Retroviruses derive their name from a feature of their life cycle that makes them unique in biology: their RNA must be transcribed “backward” into DNA for them to propagate. This unusual process is accomplished by an enzyme called reverse transcriptase. The enzyme was discovered in the particles of viruses such as the Rous sarcoma virus in 1970 by David Baltimore of the Massachusetts Institute of Technology and by Satoshi Mizutani and Howard M. Temin of the University of Wisconsin. The discovery was important on several counts. It scuttled the widely held misconception that genetic information could flow only from DNA to RNA. It triggered a surge of research on retroviruses by clarifying the previously obscure mechanism of their replication. And it provided an essential reagent for the developing technology of genetic engineering with recombinant DNA.
The life cycle of a retrovirus is a marvel of cooperation between parasite and host. The success of virus infection depends on the lavish hospitality offered by the cell, and yet the virus retains much authority to control events. During the early hours of infection the viral RNA genome is transcribed into DNA by reverse transcriptase. The viral DNA is then integrated into the cell’s genome, with the result that viral genes are replicated along with cellular genes and are expressed by the machinery of the cell.
In many cases a retrovirus infection is innocuous to the cell. The virus acquires a new and potentially enduring home; new virus particles are manufactured and leave the cell, and yet the cell suffers no damage. The partnership can go awry, however, as a result of either of the two kinds of viral oncogenesis mentioned above. If the virus carries an oncogene, the activity of the gene can convert the cell to cancerous growth. If the virus lacks an oncogene, the integration of the viral DNA can interfere with a cellular gene at or near the point of insertion; in other words, the insertion can cause a mutation in the host cell’s genome. Mutations at certain sites may engender cancerous growth. The induction of tumors by oncogenes and induction by the consequences of integration appear at first to be quite dissimilar events, but I shall show below that they are intimately related.
The src Gene
The oncogene of the Rous sarcoma virus was the first to yield to experimental analysis. An important early step was taken in 1970, when G. Steven Martin of the University of California at Berkeley identified temperature-sensitive “conditional” mutations that affect the ability of the virus to transform cells in culture. A conditional mutation is a powerful tool because it makes possible the reversible inactivation of a gene. When cultured cells infected with temperature-sensitive Rous sarcoma viruses are maintained at a “permissive” temperature, they are transformed. When the temperature is shifted to a higher, “restrictive” one, within hours the cells regain a normal appearance, only to be transformed once more when the temperature is again lowered. The interpretation is that at a restrictive temperature a mutated gene is inactivated. Transformation, then, is due to the action of a gene, which must be expressed continuously to maintain the cancerous state. (In most cases the elevated temperature probably does not act directly on the gene itself. Instead the mutation alters the structure of the protein product of the gene, with the result that the activity of the protein is impaired by the restrictive temperature.)
The gene first glimpsed by Martin is now called src (for sarcoma, the tumor it induces); it is the oncogene of the Rous sarcoma virus. The src gene was soon made more tangible by Peter H. Duesberg of Berkeley and by Charles Weissmann, Martin Billeter and John M. Coffin of the University of Zurich. They worked with strains of the Rous sarcoma virus that had been isolated by Peter K. Vogt of the University of Southern California. The strains are “deletion” mutants that have lost the oncogene and are therefore incapable of inducing tumors or transforming cells in culture. Duesberg and Weissmann and his colleagues fragmented the genomes of deletion mutants and of wild-type (oncogenic) viruses with the enzyme ribonuclease. By determining which fragment was missing in the mutants they were able to identify the oncogene as a segment of RNA near one end of the Rous sarcoma virus genome.
In the past few years the powerful new techniques of genetic engineering have been exploited to define oncogenes more precisely and to test their cancerous potential. DNA can now be cut into fragments at specific sites with the aid of a battery of enzymes called restriction endonucleases. Particular fragments can be grown in quantity in bacteria, then reisolated and inserted into cultured cells, where the genes carried by the DNA are expressed. In this way one can cut viral DNA into pieces that each carry a single gene and learn which of the pieces cause transformation. Analysis of the DNA of the Rous sarcoma virus has revealed a single gene capable of transforming cells; the gene encodes a single protein product. The implication is that one gene, by directing the synthesis of one protein, can bring about the changes characterizing a cancer cell. To know that protein and how it acts is to have in view the events that can generate a malignant tumor.
