Mutations That Cooperate with Nf1 Inactivation in Leukemogenesis Influence Therapeutic Response to MEK Inhibition.

Abstract

Hyperactive Ras is a common feature of many cancers, including myelodysplastic syndrome (MDS), myeloproliferative disease (MPD), and acute myeloid leukemia (AML). Conditional inactivation of the Nf1 tumor suppressor, which encodes a GTPase activating protein that negatively regulates Ras, in murine hematopoietic cells in Mx1-Cre, Nf1flox/flox mice induces MPD with 100% penetrance. This MPD does not evolve to AML, which implies that cooperating mutations are necessary for transformation to acute leukemia. Upon infection with MOL4070LTR, a retroviral insertional mutagen, ∼25% of Mx1-Cre, Nf1flox/flox mice develop AML. The Ras effectors MEK and ERK are hyper-phosphorylated in Nf1 mutant MPDs and AMLs with Nf1 inactivation. We investigated the effects of CI-1040, a highly selective inhibitor of the MEK kinase, on the growth of hematopoietic progenitor and blast colonies in methylcellulose cultures stimulated with GM-CSF. Blast colony formation from Nf1 mutant AML cells was abrogated at much lower CI-1040 concentrations (0.25 μM) than wild-type or Nf1 mutant MPD cells (50 μM) suggesting a therapeutic index. Nf1 mutant mice with MPD that were treated with CI-1040 showed no improvement in leukocyte counts or survival despite transient MEK inhibition in vivo. In contrast, mice transplanted with three unique Nf1 deficient AMLs responded to CI-1040 treatment with decreased leukocyte counts and markedly prolonged survival compared to vehicle treated controls (24 versus 7 days in the vehicle-treated cohort; OR 3.5 95% CI 3.0–3.8). Despite lower white blood cell counts and prolonged survival, all of the leukemic mice receiving CI-1040 eventually developed reemergence of peripheral blood blasts and died from AML. Leukemias that relapsed during CI-1040 treatment are remarkably less sensitive to CI-1040 in vitro than the vehicle treated AMLs, and do not respond to treatment in secondary recipients. This cellular resistance is not due to an acquired change in kinase sensitivity to CI-1040 as we observe equivalent inhibition of pERK in sensitive and resistant AMLs that are exposed to a range of CI-1040 concentrations. Importantly, when compared with untreated leukemias, CI-1040-resistant AMLs contain new retroviral integrations. Cloning the integration sites from sensitive and resistant Nf1 deficient leukemia pairs resulted in identification of several candidate resistance genes including members of the RasGRP family and Mapk14, which encodes p38. Inhibition of p38 with SB 202190, a relatively selective inhibitor, confers resistance to CI-1040 in sensitive leukemias. Our data support the idea that MEK is a relevant biochemical target in myeloid leukemia, and show that cooperating mutations strongly modulates response. CI-1040 and related MEK inhibitors merit further investigation in myeloid malignancies. This model provides a tractable system for identifying cooperating mutations that result in progression to acute leukemia and modulate drug sensitivity.