Azacitidine-panobinostat activates NF-kb signaling in the bone marrow microenvironment to chemosensitize KMT2A rearranged pediatric AML

Abstract

Abstract KMT2A rearrangements (KMT2Ar), a hallmark of pediatric acute leukemia, are present in 20% of pediatric acute myeloid leukemia (AML). Targeted therapy options are needed to improve outcomes of patients with KMT2Ar who have a 5-year event-free survival rate of 45% due to poor response to chemotherapy. We demonstrated that a combination of DNA hypomethylating agent azacitidine (aza) and histone deacetylase inhibitor panobinostat (pano) sensitizes KMT2Ar patient-derived xenograft (PDX) models to chemotherapy in vivo (Lehner et al., 2023). To decipher the molecular mechanism by which aza-pano overcome bone marrow microenvironment induced chemoprotection, we conducted transcriptome analysis using 3 KMT2Ar and 1 non KMT2Ar PDX cells harvested from mice treated with aza-pano. A supervised analysis identified 622 differentially regulated genes with log2FC>1. While gene set enrichment analysis (GSEA) revealed transcriptional activation of IL2-STAT5 and TGF-b signaling in KMT2Ar, transcripts modulated by aza-pano suppressed NF-kB pathways in non KMT2Ar PDXNTPL-511. These results indicate that aza-pano mediates hyperactivation of inflammatory signaling KMT2Ar AML. We validated upregulation of NF-kB target genes (IL1B, NFKB1), canonical NF-kB inhibitors (NFKBIA, TNFAIP3) and non-canonical NF-kB signaling mediator (RELB) by qRT-PCR in 3 KMT2Ar PDX models. In PDX DF-5, with KMT2A-MLLT10 fusion, we observed alteration of RelB targets TIFAB, MEF2C. Aza-pano increased RelB protein by 3.1-fold validating upregulation of non-canonical NF-kB signaling pathway in PDX NTPL-146 with KMT2A-MLLT1 fusion. Notably, aza-pano upregulated STAT1 in KMT2Ar PDX and suppressed phospho-STAT1 in non KMT2Ar NTPL-511. We found aza-pano treatment correlates with RelA activation (1.6-fold) and 2.8-fold higher phosphorylation of STAT1 in DF-5. As the bone marrow microenvironment regulates inflammatory signaling, we used a multi-cell coculture model with stromal, endothelial and mesenchymal stem cells to quantify phospho-STAT1 expression by intracellular flow cytometry. Phospho-STAT1 was increased by 2.5- and 3.1-fold in PDX NTPL-377 with KMT2A-MLLT3 and NTPL-146 respectively. We validated aza-pano induced sensitization by targeting upstream (interferon-a, IFN-a) and downstream regulators of IFN signaling with PDX harboring 3 unique KMT2A fusions. If cell death induced by aza-pano is regulated by IFN signaling, then exogenous pre-treatment with recombinant human IFN-a protein should phenocopy aza-pano mediated chemosensitization in multi-cell coculture. While IFN-a sensitized NTPL-146 and DF-5 to cytarabine (cyt) and daunorubicin (dauno), exogenous IFN-a failed to mimic aza-pano mediated sensitivity in NTPL-377, indicating aza-pano mediated sensitivity is dependent on IFN-a signaling in NTPL-146 and DF-5 but not NTPL-377. If cell death induced by aza-pano is regulated by IFN signaling, then inhibiting a downstream regulator of IFN receptor signaling with ruxolitinib (rux) should reverse chemosensitivity. Following inhibition of JAK signaling aza-pano induced sensitivity to dauno was significantly reversed in DF-5, supporting the hypothesis that aza-pano induces cell death in a JAK-dependent manner. While the combination of rux and aza-pano phenocopied chemosensitization in NTPL-377, inhibiting JAK signaling enhanced aza-pano mediated sensitization in NTPL-146. Thus, our results demonstrate that aza-pano induced sensitization is dependent on JAK signaling in NTPL-377, not in NTPL-146. To further validate aza-pano activates an immune response to suppress leukemogenesis, we compared the therapeutic efficacy in immunocompetent and immunocompromised murine models. Aza-pano (2.5 mg/Kg each, Qd5) treatment in KMT2A-MLLT1 bearing immune-deficient mice (NBSGW; median survival =31 days) had reduced efficacy than in mice with an intact immune system (C57BL/6, median survival =38.5 days). Aza-pano treatment followed by cyt (50 mg/Kg, Qd5) and dauno (1.5 mg/Kg, Qd3) in KMT2A-MLLT1 bearing immunocompetent mice (C57BL/6; median survival =45 days) had enhanced therapeutic efficacy than in immune-deficient mice (NBSGW; median survival 32 days). These data confirm that the presence of functional immune system promotes survival by aza-pano. Thus, transcriptional activation of inflammatory signaling and an immune response is responsible for therapeutic efficacy of aza-pano. Mechanistically, we have shown that aza-pano primes KMT2Ar AML by hyperactivation of NFkB signaling.