The evolution of transcription-initiation sites.

Unlike the situation in prokaryotes, most eukaryotic messenger RNAs contain a moderately long 5' untranslated region (UTR). Such leader sequences impose a burden on eukaryotic genes by providing substrate for the mutational origin of premature translation-initiation codons, which generally result in defective proteins. To gain an insight into the expansion of 5' UTRs in eukaryotic genomes, we present a simple null model in which the evolution of transcription-initiation sites is entirely driven by the stochastic mutational flux of core-promoter sequences and premature translation-initiation codons. This model yields results consistent with a variety of heretofore disconnected observations, including the form of length distributions of 5' UTRs, the relatively low variance in UTR features among distantly related eukaryotes, the universal reliance on relatively simple core-promoter sequences, and the elevated density of introns in the 5' UTR. We suggest that the reduced effective population sizes of most eukaryotes impose a population-genetic environment conducive to the movement of core promoters to random positions, subject to the constraint imposed by the upstream accumulation of premature translation-initiation codons. If this hypothesis is correct, then selection for gene-specific regulatory features need not be invoked to explain either the origin of lengthy eukaryotic 5' UTRs or the 1,000-fold range of 5'-UTR lengths among genes within species. Nevertheless, once permanently established, expanded 5' UTRs may have provided a novel substrate for the evolution of mechanisms for posttranscriptional regulation of eukaryotic gene expression. These results provide a further example of how an increase in the power of random genetic drift can passively promote the evolution of forms of gene architecture that ultimately facilitate the evolution of organismal complexity.