Response

Summary  The posited universal cause of codon disappearance, extreme genomic % GC, is apparently not required for codon reassignment. Rigorously interpreted, this does not eliminate codon disappearance as a contributor to reassignment, but it implies that all reassignments are unlikely to be so caused. Even more significantly, the rationale for the axiom that reassigned codons must first disappear has been eliminated. That is, it has been asserted that a codon with two meanings would be lethal (Osawa and Jukes 1989). A complete inability to distinguish serine and leucine is of course lethal. However, a state of reduced ambiguity in which CUG means both serine and leucine not only stably exists in wild-type organisms in which leucine-to-serine reassignment has occurred, but such ambiguity may even have a favorable, rather than a lethal, phenotype. The potential list of possible ambiguous intermediates has been expanded by the discovery of the multiple amino acid specificity of Candida Ser- and Leu-tRNA^Ser (Suzuki et al. 1997). Other means of making codons ambiguous, such as ribosomal ambiguity or unusual concentrations or sequences of particular tRNAs, are easily envisioned. We look forward to further fossil ambiguous states and further elucidation of their phenotypes. From such data we may ultimately be able to deduce the forces that occasionally drive modern codons from one meaning to another.