Identifying novel 'druggable’ targets via Npm1A-turboid fusion and mass spectrometry to overcome genetic or adaptive resistance to menin inhibitors in mtNPM1 AML

Abstract In mutant (mt) NPM1 AML cells, treatment with Menin inhibitors (MI) disrupts the binding of Menin to MLL1, leading to reduced levels/activity of HOXA9 and MEIS1 and other MLL1 target genes. Although able to induce complete remissions, most patients either fail to respond to MI or relapse after achieving a remission. MI-resistance in AML with mtNPM1 is either due to mutations in the MI-binding domain of MEN1, or more commonly, due to adaptive, dysregulated epigenome and transcriptome regulated by MLL1, regardless of MI mediated disruption of menin-MLL1 binding. Here, we first probed the nongenetic mechanisms of MI-resistance in a MI-adaptive tolerant/resistant mtNPM1 (OCI-AML3 MITR) AML cell line (lacking menin mutations) that we created. As compared to the parental OCI-AML3 (LD50 55 nM), OCI-AML3 MITR cells exhibited a LD50 to the MI SNDX-50469 of 1 μM. These MITR cells were cross resistant to other MIs, including ziftomenib and DS1594b. Treatment with revumenib may also cause emergence of hot spot mutations in menin (e.g., S160T, M327V, M327I, G331D, G331R, and T349M) exhibiting reduced affinity to MI-binding. Therefore, utilizing CRISPR/Cas9, we also generated an OCIAML3 knock in (KI) model expressing a biallelic MEN1 M327I mutation. OCI-AML3 MEN1-M327I cells were resistant to SNDX-50469, ziftomenib, and DS1594b, but sensitive to the second-generation MI, bleximenib, as previously reported.To identify novel 'druggable’ targets in mtNpm1 AML cells either sensitive or resistant to MI, we knocked in TurboID by CRISPR/Cas9 into the C-terminus of the mtNpm1 gene in OCI-AML3 cells. Mass spectrometry of proteins from these cells helped identify previously unknown, putative, interactors of mtNpm1. Besides known interactors such as XPO1 and KMT2A, we discovered SMARCA2, eIF4A, AURKA, CDK9, PLK1, IRAK4, PARP, and SF3B1 as mtNPM1 interactor proteins. Next, utilizing the OCI-AML2 Npm1A KI model we had generated, we determined the sensitivity of targeting the mtNpm1-TurboID identified novel mtNPM1 interactor proteins in mtNpm1 expressing AML cells. Rocaglamide A (RocA), an eIF4A inhibitor, and talazoparib, a PARP inhibitor, induced greater loss of viability in OCI-AML2-mtNpm1 cells relative to the parental OCI-AML2 cells. Moreover, co-treatment with MI and RocA or talazoparib induced synergistic lethality in MI-sensitive mtNpm1 models. When combined with a BET inhibitor (pelabresib) or a novel dual BET/HAT inhibitor (NEO2734/EP31670), both RocA and talazoparib induced synergistic lethality in the sensitive OCI-AML3 and OCI-AML2-Npm1A KI, as well as the MI-resistant (OCI-AML3-Menin-M327I or the OCI-AML3 MITR) cells. In the MI-sensitive AML cell models described above, RocA treatment induced log2 fold-reduction in mRNA expression of cytokines and stem cell markers, FLT3, IL7R, CD244, and S100A8/A9. Furthermore, RNA-Seq analysis showed reduced expressions of the HALLMARK MYC TARGETS V1 and V2 and REACTOME TRANSLATION gene sets in both MI-sensitive and resistant models, whereas RocA treatment significantly induced the HALLMARK APOPTOSIS pathway exclusively in MI-resistant AML cells. RocA treatment also induced log2 fold-expression of TP53 in MI-resistant OCI-AML3 AML cells. Notably, treatment with EP31670 and RocA also downregulated myeloid transcription factors such as c-Myc, c-Myb, and PU.1, while inducing p21 and p27 protein expressions. Ex vivo treatment with EP31670 and RocA or talazoparib induced synergistic loss of viability in MI (SNDX-50469)-resistant, patient-derived (N = 3) mtNpm1 AML cells. In the PDX model of luciferized mtNpm1 + FLT3-ITD AML cells engrafted in NSG mice, compared to vehicle control, monotherapy with EP31670 (7.5 mg/kg QD), talazoparib (0.25 mg/kg QD) or RocA (0.5 mg/kg 3x/wk) significantly reduced AML burden. Collectively, these findings highlight the newly discovered, preclinical efficacy of inhibitors of eIF4A and PARP against mtNpm1 AML cells and support the rationale for further evaluating the efficacy of combinations involving these agents.