Logo IMG


Ode to the Code

Brian Hayes


The idea that the genetic code is evolving under pressure to ameliorate errors—or indeed that it is evolving at all—has not won universal assent. Some cogent objections were set forth as early as 1967 by Carl R. Woese of the University of Illinois at Urbana-Champaign. Among other points, he noted that if a trait is actively evolving, you would expect to see some variation. In particular he called attention to the various "extremophiles" that live at high temperature, high salt concentration, and so on. These organisms tend to have unusual proteins and unusual nucleic acids, but they all have the standard genetic code.

The few variant codes known in protozoa and organelles are thought to be offshoots of the standard code, but there is no evidence that the changes to the codon table offer any adaptive advantage. In fact, Freeland, Knight, Landweber and Hurst found that the variants are inferior or at best equal to the standard code. It seems hard to account for these facts without retreating at least part of the way back to the frozen-accident theory, conceding that the code was subject to change only in a former age of miracles, which we'll never see again in the modern world.

Another challenge to the error-reduction hypothesis is the difficulty of showing causation in an evolutionary context. Even if the pattern of codon assignments is consistent with such a mechanism, the same pattern might have arisen in some other way.

Computer experiments like Alff-Steinberger's and Freeland's reveal nothing about pathways of evolution. A program churning out a million random genetic codes is not what you expect to see in nature. To simulate the step-by-step process of mutation and selection is much more demanding; after all, the biosphere has been working at it for a few billion years. Nevertheless, models of this kind are being attempted. Guy Sella and David H. Ardell of Stanford University are running a simulation that includes both a nucleic acid genotype and a protein phenotype, linked by a mutable genetic code. They point out that change can be introduced into the genetic code without utterly disrupting cell metabolism if there are multiple codons for a given amino acid, and some of them fall into disuse; these rarely used codons are then free to take on new roles. The mechanism is analogous to the gene duplication that often precedes evolutionary divergence of proteins: One copy of the gene carries on the original function, allowing the other to explore new territory. Thus degeneracy or redundancy is not just an accidental feature of the code but is necessary to allow scope for evolution.

comments powered by Disqus


Subscribe to American Scientist