Proc. Nati. Acad. Sci. USA Vol. 89, pp. 6393-6397, July 1992 Biochemistry Unmutated proto-src coding region is tumorigenic if expressed from the of Rous sarcoma virus: Implications for the - hypothesis of (retroviral oncogenicity from transcriptional activation/mutated and cancer) YUE WU*, HONG ZHOUt, AND PETER DUESBERG*t *Department of Molecular and Cell Biology, University of California, Berkeley, CA 94720; and tLaboratory of Molecular Oncology, National Cancer Institute-Frederick Cancer Research Facility, Building 469, Room 203, Frederick, MD 21701 Contributed by Peter Duesberg, March 25, 1992

ABSTRACT The transforming (onc) genes of oncogenic genes if they have suffered (1, 4-9). Indeed, retroviruses share most or all of their coding sequences with mutated proto-onc genes have been found in some tumors. normal cellular genes termed proto-onc genes. The viral genes However, the hypothesis has been difficult to prove, because differ from proto-onc genes in virus-specific promoters and in mutated proto-onc genes from tumors do not transform various point mutations and substitutions ofcell-derived coding diploid cells upon transfection (10) or introduction into the regions. In view of the structural similarities between viral germ line of transgenic animals (11). An apparent exception oncogenes and cellular proto-onc genes, the hypothesis has is the evidence that point-mutated proto-ras genes from been advanced that proto-onc genes become cellular cancer certain would transform the aneuploid 3T3 mouse genes if they have suffered mutations. Indeed, point mutations cell line (5, 6). and substitutions have been observed in the proto-onc genes of By contrast, retroviral oncogenes are inevitably oncogenic some cancers. However, the hypothesis has been difficult to in diploid cells or animals (1, 2). Due to the popularity of the prove because mutated proto-onc genes from tumors do not proto-onc gene-mutation hypothesis, even the transforming transform diploid cells. Moreover, owing to the popularity of function of viral onc genes is thought to derive from muta- this hypothesis, even viral oncogenes are thought to derive tions that set apart the coding regions of viral onc genes from transforming function from mutations of this cell-derived corresponding proto-onc genes (8, 9, 12, 13). For example, coding region. A competing hypothesis proposes that enhanced the transforming function of viral src genes (13-15) and viral expression from retroviral promoters is necessary and suffi- ras genes (1, 7, 12) has been interpreted as the result of point cient for oncogenic function of proto-onc genes. To distinguish mutations that set apart the encoded by the viral and between these hypotheses we have tested tumorigenicity of corresponding cellular genes. RpSV, a synthetic retrovirus with the normal proto-src coding However, transforming function of typical retroviral onc region in a vector derived from Rous sarcoma virus (RSV). In genes, including src, myc, and ras genes, was recently shown addition, we have tested the role of RSV-specific src point to depend on retroviral promoters rather than on the muta- mutations on the tumorigenicity of RpSV. It was found that tions that set apart the coding regions of viral and proto-onc RpSV with an unmutated proto-src coding region is tumori- genes (3, 16-21). These promoters increase expression about genic in chickens and that tumorigenicity is enhanced by 100-fold compared to the corresponding proto-onc genes (11, RSV-specific src point mutations. It is concluded that retroviral 21, 22). In addition, transformation of 3T3 cells by point- promoters are essential for the transforming function of viral mutated proto-ras genes from tumors was recently shown to oncogenes and that certain point mutations merely supplement reflect expression artifacts generated by the transfection their transforming function. Thus retroviral one genes are not assay, rather than a genuine function ofnative, point-mutated models for the hypothesis that mutated, but transcriptionally proto-ras genes from tumors. These artifacts include substi- normal, proto-onc genes of certain tumors are cancer genes. tution of native cellular proto-ras control elements and ofthe native proto-ras promoter by heterologous counterparts and Most or all of the coding regions of retroviral transforming concatenization of proto-ras genes during transfection (11, (onc) genes are derived from cellular genes termed proto-onc 21). genes (1). The viral genes differ from proto-onc genes in In view of this, we and others have advanced the hypoth- virus-specific promoters and various point mutations, dele- esis that transforming function of retroviral onc genes is due tions, and substitutions of their coding sequences (2). For to the 100-fold transcriptional activation ofproto-onc-derived example, the transforming src gene of Rous sarcoma virus coding sequences by retroviral promoters (2, 11, 18, 20, 21). (RSV) and of three other strains of avian sarcoma viruses Certain virus-specific or cellular mutations in the coding differs from the proto-src gene of normal cells in (i) virus- sequence merely enhance this transforming function derived specific promoters, (ii) a virus-specific carboxyl terminus from heterologous promoters but would not be able to create that replaces the six carboxyl-terminal amino acids, including it on their own. Tyr codon 527, of proto-src by heterologous virus- or cell- To put our hypothesis to a further test, we ask here whether derived sequences, and (iii) scattered, virus strain-specific the native proto-src coding region artificially introduced into point mutations within the 514 codons that viral src and a retrovirus vector would be tumorigenic in animals and proto-src have in common (1, 3). whether virus-specific point mutations would enhance the In view of the structural similarities between retroviral transforming function of such a virus. Previously, we had oncogenes and cellular proto-onc genes, the hypothesis has found that RpSV, a synthetic virus with a native proto-src been advanced that proto-onc genes become cellular cancer coding region (see Fig. 1), is sufficient to transform primary

