Internal ribosomal entry site lacks secondary structure

The search for mechanisms of translational regulation has yielded many experimentally identified internal ribosome entry sites (IRES). Because of the lack of sequence similarity among the experimentally IRESs, it is widely assumed that IRESs posses stable secondary structure allowing them to interact with the components of the translation machinery. Contrary to this view, here we show that IRES activity in nine yeast IRESs, mapped to 60 nt immediately upstream of the initiation AUG, is strongly associated with the lack of secondary structure of IRESs. Furthermore, the reverse complements of these IRESs, with their secondary structure more stable than those of the IRESs, exhibit little IRES activity. The generality of this association is exemplified by the observation that, in the natural _vpu-env_ bicistronic mRNA in HIV-1, the mRNA segment (60 nt) immediately upstream of the initiation AUG of _env_ has the weakest secondary structure among all dominant HIV-1 mRNA species. These results suggest a unified model of alternative translation initiation