The Length Evolution of 5´ Untranslated Regions - The Stochastic Model Revisited

The length evolution of 5´ untranslated regions (5´UTRs) is a fundamental issue in molecular evolution. 5´UTRs are important for both transcriptional and translational regulations. These regulatory regions are suggested to elongate with increased organismal complexity and decreased effective population size. However, the factors that determine the lengths of 5´UTRs remain elusive. A stochastic model has been proposed to explain the length evolution of 5´UTRs. This model posits that the lengths of 5´UTRs change due to the stochastic process of gains/losses of the transcription initiation signals, with upstream start codons (uAUGs, a translation repressor signal) as the major constraint to suppress 5´UTR elongation. This model predicts that species with larger effective population sizes have shorter 5´UTRs because the high efficiency of selection can effectively restrain deleterious elongation of 5´UTRs in these species. However, here we demonstrate that Drosophila melanogaster has significantly longer 5´UTRs than mouse, chicken, frog, and zebrafish despite the insect´s very large effective population size. We further show that the exceptionally long 5´UTRs in D. melanogaster do not result from relaxed selection pressure on uAUGs, biased nucleotide composition, or lineage-specific gains/losses of genes. Interestingly, two other insect species-honeybee and mosquito also have comparably long 5´UTRs, which are longer than those of other invertebrates. Therefore, the stochastic model cannot explain the differences in 5´UTR length across species. Rather, lineage-specific factors may play an important role in this regard. The generality of the stochastic model thus needs to be re-considered.