Replication-competent adenoviral vectors are potentially far more efficient than replication-defective vectors. However, for reasons of safety, there is a need to restrict viral replication both spatially, by limiting replication to certain cell types, and temporally. To control replication temporally, we have developed a system, based on the small-molecule dimerizer rapamycin, for regulating the replication of adenoviral vectors. In this system, one adenoviral vector, AdC4, expresses transcription factors whose activity is regulated by the non-immunosuppressive rapamycin analog AP21967. A second vector, Ad(Z12-I-E1aE1b19k), contains E1 genes placed downstream of binding sites for the regulated transcription factor. Co-infection of several cell lines by the vector pair leads to dimerizer-dependent E1 expression and an increase in viral replication, as shown by Southern blots and replication assays. Furthermore, expression of a reporter gene from a replication-defective vector, Ad-GM-CSF, can be augmented by up to 18-fold by co-infection with the pair of conditionally replicating vectors in the presence of dimerizer. Similar results are obtained when the vectors are directly injected into subcutaneous HT1080 xenograft tumors in nude mice. We believe that vectors based on this principle will be a useful additional tool to enhance efficiency and safety of gene delivery for anti-cancer therapy.

High-level systemic delivery of viral vectors to tumors has proved problematic as a result of immune neutralization, nonspecific adhesion, and clearance of circulating viral particles. Some cell types localize to tumors in response to particular biological properties associated with tumor growth. Their use to deliver viral vectors to tumors would allow precious viral stocks to be protected until they can be released at high local concentrations. Here, we describe a mechanism by which retroviral vector production by T cells can be regulated by a tumor-specific trigger through engagement of a chimeric immune receptor (CIR) with its target antigen. The virus that is released from the T cells can also be transcriptionally targeted. Finally, we show that it is possible to use vector-loaded, antigen-triggered human T cells as therapeutic, tumor-specific vector delivery cells in models of both local intratumoral and systemic delivery to both lung and liver metastases. This strategy incorporates multiple levels of targeting into the delivery system at the stages of surface targeting, viral production, and gene expression.

In the version of this article initially published, 'n' was incorrectly defined on p. 981, the first line after equation 3, as the "apparent number of osmotically active moles per cell." The correct phrase is, "apparent moles of osmotically active particles per cell." The error has been corrected in the PDF version of the article.