Identification of a Retroviral Transforming Gene

Our current understanding of the molecular origins of cancer derives from many sources, such as the recognition that many chemical carcinogens act as mutagens, the discovery of tumor suppressor genes inactivated in familial cancers, and the identification of tumor virus genes that can cause the alteration to malignancy ("transformation") (see GENES 2000: 29.4 Transforming viruses carry oncogenes). Here I will focus on one of these threads, the identification of the transforming (cancer-causing) gene of Rous sarcoma virus (RSV). I first review the work on RSV that suggested that its genome might include a gene responsible for transformation. I then describe the experiments that led to the identification of the src gene, the RSV gene responsible for the induction and maintenance of transformation. Finally I briefly describe how work on src has enriched our understanding of signaling in normal cells, the mechanism of malignant transformation, and the role of genetic change in human cancer.

Background

In 1908 Ellerman and Bang reported that an avian leukosis could be transmitted by a filterable agent, that is, by a virus (Ellerman and Bang, 1908). Although this report now stands as the first description of a tumor virus, at that time leukemia was not regarded as a form of cancer, and their paper did not arouse great interest or opposition. Three years later Peyton Rous described the discovery of the virus that is now called Rous sarcoma virus (Rous, 1911). Rous’s report that a virus could induce "authentic" cancers met with considerable skepticism, because cancers were believed to be of local origin, and not dependent on infection: one oncologist told him that "this can’t be cancer, because you know its cause". Many argued that virus production was a consequence of tumor growth, and not its cause. Thus in 1928 Boycott (Boycott, 1928) wrote that "all the evidence seems to concur in indicating that the Rous virus arises de novo in each tumor." Although RSV does indeed induce tumors upon injection into susceptible birds, it does not spread by infection in natural populations, and there was a germ of truth in Boycott’s statement. Thanks to the work of Bishop and Varmus, described in the following essay, we now know that transforming viruses do indeed arise "de novo " by recombination between viral and cellular genomes. In the case of RSV, this recombination event presumably occurred in the tumor from which Peyton Rous first isolated the virus. In any event Boycott’s comment predated an understanding of the origin of retroviral transforming genes by some fifty years.

And so for several decades these avian viruses were regarded as curiosities with no relevance to mammalian cancer. But by the 1950s it had become apparent that a variety of viruses could cause tumors in susceptible animal hosts. Renato Dulbecco and his colleagues at Caltech realized that animal tumor viruses could provide an entry to an understanding of cancer at the molecular level. The first challenge was the development of an in vitro system in which the mechanism of malignant change could be studied outside the animal host. There were already several reports that chicken embryo fibroblasts (CEF) could be morphologically altered by RSV (Halberstaedter, Doljanski, and Tenenbaum, 1941; Lo, Gey, and Shapras, 1955; Manaker and Groupé, 1956). This morphological alteration was used by Temin and Rubin (Temin and Rubin, 1958) as the basis for the focus assay, which formed the foundation for all subsequent studies on transformation by RSV. In this assay, infection of a single cell by a single infectious particle results in the formation - by cell division and successive rounds of infection - of a cluster or "focus" of morphologically distinct cells. Cells from these foci produce tumors in vivo and are said to have undergone malignant transformation, or simply, "transformation". In cell culture they display altered growth properties, such as the ability to grow independently of anchorage when suspended in a semi-solid agar medium.

A key issue in those early days was whether the tumor viruses were perpetuated with the transformed cells, and if so, how the viral genome was maintained. A model was provided by the phenomenon of bacterial lysogeny, characterized by André Lwoff and his colleagues in the 1940s and 1950s. As early as 1955, Harry Rubin showed that each cell in a Rous sarcoma virus-induced tumor could release infectious virus (Rubin, 1955). Moreover, in contrast to the situation observed with temperate (lysogenic) bacteriophages, the virus-producing tumor cells survived. He therefore suggested that "The virus plays a direct and continuing role in perpetuating the cell in its malignant state." In the mid-1960s Temin made the controversial proposal that the viral genome is perpetuated as an integrated DNA (a provirus) (Temin, 1964); however the physical demonstration of integrated viral genomes in transformed cells, and an understanding of how these genomes are generated and integrated, came only much later, and is a separate story.

The finding that tumor viruses were permanently associated with transformed cells raised the possibility that transformation might result from the expression of viral gene products. In the case of Rous sarcoma virus, the first key observation was made by Howard Temin in 1960. He showed that a mutant of Rous sarcoma virus could cause the production of morphologically distinct, fusiform or spindle-shaped cells, readily distinguishable from the rounded cells that resulted from infection by wild-type virus (Temin, 1960). Temin therefore concluded that the morphology of the transformed cell is controlled by a genetic property of the virus.

