A BASIS FOR NEW APPROACHES TO THE CHEMOTHERAPY OF AIDS - NOVEL GENES IN HIV-1 POTENTIALLY ENCODE SELENOPROTEINS EXPRESSED BY RIBOSOMAL FRAMESHIFTING AND TERMINATION SUPPRESSION

Several previously unnoticed genes in the human immunodeficiency virus type 1 (HIV-1), potentially encoding selenoproteins, have been discovered by analyzing the genomic RNA structure and its relation to novel open reading frames. We have found a number of new potential RNA pseudoknots, including one in the long terminal repeat, several that coincide with highly conserved enzyme active site sequences in the pot coding region, and one in the env coding region. These pseudoknots can potentially direct the synthesis of selenocysteine (SeC) containing -1 frameshift fusion proteins. This is possible because we have found potential SeC insertion sequences (SECIS) in the RNA of HIV and other retroviruses; such structures are known to be necessary and sufficient for the incorporation of SeC at UGA ''stop'' codons anywhere in a eukaryotic mRNA. In several locations, UGA codons in the -1 reading frame are highly conserved across a broad spectrum of primate immunodeficiency viruses. Due to the degeneracy of the genetic code, this conservation cannot be explained by evolutionary selection of the pol gene protein sequence alone. Such observations, combined with the conservation of the associated reading frames, strongly suggest that these are real genes, and thus that the pseudoknots are also real. A protease pseudoknot-directed -1 frameshift fusion protein contains a highly conserved SeC codon and has significant similarities to a number of DNA binding proteins, including papillomavirus E2 proteins, suggesting it may be a virally encoded repressor of HIV transcription when cleaved by protease from the rest of the gag-pol gene product. A reverse transcriptase (RT) frameshift fusion protein replaces the RT active site with a highly conserved SeC-containing module. An integrase frameshift, fusion protein contains the N-terminal integrase DNA-binding domain and a potential ATP-binding ''GKS'' motif; it has significant similarities to several helicases, but no SeC codons. A potential frameshift fusion protein from env has one SeC codon, but not in a highly conserved position. SeC incorporation could extend the nef gene product by 33 residues through the C-terminal UGA codon without frameshifting, potentially leading to substantial SeC utilization in infected cells. Significantly, a characteristic decline in plasma Se has been observed in ARC and AIDS patients; this can contribute to the pathology of AIDS, because Se is a constituent of glutathione peroxidase and one of the enzymes involved in thyroid T3 hormone synthesis. Our results suggest the possibility that the symptoms of progressive Se depletion in AIDS (e.g., impairment of antioxidant status) are not only due to malabsorption of Se, but also due to sequestration of Se in viral proteins, particularly within infected cells. We have also found potential SECIS elements encoded in the mRNA of other viruses, e.g., polio and coxsackie B, which is inhibited by Se compounds in vivo. Independent of questions raised regarding the possible roles of Se, these potential novel gene products should offer various opportunities for new approaches to anti-HIV therapy.