Until recently, simplicity and uniformity of design appeared to be universal features of retroviruses. Most retroviruses thrive with a few open reading frames in their genomes, all devoted to major constituents of virus particles: core proteins, envelope glycoproteins, and virionassociated catalytic proteins (reverse transcriptase, integrase, and protease). In the life cycle of the usual retrovirus, there is a straightforward division of labors: early events depend largely upon viral reverse transcriptase and integrase, whereas in the later stages components provided mainly by the host are harnessed for production of viral proteins and progeny particles (Varmus and Swanstrom 1982, 1985).

Over the past few years, close study of two types of human retroviruses, the T-cell leukemia viruses (HTLVI and-II) and the immunodeficiency viruses (HIV-1 and-2), has uncovered a multiplicity of unexpected viral regulatory functions, now forcing a revision of this traditional view of the retrovirus life cycle. The newly discovered functions are fascinating mechanistically and mysterious evolutionarily. Moreover, they offer promise of understanding some perplexing physiological aspects of infection with major clinical consequences—latency and reactivation of HIV and inefficient leukemogenesis by HTLV.