Analysis of Retroviral Vector Insertion Sites after T-Cell Directed Gene Therapy.

Gene transfer into peripheral blood lymphocytes has several potential applications including the correction of genetic diseases and therapeutic approaches for HIV-1 infection and cancer. Integrating gene transfer system based on murine oncoretroviruses are a convenient tool for such strategies. However, the recent occurrence of uncontrolled clonal T cell expansions in two patients treated with retroviral gene transfer for X-linked severe combined immune deficiency has raised the concern of the risk of insertional oncogenesis associated with the clinical use of integrating viral systems. In vitro studies have indicated that murine viral vectors tend to integrate in the vicinity of transcription start regions of the genome, thus providing a possible mechanism for oncogene activation, however, data from clinical gene transfer trials is lacking.

We are following patients affected with adenosine deaminase (ADA) deficiency who have received T-lymphocyte-directed, retroviral-mediated gene transfer starting in 1990. The first treated patient received the last infusion of gene-corrected cells 12 years ago, has never shown any sign of lymphoproliferation and still carries ~20% of gene-corrected peripheral blood lymphocytes. We set out to study the integration sites in the cells of this patient with the aim of mapping the regions involved by retroviral integrations, determining their localization with respect to known genes, and assessing whether a preferred pattern could be defined. Genomic DNA was prepared from stored lymphocyte samples dating 1991, 1992, 1995, 1998, 2000, and 2003. By inverse PCR and ligation-mediated PCR, we have identified ~860 bona fide insertion sites. Search for homology within the human genome using BLAT returned ~330 unique hits that involved a variety of genes, including transcription factors and oncogenes (e.g. RUNX1, STAT5, FYN). To evaluate the distribution pattern of these integration sites, 2000 randomly generated data sets of genomic coordinates were assembled and their distribution relative to annotations of the human genome was analyzed. A preliminary comparison of the random distribution to our experimental samples showed that retroviral integrations in cells obtained from the patient were significantly skewed toward regions within 2 kb of genes (p<0.002) and CpG islands (p<0.001).

These results suggest that, similar to what observed in murine fibroblast and human cancer cell lines, transcriptionally active regions of the genome may be preferred targets of retroviral vectors in human primary T lymphocytes. At the same time, our observations show that the resulting integration events are compatible with long-term, event-free in vivo survival of gene-modified cells in clinical settings.