The protein encoded by src is known, owing largely to the work of Raymond L. Erikson of the University of Colorado School of Medicine and his colleagues. They began by identifying a protein that is synthesized in the test tube under the instructions of the wildtype Rous sarcoma virus genome but not under the instructions of the genome of a deletion mutant lacking src. Then they raised rabbit antibodies to a putative src protein by inducing tumors in rabbits with the Rous sarcoma virus. The antibodies combined specifically with the protein synthesized in the test tube and also with an identical protein in cells transformed by src. These findings persuasively identified a protein that is encoded by src and is responsible for the effects of the gene. The protein was designated pp60v-src; the “pp” signifies that it is a phosphoprotein (a protein to which phosphate groups are attached), the “60” refers to its molecular weight of 60,000 daltons and “v-src” indicates its genetic origin is the viral gene src.
A Cancer Enzyme
How does the protein product of the src gene convert a cell to cancerous growth? That seemed to be a daunting question when the protein was isolated. Yet a first answer came quickly when it was discovered that pp60v-src is a protein kinase: an enzyme that attaches phosphate ions to the amino acid components of proteins in the reaction known as phosphorylation. The discovery was made by Erikson and his colleague Mark S. Collett and, independently, by Arthur Levinson, working with Harold E. Varmus and me in our laboratory at the University of California School of Medicine in San Francisco.
Soon thereafter Tony Hunter and Bartholomew M. Sefton of the Salk Institute for Biological Studies reported that pp60v-src attaches phosphate ions specifically to the amino acid tyrosine. That put pp60v-src outside the known classes of protein kinases, which phosphorylate the amino acids serine and threonine. The phosphorylation of tyrosine has turned out to be a common characteristic of oncogene-encoded enzymes; surprisingly, it also has a role in regulating the growth of normal cells.
Not many years ago phosphate appeared to most biologists to be a mundane material and its transfer to proteins a humble event. Now it is clear that the phosphorylation of proteins is one of the central means by which the activities of growing cells are governed. One enzyme, by phosphorylating a number of proteins, can vastly alter the functioning of a cell. In the case of pp60v-src two modes of action have been proposed. The enzyme could phosphorylate a single protein, precipitating a cascade of events that together give rise to the properties of a cancer cell; alternatively, the enzyme could phosphorylate numerous proteins, directly affecting the functions of each of them and perhaps precipitating secondary events or even cascades in turn. What little is known at the moment makes it seem likely that the second alternative correctly describes the action of pp60v-src.
Can the phosphorylation of tyrosine subunits in cellular proteins account for the ability of src to induce tumors? Hunter and his colleagues have shown that the amount of phosphorylated tyrosine in a cell increases approximately tenfold as a result of transformation by src. The increase is regarded as a manifestation of the activity of pp60v-src. The critical questions now are: What cellular proteins are phosphorylated by the enzyme and what are their functions? There are only a few clues, none of which can yet account for the unrestrained growth of the tumors induced by src. The pursuit of targets for pp60v-src is under way in many laboratories.
Site of Action
One approach is to find out where in the cell pp60v-src acts, in the hope of learning what proteins it affects and what those proteins do. Early studies indicated that the products of viral oncogenes might take up residence in the nucleus of the cell, where they could meddle directly with the apparatus responsible for replicating the cellular DNA and so drive the cell to unrestrained growth. Experiments by Hartmut Beug and Thomas Graf of the Max Planck Institute for Virus Research in Tübingen showed, however, that the effects of the src protein can be detected even in cells from which the nucleus has been removed. It came as no surprise, then, when several workers found that little if any of the pp60v-src in transformed cells is in the nucleus. Most of the protein is at the other extreme of the cell: it is bound to the plasma membrane, the thin film that encloses the cell and mediates its interactions with the outside world. Many cell biologists have argued that the control of cell growth may originate at the plasma membrane and its associated structures.