The publication costs of this article were defrayed in part by page charge Abbreviations: RSV, Rous sarcoma virus; RpSV, proto-onc sarcoma payment. This article must therefore be hereby marked "advertisement" virus. in accordance with 18 U.S.C. §1734 solely to indicate this fact. tTo whom reprint requests should be addressed.

6393 Downloaded by guest on September 28, 2021 6394 Biochemistry: Wu et al. Proc. Natl. Acad. Sci. USA 89 (1992) avian fibroblasts in vitro but is slower than RSV, requiring Reverse of Virus RNA to cDNA in Vitro. 9-12 days compared to 4-5 days, in cell transformation (3). About 0.2 A260 unit (measured in 0.1% SDS) of purified virus Here we show that RpSV is tumorigenic in chickens after were incubated at 40°C for 24 hr in a solution containing the 12-15 days, compared to 4-5 days for RSV, and that tumor- four dNTPs each at 10 ,uM, 50 mM NaCI, 10 mM Tris HCl at igenicity is enhanced by . Further, we have pH 7.9, 10 mM MgCI2, 1 mM dithiothreitol, 2.5 ,ug of proved that RpSV tumorigenicity is due to the input virus oligo(dT) per ml, and 0.25% Nonidet P-40 as described (24). rather than to spontaneous mutation of the virus in the The cDNA was isolated from the reaction mixture by phenol animal. extraction and ethanol precipitation. In some experiments the viral RNA was eliminated from the cDNA by treatment for 4 hr at 40°C in 0.3 M NaOH. MATERIALS AND METHODS Specific Amplification of cDNA by the Polymerase Chain Construction of Recombinant Viruses. RpSV has been Reaction (PCR). The cDNA from about 0.04 A260 unit of constructed from a DNA provirus of RSV, pJD100 (23), by purified RpSV or RSV virus was incubated in 100 Al of 100 substituting the coding region ofsrc from an Nco I site at the ,uM each of the four dNTPs, 10 mM Tris HCl at pH 8.3, 50 AUG start codon to a Pvu II site 36 base pairs (bp) upstream mM KCl, 1.5 mM MgCl2, 0.001% gelatin, and 1.0 ,uM each of of the TAA by a sequence that includes the two specific primers as recommended by the manufacturer complete coding region of proto-src (Fig. 1). The proto-src (Perkin-Elmer/Cetus). One of the primers was a 25-mer of DNA extended from the Nco I site at AUG to a Sac I site 10 the src sequence of RSV from position 8598 to position 8622 bases downstream of the TAG stop codon of proto-src (3). just 5' ofa Mst II site at position 8635; the other was a 27-mer The resulting recombinant virus contains the complete cod- complementary to a sequence of RSV 3' of src from position ing region ofproto-src (Fig. 1). RpSV with a RSV-derived src 8899 to position 8925 (Fig. 1) (1). After 2 min at 94°C, 2.5 units mutation in src codon 95, termed RpSV 95, was constructed ofTaq DNA polymerase was added. The solution was carried by replacing a 2906-bp Kpn I/Mlu I fragment of RpSV that through 45 cycles ofdenaturation for 1 min at 94°C, annealing includes the Arg codon 95 of proto-src (Fig. 1) with the for 1 min during a temperature shift from 94°C to 550C and by corresponding fragment from RSV that encodes Trp in codon 2 min at 550C, followed by polymerization for 3 min at 72°C. 95 (1). For this purpose, RpSV was digested with Kpn I and Finally, the mixture was incubated at 72°C for 10 min to Mlu I and the resulting 9266-bp fragment was ligated with the complete any incomplete DNA products (25, 26). The PCR Kpn I/Mlu I fragment from the RSV clone pJD100 (3). RpSV products were analyzed by electrophoresis in 1.7% agarose in 338, with aRSV-specific point mutation in src codon 338, was 40 mM Tris-acetate/1 mM EDTA containing 0.5 Ag of derived from pJD100 in two steps: First, a RSV-specific Mst ethidium bromide per ml (25). II site 5' of Mlu I was eliminated by replacing the 2906-bp Kpn-Mlu sequence by a 765-bp counterpart from A phage. Next the 249-bp Mst 11-bordered viral src carboxyl terminus RESULTS was replaced by the 312-bp equivalent of RpSV (Fig. 1). Transformation of Cells in Culture by Synthetic Viruses Subsequently, the A phage-derived Kpn-Mlu sequence was Carrying Proto-src Coding Regions with RSV-Speciflc Point replaced by the original 2906-bp Kpn-Mlu sequence of Mutations. To determine whether RSV-specific src point RpSV. RpSV 95-338 with RSV-specific src codons 95 and mutations other than those affecting the carboxyl terminus of 338 was made from the immediate precursor construct of proto-src-e.g., the RSV-specific src codons 95 and 338- RpSV 338 by replacing the A-derived Kpn-Mlu sequence play a role in transforming function, such mutations were with the original 2906-bp sequence of RSV. introduced singly or together into the proto-src coding region RpSV ATG 95 338 TAG

r, l.z.. -.-I I K MI m SpP

'N'.

