Open Access

Juliane Grimm, Raj Bhayadia, Lucie Gack, Dirk Heckl, Jan-Henning Klusmann

• Individuals with Down syndrome (DS) are predisposed to developing acute megakaryoblastic leukemia (ML-DS) within their first years of life [1]. Although, ML-DS is associated with a favorable prognosis, children with DS often experience severe toxicities from chemotherapy [2]. This highlights the unmet need for targeted therapies with improved risk profiles in this entity. Consequently, the aim of this study was to investigate a novel therapeutic approach specifically tailored to intervene with hallmarks of ML-DS leukemogenesis. The evolution of ML-DS occurs in a step-wise process originating from pre-malignant transient abnormal myelopoiesis (TAM) [3]. The molecular mechanisms underlying the progression from TAM to ML-DS are not fully understood. However, it was previously shown that epigenetic changes play a pivotal role in ML-DS leukemogenesis. The lysine demethylase LSD1 was identified as a crucial player in this process, as LSD1-driven gene signatures become activated in ML-DS [4]. Accordingly, RNA-sequencing analysis of pediatric acute myeloid leukemia (AML) subtypes revealed that LSD1 was highly expressed in acute megakaryoblastic leukemia (AMKL), and especially in TAM and ML-DS patients (Supplementary Fig. 1). LSD1 is essential for hematopoiesis, particularly during granulocytic and erythroid differentiation [5], and was shown to contribute to differentiation blockade in different AML subtypes [6,7,8]. Consequently, various irreversible LSD1 inhibitors have been developed, with some currently undergoing clinical trials for AML [9]. Therefore, we sought to investigate the rational use of LSD1 inhibitors in pediatric AMKL. The non-DS-AMKL cell line M-07e and the ML-DS cell line CMK were highly sensitive to irreversible LSD1 inhibition (IC50M-07e = 9.1 nM; IC50CMK = 38.8 nM; Supplementary Fig. 2A). Testing serial dilutions of the irreversible LSD1 inhibitor in non-DS-AMKL and ML-DS patient samples expanded via xenotransplantation (see Supplementary Table 1 for patient characteristics), both entities were equally sensitive to LSD1 inhibition (non-DS-AMKL: IC50#1 = 15.0 nM, IC50#2 = 2.0 nM; ML-DS: IC50#1 = 31.2 nM, IC50#2 = 17.1 nM, IC50#3 = 3.8 nM). All dose-response curves plateaued at a certain LSD1 inhibitor concentration (Supplementary Fig. 2B). The non-linear relationship between cytotoxicity and dosage points toward proliferation arrest and differentiation in response to LSD1 inhibition. In line with this, we observed myeloid differentiation upon visual inspection (Supplementary Fig. 3A) and upregulation of the myeloid markers CD86 and CD11b after 3 days of LSD1 inhibitor treatment (Supplementary Fig. 3B). These results revealed a potent proliferation block and induction of differentiation in non-DS-AMKL and ML-DS samples, however, the therapeutic efficacy of LSD1 inhibition may be limited by its non-linear dose-response relationship. Consequently, we aimed to design a rational drug combination to increase its anti-leukemic effects. Another hallmark of ML-DS development is the acquisition of activating mutations in Janus kinases (JAK) and cytokine receptors [4], promising potent anti-leukemic effects of the combination of LSD1 inhibition and the JAK1/JAK2 inhibitor ruxolitinib, as it was previously proposed for JAK2V617F mutated myeloproliferative neoplasms, secondary AML and a CSF3Rmut/CEBPαmut AML model [10,11,12]. Accordingly, pre-treatment with 350 nM LSD1 inhibitor for 3 days followed by exposure to serial dilutions of ruxolitinib led to synergistic growth inhibition in non-DS-AMKL and ML-DS cell lines (Supplementary Fig. 4), as well as in all ML-DS patient samples (Fig. 1A). The combination of LSD1 inhibition and ruxolitinib proved to be very effective in non-DS-AMKL blasts, however, with only additive cytotoxic effects in one of the two patient samples (Fig. 1A). Drug synergy in the ML-DS samples was confirmed when calculating the Bliss synergy scores (Fig. 1B). Interestingly, samples ML-DS #1 (JAK1mut) and #2 (wild-type for JAK1, JAK2, and JAK3, Supplementary Fig. 5) showed particularly high synergy scores (ML-DS #1 synergy score = 10.4; ML-DS #2 synergy score = 15.6; Fig. 1B). Contrary, the JAK3mut patient sample ML-DS #3 (Supplementary Fig. 5) only displayed mild drug synergy between LSD1 inhibition and ruxolitinib (synergy score = 2.0; Fig. 1B). Consequently, as ruxolitinib is a JAK1/JAK2 inhibitor, synergistic anti-leukemic effects seem to depend on JAK mutational status, which must be considered in future pre-clinical and clinical testing of this drug combination for ML-DS patients.