Actionable Findings from an Unbiased Drug Screen for Novel Single Agent and Combination Therapies Against AML with Mecom Re-Arrangement

EVI1 (Ecotropic Viral Integration Site 1) is a zinc finger transcriptional regulator encoded from the MECOM (MDS1 and EVI1 Complex) locus at chromosome 3q26.2. Aberrant EVI1 overexpression is observed in approximately 10% of de novo AML, of which approximately half are due to chromosomal rearrangement involving either inv3(q21;q26.2) or t(3;3)(q21;q26.2), where the distal GATA2 hematopoietic enhancer is relocated and drives EVI1 expression. In addition to the resulting haploinsufficiency of GATA2, monosomy 7 and activating mutations in the RAS pathway and SF3B1 commonly co-occur in 3q26.2-rearranged (r) AML. AML with EVI1 overexpression generally have a stem-like phenotype, are refractory to current therapy options and exhibit poor overall survival. Among the targets of EVI1 are the ERG, MYC, KIT, Bcl-xL and MPL genes. GATA2 repression, coupled with increased EVI1 activity, promotes AML progression and an aggressive phenotype in 3q26.2-rearranged AML. Therefore, there is a glaring need to develop and test novel targeted therapies with improved efficacy in repressing EVI1 and improving survival of AML with EVI1 overexpression.

To identify druggable vulnerabilities in 3q26.2-r AML, we conducted an unbiased high-throughput drug screen, utilizing a library of 2480 oncology-focused, mechanistically annotated drugs (referred as NCATS Mechanism Interrogation Plates or-MIPE 5.0) (Cancer Cell. 2021; 39:566-579), against seven AML cell lines, i.e., four with 3q26.2-r: UCSD-AML1, HNT-34, AML191 and AML194, and three non-3q26.2-r AML: SET-2, MV4-11 and OCI-AML3. The library exploits mechanistic redundancy by including multiple inhibitors against well-validated oncology targets, while simultaneously encompassing mechanistic diversity, altogether targeting more than 850 distinct mechanisms of action. We exploited this redundancy to identify druggable, target-level dependencies in 3q26.2-r AML, both in terms of absolute potency and relative activity as compared to other AML subtypes. BRD4 was identified as an absolute dependency in 3q26.2-r AMLs. This was consistent with previous reports that BET inhibitors (e.g., OTX015, mivebresib or ABBV-075 and JQ1) are effective against 3q26.2-r AML cell lines, patient-derived (PD) AML cells and PDX models. Comparison of 3q26.2-r versus other AML cell lines also identified XIAP, mTOR, PIK3CA and Bcl-xL as druggable vulnerabilities in 3q26.2-r AML.

In follow-up experiments, XIAP/cIAPs inhibitors birinapant (10-1000 nM) or SM-164 (30-1000 nM), chosen based on the MIPE screen outcomes, induced significantly more dose-dependent apoptosis in 3q26.2-r versus the other AML cell lines. This was associated on Western analyses with decline in XIAP, c-IAP1/2, p-ERK1/2, MCL1 and Bcl-xL, but increased protein levels of cleaved caspase-3 and PARP in UCSD-AML1 and AML191 cells. Mivebresib (50-250 nM) or OTX015 (100-500 nM) also induced dose-dependent apoptosis, and reduced EVI1, c-Myc, c-Myb, XIAP, c-IAP1/2, CDK4/6 and Bcl-xL while increasing protein levels of HEXIM1 and cleaved PARP. Consistent with this, co-treatment with birinapant with mivebresib was synergistically lethal in UCSD-AML1 and AML191 cells, with Delta Synergy scores > 1.0 calculated by the ZIP method. Treatment with the dual mTOR/PIK3CA inhibitor NVP-BGT226 (1-30 nM) or navitoclax or Bcl-xL-specific BH3 mimetic A-1155463 also exerted lethality and synergistically induced apoptosis with mivebresib in AML cells with inv3/t(3;3). We previously reported that the TBL1/β-catenin antagonist tegavivint induced apoptosis in 3q26.2-r PD AML cells and PDX models. Co-treatment with birinapant and tegavivint also synergistically induced apoptosis in 3q26.2-r AML cells. Finally, in the xenograft model of AML194-Luc cells in NSG mice, monotherapy with birinapant (30 mg/kg/t.i.w, Intraperitoneal), vs vehicle control, significantly reduced the AML burden, without causing toxicity. Additionally, compared to each drug or vehicle control, co-treatment with birinapant and the BETi OTX015 (30 mg/kg/day, by oral gavage) was more effective in reducing AML burden in the xenograft model. These findings demonstrate promising preclinical activity of IAP protein inhibition against the cellular models of AML with inv3/t(3;3) with EVI1 overexpression, supporting the rationale to further evaluate in vivo efficacy of birinapant and/or BETi-based combinations against this AML sub-type.