PARP1 inhibition sensitizes AML to CD123 CAR-NK cells via immunophenotypic remodeling

Abstract Background:Acute myeloid leukemia (AML) is an aggressive hematologic malignancy driven by leukemic stem cells (LSCs) with enhanced resistance to chemotherapy. Chimeric antigen receptor (CAR) T and NK cell therapies show promising results in lymphatic leukemias with B-cell origin. In AML, the efficacy of CAR-lymphocytes is hampered by the heterogeneous expression of putative target antigens on LSCs. CD123 is one of the most consistently expressed antigens on LSC and is investigated in this study as target of CAR-NK cells. Latter were chosen due to their ability to co-target further antigens, among which NKG2D-ligands (NKG2DL), that can be induced on LSC by concomitant treatment with PARP1 inhibition (PARP1i).Methods: CD123 CAR-NK cells were generated via lentiviral transduction of cytokine-activated NK cells from healthy donors and expanded feeder cell-free for 10 to 21 days. Cytotoxicity assays were performed using AML cell lines (n=3) and primary AML samples (n=7) at varying effector-to-target (E:T) ratios, with or without PARP1i (AG-14361) pre-treatment. Killing efficacy and the phenotype of residual populations were analyzed via flow cytometry, and colony-forming unit (CFU) assays were conducted with the remaining cells to evaluate residual clonogenic potential after combination treatment (n=3). In vivo, a patient-derived xenograft (PDX) model was established by intravenous injection of NSG mice with primary AML cells. Mice received +/-PARP1i for 5 days, followed by two intravenous infusions of CD123 CAR-NK cells or NK cells, along with daily IL-2 support. AML burden, LSC persistence, and CAR-NK cell distribution were assessed by flow cytometry in peripheral blood, spleen, liver, and bone marrow at endpoint (Day 10 after first NK cell treatment).Results: CD123 CAR-NK cells demonstrated strong, antigen-specific killing in vitro. Cytotoxicity was high against CD123high AML cell lines (e.g., Molm-13) and patient samples, but limited in CD123low lines (e.g., HL-60) and CD123low AML samples (<55% CD123 expression). In vivo, CD123 CAR-NK cells substantially reduced leukemic burden in peripheral tissues: PB showed a 15-fold decrease (from 1.5% to 0.1% CD123+ cells), spleen a 10-fold decrease (1% to 0.1%), and liver a 6.5-fold decrease (13% to 2%). However, the targeting of AML cells in the bone marrow was limited, with only a 1.1-fold decrease (from 11% to 10%). Notably, residual LSCs persisted in BM, spleen, and liver, highlighting a key limitation of CAR-NK monotherapy.Therefore, we investigated the combination of PARP1i with CD123 CAR-NK cells.In vitro,PARP inhibition modulated the AML immunophenotype, amongst others significantly upregulating NKG2DL and TRAIL-R, while downregulating HLA-ABC, thereby increasing susceptibility to NK cell surveillance. Indeed, PARP1i pre-treatment significantly improved CD123 CAR-NK cell-mediated cytotoxicity, especially at low E:T ratios (1:8), where we observed an increase in killing efficiency of more than 2-fold compared to CD123 CAR-NK cells alone. CFU assays following cytotoxicity assays demonstrated an 8-fold reduction in clonogenic AML potential with the combination therapy of PARP1i and CD123 CAR-NK cells compared to untreated primary AML cells. In vivo experiments evaluating the combination of PARP1i with CD123 CAR-NK cells in mice engrafted with primary AML samples are ongoing and results will be presented at the meeting.Conclusions: CD123 CAR-NK cells mediate potent, antigen-specific killing of CD123high AML cells and reduce leukemic burden. However, especially LSC elimination and bone marrow targeting remain suboptimal. PARP1 inhibition reprograms AML immunophenotypes to enhance NK cell recognition. The combination of PARP1i with CD123 CAR-NK therapy represents a promising strategy to overcome immune escape and improve eradication of both bulk AML and therapy-resistant LSCs.