Replication and Transformation Properties of RSV are Separable

During the decade that followed the biology of Rous sarcoma virus was intensively investigated. Two important findings emerged from work in the laboratories of Harry Rubin, Howard Temin, Peter Vogt and Saburo and Teruko Hanafusa, and are illustrated in Figure 1. First, Rubin and Vogt isolated replication-competent viruses (the RIFs, or Rous-interfering factors, and RAVs, or Rous-associated viruses) that were clearly related to RSV, but did not transform the infected cells (Rubin and Vogt, 1962). This finding again raised the possibility that the ability to transform was a specific genetic property of Rous sarcoma virus that distinguished it from its relatives, and was independent of the replication cycle of the virus. Second, one strain of Rous sarcoma virus, the Bryan strain, proved to be replication-defective (Hanafusa and T. Hanafusa, 1963). Cells infected with the Bryan virus became transformed but, as a result of the viral replication defect, yielded only non-infectious viral particles. Secondary "superinfection" of these transformed cells with a non-transforming "helper virus" (such as one of the Rous-associated viruses) could, however, rescue virus production in these cells. Thus the production of infectious virus was not necessary for transformation. Taken together, these two findings suggested that virus replication and malignant transformation might be separable genetic properties of Rous sarcoma virus.

Clearly the definitive way to test the idea that transforming ability was a distinct genetic function of the virus would be to isolate transformation-defective mutants of RSV. Two strains of RSV, the Schmidt-Ruppin and Prague strains, had been shown to be replication-competent (that is, non-defective). These virus strains were evidently good substrates for genetic analysis, because clonal stocks of mutant viruses could be readily isolated and propagated. One approach would be to isolate mutants non-conditionally defective for transformation, which would biologically resemble the RIFs and RAVs, and such mutants were identified by Alice Goldé (Golde, 1970) and by Toyoshima, Friis and Vogt (Toyoshima, Friis, and Vogt, 1970). A second approach would be to look for temperature-sensitive (ts ) mutants. One advantage of such mutants is that they allow the mutant function to be switched on and off by temperature-shifts. The systematic use of ts mutants had been pioneered by Edgar in studies on the replication cycle of bacteriophage T4 (Epstein, 1963). Moreover Mike Fried had isolated a temperature-sensitive mutant of polyoma (a DNA tumor virus) that was unable to initiate transformation at the non-permissive temperature (Fried, 1965). I was familiar with the utility of ts mutants from my graduate work in Sydney Brenner’s lab. So when I moved to Harry Rubin’s lab in 1968, the isolation of ts mutants seemed like a plausible strategy to identify a transforming function of RSV. Peter Vogt’s laboratory was also looking for ts mutants, and in 1969 reported the isolation of two temperature-sensitive mutants of the avian sarcoma virus B77 (Toyoshima and Vogt, 1969). These mutants however were defective in virus replication, and thus did not define an independent transforming function.

The experiment

To isolate ts transformation-defective mutants of RSV, I subjected a stock of Schmidt-Ruppin RSV to mutagenesis with the mutagen N-methyl-N’-nitro-N-nitrosoguanidine ("nitrosoguanidine"). The survival of infectious progeny was about 10-3. (I chose to use this "withering dose of chemical mutagen" (Bishop, 1985) because I was aware that much of the decrease in infectivity would be due, not to mutations, but to interaction of the mutagen with protein components of the virus particle or to modifications of the genome that directly block replication; in any event I never determined whether this drastic mutagenesis procedure was necessary for the isolation of mutants). A simple screen sufficed to examine surviving virus for temperature-sensitive mutants. The mutagenized virus was used to infect susceptible CEF, which were then plated in agar suspension at 36°C. Clonal stocks generated by picking the transformed colonies onto monolayer cultures were then tested for their ability to form foci at 36°C and 41°C. Six of the two hundred and sixty clones tested were unable to produce foci at 41°C. Because the frequency of ts mutants amongst the survivors was only 2%, I decided not to worry whether the mutants contained multiple mutations, and went on to characterize one of them, which I imaginatively named T1.

Was the mutant defective only in its ability to transform, or did the mutation also affect the ability to replicate? To test the ability of the mutant virus to replicate at the non-permissive temperature, I infected CEF with wild-type or mutant virus and then held the infected cultures at 36°C or 41°C. As shown in Figure 2, cells infected with wild-type virus became morphologically transformed at both temperatures, whereas the cells infected with mutant virus became transformed at 36°C and not at 41°C. However, as shown in Figure 3, the mutant virus replicated at the same rate as wild-type virus, both at 36°C and 41°C. Moreover the morphologically normal cells infected by the mutant at 41°C became resistant to superinfection by wild-type virus. Resistance to superinfection by RIF- or RAV-infected cells was known to result from blockade of virus receptors by the envelope protein of the virus. The high degree of superinfection resistance exhibited by the mutant-infected cells indicated that almost all of the cells in the culture were infected. Thus the mutant virus could replicate at the non-permissive temperature without inducing morphological transformation.

