The DNA Provirus: Howard Temin's Scientific Legacy

In 1958, Howard Temin and Harry Rubin developed the first quantitative assay for the transformation of cultured cells by tumor viruses. Prior to their work, Rous sarcoma virus had become the first generally accepted tumor virus because of its ability to cause sarcomas following inoculation of chickens. In order to address the fundamental questions of tumor virology, an in vitro assay that allowed virus-induced transformation to be analyzed and quantitated under controlled and reproducible experimental conditions was needed. This was provided by Temin and Rubin, whose assay for cell transformation by Rous sarcoma virus remains the standard assay used in studies of transformation induced not only by tumor viruses but also by chemicals, radiation, and isolated oncogenes of both viral and cellular origins. By permitting quantitative studies of virus replication, the focus assay provided the fundamental tool needed for analysis of the virus life cycle, including Temin’s subsequent experiments from which the provirus hypothesis was derived. The availability of the focus assay also allowed other workers to isolate Rous sarcoma mutants defective in transformation, leading to the identification of the src oncogene. The focus assay has similarly been applicable to studies not only of other tumor viruses but also of cellular oncogenes. Developed over 35 years ago, Howard Temin’s assay for Rous sarcoma virus thus remains a standard tool in virology and molecular oncology.

The development of the focus assay provided the means for studying both cell transformation and the replication of Rous sarcoma virus (RSV). As Howard Temin proceeded with these studies, he made a series of unexpected observations indicating that the replication of RSV was fundamentally different from that of other RNA-containing viruses. On the basis of these findings, he proposed the DNA provirus hypothesis, which stated that the viral RNA was copied into DNA in infected cells. The resulting DNA provirus was then replicated and stably inherited as part of the infected cell's genome. The DNA provirus hypothesis was based on several different types of experimental evidence, which are presented in the 1964 paper. Studies of cell transformation using mutants of RSV indicated that the morphology of transformed cells was determined by genetic information from the virus. On the basis of the observations, he proposed that the viral genome was present in infected cells in a stably inherited form, which he called a provirus. Evidence that the provirus is DNA was then derived from experiments with metabolic inhibitors. Temin sought further evidence for the proposal that the provirus was a DNA copy of the viral RNA genome, by using nucleic acid hybridization to detect viral sequences in infected-cell DNA, but the sensitivities of the available techniques were limited, and the data were unconvincing. The DNA provirus hypothesis was thus proposed on the basis of genetic experiments and the effects of metabolic inhibitors.

During the 1960s, Howard Temin pioneered different routes for testing his hypothesis of the DNA provirus. In the latter part of the decade, he and David Boettiger perfected an experimental protocol to test whether or not 5'-bromodeoxyuridine (5BUdR) was incorporated into retroviral replicative intermediates. This analog of thymidine had previously been shown to be incorporated into DNA, and being brominated, it rendered the DNA that contained it more sensitive to visible light than unhalogenated DNA would be. Boettiger and Temin induced chicken cells to become stationary by withdrawing serum from their cultures. These nonproliferating cells did not efficiently incorporate 5BUdR into their chromosomal DNA and thus were not sensitive to exposure to visible light. Temin and Boettiger infected the stationary cells, treated them with 5BUdR, exposed them to increasing doses of visible light, plated the treated chicken cells on a feeder layer of rat cells resistant to infection, and provided serum. This experiment, along with its described controls, indicated that in stationary cells, infecting retroviruses do synthesize DNA intermediates that can be tagged by the 5BUdR they incorporate. This finding was the most direct proof of the existence of a DNA provirus until that proviral DNA was isolated and detected 5 years later. This finding formed the basis for later studies in which Howard Temin and his colleagues explored the structure of the replicative intermediate and found that newly infected cells must pass through mitosis for infection to be completed.

