Rationale for Combining the BCL2 Inhibitor Venetoclax with the PI3K Inhibitor Bimiralisib in the Treatment of IDH2- and FLT3-Mutated Acute Myeloid Leukemia

In October 2020, the FDA granted regular approval to venetoclax (ABT-199) in combination with hypomethylating agents for newly-diagnosed acute myeloid leukemia (AML) in adults 75 years or older, or in patients with comorbidities precluding intensive chemotherapy. The treatment response to venetoclax combination treatment, however, may be short-lived, and leukemia relapse is the major cause of treatment failure. Multiple studies have confirmed the upregulation of the anti-apoptotic proteins of the B-cell lymphoma 2 (BCL2) family and the activation of intracellular signaling pathways associated with resistance to venetoclax. To improve treatment outcome, compounds targeting anti-apoptotic proteins and signaling pathways have been evaluated in combination with venetoclax. In this study, the BCL-XL inhibitor A1331852, MCL1-inhibitor S63845, dual PI3K-mTOR inhibitor bimiralisib (PQR309), BMI-1 inhibitor unesbulin (PTC596), MEK-inhibitor trametinib (GSK1120212), and STAT3 inhibitor C-188-9 were assessed as single agents and in combination with venetoclax, for their ability to induce apoptosis and cell death in leukemic cells grown in the absence or presence of bone marrow stroma. Enhanced cytotoxic effects were present in all combination treatments with venetoclax in AML cell lines and AML patient samples. Elevated in vitro efficacies were observed for the combination treatment of venetoclax with A1331852, S63845 and bimiralisib, with differing response markers for each combination. For the venetoclax and bimiralisib combination treatment, responders were enriched for IDH2 and FLT3 mutations, whereas non-responders were associated with PTPN11 mutations. The combination of PI3K/mTOR dual pathway inhibition with bimiralisib and BCL2 inhibition with venetoclax has emerged as a candidate treatment in IDH2- and FLT3-mutated AML.


Introduction
With a median age of 68 years at diagnosis, acute myeloid leukemia (AML) is predominantly a disease of the elderly. The majority of these patients are not eligible for intensive chemotherapy with curative intent, and the standard of care has been treatment with hypomethylating agents (HMA) [1,2]. The addition of the BCL2 inhibitor venetoclax  to the HMA backbone leads to an increase in response and survival rates [3]. In October 2020, the Food and Drug Administration granted regular approval to venetoclax (VENCLEXTA ® , AbbVie Inc., North Chicago, IL, USA and Genentech Inc., San Francisco, CA, USA) in combination with HMA including azacitidine, decitabine, or with low-dose cytarabine (LDAC) for newly-diagnosed AML in adults 75 years or older, or in patients with comorbidities precluding intensive chemotherapy. The combination of venetoclax

