: the search for the limits of Darwinism. Free Press, New York.
⁴Behe, M. J., 2010 “Experimental Evolution, Loss-of-function Mutations, and ‘The First Rule of Adaptive Evolution.'” Quarterly Review of Biology 85: 1-27.
⁵Bridgham, J. T., E. A. Ortlund, and J. W. Thornton, 2009 “An epistatic ratchet constrains the direction of glucocorticoid receptor evolution.” Nature 461: 515-519.
⁶See my comments on Thornton’s work here, here, and here.