Modeling the Genotoxicity of Viral Vector Integration in a Tumor Prone Hematopoietic Stem Cell Transplantation Model.

Insertional mutagenesis represents a major hurdle to successful gene therapy and mandates for sensitive pre-clinical assays of genotoxicity. Cdkn2a^{−/ −} mice are defective for p53 and Rb pathways, and are susceptible to a broad range of cancer-triggering genetic lesions. We developed an in-vivo genotoxicity assay, based on transplantation of Cdkn2a^{−/ −} hematopoietic stem cells (HSCs), treated or not with prototypical retroviral (RV) and lentiviral (LV) vectors. In our rationale if RV or LV treatment is genotoxic, transplanted mice will show a significantly earlier tumor onset. Using this approach, we detected a dose-dependent acceleration in tumor onset in the mice transplanted with RV treated cells. We compared the RV and LV integration site distribution in pre-transplant cells and tumors from the transplanted mice. As expected, RV integrates close to gene promoters in pre-transplant cells and tumors. Moreover, we found that RV preferentially targeted genes encoding for transcription factors, kinases and genomic positions previously described as Common Integration Sites (CIS). CIS are genomic regions targeted at high frequency in tumors by retroviral integrations and probably map within or close to proto-oncogenes activated upon integration. Interestingly, in tumors, RV insertions at CIS and cell cycle genes were further enriched and associated to early lymphomagenesis. Remarkably, LV tested in the same conditions, did not show any tumor acceleration, and targeted CIS much less frequently than RV in pre-transplant cells or tumors, and did not show selection for integrations at any specific gene class. This is the first evidence that prototypic LV have low oncogenic potential, and provides a major rationale for their application to HSC gene therapy.

To dissect the role in lymphomagenesis of the strong enhancers in the RV LTRs and the different integration site selection of each vector, we tested an RV with enhancer deleted LTRs (SIN RV) and a moderate promoter in internal position, an LV with a strong RV-like enhancer-promoter into the LTRs or in internal position. Our results show that:

• SIN RV shows reduced effect on lymphomagenesis acceleration with respect the conventional RV;

• LV with RV like LTRs treatment significantly accelerates lymphomagenesis, underlining important role of RV-LTRs in oncogenesis;

• LV with RV-like enhancer in internal position show a reduced genotoxicity.

These data show that the type of genetic elements used (strong enhancers or moderate promoter) and their position in the vector genome (LTR or internal positions) significantly influence the vector safety profile. We are currently mapping the integration sites in pre-transplant cells and tumors marked by each vector to identify the genes targeted by the integrations and elucidate the potential mechanism of deregulation.

The described model provide an important tool to compare the risk of insertional mutagenesis of different integrating vectors and provides a platform to test different vector types, designs and safety improvements.