We describe experiments that show that simian virus 40 (SV40) T antigen is required fQr viral excision from host chromosomes at some point prior to or during the homologous recombination events that create circular wildtype virus. Two recombinant SV40-pBR322 plasmids were constructed such that homologous recombination across similar-sized but different duplications of SV40 would reconstitute wild-type viral DNA. One plasmid (pSVED) Cells that can be infected productively by simian virus 40 (SV40), such as African green monkey cells, are thought to be permissive for the lytic cycle because they express one or more chromosomal genes that specify factor(s) that are essential for viral replication. In addition, studies of permissive cells infected with temperature-sensitive mutations in the SV40 A gene (3) demonstrated that SV40 large T antigen is also required for successful initiation of viral replication. In contrast, SV40 can neither replicate in nor productively infect cells lacking the permissive factor(s), although viral sequences can persist in these nonpermissive cells if they become integrated and subsequently carried in the host's genome in a proviral form (3).Though the mechanism for SV40 integration is poorly understood, recent sequencing of parental DNAs indicates that small DNA homologies at the crossover points between virus and chromosome may play some role in this process (2, 4). Integration of SV40 into the chromosomes of nonpermissive cells generally results in a tandem duplication of the inserted proviral sequences though nontandemn, single-copy, proviral inserts have been described (1, 5-7).The association between proviral and chromosomal sequences in these SV40-transformed cells can be destroyed by inducing cell fusion to cells permissive for SV40 replication, such as those of the simian line CV-1 (3). When transformed cells containing tandem duplications of SV40 are fused to permissive cells, freely replicating, form I SV40 DNA molecules are detected in 10-50% of the resulting heterokaryons and these forms are generated via homologous recombination across the tandem duplications (8, 9).Botchan et al. (8) proposed that rather than providing a specific factor for viral excision, fusion to permissive cells induced viral excision by establishing within the heterokaryons all of the components necessary for the initiation of in situ proviral replication. They found that an increase in high molecular weight viral DNA still associated with the chromosome preceded viral excision and they proposed that upon initiation of replication at a given proviral locus, multiple rounds of DNA synthesis occurred to form a localized "onion skin" of amplified sequences. They suggested that resolution of this aberrant, polytenized region would require the intervention of repair enzymes that either directly or indirectly lead to the excised forms. It was clear that this sort of in situ amplification model could also be used to explain the spontaneous induction of virus in "semi-permissive" lines as well as t...