Inspection of the plasma membrane of cells transformed by src has provided the first correlation between the action of pp60v-src on a specific cellular protein and one of the typical changes in structure and function seen in cancer cells. By means of specialized techniques of photomicroscopy Larry R. Rohrschneider of the Fred Hutchinson Cancer Research Center in Seattle was able to demonstrate that pp60v-src is concentrated in adhesion plaques: regions of the membrane that adhere to solid surfaces. In cancer cells the adhesion plaques are dismantled; the resulting decrease in cell adhesion may contribute to the ease with which most cancer cells break away from their tissue of origin and metastasize to other sites.
Rohrschneider’s findings suggested that pp60v-src might dismantle adhesion plaques by phosphorylating one of their component proteins, or perhaps several of those proteins. Pursuing that suggestion, Sefton and S. J. Singer of the University of California at San Diego showed that pp60v-src phosphorylates a tyrosine unit in vinculin, a protein that is a constituent of normal adhesion plaques and becomes dispersed throughout the cell following transformation by src. It seems reasonable to suggest that the phosphorylation of vinculin precipitates the dismantling of adhesion plaques, but the importance of such events in the unruly behavior of cancer cells has yet to be established.
Once it was thought that the oncogenic effects of viruses might be ancillary manifestations of viral genes whose main function is to assist in the production of new virus particles. Now it is clear that the replication of retroviruses proceeds normally in the absence of oncogenes. How then can one explain the wide occurrence of oncogenes in retroviruses and their apparent conservation in the course of evolution? A decade of investigation has furnished a surprising answer. Retrovirus oncogenes are merely cellular genes in another guise, passengers acquired from the animals in which the viruses replicate. The discovery that cells too have oncogenes has implications extending far beyond tumor virology.
The Origin of Oncogenes
In 1972 Dominique Stehelin, Varmus and I set out to explore the “oncogene hypothesis” proposed by Robert J. Huebner and George J. Todaro of the National Cancer Institute. Seeking one mechanism to explain the induction of cancer by many different agents, Huebner and Todaro had suggested that retrovirus oncogenes are a part of the genetic baggage of all cells, perhaps acquired through viral infection early in evolution. The oncogenes would be innocuous as long as they remained quiescent. When stimulated into activity by a carcinogenic agent, however, they could convert cells to cancerous growth. We reasoned that if the hypothesis was correct, we might be able to find the src gene in the DNA of normal cells.
The DNA of vertebrates includes tens of thousands of genes. To search for src amid this vast array Stehelin fashioned a powerful tool: radioactive DNA copied solely from src by reverse transcriptase. The copied DNA served as a probe with which to search for cellular DNA with a nucleotide sequence similar to that of src. The search was carried out by molecular hybridization, in which chains of a nucleic acid (either DNA or RNA) hybridize, or form complexes, with nucleic acids to which they are related. We were exhilarated (and more than a little surprised) to learn that Stehelin’s copy of src could hybridize with DNA from uninfected chickens and other birds. Deborah H. Spector joined us and went on to find DNA related to src in mammals, including human beings, and in fishes. We concluded that all vertebrates probably possess a gene related to src, and it therefore seemed the Huebner-Todaro oncogene hypothesis might be correct.
On closer inspection, however, the gene we had discovered in vertebrates proved not to be a retrovirus gene at all. It is a cellular gene, which is now called c-src. The most compelling evidence for this conclusion came from the finding that the protein-encoding information of c-src is divided into several separate domains, called exons, by intervening regions known as introns. A split configuration of this kind is typical of animal-cell genes but not of the genes of retroviruses. Apart from their introns, the versions of c-src found in fishes, birds and mammals are all closely related to the viral gene v-src and to one another. It appears the vertebrate src gene has survived long periods of evolution without major change, implying that it is important to the well-being of the species in which it persists.
