Retroviral targeting to tumour antigens

The development of efficient, cell surface targeted retroviral vectors is critical for successful gene therapy as gene delivery to non-target cells may be harmful and would deplete the pool of viral particles. To date, the only surface-targeting strategies that have allowed efficient infection by retroviral vectors in vivo are those that have limited the tropism of amphotropic murine leukaemia virus (MLV-A), which can infect cells of many mammals, by modification of the envelope glycoprotein. To this end we have explored tumour targeting of vectors based on MLV-A by modification of the retroviral surface protein (SU) backbone of the envelope chimera. The first approach used a receptor co-operation strategy to target human tumour cells by linking single chain antibodies (scFv) recognising tumour antigens (carcinoembryonic antigen (CEA) and high molecular weight melanoma-associated antigen (HMWMAA)) via proline-rich spacers to the amphotropic murine leukaemia virus surface protein. This approach showed selective targeting to both CEA and HMWMAA in vitro. The second approach used a protease targeting strategy to target tumour cells expressing CEA. We fused a single-chain variable fragment (scFv) directed against CEA to the amphotropic murine leukaemia virus envelope. A proline-rich hinge and matrix metalloprotease cleavage site linked the two proteins. Following attachment to CEA, MMP cleavage of the envelope at the cell surface removed the scFv and proline-rich hinge allowing infection. This approach showed selective targeting to carcinoembryonic antigen (CEA) both in vitro and in vivo with up to 10% infection of cells within a CEA-positive tumour xenograft. No infected cells were detected after delivery of targeted vectors to CEA-negative tumour xenografts. Intraperitoneal injection of amphotropic producer cells resulted in transduction in spleen, liver and kidney, which was not detected when targeted producer cells were used. These results demonstrate the feasibility of using targeted retroviral vectors for in vivo gene delivery and highlight the safety benefits of targeted vectors that do not infect other host tissues.