Raf revertant cells resist transformation by non-nuclear oncogenes and are deficient in the induction of early response genes by TPA and serum.

A revertant cell line was generated from v-raf transformed NIH/3T3 fibroblasts. These cells, termed CHP25, express a functional v-raf oncogene. However, they are non-tumorigenic, do not form colonies in soft agar and possess a flat morphology. CHP25 cells are resistant to re-transformation by sis, ras, tyrosine kinase- as well as serine/threonine kinase-encoding oncogenes suggesting that Raf functions downstream of most membrane associated signal transducers. In contrast to v-raf transformed cells, in which the endogenous Raf-1 protein kinase is constitutively activated, v-Raf in CHP25 cells does not activate endogenous Raf-1 kinase. Since mitogen regulation of Raf-1 kinase in CHP25 cells is intact, we conclude that CHP25 cells are blocked at the level of Raf-1 substrate phosphorylation. Consistent with this interpretation CHP25 cells show specific alterations of early gene induction. The serum induction of c-fos and junD as well as the serum and TPA (12-O-tetradecanoylphorbol-13-acetate) induction of junB and egr-1 are almost completely abolished. Only v-fos can transform CHP25, whereas c-fos, v-myc, c-jun and junB are ineffective. These data suggest that the lesion responsible for the revertant phenotype of CHP25 cells is the inability to activate the AP-1 complex. We conclude that Raf-1 signaling is essential for transformation of NIH/3T3 cells by peripheral oncogenes and for regulation of a subset of early response genes by TPA and serum growth factors.

Isolation of rsp-1, a novel cDNA capable of suppressing v-Ras transformation.

Using an expression cloning assay, we have isolated a novel cDNA, referred to as rsp-1, which suppresses the v-Ras-transformed phenotype. When introduced into NIH 3T3 fibroblasts under the control of a metallothionein promoter, rsp-1 confers resistance to v-Ras, but not to v-Mos or v-Src, and inhibits growth of the cells. The rsp-1 cDNA contains a 831-bp open reading frame encoding a 277-amino-acid leucine-rich protein. The rsp-1 cDNA exhibits no significant homology to sequences in the DNA data bases. However, searches of the protein data bases revealed that it contains a series of leucine-based repeats which are homologous to the leucine repeats found in the regulatory region of the yeast adenylyl cyclase. rsp-1 specific RNA is detectable in a wide variety of cell lines and tissues, and the gene is conserved among eukaryotic species. These data suggest that rsp-1 plays a role in Ras signal transduction.

Long-term in vivo expression of genes introduced by retrovirus-mediated transfer into mammary epithelial cells.

Nonimmortalized mouse mammary epithelial cells expressing Escherichia coli beta-galactosidase from a murine amphotropic packaged retroviral vector were injected into the epithelium-divested mammary fat pads of syngeneic mice. Mammary glands formed from the injected mammary epithelial cells contained ductal and lobular cells, both of which expressed beta-galactosidase when examined in situ more than 12 months later. These results indicate that stable recombinant gene expression can be achieved in vivo in the mammary gland without altering the growth properties of normal mammary epithelium.

Increased sensitivity of nontumorigenic fibroblasts expressing ras or myc oncogenes to malignant transformation induced by 5-aza-2'-deoxycytidine.

The hypomethylating chemotherapeutic drug 5-aza-2'-deoxycytidine (5AzadC) has been shown to induce cell differentiation in some systems, while promoting neoplastic transformation in others. Using both in vitro and in vivo models, we have explored the relationship between oncogene expression and the susceptibility of cells to malignant transformation by 5AzadC. The study involved several nontumorigenic subclones of NIH3T3 fibroblasts, including cells transfected with deregulated c-myc, as well as phenotypic revertants expressing v-Ki-ras or long terminal repeat-activated c-Ha-ras. Transient 5AzadC treatment of the oncogene-bearing cell lines was associated with a rapid and efficient neoplastic transformation. In some cases, over 50% of the cell population exhibited loss of contact inhibition of growth within 1 week of treatment. The transformants were capable of forming s.c. tumors and experimental lung metastases in recipient nude mice. In contrast, 5AzadC failed to induce malignant properties in control 3T3 cultures transfected with the bacterial neor gene; rather, treatment of these cells was associated with differentiation into adipocytes and myotubes. The differential response to 5AzadC was also observed in vivo, in mice first inoculated s.c. with the premalignant cells and then treated with 5AzadC 24 h later. In agreement with the in vitro model, tumor development in mice correlated with the presence of cells with activated ras or myc oncogenes. Cytidine analogs that do not inhibit DNA methylation (i.e., 6-azacytidine and 1-beta-D-arabinofuranosyl cytosine) had no effect on cell phenotype. The results indicate that exposure of cells to 5AzadC can lead to tumor progression both in vitro and in vivo and suggest that preexisting alterations in oncogene expression may facilitate the evolution of cancerous growth induced by hypomethylating agents.