The mystery presented by c-src deepened with the discovery that the gene not only is present in normal cells but also is active in them, that is, the gene is transcribed into messenger RNA and the RNA is translated into protein. Molecular hybridization with Stehelin’s radioactive copy of v-src brought the RNA to view first, in both avian and mammalian cells. The protein was more elusive, mainly because it is synthesized in very small quantities in most cells. Success came when we and others probed for the cellular protein with antibodies prepared originally for the pursuit of the viral transforming protein, pp60v-src. The cellular protein isolated with the aid of these antibodies proved to be virtually indistinguishable from the viral protein, and it was therefore named pp60c-src. The two proteins are similar in size and chemical structure; both catalyze the phosphorylation of tyrosine and both are tightly bound to the plasma membrane of cells (transformed cells in the case of pp60v-src, normal cells in the case of pp60c-src). It is as if the two proteins were designed for the same purpose, even though one is a viral protein that causes cancer and the other is a protein of normal cells.
Cellular Oncogenes
The findings with respect to src were the first hint of a generalization whose extent and significance have yet to be established. Of 17 retrovirus oncogenes identified to date, 16 are known to have close relatives in the normal genomes of vertebrate species. Most of these cellular relatives of viral oncogenes obey the principles first deduced for c-src. They have the structural organization of cellular genes rather than of viral genes; they seem to have survived long periods of evolution, and they are active in normal cells. To account for these facts and for the remarkable similarity between retrovirus oncogenes and their normal cellular kin most virologists have settled on the idea that retrovirus oncogenes are copies of cellular genes. It appears the oncogenes were added to preexistent retrovirus genomes at some time in the not too distant past. How and why retroviruses have copied cellular genes is not known, but there is reason to think the copying continues today, and it may even be possible to recapitulate the process in the laboratory.
The vertebrate genes from which retrovirus oncogenes apparently arose were at first called proto-oncogenes, to emphasize their evolutionary significance and to avoid implying that the cellular genes themselves have oncogenic potential. Now it is clear that they do have such potential. They are cellular oncogenes. The investigations that justify this designation began with this question: If retrovirus oncogenes are merely copies of genes found in normal cells, how can one account for the devastating effects of the viral genes on infected cells? Two explanations have been offered. The mutational hypothesis proposes that viral oncogenes differ from their cellular progenitors in subtle but important ways as a result of mutations introduced when the cellular genes were copied into the retrovirus genome. For example, the apparently similar enzymatic activities of pp60v-src and pp60c-src might actually have different targets in the cell and might therefore have very different effects on cellular behavior. The alternative dosage hypothesis suggests that retrovirus oncogenes act by brute force, overburdening cells with too much of what are essentially normal proteins carrying out normal functions. In this view the genesis of cancer by retrovirus oncogenes is related to the amount of the viral proteins rather than to any unique properties they have.
It is too early to know which of these views is correct, but initial indications favor the dosage hypothesis. First, the doses of retrovirus transforming proteins are unquestionably large. The signals directing the activity of retrovirus genes are quite powerful, with the result that the amount of protein produced from a viral oncogene is far larger than the amount usually produced from the corresponding cellular gene; it is clearly possible that the cell might be overwhelmed. More important evidence has come from attempts to test a central prediction of the dosage hypothesis: If retrovirus oncogenes and cellular oncogenes are indeed identical in function, it should be possible to find circumstances under which the cellular genes themselves can induce cancerous growth.
Cellular-Gene Oncogenesis
The first test of this prediction came from the remarkable experiments of Hidesaburo Hanafusa and his colleagues at Rockefeller University. Hanafusa found strains of the Rous sarcoma virus that had lost large portions of the src gene (but not all of it) and were therefore incapable of inducing the characteristic sarcoma in experimental animals. When Hanafusa injected the crippled viruses into chickens and then recovered the virus particles manufactured in the infected cells, he was astonished to discover that the v-src gene of the virus had been reconstituted. Apparently genetic material from the c-src gene was recombined with the viral genome while the virus was growing in the birds. The virus bearing the reconstituted gene was again fully capable of causing tumors, even though as much as three-fourths of its oncogene had just been acquired from a cellular gene. Hanafusa was able to repeat this extraordinary exercise at will, in quails as well as in chickens. His findings lent weight to the idea that the functions of c-src and v-src are the same, but many tumor virologists were unpersuaded in the absence of more direct evidence for the tumorigenic potential of the cellular genes.