5 primer TAG TAA r' '.1 I .., , e-

Li

M Ps S.P rN.A 395 bp

FIG. 1. Genetic structure ofthe recombinant proto-src gene ofRpSV. ATG and TAG mark the proto-src coding region (high box), the arrows mark the proviral long terminal repeats including the viral promoters, and the lower white boxes are noncoding regions (3). The positions of src codons 95 and 338 and of several restriction enzyme sites are indicated. Restriction enzyme sites are abbreviated as follows: K, Kpn I; M, Mst II; Ml, Mlu I; P, Pvu II; Ps, Pst I; S, Sac I. The lower panel shows the 3' terminal proto-src region that was amplified for size and sequence analysis (see text). Downloaded by guest on September 28, 2021 Biochemistry: Wu et al. Proc. Natl. Acad. Sci. USA 89 (1992) 6395 of RpSV. For this purpose, the amino-terminal domain ofthe effective enhancer of transforming function of a proto-src proto-src coding region of RpSV, including codon 95, and an coding region under a retroviral promoter (3, 13), we inves- internal domain, including codon 338, were exchanged for the tigated whether the sequence ofthis region had been mutated RSV equivalents (Fig. 1) (Materials and Methods). The in viruses recovered from tumors. resulting proviruses, termed RpSV 95, RpSV 338, and RpSV Viruses were recovered from the sarcomas of six chickens 95-338, were transfected onto quail embryo fibroblasts as inoculated with RpSV. The sarcomas were washed with 75% described (3). In parallel, quail cells were transfected with ethanol, excised, and dissected with scalpels. The minced proviruses of RpSV and RSV. RSV began to transform cells tissue was washed three to five times by vigorous stirring on day 4. Foci of RpSV-transformed cells were observed on with 10 vol of buffered saline at 0-4°C to eliminate blood day 6 for RpSV 95-338, on day 7 for RpSV 95, on day 8 for cells. Subsequently, the tissue was stirred with trypsin at RpSV 338, and on day 9 for RpSV. Cells transformed by the 36°C and single cells were decanted about three times at three recombinant RpSVs were all more refractile than those 20-min intervals and propagated in 10-cm Petri dishes as transformed by RpSV. Within 2 weeks most cells of cultures described (3). The growth media were then collected for 1-2 transfected by RpSV 95, RpSV 338, and RpSV 95-338 were weeks and the viruses were purified by sucrose gradient transformed. By contrast, only about halfthe cells in cultures ultracentrifugation (3). The viral were then transcribed transfected by the provirus of RpSV were transformed, in vitro to cDNA by viral reverse transcriptase (Materials confirming previous observations (3). It is concluded that and Methods). A sequence of the cDNA from six some RSV-specific src point mutations enhance transforming RpSVs recovered from sarcomas from a position correspond- function of RpSV in vitro. ing to nucleotide 8598 in the src gene ofRSV to position 8925, Tumorigenicity of RpSV, RpSV 95, RpSV 338, and RpSV which is located downstream ofthe src coding region ofRSV 95-338 in Newborn Chickens. Next the tumorigenicity of (1), was then amplified by the PCR with specific primers and RpSV and of the RpSV mutants 95, 338, and 95-338 was for its tested in newborn chickens. For this purpose the viruses analyzed size and sequence (Materials and Methods). were purified by sucrose gradient ultracentrifugation from In parallel, the cDNA ofwild-type RSV and the cloned DNA the growth media of cultures of transformed quail cells (27). of RpSV were amplified with the same primers. The media were collected at daily intervals (28). One-day-old The sequence between our primers is expected to measure chickens were injected under the wing web with 50-150 u4 of 395 for intact, unmutated RpSV and 328 nucle- purified virus in physiological saline buffer, derived from otides for RSV (Fig. 1). As can be seen in Fig. 3, all RpSV src 10-20 ml ofculture medium. Sarcomas were detected on day had the same, expected size of about 395 nucleotides 7 after injecting RpSV 85-338, on day 9 after injecting RpSV as the RpSV control, indicating that their proto-src carboxyl 85 and RpSV 338, and on days 12-15 after injecting RpSV termini were not mutated with regard to their size. The RSV (Fig. 2). Sarcomas were observed on day 4 after injecting 3' terminal src sequence also had the predicted size (Fig. 3). RSV. Thus, incubation periods from inoculation of RpSV Next we asked whether the 3' terminal proto-src coding mutants to tumorigenicity varied with distinct proto-src mu- region, particularly the carboxyl-terminal Tyr codon 527 (13), tations, and all proto-src viruses were slower in causing had been mutated in any of the tumorigenic RpSVs. For this tumors than RSV. Since all tumors caused by synthetic purpose most of the sequences amplified by the above viruses carrying proto-src coding regions had appeared later than those caused by wild-type RSV, it may be argued that 1 2 3 4 5 6 7 8 9 the longer incubation periods to tumorigenicity reflected time required to generate oncogenic mutants. -1353 Sequence Analyses of RpSVs from Viral Tumors. To deter- -1078 mine whether tumorgenicity ofproto-src viruses was genuine -872 or due to mutations of proto-src, the proto-src sequences of RpSVs recovered from tumors were analyzed. Because mu- -603 tation ofthe proto-src-specific carboxyl terminus is the most 395- 328- -310 -281