Although experiments with metabolic inhibitors continued to provide evidence that Rous sarcoma virus (RSV) replicated via a DNA provirus, it was the discovery of a viral enzyme capable of carrying out the synthesis of DNA from an RNA template that finally led to widespread acceptance of the DNA provirus hypothesis in 1970. This paper by Temin and Mizutani, published together with a similar paper by David Baltimore, unambiguously demonstrated a biochemical mechanism for synthesis of the DNA provirus. The existence of RNA-dependent DNA polymerase clearly showed that the "central dogma" could be reversed and brought widespread acceptance of the new mode of information transfer that Howard Temin had predicted in 1964. The key to detecting reverse transcriptase was looking for the enzyme in virus particles rather than in cells. In addition, Temin and Mizutani had found that formation of the RSV provirus in infected cells did not require new protein synthesis following virus infection. Moreover, the virion DNA polymerase activity was dependent upon the presence of intact viral RNA, indicating that the virion enzyme catalyzed the synthesis of DNA from an RNA template. As Temin and Mizutani concluded in their paper, these results provided "strong evidence that the DNA provirus hypothesis is correct". They also pointed out that the discovery of reverse transcriptase not only had major implications for understanding viral carcinogenesis but would also change the way it is thought about information transfer in biological systems.

Following the discovery of reverse transcriptase and widespread acceptance of the provirus hypothesis, Howard Temin turned his attention to the possible roles of RNA-directed DNA synthesis (reverse transcription) in the cells of healthy organisms. He predicted that this unique mode of information transfer was not restricted to viruses but would also be found in healthy cells, where the transfer of information from RNA to DNA might play a role both in normal development and in generating the mutations responsible for non-virus-induced carcinogenesis. These considerations formed the basis of the protovirus hypothesis, as discussed in this 1971 editorial. The central predictions of the protovirus hypothesis, the existence of reverse transcriptase and RNA->DNA information transfer in cells as well as in viruses, have been thoroughly substantiated. Reverse transcriptases not only are widely distributed in eukaryotes (including yeasts, plants, and animals) but also have been found in bacteria. In mammals, transposable elements that move via RNA->DNA information transfer (retrotransposons) constitute approximately 10% of genomic DNA. Also as predicted in the protovirus hypothesis, mutations induced by the movement of retrotransposons can contribute to the development of cancer both by activating oncogenes and by inactivating tumor suppressor genes. The retrotransposons that have been described to date appear to be parasitic self-replicating elements that transpose to random sites throughout the genome. However, much remains to be learned concerning the molecular mechanisms of development and differentiation, and a normal physiological role for RNA->DNA information transfer may still await discovery.

This chapter first reviews Howard Temin's thoughts on the origin of cancer genes, who had correctly predicted many aspects of the current understanding of the molecular alterations responsible for the development of malignant tumors. The chapter discusses some of the functions of proto-oncogenes in controlling normal cell growth and differentiation as well as the possibility that oncogene proteins provide novel targets for cancer chemotherapy-areas of current interest which represent direct outgrowths of some of Howard's seminal contributions to cancer research. All three key elements of Howard's hypothesis for the origin of cancer genes have proven correct. First, mutations play a critical role in converting protooncogenes to oncogenes as well as in inactivating tumor suppressor genes. Second, similar oncogenes are found in both strongly transforming retroviruses and nonvirus-induced tumors, including human cancers. Third, reverse transcription is involved in at least some mutations of proto-oncogenes and tumor suppressor genes in non-virus-induced tumors as well as being responsible for the incorporation of oncogenes into retroviral genomes. The discovery that oncogenes are mutated versions of normal cell genes (proto-oncogenes) focused attention on the roles of proto-oncogenes in normal cells and on the nature of the molecular alterations that convert proto-oncogenes to oncogenes. The chapter concludes with a brief consideration of the possibility that oncogenes provide novel targets against which drugs with increased selectivity for cancer cells could be designed. Posttranslational modification of Ras proteins by addition of a farnesyl isoprenoid is required for their membrane association and biological activity.

Genetic dowries contain two sorts of genes that govern the proliferation of cells: one sort provides accelerators to activate the engines of the cell, and the other sort provides brakes. In this chapter, the author briefly narrates a story rich with illustrations of how science proceeds, of what it can and cannot do. It is a story that can be accessible by even the most general reader. It is a story that does great credit to the legacy of Howard Temin. The story begins with cells, those microscopic, irreducible, living bricks from which bodies are constructed. The ability of cells to multiply lies at the root of life, but it also prefigures the baleful threat of cancer. Giant strides were made to view cancer as a disease of individual cells and to study this disease not in an animal but in a petri dish, and also to exploit viruses that rapidly and reproducibly convert cells to cancerous growth. Bodies are divided into two lineages one lineage, the somatic lineage (from "soma," for body), and another lineage, the germinal lineage or germ line. Most cancer genes arise from damage in the somatic lineage and thus affect only a single individual. The discovery and exploration of proto-oncogenes provided a new view of the cancer cell, rich in detail and prospect.