Isolation of a new class of 'flat' revertants from ras-transformed NIH3T3 cells using cis-4-hydroxy-L-proline.

A new class of nontransformed revertant cells has been isolated from the ras-transformed cell line DT using cis-4-hydroxy-L-proline (CHP) as a selective agent. The new revertants, CHP 9CJ and CHP CB4, each contain two copies of the v-Ki-ras gene, elevated levels of phosphorylated p21ras protein, and rescuable transforming virus, indicating that the revertant phenotype observed in these cells does not result from inactivation of v-Ki-ras or inhibition of its expression. Both CHP 9CJ and CHP CB4 revertants show a greatly reduced ability to form colonies in soft agar and to produce tumors in syngeneic mice. CHP 9CJ cells are resistant to retransformation by ras and by additional oncogenes that do not encode tyrosine kinases. A comparison of oncogene resistance patterns in these CHP-derived revertants with those from our original ouabain-derived revertants fos C-11 and F-2 indicates that oncogenes may be divided into four general groups. Oncogenes that encode proteins structurally related to p21ras comprise the first group. The second group contains only tyrosine kinase-encoding oncogenes. The third group is composed of 'nuclear', e.g. fos, and 'cytoplasmic' serine-threonine-encoding oncogenes such as mos and raf. The fourth group contains the oncogenes sis and fms.

Efficient induction of focus formation in a subclone of NIH3T3 cells by c-myc and its inhibition by serum and by growth factors.

In both experimental and spontaneous tumors, c-myc expression is often enhanced following its amplification or its rearrangement adjacent to a strong promotor/enhancer. However, c-myc by itself does not induce foci efficiently in fibroblast cultures. The effect of high levels of c-myc expression from a retroviral construct was investigated in several rodent fibroblast cell lines grown in medium containing 10% fetal calf serum or in serum-free PC-1 medium. c-myc-infected NIH3T3 clone 7 cells exhibited efficient quantitative focus formation when grown in PC-1 medium, whereas foci were not detected when grown in serum-supplemented medium. NIH3T3 clone 7 was the only cell line found to be sensitive to c-myc; other clones of NIH3T3 or other rodent fibroblast cell lines proved to be resistant to c-myc focus formation. At least two major types of morphologically distinct c-myc-induced foci were observed; the first was similar to ras-transformed foci induced in NIH3T3 and other fibroblast cell lines, and the second type was composed of adipocyte-like cells similar to NIH3T3 L1 cells. The c-myc infected cells cloned from these two types of foci expressed high levels of retrovirus-derived c-myc RNA and exhibited elevated levels of immunoreactive myc protein, as detected by immunofluorescent staining with an anti-myc polyclonal antibody. c-myc-transformed clones displayed only a limited ability to grow in soft-agar in the presence of serum and were not tumorigenic in nude mice. Focus formation by c-myc was quantitatively inhibited by the addition of interferon alpha + beta (INF alpha, beta), tumor necrosis factor alpha (TNF alpha) or transforming growth factor beta 1 (TGF beta 1) to the serum-free PC-1 medium, and, in correlation, NIH3T3 clone 7 cells produced the lowest level of endogenous TGF beta of the various cell lines tested.

Differential growth factor expression in transformed mouse NIH-3T3 cells.