FIG. 3. The 3' terminal proto-src coding regions of RpSVs recovered from six viral tumors. Viruses from sarcomas of chickens injected with RpSV were purified by sucrose gradient ultracentrif- Nk ugation. An in vitro reverse transcriptase reaction was carried out to make cDNA. Two primers flanking the 3' terminal proto-src se- quence of RpSVs were used to amplify the sequence in PCR reactions (see text and Fig. 2). The DNA products were analyzed by FIG. 2. Sarcoma induced by injecting RpSV into a 1-day-old electrophoresis in a 1.7% agarose gel containing 0.5 Jtg of ethidium chicken. The supernatant medium from cultured quail cells trans- bromide per ml. Lanes 1-6, cDNAs from RpSVs recovered from six formed by RpSV was collected and virus particles were purified by separate chicken sarcomas; lane 7, cDNA from RSV amplified with sucrose gradient ultracentrifugation. About 50-150 ,ul of purified the same primers as for RpSV (see text); lane 8, DNA from the virus was injected into the wing web of a 1-day-old chicken. The plasmid containing the provirus of RpSV amplified as above; lane 9, sarcoma shown above became detectable 12 days after injection and DNA size markers (in nucleotides) of phage 4'x174 replicative form was photographed 3 weeks after the injection. DNA digested with Hae III. Downloaded by guest on September 28, 2021 63% Biochemistry: Wu et al. Proc. Natl. Acad. Sci. USA 89 (1992) primers-e.g., 292 nucleotides from a Pst I site at position in the coding region, there are no precedents or models for 8662 to aMst II site at position 8884 (see Fig. 1)-were cloned how mutations that do not enhance transcription would into the polylinker of a pUC plasmid (Pharmacia pSL1180) convert proto-onc genes to cancer genes. Indeed, none ofthe and analyzed. As can be seen in Fig. 4, the 3' terminal mutated proto-onc genes observed in cancer cells is tran- sequences of three RpSVs from three different tumors were scriptionally activated like retroviral onc genes (11). unchanged in their coding regions, each carrying the native (ii) The probability of the mutations that are postulated to Tyr in codons 527. It is concluded that the unmutated convert proto-onc genes to cancer genes, such as point proto-src coding region of the cell is tumorigenic in a RSV- mutations and truncations, is much higher than the incidence derived retrovirus vector. of cancer. For example, the probability of a point mutation per nucleotide per mitosis is about 1 in 109 (2, 11, 30, 34). Given this point and 109 nucleotides per human DISCUSSION DNA (38), a cell with any possible point mutation of the Retroviral Promoter Sufficient to Convert Proto-src Genes to human would be found in 109 cells. Since the human Cancer Genes. Our experiments have demonstrated that body consists ofmore than 1014 cells (2, 30), this corresponds enhanced expression of the normal proto-src coding region to a minimum of over 100,000 hypothetical cancer cells from a RSV-derived retroviral promoter is sufficient for (1014:109 = 105) per human body at any given time. Consid- tumorigenicity. Further, we have demonstrated that tumor- ering that many proto-onc genes are said to be activated by igenicity of the retrovirus-promoted proto-src is point mutation and some like proto-ras even by >50 different enhanced by RSV-specific src point mutations other than that point mutations (7, 16, 39), the number of cancer cells per that alters Tyr codon 527. These experiments confirm and human body would be correspondingly higher. extend studies with other proto-onc genes, which have dem- It is possible, however, to reconcile numerically the high onstrated that about 100-fold increased expression from probability of such mutations with the low probability of retrovirus promoters is sufficient for transforming function of cancer by assuming cooperations of multiple mutated proto- diploid cells and tumorigenicity. Examples include retrovi- onc genes and anti-oncogenes in the same cell. This has been rus-promoted coding regions of proto-ras (3, 16, 18, 20, 21) done in the case of colon cancer (10, 35, 40) and some other and proto-myc (17, 19, 29). Thus, our experiments support tumors (10, 11, 36). However, these ad hoc assumptions the hypothesis that retroviral onc genes derive transforming create new unresolved problems. They assume that mutated function from strong viral promoters rather than from muta- proto-onc genes, as, for example, point-mutated proto-ras in tions that affect the coding sequence of native proto-onc colon cancer, cause cancer by second- and higher-order genes (2, 11). Such mutations may merely enhance the mechanisms of , although all known retroviral transforming function derived from retroviral promoters. onc genes, including the retroviral ras genes, are first-order Retrovirus Model Casts Doubts on the Somatic Gene- that are sufficient for carcinogenesis (2, 11). Mutation Hypothesis of Cancer. Retroviral onc genes provide (iii) Nothing would be evolutionarily more implausible for as yet the only proven examples that mutated cellular genes a multicellular organism than a battery of cellular oncogenes can cause cancer. Therefore they are considered the most that can each be activated to cancer genes by point mutations critical support for the hypothesis that mutation of cellular and other mutations, because just a single cancer cell is genes causes cancer (8, 30-32). The hypothesis that mutated sufficient to initiate a clonal cancer (2, 30). Moreover, it cellular genes cause cancer is based on two kinds of circum- would be unlikely that such common mutations as point stantial evidence: (i) many carcinogens are in mutations and substitutions could activate genes that have animals (30) or bacteria (33, 34) and (ii) the coding regions of been optimized over billions of years of evolution. On the the proto-onc genes of some cancers sometimes have muta- contrary, mutations are typically silent or inactivating (11, 41, tions that are similar to those ofthe coding regions ofrelated 42). retroviral onc genes (1, 4-9, 35-37). However, our results (iv) There is no functional proof for cellular cancer genes. cast doubt on the hypothesis that mutations that do not (a) Mutated or "activated" proto-onc genes from tumors do enhance transcription like retroviral promoters can "acti- not transform normal diploid cells upon transfection or in vate" proto-onc genes to cancer genes for several reasons. transgenic animals (10, 11). According to Stanbridge (10), (i) Since retroviral onc genes derive transforming function ". . . despite intensive efforts to transform normal human from strong retroviral promoters rather than from mutations fibroblasts or epithelial cells with varying combinations of