This chapter focuses on the abiding interest in the intersection of mammalian development and human cancer. In this essay, a mouse model for intestinal cancer is discussed. To answer the question of what are the molecular and biological intersections of mammalian development and human cancer, one determines what neoplastic processes occur spontaneously in a mutant that has lost the function of a single developmental gene. In human cancer genetics, this case is approximated by families in which a mutant allele inherited in heterozygous form predisposes the individual to a particular cancer syndrome. The tumor lineage has somatically lost the remaining normal allele by one of several possible mitotic mechanisms. A full exploration of the intersection with normal mammalian development of the loss-of-function neoplasms requires analysis of embryos homozygous for the mutant allele. The chapter focuses on two genes: Apc on chromosome 18, in which the Min nonsense allele has been induced by ethylnitrosourea, and the first modifier-of-Min locus, Mom-1, on chromosome. In sections of Min-induced tumors, the expression of cellular markers for three differentiated cell types of the intestinal epithelium is detected by immunohistochemistry: the Paneth cell (lysozyme), the enteroendocrine cell (serotonin), and the enterocyte (fatty acid-binding protein). The properties of the tumor can in principle be altered by changes either in the provirus or in the source of any differential growth factors.

The v-Rel oncoprotein is the transforming protein of the replication-defective avian Rev-T retrovirus, which is derived from the Rev-A replication-competent helper virus. Although specific Rel proteins can be regulated in several ways all Rel proteins are likely to be regulated by subcellular location. v-Rel and c-Rel differ in several ways. The deletion of C-terminal sequences in v-Rel has three functional consequences. First, v-Rel and C-terminally truncated forms of c-Rel are located in the nuclei of chicken fibroblasts, whereas full-length c-Rel is located in the cytoplasm, probably because these truncated Rel proteins interact less strongly with IKB-α than does full-length c-Rel. Second, the C-terminal truncation in v-Rel has removed c-Rel sequences that can act as strong transcriptional activation domains. Third, v-Rel and C-terminally truncated forms of c-Rel are highly transforming. The C-terminal deletion in v-Rel appears to be the most important mutation for activating the oncogenicity of c-Rel. Chicken spleen cells transformed in vitro by wild-type v-Rel usually grow as large multicellular clumps. It is fairly clear that most nuclear oncoproteins, including Myc, Ets, Fos, Jun, Myb, and v-Rel, need to bind to DNA and activate transcription to effect malignant transformation. There have been several reports of rearranged rel family genes associated with human lymphoid cancers. Through a recombination event, human c-rel is fused to an unknown gene termed nrg (for non-rel gene) in a cell line derived from a pre-T diffuse large-cell lymphoma, and c-rel has been amplified in two follicular large-cell lymphoma cell lines.

Cancer is a formidable challenge, and one should have no illusions that major breakthroughs in the prevention, diagnosis, and treatment of the disease will come easily. Nevertheless, several converging lines of research are providing exciting new opportunities for clinical application. Some of these areas are summarized in this chapter. The goal is to identify a few areas in which the molecular foundations of oncology are supporting ideas that are likely to have clinical ramifications in the near future. The molecular dissection of colon cancer continues to provide new insights into the cascade of genetic events that lead to full-blown malignant transformation. Microsatellite alterations are associated with small-cell lung cancer, head and neck tumors, bladder cancer, and most likely other tumors as well. Diseases targeted for innovative trials of cytokine gene-transfected tumor cells include breast, ovarian, renal, and colon cancers, and development of their use against melanoma continues. The suicide vector model serves as one template for further development in other cancers as well, especially those compartmentalized in relatively sequestered sites, e.g., intraperitoneal ovarian cancer or primary central nervous system lymphoma, an opportunistic cancer associated with AIDS. Molecular medicine continues to uncover the broad principles that govern the interplay between internal and external factors that modulate net gene expression and cellular function for many cell types. If the tradition of excellence and the balance between creativity and scientific rigor exemplified by the work of Howard Temin is maintained, progress against cancer is assuredly made.

