Development and application of non-integrating lentiviral vectors for gene therapy

Lentiviruses stably integrate their genome into the host genome. Although this feature can be advantageous for long term transgene expression, it also has the potential to cause mutagenesis and cell transformation. To address this problem, thereby improving the safety of lentiviral gene therapy, non-integrating lentiviral vectors (NILVs) were developed. NILVs were generated by mutating the cis-acting sequences that interact with integrase (att sites) or by mutating specific residues in integrase in different domains (catalysis - D64V, strand transfer – Q148A, K264R, K266R, K273R, DNA or chromatin binding - N120L, W235E). Relevant mutations were then combined in order to improve the safety of these vectors. It was shown that all mutant vectors were efficiently produced and mutations did not affect infectivity. In contrast to dividing cells, differentiated muscle cells infected with NILVs show stable transgene expression over time without degradation of episomal viral DNA. The vectors were also tested in vivo by intramuscular injection in neonate mice. Transgene expression from muscle cells was maintained for 8 months using both integrating and NILVs. The vectors were then tested in a haemophilia B disease model. It was shown that plasma levels of FIX produced by muscle cells infected with integrating lentiviral vectors were above the therapeutic threshold. However, expression from NILVs was lower. This was studied in detail and it was found that integrating lentiviral vectors are transcriptionally more active than NILVs. A comparison of expression levels revealed that integrated lentivectors express more transgene protein per vector copy than NILVs and AAV vectors, but both episomal vectors display similar levels of transgene expression per vector copy. In conclusion, NILVs have the potential to be used as tools for prolonged transgene expression in non-dividing muscle cells or transient expression in dividing cells. However, vectors may need to be optimised if high expression levels are required