Now such evidence is at hand. The research groups of George F. Vande Woude and Edward M. Scolnick of the National Cancer Institute exploited the techniques of genetic engineering to isolate three cellular oncogenes (one from mice and the other two from rats) and to show directly that these genes can induce cancerous growth in cultured cells. The feat was accomplished by attaching to the cellular genes a viral “promoter,” a DNA-encoded signal that helps to regulate the expression of a nearby gene. In accordance with the dosage hypothesis, when the src-promoter complex was introduced into cells, some of the cells were transformed as if they had received a viral oncogene, whereas what they had received was a cellular gene under viral orders to work harder than usual. Moreover, cells transformed by the two rat cellular oncogenes could be shown to make very large quantities of the proteins encoded by the genes, again in accordance with the dosage hypothesis.
Why should an overabundance of a normal protein wreak such havoc? The question can be answered with assurance only when the role of cellular oncogenes in the orderly affairs of normal cells is understood. Perhaps cellular oncogenes are part of a delicately balanced network of controls that regulate the growth and development of normal cells. Excessive activity by one of these genes might tip the balance of regulation toward incessant growth.
There is evidence that the activities directed by viral and cellular oncogenes do help to control the growth of normal cells. Whereas at first the phosphorylation of tyrosine by pp60v-src seemed to be an anomalous process whose foreign nature might underlie the cancerous response to src, that view was reversed when Stanley Cohen of the Vanderbilt University School of Medicine found a role for tyrosine phosphorylation in the housekeeping of normal cells. Having discovered and purified a small “epidermal growth factor” whose binding to the surface of cells stimulates DNA synthesis and cell division, he considered how the signal for these events might be transmitted from the exterior of the cell to the interior. Cohen first showed that the binding of epidermal growth factor to cells brings about phosphorylation of proteins; prompted by the findings with pp60v-src, he ultimately found that the phosphorylation elicited by epidermal growth factor specifically affects tyrosine. Other workers have since shown that some proteins phosphorylated in response to epidermal growth factor can be phosphorylated by pp60v-src. A normal stimulant of cell division (epidermal growth factor) and an abnormal one (pp60v-src) thus appear to play on the same keyboard. The implication is that tyrosine phosphorylation by pp60c-src has a part in regulating the growth of normal cells.
In Search of a Unified Theory
Retroviruses do not seem to be a major cause of human cancer. They may nonetheless have pointed the way to the central mechanisms by which the disease arises. It is generally thought that cancer begins with damage to DNA, although the exact nature of the damage is in dispute. How might the damage cause cancerous growth? Most recent efforts to answer the question in a way that might apply to all forms of cancer have invoked the existence of “cancer genes”: components of the normal cellular genome whose activity is unleashed or augmented by carcinogens of various kinds and is then responsible for sustaining the undisciplined behavior of cancer cells. In this scheme cancer genes are viewed not as alien intruders but as normal, indeed essential, genes run amok; the damage done by a carcinogen turns friend into foe, perhaps by acting directly on the cancer gene or perhaps by crippling a second gene that normally polices the activity of the cancer gene.
Medical geneticists may have detected the effects of cancer genes years ago, when they first identified families whose members inherit a predisposition to some particular form of cancer. Now, it appears, tumor virologists may have come on cancer genes directly in the form of cellular oncogenes. In their viral form these genes are tumorigenic, and Vande Woude’s and Scolnick’s results imply that the cellular genes can also transform cells. It is therefore easy to imagine that cancer genes and the cellular oncogenes revealed by retroviruses are one and the same. The oncogene hypothesis has been restaged with the actors now cellular rather than viral genes; the dosage hypothesis serves to explain why the augmented activity of a normal cellular gene might cause cancer.
