Retrovirus variation and evolution

Sixty-seven years ago in Toronto at the 39th Annual Meeting of the American Society of Naturalists, H. J. Muller (1922) in a theoretical talk entitled Variation due to Change in the Individual Gene discussed the newly described d'HCrelle bodies, which he stated fulfills our definition of a gene, and Gratia's finding that these bodies could change while still retaining self-propagable nature. Muller then speculated that if these d'Herelle bodies were really genes, fundamentally like our chromosome genes, they would give us an utterly new angle from which to attack the gene problem. Now of course, we call d'HCrelle bodies phage or viruses, and we appreciate the prescience of Muller's predictions. In my article, by contrast to Muller, I am going to reverse this prediction and discuss experiments on genetic processes in certain viruses, not to find out more about the nature of the genetic processes per se, but to find out more about the viruses themselves. The viruses I shall discuss are retroviruses, RNA viruses with a DNA replicative intermediate. Thus, retroviruses have RNA and DNA genomes at different times in their life cycles. In their replication, retroviruses use the cellular machinery for transcription of RNA in going from the DNA phase to the RNA phase, and viral reverse transcriptase to go from the RNA phase to the DNA phase. Reverse transcription is also used by other genetic elements that code for reverse transcriptase and also by sequences that do not code for reverse transcriptase (processed or cDNA genes). The elements that code for reverse transcriptase include retroviruses, hepadnaviruses (the human DNA virus that causes hepatitis B and its relatives), caulimoviruses (the plant DNA virus cauliflower mosaic virus and its relatives), retrotransposons (cellular movable genetic elements with long terminal repeats), and retroposons (cellular movable genetic elements that do not have long terminal repeats) (Fig. I). Thus, there are many different genetic elements that code for reverse transcriptase, and there must have been a great deal of evolution involved in their formation. Recently, an extensive phylogenetic tree of the retroviruses and some retrovirus-related elements was published (McClure et al. 1988). This phylogenetic tree shows the extent of relationships among retroviruses and retrovirus-related elements and various postulated recombinational and gene conversion events in their evolution. McClure et al. state that the differences between one of the proteins, the acid protease, of human immunodeficiency virus (HIV) and visna virus (a virus of sheep) are as great as the differences between the acid proteases of fungi and mammals. They further state that the acid proteases of human T-cell leukemia virus-1 and human immunodeficiency virus are as divergent as the acid proteases of a typical mammal and of a hypothetical prokaryotic ancestor. There is less than 50% nucleotide sequence identity between HIV1 and HIV-2. Recently, phylogenetic trees have been published for these human immunodeficiency viruses. They show a very extensive evolution in apparently a very short period of time, possibly only 40 years for the divergence of HIV-1 and HIV-2 from a common ancestor (for example, see Smith et al. 1988). My laboratory has taken advantage of new developments in genetic engineering of retroviruses to determine the rate of different mutations during retrovirus replication in order to gain a basis for understanding the evolution of these and other elements that use reverse transcription. So far the experiments have been carried out in only a limited fashion with one subfamily of retroviruses, but the results appear to be consistent with other observations. The oncoretrovirus life cycle involves the following steps: (I) attachment of virus particles to specific cellular receptors; (2) entrance of virus into a cell; (3) reverse transcription of the input viral RNA genome in the cytoplasm to form an unintegrated viral DNA genome; (4) transport of the unintegrated viral DNA genome to the cell nucleus; (5) integration of the viral DNA into host cell DNA; (6) transcription of the integrated viral DNA; (7) splicing of some of the viral RNA transcripts; (8) transport of some of the full length and spliced viral RNAs to the cytoplasm; (9) translation of the viral mRNAs and processing of the protein products; (10) packaging or encapsidation of some of the full length viral RNA; and (1 1) budding of the progeny virus particles from places on the cell plasma membrane modified by the insertion of viral glycoproteins. Some of these steps are carried out exclusively by viral proteins (for example, reverse transcription and packaging), while others involve cellular proteins (for example, entrance and transcription). Mutational, especially deletional, analysis has been used to define the sequences in the viral nucleic acids used for each step in this replication, and a map of the viral cis-acting sequences has been generated (Fig. 2) (Weiss et al. 1985; Varmus 1988). Almost all of the cis-acting sequences of retroviruses are at the ends of the viral nucleic acid molecules. Thus, it has been easy to make oncoretrovirus vectors (Temin 1986, Eglitis and Anderson 1988). Such vectors can range in size from 2 to 10 kbp. They can be simple or very complex. For example, a vector we have used consists of a backbone of about 1 kbp of DNA from one retrovirus with one of its long terminal repeats modified and a complete transcription unit with two tandem promoters inserted in the opposite orientation to the retrovirus sequences (Dornburg and Temin 1988). This vector consists of parts of four different viruses (spleen necrosis virus, Moloney murine leukemia virus, SV40, and herpes simplex virus type 1) as well as a gene from Escherichia coli. It forms a nonretroviral transcription unit in cellular chromosomes. Most oncoretrovirus vectors, except some of the very largest, are defective for virus replication because they do not code for retroviral proteins. However, retroviral proteins can be supplied in trans by a helper or packaging cell line, which has coding sequences for these viral proteins integrated into the cell genome. We have recently constructed such helper cell lines with separate gag-pol and env expression units using heterologous promoters and poly(A) addition sequences (Fig. 3). These expression units have almost no nucleotide