The occurrence of leukaemia in three patients in a gene-therapy trial for X-SCID (refs 2–4) has highlighted an adverse effect of insertional mutagenesis as a delivery technique for the therapeutic gene. In our investigation, we expressed both LMO2 and IL2RG transgenes in the same retroviral vector that was used in the clinical gene-therapy trials; those trials found that retroviral integration of the corrective IL2RG occurred near the locus of the LMO2 oncogene5. We concluded that this integration may have upregulated the expression of LMO2 and triggered the leukaemia cases in the gene-therapy trials.

Woods et al. used a murine model in which stem cells from wild-type or mutant IL2RG−/− mice were transduced with very high amounts of human IL2RG by means of a lentiviral vector1. They found that these transplanted mice developed a high incidence of tumours. To the best of our knowledge, however, IL2RG has never been reported to act as an oncogene in human T-ALL, and murine studies should be interpreted with caution when extrapolating to humans. Although T-cell development is very similar in mice and humans at the genome-wide level6,7, there are differences in individual genes: for instance, mutant IL2RG−/− mice lack T, B and natural killer (NK) cells, whereas human X-SCID patients have normal numbers of B cells.

Mouse–human hybrid fetal thymus organ culture (FTOC) is often used to study human T-cell development ex vivo. In FTOC experiments, we found that LMO2-transduced cells expressed the early thymocyte marker CD1a less frequently than untransduced cells derived from the same culture (22% compared with 57%), whereas a higher proportion of LMO2-transduced cells retained the progenitor marker CD34 (12% compared with 2–4%, representative of five different FTOC experiments). Later, during T-cell development, this block was even more pronounced (58% of cells in the subsequent immature CD4 single-positive stage, compared with 22% in controls). These results indicate that progenitor cells ectopically expressing LMO2 are severely hampered in T-cell development.

The percentage of immature CD4 single-positive (37% compared with 38%) and the later CD4+CD8+ double-positive cells (6% compared with 9%) was similar in untransduced and IL2RG-transduced populations, indicating that retroviral-mediated expression of IL2RG does not affect T-cell development. LMO2-transduced cells developed normally into B, NK and myeloid cells.

The very high expression of lentiviral transgenic IL2RG in the transplanted mice studied by Woods et al.1 may have contributed to development of T-cell lymphomas. In contrast, IL2RG concentrations were slightly lower than normal in the human X-SCID trials after treatment using retroviral vectors. The phenotype of the mouse tumours is very immature and different from that of the T-ALL that occurred in the three X-SCID patients. Unfortunately, Woods et al. do not indicate whether the tumours were clonal, whether they expressed IL2RG, or whether JAK3 was activated; it is possible that the IL2RG gene might itself be mutated — for example, as a result of errors in lentiviral reverse transcription, but details of the insertion sites recovered are not given.

In our view, the retroviral-mediated expression of IL2RG in haematopoietic precursors does not represent a leukaemogenic event. This γ-chain is highly expressed in normal haematopoietic CD34+ precursors and during all stages of T-cell development, whereas LMO2 is only expressed in progenitor cells and the earliest thymocytes6. Furthermore, IL2RG expression in treated X-SCID patients is within the normal range5. Persistent activation of JAK3 was not detected in these patients, indicating that IL2RG function was normal in the setting of retroviral expression5.

Taken together, these observations argue against IL2RG acting as an oncogene in human gene therapy, although our experiments cannot rule out the possibility that a therapeutic transgene might provide an inappropriate growth stimulus when expressed at high dose or at an inappropriate time. Instead, it is likely that restored IL2RG expression allows T-cell development to progress to stages at which LMO2 would normally be completely downregulated but might contribute to leukaemogenesis if ectopically expressed.