In the paper by Shimotohno, Mizutani, and Temin, the authors describe the identification, cloning, and sequencing of the long terminal repeats (LTRs) and the abutting cellular DNA of infectious proviruses of spleen necrosis virus. The LTRs of retroviruses are direct repeats of sequences of DNA at the ends of the provirus. The 3' terminus of the 5' LTR should therefore be identical to the 3' terminus of the intact provirus; similarly, the 5' terminus of the 3' LTR should be identical to the 5' terminus of the intact provirus. A knowledge of this structure along with the determined DNA sequences allowed Howard and his colleagues to identify the junctions of proviral DNA and the cellular DNA into which the proviral DNA had integrated. They made two striking observations: the proviral DNA lost two nucleotides from each end prior to its integration, and five nucleotides of the cellular DNA were duplicated at the site of integration. These two observations contributed substantively to the basis for their hypothesis that retroviruses have evolved from cellular movable genetic elements that often share the features. Howard Temin continued to study and elucidate the mechanism of retroviral replication through genetic and biochemical dissections of the products of replication during a single cycle of infection. These studies allowed him to validate his provirus hypothesis and to reconstruct pathways by which retroviruses assimilate proto-oncogenes. The resemblance of the provirus to transposable elements indicates that retroviruses may have evolved from such elements.

This chapter focuses on the development of one's understanding of carcinogenesis mediated by avian retroviruses, to which Howard Temin contributed so much. This understanding, coupled with the associated study of retroviruses in general, has been essential to dealing effectively with human disease. It also outlines current appreciation of human tumor viruses, using Epstein-Barr virus (EBV) as a model. Human tumor viruses are less efficient pathogens than most of the well-studied avian oncogenic retroviruses. Howard Temin refined a transformation assay for Rous sarcoma virus (RSV) such that it became a standard method for detecting and measuring the transforming abilities of different viruses in cell culture. Avian leukosis viruses (ALVs) provide one example of weakly transforming viruses. These viruses formerly often infected commercial flocks of chickens, in which they were propagated either vertically or horizontally. Four human tumor viruses-EBV (a herpesvirus), hepatitis B virus (HBV), human papilloma virus types 16 (HPV-16), -18, -31, and -33, and human T-cell leukemia virus type 1 (HTLV-1; a retrovirus)-have been studied sufficiently to be considered in this chapter. A fifth, hepatitis C virus, has been identified, but virologic studies of it are only now beginning. The outcome of infection with some human tumor viruses can be profoundly affected by the immune response of the host. The four human tumor viruses (EBV, HBV, HPV, and HTLV-1 ) vary in the cell types they infect, in the times between infection and development of cancers, and in the mechanisms by which they induce and/or maintain proliferation of infected cells.

A part of a first group at University of California at San Francisco (UCSF) worked on the early events in retrovirus replication that result in the establishment of a provirus. Another part worked on aspects of proviral gene expression, especially mechanisms, such as ribosomal frameshifting, that permit synthesis of the viral enzymes required for forming a provirus. Yet another portion of the group studied the oncogenic consequences of proviral integration, including insertional activation and the capture of host proto-oncogenes. All of these components, arrayed around a schematic drawing of a provirus, revealed the depth of commitment to Teminism. The idea that insertional activation of proto-oncogenes might constitute the first step in a cancerous process initiated by retroviruses lacking viral oncogenes took flight with the demonstration that avian leukosis viruses augment expression of the c-myc proto-oncogene by nearby integration of its provirus. These findings stimulated similar work with the mouse mammary tumor virus (MMTV), resulting in the discovery of the Wnt-1 gene, the first cloned member of a large gene family known to govern early developmental events in diverse organisms. In the studies the first group has moved beyond the study of MMTV proviruses as insertional mutagens to consider more broadly the multistep nature of mammary carcinogenesis in hopes of understanding the genetic and physiological events that produce premalignant lesions, primary cancers, and metastases in human beings as well as mice.

This chapter gives an overview on the replication and maturation of the herpes simplex virus type 1 (HSV-1) genome, drawing parallels with phage systems. It provides an overview of what is known about HSV DNA replication and recombination under the sections Formation of circular DNA Intermediates, Formation of greater-than-unit-length replication intermediates, Resolution of branched recombination intermediates, and Cleavage of concatemers into unit-length virion DNA and packaging of unit-length virion DNA into capsids. The chapter concentrates on the helicase-primase (UL5, UL8, and UL52) and the origin-binding protein (UL9). The study of transdominant mutations has proven to be a powerful tool for characterizing specific regions of multifunctional proteins. Several lines of evidence indicate that overexpression of the wild-type UL9 protein can be inhibitory to viral replication. First, in trans-dominance assays, plasmids that express wild-type UL9 are somewhat inhibitory to plaque formation by wild-type virus. Second, complementing cell lines that contain high copy numbers of a UL9 expression plasmid do not efficiently support wild-type HSV-1 infection are reported. The third line of evidence comes from experiments in which the HSV replication proteins are expressed in insect cells from recombinant baculoviruses. Several lines of evidence suggest that viral genome maturation involves site-specific cleavage of viral DNA concatemers. Genetic analysis has provided important insights not only into the identification of viral proteins required in DNA replication and genome maturation but also into their roles in these complex processes.