Evidence in support of these ideas comes from the study of chicken retroviruses that induce lymphoma, a lethal tumor of the immune system. The chicken lymphoma viruses have no oncogenes. Why then do they cause tumors? William S. Hayward and Benjamin G. Neel of Rockefeller University and Susan M. Astrin of the Institute for Cancer Research in Fox Chase, Pa., have discovered that in tumors induced by the chicken lymphoma viruses the viral DNA is almost always inserted into cellular DNA in the immediate vicinity of a single cellular oncogene (not c-src but a more recently recognized oncogene known as c-myc). As a seeming consequence of the insertion, the expression of the cellular oncogene is greatly amplified.
These findings fit the concept of cancer genes quite well. The insertion of lymphoma-virus DNA into the host genome is analogous to mutagenesis or other forms of damage introduced by carcinogens of many kinds. The insertion apparently stimulates the activity of a gene that is known to be oncogenic when it appears (as v-myc) in a different chicken retrovirus; the stimulated action of the cellular oncogene seems to be responsible, at least in part, for the genesis of tumors. Retroviruses without oncogenes induce a variety of tumors; by identifying the site where the viral genome is inserted into cellular DNA in some of those tumors, virologists may be able to discover cancer genes not yet identified by other means.
The unveiling of cancer genes (in the form of cellular oncogenes) by retroviruses was serendipitous. Must investigators be content with the pace at which retroviruses thus offer up new oncogenes from within the cell? Apparently not. Robert A. Weinberg of M.I.T. and Geoffrey Cooper of the Harvard Medical School have broadened the search for cancer genes beyond the purview of tumor virology. They have shown that gene-length bits of DNA isolated from some tumors (tumors that were not induced by viruses) can transmit the property of cancerous growth when they are introduced into previously normal cells in culture.
Weinberg and Cooper have evidently found a way of transferring active cancer genes from one cell to another. They have evidence that different cancer genes are active in different types of tumors, and so it seems likely that their approach should appreciably expand the repertory of cancer genes available for study. None of the cancer genes uncovered to date by Weinberg and Cooper is identical with any known oncogene. Yet it is clearly possible that there is only one large family of cellular oncogenes. If that is so, the study of retroviruses and the procedures developed by Weinberg and Cooper should eventually begin to draw common samples from that single pool.
A Final Common Pathway
Normal cells may bear the seeds of their own destruction in the form of cancer genes. The activities of these genes may represent the final common pathway by which many carcinogens act. Cancer genes may be not unwanted guests but essential constituents of the cell’s genetic apparatus, betraying the cell only when their structure or control is disturbed by carcinogens. At least some of these genes may have appeared in retroviruses, where they are exposed to easy identification, manipulation and characterization.
What has been learned from oncogenes represents the first peep behind the curtain that for so long has obscured the mechanisms of cancer. In one respect the first look is unnerving, because the chemical mechanisms that seem to drive the cancer cell astray are not different in kind from mechanisms at work in the normal cell. This suggests that the design of rational therapeutic strategies may remain almost as vexing as it is today. It will be of no use to invent means for impeding the activities responsible for cancerous growth if the same activities are also required for the survival of normal cells.
However the sage of oncogenes concludes, it presents some lessons for everyone concerned with cancer research. The study of viruses far removed from human concerns has brought to light powerful tools for the study of human disease. Tumor virology has survived its failure to find abundant viral agents of human cancer. The issue now is not whether viruses cause human tumors (as perhaps they may, on occasion) but rather how much can be learned from tumor virology about the mechanisms by which human tumors arise.
Source: Reprinted with permission. Copyright (c) by Scientific American, Inc. [http://www.sciam.com/]. All rights reserved.
"Oncogenes," Microsoft® Encarta® Encyclopedia 2000. © 1993-1999 Microsoft Corporation. All rights reserved.