Discovery and Validation of Alternatives to VSV-G for Pseudotyping of Lentiviral Vectors for In Vivo Delivery of Anti-Tumor Transgenes

Though cell therapy for cancer is now widely used commercially and has been efficacious for tens of thousands of patients, conventional manufacturing methods are expensive and difficult to scale. An alternative approach is to deliver the relevant anti-tumor transgenes to T cells in vivo in the patient. Such "in vivo cell therapy" methods promise to be more scalable, with reduced cost of goods, since the same drug product can be administered to any patient. Typically, conventional cell therapy introduces anti-tumor transgenes into T cells using a lentivector pseudotyped with VSV-G. However, VSV-G is not cell type specific because its molecular target is present on a diversity of human cells, which may result in less than optimal pharmacology in vivo. Natures existing viral diversity presents the opportunity to identify alternative pseudotypes that are more optimal for in vivo cell therapy. In this study, we first performed a large-scale bioinformatic sequence search for G-proteins similar to VSV-G. We identified 166 G-proteins in the sequence search and then tested 9 in vitro for efficiency, specificity, and sensitivity of transgene delivery to various human immune cell phenotypes. We used the results from this screen to select three G-protein candidates for a pilot GFP transgene delivery study in humanized mice, using anti-CD3 antibody fragments for T cell tropism. One candidate G-protein performed significantly better than VSV-G, so we moved that candidate into tumor control studies in a humanized mouse model. This candidate G-protein was able to deliver an anti-tumor transgene to T cells, which subsequently cleared 100% of tumor burden in 100% of mice. We conclude that systematic screens for optimal lentivector designs can be used to identify optimized candidates for in vivo cell therapy.