Synergistic Pre-Clinical Activity of Targeted Inhibition of KDM1A and BET Proteins Against Human AML Blast Progenitor Cells

There is a clear need to develop novel therapies that would overcome differentiation block and eliminate AML stem/progenitor cells. Genetic and epigenetic dysregulation of enhancers regulates expressions of myeloid lineage transcriptional regulators and their target genes in AML stem/progenitor cells. LSD1 (KDM1A) is an FAD-dependent amine-oxidase that demethylates mono and dimethyl histone H3 lysine 4 (H3K4Me1 and H3K4Me2), which regulates enhancer maintenance and transcription in AML stem/progenitor cells (LSCs). LSD1 is part of the repressor complexes involving HDACs, CoREST or GFI1 that mediate transcriptional repression and differentiation block in AML blast progenitor cells (BPCs). We had previously reported that treatment with the reversible LSD1 inhibitor (LSDi) SP2509 increases the permissive H3K4Me3 mark on the chromatin, associated with induction of p21, p27 and CEBPα levels, as well as of differentiation and loss of viability of AML BPCs (Leukemia. 2014; 28: 2155-64). In the present studies, we further evaluated the anti-AML efficacy of LSD1i-based combination with BET protein inhibitor (BETi). First, we determined that tet-inducible shRNA to KDM1A depleted protein levels of KDM1A, repressed c-Myc, but de-repressed p21, CD11b (ITGAM), CD86 and CEBPα, thereby inhibiting colony growth and modestly inducing lethality in genetically diverse cultured AML cell lines. Following sgRNA-directed, CRISPR/Cas9-mediated gene-editing of LSD1 in AML BPCs, surviving clones exhibited ~50% KDM1A levels and decreased c-Myc and DNMT1 expressions compared to the control AML BPCs. Treatment with either the reversible LSDi, SP2577 (Salarius Pharma), or with the irreversible LSDi ORY-1001, disrupted binding of KDM1A with CoREST. Following LSDi treatment, ATAC-Seq analyses demonstrated significant increase in the accessible chromatin of AML BPCs (represented by gained peaks). Gained ATAC-Seq peaks also involved the chromatin of MED11/13, LY96, CEBPB, RARA, CDKN1C and CD86 genes. ChIP-Seq analysis also showed increased H3K27Ac peaks in the chromatin of CD86, ITGAM, SAMHD1, TET2, MED12 and E2F1, and a reduction of peaks in RUNX1, CDK6, KIT, CTNNB1, HOXB5, FLT3 and MEIS1. RNA-Seq analyses after LSD1i treatment also showed significant perturbations (log2 fold-change >1.25 and p<0.05) in the mRNA expressions, including those of ITGAM, LY96, CD86, SAMHD1, IRF8, APAF1, CDK6, and KIT. Gene set enrichment analysis against Hallmark and Transcription Factor-Target datasets showed positive enrichment of E2F and GFI1 targets, as well as of IL2-STAT5 and TGF-beta signaling, but significant depletion (FDR q-values <0.1) of MYC-targets and genes involved in oxidative phosphorylation. QPCR and Western analyses following LSD1i treatment confirmed significant up regulation of mRNA and protein levels, respectively, of ITGAM, CD86 and LY96 in cultured and primary patient-derived AML BPCs. This was associated with morphologic features of differentiation and inhibition of colony growth in AML cells (OCI-AML5, MOLM13, THP1 and MV4-11) (p < 0.01). We also queried for expression mimickers (EMs) through connectivity mapping of the mRNA signature following LSD1i treatment, utilizing the LINCS1000-CMap analyses. Among the top EM hits were BET protein inhibitors (BETis). Treatment with BETi or BET protein degraders (PROTACs) depleted LSD1 levels in AML BPCs. Utilizing ChIP-Seq data, we also noted that LSD1 promoter is occupied by BET protein BRD4 in the AML cells. Consistent with this, treatment with the BETi OTX015 depleted KDM1A expression in AML cells. Notably, co-treatment with LSDi (SP2577 or ORY-1001) and OTX015 induced synergistic lethality in AML BPCs, including of CD34+, CD38-, Lin- AML stem/progenitor cells (combination indices < 1.0). This was associated with greater depletion of c-Myc, c-Myb and PU.1, but greater induction of p21 and p27. LSD1i or BETi treatment significantly improved survival of the immune-depleted mice (compared to the control mice) engrafted with the AML OCI-AML5 cells or patient-derived xenograft (PDX) models of AML (p < 0.01). Collectively, these findings elucidate the molecular mechanisms and strongly support further in vivo testing and pre-clinical development of LSD1i-based combinations with BETi against AML BPCs.