We used a murine retrovirus shuttle vector system to construct recombinants capable of constitutively expressing the simian virus 40 (SV40) large T antigen and the polyomavirus large and middle T antigens as well as resistance to G418. Subsequently, these recombinants were used to generate cell lines that produced defective helper-free retroviruses carrying each of the viral oncogenes. These recombinant retroviruses were used to analyze the role of the viral genes in transformation of rat Flll cells. Expression of the polyomavirus middle T antigen alone resulted in cell lines that were highly tumorigenic, whereas expression of the polyomavirus large T resulted in cell lines that were unaltered by the criteria of morphology, anchorageindependent growth, and tumorigenicity. More surprisingly, SV40 large T-expressing cell lines were not tumorigenic despite the fact that they contained elevated levels of cellular p53 and had a high plating efficiency in soft agar. These results suggest that the SV40 large T antigen is not an acute transforming gene like the polyomavirus middle T antigen but is similar to the establishment genes such as myc and adenovirus EIa.The DNA tumor viruses polyomavirus (Py) and simian virus 40 (SV40) are both capable of transforming both primary and established cell lines in vitro (for a review see reference 70). Extensive work has shown that genes encoded by the early regions of these viruses are both necessary and sufficient for transformation (1,10,12,14,15,18,29,38,43,49,67; for a review see reference 70). Analysis of the respective roles of these genes has been complicated because the early regions are multiply spliced to specify more than one protein: Py encodes three antigens, the large T (100 kilodaltons [kDa]), the middle T (56 kDa), and the small T (22 kDa) (20,21,24,25,58); SV40 encodes two antigens, the large T (95 kDa) and the small T (20 kDa) (48,51,69). In addition, transformation assays with DNA tumor viruses have generally involved infection or transfection of a cell population and selection of a small fraction of cells with a particular phenotype. For example, this selection could be based on either focus formation, anchorage-independent growth, or growth in low serum, and thus, this type of assay does not necessarily reveal the average phenotype resulting from interaction of the oncogene with the cell. Moreover, the process by which the viral DNA segment integrates into cellular chromosomes can give rise to multiple copies and rearrangements. The resulting variation in levels of viral mRNA and gene expression can significantly affect the range of phenotypes of the transformants. Previous studies of the variation in phenotype observed after transformation by SV40 detected clonal cell lines that ranged over completely, partially, or minimally transformed (50, 56).We therefore undertook a systematic analysis of the role of the various genes encoded by the early regions of these viruses by isolating cell lines which constitutively express these proteins without prior selection ...