When one wants to modify an enzyme for a new function, as Matti Leisola explains in Chapter 10 of his new book Heretic, there are two ways to go about it.
First, one can use what is called directed evolution. The term is a bit misleading — it’s similar to evolution, in that it involves generating millions of random mutations in the gene being studied (not a single gene, but millions of copies of that gene), and then selecting for the desired function by putting the mutated genes into bacteria and asking the resulting proteins to perform the new function.
But it’s not at all like real Darwinian evolution, in that the experimenter determines what the overall goal is (evolution has no goal) and makes a plan for how to achieve it. The experimenter determines the rate of mutation, and causes all those mutations to take place in one population of bacteria, all at once, and the screening is designed by the experimenter as well, to sort through all those millions of mutations for the desired, designed change.
Second, one can use rational design, which as the name suggests is an attempt to shift the protein’s function by studying how it works, and making modifications to the protein’s sequence. The investigator hopes these changes will make the desired functional change. Discerning which changes will work takes a great deal of knowledge about that particular enzyme’s chemistry, as well as its three-dimensional structure. But even with all that knowledge, there is no guarantee that a particular change will work as planned. Much trial and error is involved, because enzyme structure and function are quite complex.
I have used both these techniques in the lab, but I am not in Matti’s league. He is a master at modifying enzymes. And one thing he knows: getting new enzymatic function won’t happen by random mutation and natural selection. He has spent decades working in the lab to engineer enzymes and knows the challenges quite well. In Chapter 10, he lays out the evidence for why random variation and natural selection, or gene duplication and neutral evolution to put it in other terms, won’t work.
First, folded functional proteins are extremely rare in sequence space. This has been shown by several labs, by measuring the mutational sensitivity of different enzymes, and using that to estimate what fraction of all possible sequences might be able to carry out each enzyme’s function. Second, there is evidence that shows that getting a completely new protein from an existing one “is unlikely to occur by evolution through a route of folded intermediate sequences.” Further, even an enzyme with similar structure but no shared chemistry with another enzyme cannot be converted to carry out the function of that seemingly related enzyme.
Part of the problem is that the combinatorial problem is so great. Matti describes attending a PhD defense for a student who had been unable to accomplish a particular protein re-engineering problem using either directed evolution or rational design. Matti asked him if he had calculated the odds of finding the right combination of mutations when only two or three specific changes were required. Since the student had not, Matti suggested that the odds exceeded twenty million to one. Put another way, Marci Reeves, Doug Axe and I have calculated that getting four mutations would take 10^15 years for E. coli, using generous estimates for population size, mutation rate, doubling time, etc.
It is true that in the lab, protein engineers can improve thermal stability, change pH stability, improve side reactions and increase enzyme activity by random walks. But all of this involves a great deal of investigator involvement. And there are limitations. There can be no leaps where three or more mutations must happen at once. There must be a selectable path the whole way. And one must be able to create enough mutants and screen them quickly to have a hope of finding the very rare ones that work.
It is possible to engineer a new pathway into organisms, that is if one chooses an organism that can tolerate the changes introduced. The metabolism as a whole can be thrown out of balance if a new activity is introduced without the proper care. Matti describes how his lab successfully created a strain of yeast that was able to produce xylitol, for which they received a patent. But it required two new enzymes to be introduced, and much labor and planning on the part of the scientists. Evolution does not have the ability to plan. It has no foresight.
It is also true that occasionally a spontaneous mutation will occur that will confer a new ability on an organism. One example was when Aerobacter aerogenes developed the ability to use xylitol as a carbon source. It turned out this was not the development of a new enzyme ability, but rather the breaking of a regulatory element, which resulted in the constitutive turning on of an enzyme already in existence. Similarly, in Lenski’s lab E. coli developed the ability to use citrate as a carbon source as the result of a DNA rearrangement. That also proved to be the turning on of a gene that was normally shut off. Finally, the development of the enzyme nylonase proved not to be the result of a frameshift, producing a whole new enzyme, but of two point mutations introduced into an existing enzyme, which improved its ability to digest nylon byproducts. Do these count as innovations? Not if by innovation you mean making something new, that did not exist before. The reconfiguring of existing things by breaking regulatory elements or rearranging DNA or changing substrate preferences does not constitute innovation, at least the kind of innovation needed to explain the diversity of enzymes that exist. No new enzymes were made.
We didn’t get here from the proverbial warm little pond by not making new things. A lot of new things were required, including 1221 new protein folds (at least as of now in SCOP). That’s 2008 superfamilies of proteins, or 4851 families. Where did they come from? According to Dan Tawfik at the Weizmann Institute, “What we lack is a hypothesis for the earlier stages, where you don’t have this spectrum of enzymatic activities, active sites and folds from which selection can identify starting points. Evolution has this catch-22: Nothing evolves unless it already exists.” In other words, according to Tawfik, the origin of proteins is “something like close to a miracle.”