At this point two possible roles could be imagined for the transforming function of the virus. They are distinguished in Figure 4. One possibility was that the function was required only to initiate transformation: that is the virus might transform by a "hit-and-run" mechanism, so that once the cell was transformed the function would be dispensable. If that were case, mutant-infected cells would be expected to remain transformed if an infection was first established at 36°C and the cells were then shifted to the non-permissive temperature. Alternatively the transforming function might be required continuously to maintain the transformed state. In the latter case, mutant-infected cells would be expected to revert to the normal phenotype after a shift to the non-permissive temperature. Temperature-shift experiments of this type indicated that the mutant-infected cells did in fact revert to the normal morphology following a shift from 36°C to 41°C, and would then re-transform when shifted back to 36°C. Similarly, the mutant-infected cells could not grow into colonies in agar suspension at 41°C even if first grown at 37°C for a few days, whereas transformed colonies did appear in cultures held at 41°C and then shifted to 37°C ( Figure 4). Thus the viral function was required continuously to maintain both morphological transformation in monolayer culture and anchorage-independent growth in suspension cultures.

The legacy

The isolation of ts transformation defective mutants of RSV raised a series of questions: can the transforming gene be identified physically? what is its protein product? and how does that protein product induce transformation? In 1970—the same year that the temperature-sensitive mutants were first described—Peter Duesberg and Peter Vogt demonstrated that the RNA genomes of wild-type replication-competent RSV strains were larger than those of non-conditional transformation-defective mutants (or non-transforming RAVs) (Duesberg and Vogt, 1970). They concluded that the transformation defect of these mutants resulted from the deletion of a gene required for transformation. Genetic crosses subsequently demonstrated that the ts mutations all fell within the region deleted in the non-conditional transformation-defective mutants, thus identifying the same transforming gene (Bernstein, 1976). Peter Duesberg’s laboratory went on to define the region deleted in the non-conditional transformation-defective mutants—the src gene—by oligonucleotide fingerprinting (Lai et al., 1973). It was not until the end of the decade, when the moratorium on cloning was over, that the RSV genome and the v-src gene were sequenced. Meanwhile, the product of the src gene was identified by Brugge and Erikson as a 60 kDa phosphoprotein that could be immunoprecipitated from RSV-transformed cells (Brugge and Erikson, 1977). One year later the Bishop and Erikson labs showed that this protein had protein kinase activity (Collet and Erikson, 1978; Levinson et al., 1978), and in 1980 Hunter and Sefton demonstrated that the kinase specifically phosphorylated protein substrates at tyrosine residues (Hunter and Sefton, 1980). Growth factor receptors also proved to have tyrosine kinase activity, providing the first biochemical link between malignant transformation and growth control in normal cells.

The isolation of temperature-sensitive mutants also made possible a detailed examination of the biochemical events that occur during transformation. Temperature-shift experiments showed that a number of membrane-associated events occurred early in the transformation process, and could occur in the absence of protein synthesis. In a classic experiment, Beug and Graf showed that cells infected with ts mutant virus could undergo morphological transformation even when enucleated, although later events were blocked by enucleation (Beug et al., 1978). The picture that emerged from these early studies was that the transformation was initiated at the plasma membrane - where the Src protein was found to reside - and that signaling pathways then conveyed signals to the nucleus.

The discovery of tyrosine kinase activity of Src made it possible to examine the nature of the signaling pathways responsible for transformation. We and others identified many Src substrates. But the fundamental raison d’être of tyrosine phosphorylation did not become apparent until the end of the 1980s, when Tony Pawson noticed a region of homology in the non-catalytic domain of Src and a related non-receptor tyrosine kinase, Fps (Sadowski, Stone, and Pawson, 1986). This region of homology, the Src homology 2 (SH2) domain, was subsequently identified in many other signaling proteins, and was shown by Pawson’s and Hanafusa’s groups to specifically recognize phosphotyrosine (Matsuda et al., 1990; Beug et al., 1978). The interaction between SH2 domains and phosphotyrosine residues is now understood to be the key step in the assembly of signaling complexes and in signal transduction at the plasma membrane (see GENES 2000: 27.9 Receptor kinases activate signal transduction pathways).

But the most significant questions raised by the definition of the v-src gene concerned its origin: why did the virus carry a gene that was not required for replication? and where did it come from? Consideration of these questions led Bishop and Varmus to look for a cellular homolog of v-src. The identification of this gene, and how this discovery led to an understanding of the molecular basis of cancer, are described in the following essay.

The Author

G. Steven Martin is Professor and Head of the Division of Cell and Developmental Biology in the Department of Molecular and Cell Biology at the University of California, Berkeley. He earned his B.A. and Ph.D. degrees at the University of Cambridge, and did postdoctoral work in the Virus Laboratory at the University of California, Berkeley. He joined the faculty at Berkeley in 1975, after working for several years at the Imperial Cancer Research Fund in London. Honors include a Guggenheim Fellowship and election as a Fellow of the Royal Society of London.