L2: c......

L4: ......

L6: ....a ...... RpSV:ctgcaggcct tcctggagga ctacttcacc tcgacagagc cccagTACca gcctggagag aacctaTAGg cctggctgct

...... L2: ...... a ......

L4: ...... a ......

...... L6: ...... a ...... *atc-c.... RpSV:ccctgcttgt gtgttggagg tcgctgagta agtacgaggc gtgacctaca attgctcaaa taatgcttct gtagaaattg

L2: ......

...... L4: ...... g*ctgag-gg

...... L6: ...... RpSV:tttagcatta ggcgtcctgc gttgctccgc gatgtacggg tcaggtataa tgtgcagttt ga~ctgagggg accatgatgt

.. L2: ......

...... L4 : ......

...... L6 : ...... RpSV:gtataggcgt caagcggggc ttcggttgta cgcggatagg aatcccctca gg FIG. 4. The 3' terminal proto-src sequences of three RpSVs recovered from chicken sarcomas. The sequences of the RpSV DNAs L2, LA, and L6 correspond to the DNAs shown in lanes 2, 4, and 6 in Fig. 3. The sequence from the original RpSV is included for comparison (1). Dots indicate identity. The Tyr codon 527 (TAC) and the stop codon (TAG) of proto-src are capitalized and highlighted. The sequences were determined by Lark Sequencing Technologies (Houston). Downloaded by guest on September 28, 2021 Biochemistry: Wu et al. Proc. Natl. Acad. Sci. USA 89 (1992) 6397 activated cellular oncogenes, the results have been uniformly 11. Duesberg, P. H. & Schwartz, J. R. (1992) Prog. negative." Even combinations of mutated proto-onc genes Res. Mol. Biol. 43, 135-204. have failed to transform cells unless linked to viral promoters 12. Tabin, C. J. & Weinberg, R. A. (1985) J. Virol. 53, 260-265. 13. Hunter, T. (1987) Cell 49, 1-4. (11, 21). By contrast, proviral DNA from oncogenic retrovi- 14. Parker, R. C., Varmus, H. F. & Bishop, J. M. (1984) Cell 37, ruses transforms diploid cells like retroviruses (1, 11). (b) 131-139. Mutated proto-onc genes of tumors do not determine the 15. Iba, H., Takeya, T., Cross, F. R., Hanafusa, T. & Hanafusa, character of a given type of tumor compared to one without H. (1984) Proc. Natl. Acad. Sci. USA 81, 4424-4428. such a gene (11, 35, 43). By contrast, retroviral oncogenes 16. Cichutek, K. & Duesberg, P. H. (1986) Proc. Natl. Acad. Sci. determine many characters of the resulting tumors (1, 44). USA 83, 2340-2344. It appears that the somatic gene-mutation hypothesis of 17. Zhou, R.-P. & Duesberg, P. H. (1988) Proc. Nail. Acad. Sci. USA 85, 2924-2928. cancer is numerically and evolutionarily implausible and 18. Cichutek, K. & Duesberg, P. H. (1989) J. Virol. 63, 1377-1383. functionally unconfirmed. Similar conclusions were reached 19. Zhou, R.