The expression of growth factor-specific mRNA transcripts and the presence of biologically active growth factors in the conditioned medium and in the cell extracts from mouse NIH-3T3 cells transformed by different oncogenes (Ki-ras, mos, src, fms, fes, met, and trk), by a DNA tumor virus (SV40), or by a chemical carcinogen (N-nitrosomethylurea) were studied. In contrast to NIH-3T3 cells or simian virus 40 (SV40)-transformed 3T3 cells, all the other transformed NIH-3T3 cell lines expressed a 4.5 kb transforming growth factor-alpha (TGF alpha)-specific mRNA transcript and secreted immunoreactive and biologically active TGF alpha ranging from 100 to 225 ng/10(8) cells/48 h. In addition, in the transformed cell lines that were secreting elevated levels of biologically active TGF alpha, there was a 75-95% reduction in the total number of epidermal growth factor receptors on these cells. A 2.6 kb TGF beta mRNA transcript and TGF beta protein in the conditioned medium (30-140 ng/10(8) cells/48 h) was also detected in those lines expressing TGF alpha. Basic fibroblast growth factor-like activity (11-50 ng/10(8) cells) was detected in the cell lysates from NIH-3T3 cells transformed with N-nitrosomethylurea or with trk, where expression of specific 6.9 and 3.9 kb mRNA transcripts for basic fibroblast growth factor could also be found. B chain (c-sis) expression of platelet-derived growth factor was present only in trk-transformed NIH-3T3 cells in which specific c-sis 6.5 and 4.6 kb transcripts were identified. In contrast, platelet-derived growth factor A chain expression of 2.9 and 2.3 kb transcripts was found in ras-, met-, mos-, and fms-transformed NIH-3T3 cells. These results suggest that the expression of different sets of growth factors is controlled in part by structurally distinct groups of transforming genes.

Detection of genes with a potential for suppressing the transformed phenotype associated with activated ras genes.

Seven morphologically nontransformed (flat) revertants with reduced tumorigenicity in vivo have been isolated from populations of Kirsten sarcoma virus-transformed NIH 3T3 cells transfected with a cDNA expression library of normal human fibroblasts. Each revertant harbors 1-10 recombinant plasmids per cell and retains a rescuable transforming virus as well as high level expression of v-Ki-ras-specific RNA and the viral oncogene product, p21v-Ki-ras. Transformed phenotypes are suppressed in cell hybrids generated by fusing each revertant to v-Ki-ras-transformed NIH 3T3 cells. From two of the revertant lines, plasmids capable of giving rise to flat secondary transfectants have been recovered. Thus, in some, if not all, of the revertants, transfected cDNAs seem to be responsible for the suppression of specific transformed phenotypes.

Transformation of an established mouse mammary epithelial cell line following transfection with a human transforming growth factor alpha cDNA.

To determine whether the enhanced expression of transforming growth factor alpha (TGF alpha) is sufficient to induce the neoplastic transformation of an immortalized population of mammary epithelial cells, we cotransfected NOG-8 cells, a cloned mouse mammary epithelial cell line, with a simian virus 40-human TGF alpha cDNA expression vector plasmid and a pSV2neo plasmid. After cotransfection, nine G418-resistant NOG-8 colonies were cloned and expanded. All clones were subsequently analyzed for TGF alpha mRNA expression by northern blot analysis, TGF alpha secretion, anchorage-dependent growth in serum-free medium, anchorage-independent growth in soft agar, and tumorigenicity in nude mice. Three TGF alpha-transfected NOG-8 clones expressed high levels of a specific TGF alpha mRNA, secreted elevated levels of TGF alpha into the culture medium (177-595 ng/10(8) cells/48 h), exhibited an enhanced growth rate, grew aggressively as colonies in soft agar, and formed undifferentiated, invasive carcinomas in nude mice. A neutralizing mouse monoclonal antibody generated against the low molecular weight human TGF alpha peptide was able to inhibit colony formation in soft agar by TGF alpha-transfected NOG-8 clones that produced high levels by TGF alpha. This inhibition suggested that TGF alpha acted through an external autocrine loop. NOG-8 cells and NOG-8 cells transfected with a pSV2neo plasmid alone secreted very low levels of TGF alpha, failed to grow as colonies in soft agar and did not form tumors in nude mice. These results demonstrate that overexpression of a human TGF alpha cDNA in immortalized, nontransformed mouse mammary epithelial cells can induce a transformed phenotype in vitro and can facilitate tumor formation in vivo.

Differential growth sensitivity to 4-cis-hydroxy-L-proline of transformed rodent cell lines.

