Genetic engineering techniques can now efficiently express T cell receptors recognizing tumor antigens on mature T cells and antitumor effects have been observed with this approach, although persistence of such mature T cells is often limiting. To enhance the potential for long-term persistence, some investigators are considering using genetic engineering to express T cell receptors targeting tumor antigen in developing T cells. To test this approach, we generated three founders of TCR transgenic (Tg) mice with specificity for the H-2b restricted immunodominant epitope of murine survivin, a molecule overexpressed by most cancers. The gene was expressed under the human CD2 promoter, which induces expression of the transgene in DN2 thymocytes and has been used successfully in the generation of multiple previous TCR transgenics. In survivin TCR Tg mice, survivin reactive T cells were predominantly CD8+ and mediated specific immune reactivity toward survivin peptides. Some antitumor reactivity was observed, but it was not potent, and the survivin reactive transgenic T cells were unable to mediate objective tumor regression of survivin bearing tumors in vivo. Surprisingly however, spontaneous T cell acute lymphoblastic leukemia (T-ALL) occurred in progeny of all three founders, beginning at 4-6 months of age and by two years of age, all survivin TCR Tg+ mice succumb to T-cell ALL. Both survivin TCR+Rag+/- and survivin TCR+Rag-/- mice develop leukemia at the same rate. The leukemic cells are CD3+, survivin TCR+, and CD8+ or CD4/CD8, and are malignant as evidenced by continous growth in vitro without growth factors and the induction of leukemia following inoculation into C57BL/6 immunocompetent recipients. Alpha gene rearrangement analysis in tumor tissues demonstrated oligoclonality, consistent with a model wherein leukemic transformation is not isolated to a rare clone, but occurs with significant frequency in developing thymocytes. Insertional mutagenesis was judged unlikely since leukemia occurred in three separate founders, and was further ruled out by insertion site analysis of one founder, which demonstrated no evidence for integration of the transgene within an oncogenic locus. We thus postulated that the survivin reactive TCR itself was oncogenic, perhaps as a result of signaling via survivin peptides expressed within the thymus. In support of this, we demonstrate that survivin is expressed in thymic tissue, and prior to the development of leukemia, the survivin-specific TCRs signal, as evidenced by NFAT nuclear translocation, and enhanced BrdU incorporation compared to non-transgenic controls. Furthermore, transgenic mice that lack beta2microglobulin, and thus are unable to present survivin peptides, have a significantly diminished rate of leukemia development (see figure). Interestingly, at least one NOTCH1 mutation was also found in all leukemias, with mutations in the PEST domain being most common (8/8), but 5’ deletions (19/25) and mutations in the heterodimerization domain also observed. We also observed T cell acute lymphoblastic leukemia with NOTCH mutations in TCRTg+ mice with specificity to other tumor-associated antigens including WT1 and gp100, albeit with lower penetrance. We conclude that genetic engineering aimed at endowing T lymphoid progenitors with the capacity to recognize tumor associated antigens expressed in the thymus could pose a risk for neoplasia, independent of insertional mutagenesis.

No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.