Molecular evolution in the gnd locus of Salmonella enterica

The gnd gene, the structural gene for 6-phosphogluconate dehydrogenase, was sequenced and analyzed in 34 isolates from different serovars of the seven subspecies of Salmonella enterica to provide comparative information on the evolution in this gene, which has been studied extensively in Escherichia coli. The gene tree obtained by the neighbor- joining method in general gave separate branches for each subspecies, with the few exceptions readily explained by recombination. There is evidence of recombination involving transfer of long (more than 400 bp) and short (30-150 bp) segments of DNA. Four of the six long-segment transfers detected are at the 5' end of the gene, and in all four cases a variant of the chi sequence is located close to the recombination junction and appears to have mediated the recombination events. We suggest that in these four cases and in a fifth case with intersubspecies transfer of the whole gnd gene, the adjacent rfb (O antigen) locus may have been transferred in the same event. The estimates of the number of synonymous substitutions per synonymous site, KS, and the number of nonsynonymous substitutions per nonsynonymous site, KA, within the E. coli and S. enterica gnd genes, and also between the two species show an interesting distribution, with KS being lower toward the ends of the gene and KA in particular being lower in the first than in the second domain. In S. enterica, synonymous sites also seem to be subjected to negative selection. The ratio of KA to KS was higher within S. enterica and E. coli than between them, which may indicate that intraspecies variation is essentially between clones and that mildly deleterious mutations can be fixed within clones, which would thus raise KA within species.