The effect of 4-cis-hydroxy-L-proline (CHP), a proline analogue, on the anchorage-dependent and -independent growth of several transformed rodent cell lines was studied. Mouse NIH-3T3 fibroblasts transformed by a variety of different oncogenes (Ki-ras, mos, src, fms, fes, met, and trk) by a DNA tumor virus (SV40) or by a chemical carcinogen (N-methylnitrosourea) were all found to be more sensitive (50% inhibitory dose, 20 to 55 micrograms/ml) to the dose-dependent inhibitory effects of CHP on growth in monolayer culture than were NIH-3T3 cells (50% inhibitory dose, 120 micrograms/ml). CHP was generally found to be even more effective in inhibiting the growth of these transformed cells as colonies in soft agar than in monolayer cultures. In addition, rat embryo fibroblasts (CREF) and normal rat kidney fibroblasts (NRK) after transformation with a Ki-ras oncogene exhibit a similar increase in their sensitivity to CHP-induced growth inhibition. Treatment of NRK cells with transforming growth factor alpha (TGF-alpha) and beta (TGF-beta), which reversibly induces phenotypic transformation of these cells, increases their sensitivity to CHP to a level comparable with that observed in Ki-ras-transformed NRK cells (K-NRK). The growth inhibitory effects of CHP are reversible, since removal of CHP results in a normal resumption of cell growth. CHP uptake occurs primarily through the Na+- and energy-dependent neutral amino acid transport A system, which is 6- to 7-fold more elevated in K-NRK cells compared with NRK cells. Treatment of NRK cells with TGF-alpha and/or -beta increases the uptake of [3H]methylaminoisobutyric acid on the A system to a level that is similar to that found in K-NRK cells. The functions of the Na+/K+ and Na+/H+ exchange systems are apparently necessary for the enhanced A system activity, since ouabain and amiloride can inhibit the uptake of [3H]methylaminoisobutyric acid in K-NRK cells and in NRK cells treated with TGF-alpha and/or -beta. The activity of the A system is specifically increased in K-NRK and in TGF-alpha- and/or -beta-treated NRK cells, since the other two major neutral amino acid uptake systems, the ASC and the L systems, and the Ly+ system for basic amino acid uptake show no apparent changes in their activity in NRK cells after treatment with TGF-alpha and/or -beta or in these cells after transformation with the Ki-ras oncogene. These results suggest that the differential growth sensitivity to CHP of transformed rodent cells and of normal fibroblasts treated with TGF-alpha and/or -beta is due in part to an elevated uptake of this amino acid analogue on the neutral amino acid transport A system.

A system for deriving revertants of oncogene-transformed human cells.

We have previously shown that some transformed derivatives of the human osteosarcoma-derived cell line HOS are killed by treatment with 1 microM ouabain at pH 8.2, whereas their nontransformed counterparts are relatively unharmed by the same conditions. HOS cells transformed by v-Ki-ras and RAS, v-fms, or MET are susceptible to 1 microM ouabain while those transformed by v-fes are not. Here we describe the adaptation of this differentially cytotoxic effect as a method to enrich for cells which revert to a nontransformed phenotype. We have optimized parameters which increase the differential cytotoxicity, including pH and potassium concentration during and subsequent to ouabain treatment. The efficiency of this procedure was tested in mixed cell experiments where model populations were constructed consisting of HOS cells mixed with an excess of v-Ki-ras-transformed HOS cells. Two successive OAK treatments (ouabain/alkaline/K+-free) were sufficient to recover nontransformed cells free of ras-transformants as indicated by genetic markers and morphology. This HOS/ouabain system is currently being used to derive revertants of ras-transformed human cells and could facilitate the isolation of genes interacting in the pathways by which these cells are transformed.

Effects of ouabain on NIH/3T3 cells transformed with retroviral oncogenes and on human tumor cell lines.

Both murine and human cell lines transformed by the v-Ki-ras gene have been shown to be much more sensitive to the toxic effects of the cardiac glycoside ouabain than their respective controls. This differential toxicity has previously been used in the isolation of flat revertant clones from populations of Kirsten murine sarcoma virus transformed NIH/3T3 cells. Here, we have undertaken a further characterization of this phenomenon in murine and human tumor cells. Two different techniques, a 51Cr-release assay and a quantitative Crystal violet elution assay, have been employed to compare the sensitivities to ouabain of normal and v-Ki-ras-transformed NIH/3T3 cells. In each assay, ras-transformed NIH/3T3 cell lines displayed an increased sensitivity to ouabain as compared to the parental NIH/3T3 cell line, both in dose-response and in time-course experiments. In a separate study, ouabain was also able to inhibit the growth in semi-solid medium of 2 v-Ki-ras-transformed NIH/3T3 cell lines (DT and K-NIH) in a dose-dependent fashion. The same concentrations of ouabain were effective in both the 51Cr-release and Crystal violet assays. To address the question of whether increased sensitivity to ouabain is a specific result of transformation with the ras oncogene or is a common event which accompanies transformation by other oncogenes, we have screened a variety of transformed NIH/3T3 derivatives. All of these lines displayed an increased sensitivity to ouabain when compared to the parental NIH/3T3 cell line.