For two decades, Howard Temin analyzed retroviral replication in detail in order to elucidate the high rate of genetic variation inherent to this process. In "Retrovirus Variation and Reverse Transcription: Abnormal Strand Transfers Result in Retrovirus Genetic Variation," he summarized his understanding of the kinds of retroviral variation and proposed a mechanism by which several of them might occur. Virologists have appreciated for many years that viruses with RNA genomes vary rapidly and that retroviruses do, too. One contribution to this high rate is thought to be inherent in retroviral RNA replicases: they lack the editing activities of DNA replicases, which serve to limit some types of errors potentially made by nucleic acid replicases. One elegant contribution to the resolution of the problem of multiple rounds of viral replication was the development and application of helper cells by Howard Temin and his colleagues. They developed cells that constitutively express the structural genes of spleen necrosis virus, a simple retrovirus. Howard, his students, and his postdoctoral researchers developed several protocols for selecting and enumerating the frequency of mutations generated in these viral vectors after one round of their replication. The identities of the mutants were revealed by sequencing their DNA after its rescue from the infected cells. Howard's hypothesis is appealing because it provides a single explanation for a variety of both simple and complex mutations known to arise in retroviruses. It also helps explain how this family of viruses can in general acquire cellular genes and in particular transduce oncogenes.

In this chapter, the author discusses two topics: the long-term association of virus and host as revealed by the study of endogenous proviruses and the short-term "evolution" that characterizes the association of some viruses, most notably human immunodeficiency virus (HIV), with their host. Ancient endogenous proviruses were inserted into the germ line of a species prior to separation of that species from related species. To understand the host-provirus relationship better, the proviruses of mice is chosen as a model, in particular the C-type proviruses, which form the largest group related to known viruses. In some species of mice, the predominant endogenous proviruses are recombinant relative to those that had been studied in laboratory mice. Infection of the germ line is a rare event, yet endogenous proviruses have many times been independently fixed in the genomes of mice and other species. A final point concerning the evolution of endogenous viruses is an interesting correlation between virus lifestyle and endogenization. Retroviruses differ considerably in the amount of genetic variation observed among individuals within species. Divergence of up to 25% within the most variable regions (in env) has been observed, even when infection was initiated with known clonal virus. These observations are consistent with the evolution of HIV in vivo into a complex quasispecies. The high rate of replication of HIV during the entire course of the infection has important consequences for understanding genetic variation and its role in evolution, pathogenesis, and therapy.

This chapter summarizes some current concepts on human retroviruses, and focuses on three topics: human T-cell lymphotrophic leukemia virus types I (HTLV-I), some approaches to the inhibition of human immunodeficiency virus (HIV) replication, and the pathogenesis of AIDS-associated Kaposi's sarcoma (KS). HTLV-I is now known to cause some adult Tcell leukemias (ATLs)/lymphomas, tropical spastic paraparesis/ HTLV-I-associated myelopathy (TSP/HAM) (a neurological disease resembling multiple sclerosis), and several apparently autoimmune disorders, including some polymyositis, rheumatoid arthritis-like disorders, uveitis, bronchitis, and mild immune impairment associated with bacterial dermatitis in infants. Most studies favor the concept of an autoimmune mechanism for the demyelinating neurological disorder TSP/HAM. Steadily (even if slowly) accumulating knowledge about the pathogenic mechanisms by which HIV causes AIDS and abundant information on the replication cycle of HIV provide new ideas for different therapeutic and preventive approaches which some believe have not yet been fully exploited or sufficiently pursued. Most testing for new anti-HIV therapy today is based on targeting the enzymes of HIV (RT, protease, integrase, and RNase H). KS remains the most important tumor associated with HIV-1 infection in terms of both frequency and the suffering it produces.

