Myeloid-Lineage Enhancers Drive Oncogene Synergy and Represent a Novel Therapeutic Target in CSF3R-CEBPA Mutant AML

Acute Myeloid Leukemia (AML) results from the stepwise accumulation of mutations from distinct functional classes, ultimately culminating in malignant transformation. Based on their oncogenic activity, mutations can be classified into three distinct groups. Class I mutations activate signaling pathways, produce uncontrolled proliferation, and in isolation produce a myeloproliferative phenotype. Class II mutations result from point mutations or chromosomal translocation events in lineage determining transcription factors, producing differentiation arrest and myelodysplasia in isolation. A classic example of oncogene synergy between distinct mutational classes can be found in the co-occurrence of mutations in the transcription factor CCAAT-enhancer binding protein alpha (CEBPA) with mutations in colony stimulating factor receptor 3 (CSF3R). Mutations in CEBPA occur in approximately 10% of AML where they block differentiation and convey favorable risk. In contrast, CSF3R mutations lead to constitutive receptor activation and uncontrolled neutrophil proliferation. In the absence of co-occurring Class II mutations, membrane proximal CSF3R mutations produce the myeloproliferative neoplasm chronic neutrophilic leukemia (CNL). Interestingly, patients with CEBPA mutant AML that also harbor an oncogenic CSF3R mutation have worse prognosis than those with wild type CSF3R. However, the mechanism underlying this oncogene synergy remains unknown.

To model the co-occurrence of these mutations, we expressed CSF3R^T618I (The most common membrane proximal CSF3R mutation) in fetal liver hematopoietic stem cells harboring compound heterozygous CEBPA mutations in the endogenous allele (CEBPA^K/L). Mice transplanted with mutant CEBPA alone developed a long latency AML with a median survival of 60 weeks. In contrast, mice transplanted with mutant CSF3R^T618I/CEBPA^K/L cells developed a much more rapid AML with a median survival of 13 weeks. These results were corroborated in an orthogonal model in which mutant CSF3R and a C-terminal mutant CEBPA were retrovirally expressed prior to bone marrow transplant.

To dissect the underlying mechanism, we performed a comprehensive transcriptomic and epigenetic analysis on cells expressing each mutation in isolation as well as the combination. This analysis revealed that mutant CSF3R activates a distinct set of enhancers that regulate genes associated with differentiation and drive neutrophil differentiation. Co-expression of mutant CEBPA blocks the activation differentiation-associated enhancers but is permissive to those associated with proliferation. Differentiation but not proliferation-associated enhancers are bound by wild type CEBPA. Thus, the dominant negative impact of mutant CEBPA at these enhancers explains its differential impact on differentiative and proliferative transcriptional programs. Enhancer activation precedes promoter activation and CEBPA mutations are thought to represent early events in AML initiation. The epigenetic mechanism underlying the observed oncogene synergy argues that CEBPA mutations must occur prior to CSF3R to impact differentiation. We therefore developed a retroviral vector system enabling temporal control of Cre-mediated oncogene expression. Using this system, we found that only when mutant CEBPA is expressed prior to mutant CEBPA is differentiation arrest observed. Furthermore, AML develops in vivo only when mutant CEBPA is expressed prior to mutant CSF3R.

To develop novel therapeutic strategies for this subclass of AML with adverse prognosis, we performed medium throughput drug screening on CSF3R/CEBPA mutant AML cells and identified sensitivity to inhibitors of JAK/STAT signaling as well as Lysine Demethylase 1 (LSD1). In other subtypes of AML, LSD1 inhibitors activate enhancers associated with differentiation. We confirmed that LSD1 inhibition promotes neutrophilic differentiation in CSF3R/CEBPA and through epigenetic and transcription profiling establish that this occurs via the reactivation of differentiation-associated enhancers. We further found that the combination of ruxolitinib (JAK/STAT inhibitor) and GSK2879552 produce a complete hematologic response and double median survival in mice harboring CSF3R/CEBPA mutant AML. Thus, the combination of JAK/STAT and LSD1 inhibitors represents and exciting therapeutic strategy for CSF3R/CEBPA mutant AML.