3. Effect of Retroviral and Lentiviral Vector Integrations on Transcription of Flanking Genes

Abstract

Listen

Translate article

In mouse, retroviral infection triggers leukemias by activation of cellular proto-oncogenes. This is usually mediated by promoter/enhancer sequences in the proviral LTR that recruit ubiquitous and/or cell type-specific transcription factors, and up-regulate expression of flanking genes, up to 300 Kb away from the integration site, and independently of orientation. Unfortunately, Retroviral Vectors (RV) used in gene therapy also triggered unwanted oncogenesis, indicating that the risk of RV insertional mutagenesis had been underestimated, and raising concerns for their further use in gene therapy. In addition RV, Lentiviral (LV) and Adeno-Associated vectors have the tendency to integrate in the proximity of expressed genes. Thus, these safety concerns apply to all integrative vectors, in particular to the closely related LV that have been proposed as improved versions of RV for several applications. The actual frequency and extent of transcriptional de-regulation occurring at randomly picked retroviral integration sites is currently unknown. In fact, integration sites retrieved from tumors are likely the result of in vivo selection for proto-oncogene activation, and it has proven very difficult to estimate the frequency of such events in unselected target cells. This information is critical to properly assess the insertional mutagenesis risk of each vector. Thus, we embarked in testing and comparing the effect of non-selected RV and LV integration on the expression of flanking genes. We generated three panels of hematopoietic cell clones, each carrying a single insertion of either a conventional RV expressing GFP from the LTR, a self inactivating (SIN)-RV expressing GFP from an internal PGK promoter, and a third-generation SIN-LV expressing GFP from the same PGK promoter. For each vector, we isolated 150 clones, and retrieved genomic DNA flanking the proviral insertion in 50 of them by linear amplification-mediated PCR. So far, sequence analysis has allowed in silico mapping of 40 single integrations. Even if our study requires a relatively small sample size, it is apparent that integration site selection differed between RV and LV; RV integrated within regions of lower gene density with respect to LV; 66% of RV insertions were intergenic, while 83 % of LV insertions were intragenic; 50% of RV insertions were found between 0.5–5kb from the closest transcription start site, as compared to only 5% of LV insertions. We then selected all the genes (SeqRef database) contained in a 600 kb interval centered on each integration site (344 up to now). We are currently comparing their relative expression levels in all cell clones by microarray technology, using custom-made microchips spotted with the selected sequences, and probing with total cDNA obtained from each clone and a pool of control untransduced ones. This analysis will indicate the frequency and extent of transcriptional perturbation around RV and LV integration. We expect to find significant differences between each vector, according to its viral origin and design. If so, our data will be critical to estimate the relative risk of oncogene activation in gene therapy applications.