Replication competent retrovirus (RCR) may arise through homologous recombination in virus producer cells. A seminal study by Donahue (J.Exp.Med. 1992) showed that infection with RCR may lead to development of cancer in primates. Therefore, rigorous testing of clinical grade preparations of retroviral vectors is required to exclude the presence of RCR. Many regulatory agencies require RCR-testing of ex vivo transduced cells as well. The latter requirement originates from the presumed amplification of RCR during culture, if cells are infected by very low levels of RCR. However, in a previous study we found that T lymphocytes have a very low capacity to produce RCR (Ebeling, Gene Ther. 2003).In this study we present the mechanism, which permit T cells to efficiently abort RCR-infection. Individual clones were established from the bulk cultures of RCR-infected T cells. None of the 19 clones tested produced RCR, although all clones had been infected initially, as evidenced by an LTR-specific PCR on genomic DNA. Intriguingly, only two clones contained the env gene, suggesting that part of the proviral genome had been deleted from the remainder of the clones. To establish the extent of the deletions, four PCR assays were designed, which together cover the entire proviral genome. Four clones were negative in all four PCR assays, suggesting that a very large part of the proviral genome had been deleted. One clone, #27, was positive in the PCR assay, which covers almost the entire env gene. A second clone, #36, was positive in two PCR assays, which together cover half of the pol and the entire env gene.Sequence analysis revealed that the 3′ end of the PCR product from clone #27 contained numerous G to A substitutions, which is consistent with the activity of the recently discovered antiviral APOBEC3G gene (e.g. Harris, Cell 2003, and Mangeat, Nature 2003). We therefore conclude that APOBEC3G is involved in abortion of the RCR-infection in human T cells.The extent of the proviral deletions was established through analyis of the retroviral integration sites in six clones via LM-PCR. In clones #26, #37, and #27 3′ end retroviral sequences were preceded by host DNA, definitively confirming deletion of the proviral genome. Intriguingly, in clones #26 and #31, aberrant proviral sequences were found in which part of the provirus was present in an ‘upside down' orientation. In the case of clone #31, this ‘flipped' sequence had also changed its position in the provirus. Such flipped sequences were never found during our analysis of > 156 independent integration sites of recombinant retroviruses in hematopoietic stem cells (Laufs, Blood 2003). To our knowledge, deletions or aberrant proviral configurations were never found to be associated with APOBEC3G in more than 30,000 bp of retro- or lentiviral DNA, synthesized in the presence of APOBEC3G. In summary, our work suggests the existence of a novel intracellular defense mechanism, beside APOBEC3G, in primary T lymphocytes, which leads to an efficient deletion of murine retroviral sequences. We believe that these experiments provide sufficient reason to reconsider the necessity of RCR-testing of ex vivo transduced T lymphocytes for gene therapy purposes. If this mechanism is also operational on lentiviral sequences, this may open up novel antiviral strategies for treatment of AIDS.