-P. & Duesberg, P. H. (1989) Proc. Natl. Acad. Sci. by Rous (45, 46) and Rubin (47) after studying oncogenic USA 86, 7721-7725. viruses and cancer for over 50 and 30 years, respectively. 20. Velu, T. J., Vass, W. C., Lowy, D. R. & Tambourin, P. E. Rous concluded, "A favorite explanation has been that (1989) J. Virol. 63, 1384-1392. oncogens [Rous' term for carcinogens] cause alterations in 21. Chakraborty, A. K., Cichutek, K. & Duesberg, P. H. (1991) the genes of the ordinary cells of the body . . . somatic Proc. Natl. Acad. Sci. USA 88, 2217-2221. mutations as these are termed. But numerous facts, when 22. Bishop, J. M. (1982) Adv. Cancer Res. 37, 1-32. and, 23. Wilkerson, V. W., Bryant, D. L. & Parsons, J. T. (1985) J. taken together, decisively exclude this supposition" (46) Virol. 55, 314-321. "A hypothesis is best known by its fruits. What have been 24. Goodrich, D. W. & Duesberg, P. H. (1990) Proc. Natl. Acad. those of the hypothesis? . . . It acts as a Sci. USA 87, 2052-2056. tranquilizer on those who believe in it, and this at a time when 25. Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. E., every worker should feel goaded now and again by his Seidman, J. G., Smith, J. A. & Struhl, K. (1988) Current ignorance ofwhat cancer is" (45). Likewise, Cairns (48) " . Protocols in Molecular Biology (Wiley, New York). suggests that most human cancers are not caused by con- 26. Saiki, R. K., Gelfand, D. H., Stoffel, S., Scharf, S. J., Higuchi, ventional mutagens...." R., Horn, G. T., Mullis, K. B. & Erlich, H. A. (1988) Science It is conceivable, therefore, that the gene-mutation hy- 239, 487-491. onc 27. Duesberg, P. H. (1968) Proc. Natl. Acad. Sci. USA 60, 1511- pothesis of cancer is limited to the genesis of retroviral 1518. genes from the coding regions ofcellular proto-onc genes and 28. Canaani, E., von der Helm, K. & Duesberg, P. H. (1973) Proc. retroviral promoters. Indeed, retroviruses have evolved Natl. Acad. Sci. USA 70, 401-405. dominant promoters to override cellular controls and thus 29. Pfaff, S. L., Zhou, R.-P., Young, J. C., Hayflick, J. & Dues- have the real potential to activate genes by promoter substi- berg, P. H. (1985) Virology 146, 6389-6393. tution via rare illegitimate recombination (2). There is how- 30. Cairns, J. (1978) Cancer Science and Society (Freeman, San ever no evidence for cellular promoters that could activate Francisco). heterologous cellular genes to retrovirus-like transforming 31. Pitot, H. C. (1986) Fundamentals of Oncology (Dekker, New genes. An alternative hypothesis suggests that nonviral can- York). mutation of but are the 32. Temin, H. M. (1988) Cancer Res. 48, 1697-1701. cers are not caused by singular genes 33. Ames, B. N. (1979) Science 204, 587-593. product of gross genetic imbalances of normal genes due to 34. Strauss, B. S. (1992) Cancer Res. 52. 