Pre-T cells migrate from the bone marrow into the thymus, where they mature. The most primitive T cells begin as double-negative thymocytes with no CD4 or CD8 on the cell surface. While characterization of the receptor genes is clearly important to understanding the selection process and other T-cell functions, the genes coding for the T-cell receptor (TcR) had eluded isolation. A technique of differential screening and molecular subtraction, favoring a subtraction approach partly because of previous use of the protocol to isolate virus-specific sequences is used. T-cell-specific genes involved in differentiation and function were simply set out to isolate. Encouraged by the results with the CD4 and CD8 knockout mice, researchers proceeded to generate mice lacking genes engaged in TcR-mediated signal transduction. From Howard Temin, the author learned that science has its own integrity and that one must work tirelessly, be innovative, and, above all, occasionally cherish an idea that does not attach itself to anything.

Development of gene therapy strategies for the treatment of human T-lymphocyte disorders, including AIDS, requires an in vivo system in which transduced human hematopoietic stem cells can be used to reconstitute the T-lymphoid compartment. Combination of HIV-1 pathology and CD34+ progenitor cell transduction taking place within a relatively short time in the SCID-hu mouse provides an experimental system in which to address many of the issues that are critical to successful gene therapy approaches for AIDS but are not easily tested in clinical trials. This chapter reviews the recent progress in the use of the SCID-hu system to further one's understanding of vector transduction and expression in modeling potential therapeutic approaches in stem cells. Most anti-AIDS gene therapy strategies are based on the assumption that direct cell killing is the primary pathogenic mechanism for HIV-1 CD4+ cell destruction. One of the key issues in human gene therapy has been the development of retroviral vectors with optimal gene expression in differentiated hematopoietic cells. Gene therapy strategies aimed at inhibiting HIV infection or replication require vectors that are expressed in cells infected by HIV, the human CD4+ T lymphocytes. Absence of an immune response in SCID-hu mice has two advantages. First, viral pathogenesis can be studied in the absence of an immune response, thereby shortening the time frame for onset of disease. Second, while it is likely that introduction of a foreign protein elicits a host immune response in patients, the effects of therapeutic reagents is assessable independent of the immune response.

This chapter discusses about an effective vaccine against human immunodeficiency virus (HIV). From the standpoint of vaccine development, it is becoming more and more apparent that HIV is like no other virus. Key features that have led to successful vaccines with other viruses appear to be missing and are replaced by ones that are not conducive to vaccine development. In the absence of natural immunity, one must confront several serious obstacles: (i) correlates of immunity become difficult to establish, (ii) the rationale for live attenuated or even whole inactivate vaccines is weakened because of concerns for safety, (iii) the specter that all immune responses to the pathogen may not necessarily be salutary must be resolved, and (iv) the need to better understand virulence and how to overcome it becomes paramount. Thus, the empiricism that historically has been so dominant in development of vaccines against viruses gives way to a concerted effort to understand the fine details of infection and pathogenesis and how these are balanced with the ensuing host responses. When faced with such obstacles, vaccine developers have sometimes turned to animal models. Several independent studies had demonstrated that recombinant envelope products were effective in preventing HIV infection in chimpanzees and that antibodies were the best correlate of protection. In any event, it is now evident that a great deal of momentum is required to drive an HIV vaccine to the all-important milestone of an efficacy trial. The chapter ends with Howard Temin's contribution to work on HIV vaccines.

From the mid-1980s, Howard Temin not only was keenly aware of the threat to public health posed by the expanding AIDS epidemic but also acted to limit that threat. He tried to minimize the personal and scientific conflicts among early workers on human immunodeficiency virus (HIV) so as not to obscure the national and international problem this virus presents. He also sought to contribute scientifically to the resolution of this problem, as evidenced by his "Proposal for a New Approach to a Preventive Vaccine against Human Immunodeficiency Virus Type 1" and his experimental pursuit of that proposal. In this article, Howard Temin built on his appreciation of simple and complex retroviruses to formulate a hypothesis, to develop a test of that hypothesis, and to propose an application of the hypothesis, if it were proven correct, in developing a vaccine against HIV. Howard's proposal is not conventional. It does not build on forced passage of HIV in cells in culture, an approach that seems unwise in view of this virus's capacity to generate variants in vivo. It is based on his novel biological insight. More important, it provides an approach that can be tested in animal models and thereby yields one rational approach to the development of a vaccine for HIV and eventual prevention of AIDS.