Comparative Study of Retroviral Insertions in ADA-SCID Patients Treated with PBL-GT and HSC-GT Unveils a Cell Specific Integration Profile

Analysis of integration distribution of gamma-retroviral vectors in gene therapy treated patients is crucial in order to assess the accessibility to insertion events in human genome as well as the influence of vector on host genome and clonal selection of transduced cells. To address these issues we compared through 454-pyrosequencing the distribution of MLV-vector integrations in two different gene therapy (GT) trials for ADA-SCID based on repeated infusions of transduced mature lymphocytes (PBL) or a single infusion of hematopoietic stem/progenitor cells (HSC) combined with nonmyeloablative conditioning. In the PBL-GT group we isolated a total of 2509 unique integrations derived from in vitro transduced PBL or ex vivo blood samples in 4 patients (5 to 7 years after last infusion). In the HSC-GT group we retrieved 1064 integrations from in vitro transduced CD34+ cells and ex vivo purified blood cells in 3 patients (2 to 5 years after GT). In both groups MLV vector displays the classical non-random distribution favoring gene-dense regions and TSSs. However, integrations from PBL-GT displayed a significant preference for genes highly expressed in T cells and involved in T cell specific functions (proliferation, differentiation and development) and signaling pathways (IL-2, IL-15, TCR) as compared with insertions from HSC-GT (IPA software). The T-cell specific pattern was observed both in in vitro and ex vivo integrations, and did not result in clonal selection or in vivo skewing. On the other hand, in T cells derived from patients treated with HSC-GT, these genes were not represented and hit genes were involved in a wider range of functional categories reflecting the stemness of the transduced precursors from which they derived. CIS were also specifically different between the two groups of patients. For example, LMO2 was an hotspot both before and after infusion in HSC-GT patients while we did not detect any integration in this locus in PBL-GT patients. Comparison of the expression profile of hit genes profile by Affymetrix analysis revealed that these differences could be partly related to their expression state at the time of transduction in transduced PBL vs CD34+cells. In addition, we found a preferential distribution of PBL-GT integrations in regions with a higher density of T cells hypersensitive sites as compared to HSC-GT, suggesting a possible link with T-cell specific accessibility of genome. Overall our data show in ADA-SCID patients treated with GT clues of a cell-specific integration profile with no particular in vivo biases before infusion and following in vivo selections years after GT.