Transformation by oncogenes encoding protein kinases induces the metastatic phenotype.

Oncogenes encoding serine/threonine or tyrosine kinases were introduced into the established rodent fibroblast cell line NIH 3T3 and tested for tumorigenic and metastatic behavior in T cell-deficient nude mice. Transforming oncogenes of the ras family were capable of converting fibroblast cell lines to fully metastatic tumors. Cell lines transformed by the kinase oncogenes mos, raf, src, fes, and fms formed experimental metastases and (in some cases) these genes were more efficient at metastatic conversion than a mutant ras gene. In contrast, cells transformed by either of two nuclear oncogenes, myc or p53, were tumorigenic when injected subcutaneously but were virtually nonmetastatic after intravenous injection. These data demonstrate that, in addition to ras, a structurally divergent group of kinase oncogenes can induce the metastatic phenotype.

Suppression of synthesis and utilization of tropomyosin in mouse and rat fibroblasts by transforming growth factor alpha: a pathway in oncogene action.

Two events which commonly occur during transformation of murine and avian fibroblasts by retroviral oncogenes are production of transforming growth factor alpha (TGF-alpha) and suppression of tropomyosin synthesis. TGF has been proposed as a mediator of transformation through autocrine stimulation. Suppression of tropomyosin synthesis may contribute to the transformed phenotype through destabilization of actin microfilaments and cytoskeletal derangement. To determine whether suppression of tropomyosin synthesis might be a consequence of the action of TGF-alpha we studied tropomyosin synthesis in rat (normal rat kidney) and mouse (NIH3T3) fibroblasts treated with TGF-alpha. In a serum-containing system, addition of TGF-alpha or epidermal growth factor to normal rat kidney monolayers in subnanomolar concentrations induced morphological changes consistent with transformation. These changes were accompanied by prominent suppression of synthesis of Mr 36,000 and 39,000 tropomyosins. Similar suppression was observed in NIH3T3 cells. Inhibition of tropomyosin synthesis began almost immediately after addition of TGF-alpha and became progressively more pronounced during the succeeding 48 h. Suppression of tropomyosin synthesis was correlated with reduced expression of 1.1- and 1.8-kilobase tropomyosin mRNAs in both TGF-treated normal rat kidney cells and v-Ki-ras-transformed NIH3T3 cells. Rapid onset of a specific block in utilization of newly synthesized tropomyosin for formation of cytoskeletal elements was also demonstrated following TGF-alpha treatment. The evidence suggests that this block may be a specific effect of TGF-alpha treatment and that reduced expression of tropomyosin gene products may be either an independent event or a regulatory consequence of the block to utilization. The data support the conclusion that suppression of tropomyosin synthesis in cells transformed by a number of retroviral oncogenes results from the autocrine action of TGF-alpha.

Ouabain sensitivity is linked to ras -transformation in human HOS cells.

Mouse cells transformed by the retroviral oncogene v-Ki- ras are significantly more sensitive to the toxic effects of 1mM ouabain than are their nontransformed counterparts. We have extended these findings to a human cell line (HOS). HOS cells (ATCC CRL 1543) are relatively resistant to treatment with 1 microM ouabain while KHOS cells (transformed by Kirsten murine sarcoma virus) are extremely sensitive. Two flat revertant cell lines isolated from the KHOS line and lacking the v- ras gene sequences are resistant to ouabain. This effect may be observed morphologically and can also be demonstrated by dye exclusion and plating efficiency tests. In addition, the toxic effects of ouabain may be rapidly and efficiently quantitated in a 51Cr-release assay. This differential lethality may be used to enrich the proportion of non-transformed revertants in populations of mutagen-treated transformed cells.

Cyclic AMP suppresses expression of v-rasH oncogene linked to the mouse mammary tumor virus promoter.

Clone 433.3 of NIH 3T3 cells is a stable carrier of the MMTV LTR:v-rasH chimeric DNA. Only in the presence of dexamethasone (a synthetic glucocorticoid), 433.3 cells exhibit an induced level of p21 transforming protein and phenotypic transformation. N6,O2'-dibutyryl cAMP (DBcAMP) antagonized the effect of dexamethasone in a time - and concentration - dependent manner. DBcAMP (5 X 10(-4)M) added 18 hr prior to the addition of dexamethasone (10(-7)M) almost completely blocked the hormone effect: cells contained levels of p21 20% of that in the cells treated with dexamethasone alone, and formed flat, contact inhibited monolayers. On the basis of these results together with our previous data on mammary carcinomas in vivo, we postulate that cAMP may be an intracellular suppressor acting at a regulatory locus of both cellular and viral ras genes.