249-253. chromosome abnormalities (11), as originally proposed in 35. Vogelstein, B., Fearon, E. R., Stanley, B. A., Hamilton, R., 1914 by Boveri (49). Kern, S. E., Preisinger, A. C., Leppert, M., Nakamura, Y., White, R., Smits, A. M. M. & Bos, J. L. (1988) N. Engl. J. We thank Harry Rubin for a critical review ofthe manuscript. This Med. 319, 525-532. research is supported by Outstanding Investigator Grant 5-R35- 36. Weinberg, R. A. (1991) Science 254, 1138-1146. CA39915-07 from the National Cancer Institute. 37. Bishop, J. M. (1991) Cell 64, 235-248. 38. Lewin, B. (1990) Genes IV (Oxford Univ. Press, Oxford). 1. Weiss, R., Teich, N., Varmus, H. & Coffin, J. (1985) Molecular 39. Seeburg, P. H., Colby, W. W., Capon, P. J., Goeddel, D. V. & Biology ofRNA Tumor Viruses (Cold Spring Harbor Lab., Cold Levinson, A. D. (1984) Nature (London) 312, 71-75. Spring Harbor, NY). 40. Stanbridge, E. (1990) Science 247, 12-13. 2. Duesberg, P. H. (1987) Proc. Natl. Acad. Sci. USA 84, 2117- 41. Watson, J. D. (1970) Molecular Biology ofthe Gene (Benjamin, 2124. New York). 3. Zhou, H. & Duesberg, P. H. (1990) Proc. Natl. Acad. Sci. USA 42. Puck, T. (1972) The Mammalian Cell as a Microorganism 87, 9128-9132. (Holden-Day, San Francisco). 4. Bishop, J. M. (1981) Cell 23, 5-6. 43. Barbacid, M. (1987) Annu. Rev. Biochem. 56, 779-827. 5. Tabin, C. J., Bradley, S. M., Bargmann, C. I., Weinberg, 44. Lee, W.-H., Bister, K., Pawson, A., Robins, T., Moscovici, C. R. A., Papageorge, A. G., Scolnick, E. M., Dhar, R., Lowy, & Duesberg, P. H. (1980) Proc. Natl. Acad. Sci. USA 77, D. R. & Chang, E. H. (1982) Nature (London) 300, 143-149. 2018-2022. 6. Reddy, E. P., Reynolds, R. K., Santos, E. & Barbacid, M. 45. Rous, P. (1959) Nature (London) 183, 1357-1361. (1982) Nature (London) 300, 149-152. 46. Rous, P. (1967) Science 157, 24-28. 7. Varmus, H. E. (1984) Annu. Rev. Genet. 18, 533-612. 47. Rubin, H. (1980) J. Natl. Cancer Inst. 64, 995-1000. 8. Bishop, J. M. (1987) Science 235, 305-311. 48. Cairns, J. (1981) Nature (London) 289, 353-357. 9. Cooper, G. M. (1990) Oncogenes (Jones & Bartlett, Boston). 49. Boveri, T. (1914) Zur Frage der Enstehung maligner Tumoren 10. Stanbridge, E. J. (1990) Annu. Rev. Genet. 24, 615-657. (Gustav Fischer, Jena, Germany). Downloaded by guest on September 28, 2021