Genomic structures of viral agents in relation to the biosynthesis of selenoproteins

The genomes of both bacteria and eukaryotic organisms are known to encode selenoproteins, using the UGA codon for selenocysteine (SeC), and a complex cotranslational mechanism for SeC incorporation into polypeptide chains, involving RNA stem-loop structures. These common features and similar codon usage strongly suggest that this is an ancient evolutionary development. However, the possibility that some viruses might also encode selenoproteins remained unexplored until recently. Based on an analysis of the genomic structure of the human immunodeficiency virus HIV-1, we demonstrated that several regions overlapping known HIV genes have the potential to encode selenoproteins (Taylor et al. [31], J. Med. Chem. 37, 2637-2654 [1994]). This is provocative in the light of overwhelming evidence of a role for oxidative stress in AIDS pathogenesis, and the fact that a number of viral diseases have been linked to selenium (Se) deficiency, either in humans or by in vitro and animal studies. These include HIV-AIDS, hepatitis B Linked to liver disease and cancer, Coxsackie virus B3, Keshan disease, and the mouse mammary tumor virus (MMTV), against which Se is a potent chemoprotective agent. There are also established biochemical mechanisms whereby extreme Se deficiency can induce a proclotting or hemorrhagic effect, suggesting that hemorrhagic fever viruses should also be examined for potential virally encoded selenoproteins. Ln addition to the RNA stem-loop structures required for SeC insertion at UGA codons, genomic structural features that may be required for selenoprotein synthesis can also include ribosomal frameshift sites and RNA pseudoknots if the potential selenoprotein module overlaps with another gene, which may prove to be the rule rather than the exception in viruses. One such pseudoknot that we predicted in HIV-1 has now been verified experimentally; a similar structure can be demonstrated in precisely the same location in the reverse transcriptase coding region of hepatitis B virus. Significant new findings reported here include the existence of highly distinctive glutathione peroxidase (GSH-Px)-related sequences in Coxsackie B viruses, new theoretical data related to a previously proposed potential selenoprotein gene overlapping the HIV protease coding region, and further evidence in support of a novel frameshift site in the HIV nef gene associated with a well-conserved UGA codon in the -1 reading frame.