Reduced mRNA secondary-structure stability near the start codon indicates functional genes in prokaryotes

Several recent studies have found that selection acts on synonymous mutations at the beginning of genes to reduce mRNA secondary-structure stability, presumably to aid in translation initiation. This observation suggests that a metric of relative mRNA secondary-structure stability, Z(ΔG), could be used to test whether putative genes are likely to be functionally important. Using the Escherichia coli genome, we compared the mean Z(ΔG) of genes with known functions, genes with known orthologs, genes where function and orthology are unknown, and pseudogenes. Genes in the first two categories demonstrated similar levels of selection for reduced stability (increased Z(ΔG)), whereas for pseudogenes stability did not differ from our null expectation. Surprisingly, genes where function and orthology were unknown were also not different from the null expectation, suggesting that many of these open reading frames are not functionally important. We extended our analysis by constructing a Bayesian phylogenetic mixed model based on data from 145 prokaryotic genomes. As in E. coli, genes with no known function had consistently lower Z(ΔG), even though we expect that many of the currently unannotated genes will ultimately have their functional utility discovered. Our findings suggest that functional genes tend to evolve increased Z(ΔG), whereas nonfunctional ones do not. Therefore, Z(ΔG) may be a useful metric for identifying genes of potentially important function and could be used to target genes for further functional study.