Suppression of tropomyosin synthesis, a common biochemical feature of oncogenesis by structurally diverse retroviral oncogenes.

To identify proteins whose production may be altered as a common event in the expression of structurally diverse oncogenes, we compared two-dimensional electropherograms of newly synthesized proteins from NIH/3T3 cell lines transformed by a variety of retroviral oncogenes, from cellular revertant lines, and from a line (433.3) which expresses the v-ras oncogene in response to corticosteroids. Most alterations in the synthesis of specific proteins detected by this approach appeared to be the result of selection during prolonged cultivation and were probably unrelated to the transformation process. However, we detected seven proteins whose synthesis was strongly suppressed in cell lines transformed by each of the six retroviral oncogenes we studied and whose production was fully or partially restored in two cellular revertant lines. Suppression of two of these proteins was also correlated with the initial appearance of morphological alteration during corticosteroid-induced oncogene expression in 433.3 cells. These proteins (p37/4.78 and p41/4.75) were identified as tropomyosins, a group of at least five cytoskeletal proteins. Transformation by the papovaviruses simian virus 40 and polyomavirus caused no suppression of synthesis of these tropomyosins. This indicates that suppression of tropomyosin synthesis is not a nonspecific response by cells to being forced to grow with the transformed phenotype but is specifically associated with oncogenesis by diverse retroviral oncogenes. The results are consistent with the hypothesis that the different biochemical processes initiated by expression of structurally diverse retroviral oncogenes may converge on a limited number of common targets, one of which is the mechanism which regulates the synthesis of tropomyosins.

Flat revertants derived from Kirsten murine sarcoma virus-transformed cells produce transforming growth factors.

Two flat cellular revertant cell lines, F-2 and C-11, which were originally selected from the DT line of Kirsten murine sarcoma virus (Ki-MuSV)-transformed NIH/3T3 cells, were examined for the production of transforming growth factors (TGFs). The revertant cells fail to grow in semisolid medium as colonies and exhibit a markedly reduced level of tumorigenicity in nude mice, although they are known to express high levels of p21ras, the product of the Kirsten sarcoma virus oncogene, ras, and they contain a rescuable transforming virus. TGF activity associated with the transformed, revertant, and non-transformed cell lines was measured by the ability of concentrated conditioned medium (CM) from these cells to induce normal rat kidney (NRK) and NIH/3T3 cells to form colonies in semisolid agar suspension cultures and to inhibit the binding of 125I epidermal growth factor (EGF) to specific cell surface receptors. CM from the transformed DT cells and from both the F-2 and C-11 revertants contains TGF activity, in contrast to CM obtained from normal NIH/3T3 cells. Furthermore, unlike NIH/3T3 cells, neither the DT nor the revertant cells were able to bind 125I EGF. All four cell lines were able to proliferate in serum-free medium supplemented with transferrin, insulin, EGF, and Pedersen fetuin. However, in basal medium lacking these growth factors, only DT cells and, to a lesser extent, the revertant cells were able to grow. These results suggest that the F-2 and C-11 revertants fail to exhibit all of the properties associated with transformation because the series of events leading to the transformed phenotype is blocked at a point(s) distal both to the expression of the p21 ras gene product and also to the production of TGFs and that the production of TGFs may be necessary but not sufficient for maintaining the transformed state.

Flat revertants isolated from Kirsten sarcoma virus-transformed cells are resistant to the action of specific oncogenes.

Two flat revertants have been isolated from mutagen-treated populations of Kirsten murine sarcoma virus (Ki-MuSV)-transformed NIH/3T3 cells. These revertants, which appear to be cellular variants resistant to transformation by the Ki-MuSV oncogene v-Ki-ras, contain Ki-MuSV-specific DNA, elevated levels of the v-Ki-ras gene product p21, and rescuable transforming virus. Cell hybridization studies indicated that the revertant phenotype is dominant in hybrids between revertant cells and cells transformed by Ki-MuSV or the closely related Harvey MuSV and BALB MuSV. Analysis of hybrid cells resulting from the fusion of these revertants to cell lines transformed by other retroviruses showed that the action of certain oncogenes structurally unrelated to v-Ki-ras also could be suppressed. Thus, there appear to be functional relationships and diversities among transforming genes (oncogenes) not readily apparent from their structural characteristics.