Sporadic Notch1 Mutations Found in Human T-ALL Drive T Cell Development but Not Leukemogenesis in Murine Models: Implications for Pathogenesis and gamma-Secretase Inhibitor (GSI) Therapy.

The presence of gain-of-function Notch1 mutations in more than 50% of human T-ALL patient samples is consistent with a central role for Notch1 signals in the pathogenesis of T-ALL (Weng et al., 2004, Science, 306, 269-71). Notch1 mutations are of two types:

• common mis-sense or insertional mutations in the extracellular heterodimerization (HD) domain that cause heightened proteolytic activation of Notch1 by the gamma-secretase protease complex, and

• less frequent frameshift or stop codon mutations that delete a C-terminal PEST destruction box and increase the duration of Notch1 signaling.

To investigate the significance of these mutations in the pathogenesis of T-ALL, the HD mutations L1601P, L1594P, and P12-ins (which produce weak, moderate, and strong ligand-independent increases in Notch1 signaling, respectively; Malecki et al., 2006, Mol. Cell. Biol. 12, 4642–51), and a PEST deletion (which has little intrinsic effect on Notch1 signaling) were transduced into hematopoietic progenitor cells. These cells were then used to reconstitute lethally irradiated mice, which were monitored for aberrant lymphocyte development and leukemia. The PEST deletion by itself did not produce a phenotype, consistent with prior studies (Feldman et al., 2000, Blood, 96, 1906–13). The two weakest HD mutations, L1601P and L1594P, induced ectopic T development in the bone marrow and inhibited CD19+ B-lymphoid and Lin-Sca-1+Kit-CD16/32+CD34+ granulocyte-macrophage progenitor development, but failed to induce T-ALL after more than 6 months. In contrast, the strongest HD mutation, P12-ins, induced leukemia with a penetrance of ~40%. When non-leukemogenic HD mutations were paired in cis with a PEST deletion (an arrangement found in ~15% of T-ALL that results in synergistic increases in signal strength), T-ALL occurred in a minority of animals (median time, ~4 months post-transplantation). Together, these data show that the level of Notch1 signaling required for induction of leukemia is higher than that needed for fate-determination during hematopoiesis and that most solitary Notch mutations fail to achieve the level needed for efficient induction of T-ALL. To investigate whether weak Notch alleles cooperate with other oncogenes, the L1601P and L1601P/PEST alleles were retrovirally introduced into lymphoid progenitors engineered to have constitutive Ras signaling, which has been shown to induce T-ALL (Hawley et al, Oncogene, 1995, 11, 1113–23). The L1601P and L1601P/PEST alleles shortened the mean time to leukemic presentation from 91 days to 76 and 52 days, respectively, indicating that these weak Notch1 alleles can cooperate with Ras signals to enhance leukemogenesis. GSI abolished the growth of cell lines derived from these tumors, indicating a continuing dependency on Notch1 signals. In summary, these experiments suggest that a supraphysiologic level of Notch signaling is required for efficient induction of T-ALL, but that once other oncogenic events have occurred, lower doses suffice to promote leukemogenesis cooperatively. However, even when Notch1 mutations are not initiating events, tumors remain “Notch addicted” and can be targeted pharmacologically with GSI. These findings reinforce the notion that the Notch pathway is a rational therapeutic target in T-ALL.