The research is published in the journal Science.
"Just as there are many alternate routes one might follow in driving from Boston to New York, one intrinsic property of DNA is that very many distinct mutational paths link any two variants of a gene," says lead author Daniel M. Weinreich, a research associate in Harvard's Department of Organismic and Evolutionary Biology. "Although this fact has been recognized for at least 35 years, its implications for evolution by natural selection have remained unexplored. Specifically, it is of great interest to determine whether natural selection regards these many mutational paths equivalently."
Weinreich and colleagues generated a series of mutants found along all 120 possible mutational trajectories involving the gene coding for the enzyme beta-lactamase, which in altered form can serve to inactivate antibiotics including penicillin and cefotaxime. Analyzing how well each variant protected host Escherichia coli cells against treatment with various concentrations of antibiotic, the scientists found that only a very small fraction of these pathways confer ever-increasing resistance in pathogenic microbes, and are therefore relevant to natural selection.
Resistance-granting mutations of beta-lactamase occur in a five-step process, with the 120 possible mutational paths representing all the possible ways in which these five point mutations can occur. Fully 102 of the 120 trajectories are inaccessible to natural selection because they create intermediates that are no more fit than the original gene, and of the remaining 18 Weinreich and colleagues observed that only about half actually had a significant probability of evolutionary occurrence.
"To be followed by an evolving population, natural selection requires that antibiotic resistance increase with each mutation," Weinreich says. "In contrast, most mutational paths of the enzymatic variant we examined fail to continuously increase resistance. Importantly, this is not a reflection of the fact that many more mutations reduce biological function than improve it, because in the present case each mutational path is composed exclusively of mutations known jointly to improve resistance."
Weinreich argues that this finding likely applies to most protein evolution, not just the beta-lactamase enzyme.Although many mutational paths lead to favored variants, only a very small fraction are likely to result in continuously improved fitness and therefore be relevant to the process of natural selection.
Weinreich's co-authors on the Science paper are Nigel F. Delaney, now at the Scripps Institute of Oceanography in La Jolla, Calif., and Mark A. DePristo and Daniel L. Hartl in Harvard's Department of Organismic and Evolutionary Biology. The work was funded by the National Science Foundation and the Damon Runyon Cancer Research Foundation.