Variable Susceptibility of AML Cell Lines to Venetoclax and Various Targeted Therapies
To determine the sensitivity of AML cells to different targeted compounds, AML cells were subjected to in vitro cytotoxicity assays. Seven AML cell lines and one BCL2-driven DLBCL cell line were treated for 20 h in dose escalation experiments before cell viability assessment. Our panel of AML cell lines covered the majority of the morphologic and molecular subtypes, particularly including FLT3-ITD and FLT3 wild type, NPM1 mutant and wild type, as well as TP53 wild type, mutant, hemizygous, and null cells (Table 1). We observed that the BCL2-driven and TP53 double mutant lymphoma cell line OCI-Ly1 was highly susceptible to venetoclax with IC50 of 60 nM. The AML cell lines ML-2 and MOLM-13 were susceptible to venetoclax with IC50 of 100 and 200 nM, while OCI-AML3, SKM-1, and HL-60 had IC50 of 600 nM, 1 µM, and 1.6 µM, respectively. PL-21 and MOLM-16 cells were resistant to venetoclax with IC50 > 10 µM ( Figure S1, Tables S1 and 2). The IC 50 levels of susceptible AML cells were in the range of physiological relevant concentrations. Venetoclax steady state plasma concentrations of 1.2-3.5 µM were observed in CLL patients receiving the recommended phase 2 dose of 400 mg per day [31]. In order to find effective treatment combinations, we focused on inhibitors with the potential to elicit synergistic effects in combination with venetoclax. These included targeted compounds tested in previous studies like BMI1-, MCL1-, and MEK-inhibitors [8,9] as well as BCL-XL-, PI3K-and STAT3 inhibitors as indicated in Figure 1. With respect to the MCL-1 inhibitor S63845, we found that PL-21 and MOLM-16 cells were resistant with IC50 > 10 µM, while the other AML cell lines were susceptible with IC50 of 100-500 nM. With respect to the BCL-XL inhibitor A1331825, PL-21 and MOLM-16 cells were susceptible with IC50 of 2-4 µM, while other AML cell lines were resilient with IC50 values of 5-10 µM. With respect to the dual PI3K-mTOR inhibitor bimiralisib, the PTEN deleted OCI-Ly1 cells were most sensitive with an IC50 of 1 µM, the FLT3-mutated MOLM-13 with IC50 of 2 µM, ML-2 and SKM-1 with IC50 of 3 µM, while OCI-AML3, PL-21 and MOLM-16 were resilient with IC50 of 10 µM. With respect to the BMI-1 inhibitor PTC596, ML-2, PL-21 and MOLM-16 cells had IC50 >1 µM, while the other AML cell lines were susceptible with IC50 in the range of 200-500 nM. With respect to the MEK inhibitor trametinib, HL-60, ML-2, MOLM-13, OCI-AML3 and SKM-1 cells were most susceptible with IC50 of 80-120 nM, while MOLM-16 and PL.-21 cells were resistant with IC50 of 10 µM. The IC50 levels of susceptible AML cells were in the range of physiologically relevant concentrations: Bimiralisib plasma levels were 2 µM in patients treated for solid tumors [32]; A-1331852 plasma levels were 2 µM in rat p.o. [33]; PTC-596 cmax ranged from 1 to 5 µM in patients with advanced tumors [34]; trametinib plasma levels were 20 nM in patients treated for BRAF melanoma [35]. Only for the STAT3 inhibitor C-188-9 the in vitro IC50 levels of AML cells exceeded the 2 µM plasma concentration determined in PDX mice [36]. The in vitro IC50 of C-188-9 was in the range of 4-8 µM in AML cell lines, and in the range of 8-18 µM in primary AML samples [37].

Synergistic Effects on Cell Viability in AML Cell Lines Treated with Venetoclax Combinations
Cell viability was determined in seven AML cell lines and one BCL2-driven DLBCL cell line treated with increasing dosages of single compounds and in combination using variety of targeted therapies including the BCL-XL inhibitor A1331825, the PI3K inhibito bimiralisib (PCR309), the STAT3 inhibitor C-188-9, the BMI-1 inhibitor PTC596, the MCL Figure 1. Schematic representation of signaling pathways in myeloid cells. The inducible growth factor receptor FLT3 signals via PI3K-AKT-mTOR and RAS-MEK-ERK (black arrows). FLT3-ITD, a constitutively active growth factor receptor, additionally induces PTPN11-STAT5 (red arrows). Activated cytokine receptors signal via Janus kinase (JAK)-signal transducer and activator of transcription (STAT) pathway. Signal transduction leads to inhibition of the tumor suppressor p53 and induction of the anti-apoptotic BH3 proteins BCL2, BCL-XL, and MCL1, thereby promoting proliferation and cell growth of myeloid cells. Oncogenic functions are indicated in red, tumor suppressor functions in green, chemical inhibitors in blue.

Synergistic Effects on Cell Viability in AML Cell Lines Treated with Venetoclax Combinations
Cell viability was determined in seven AML cell lines and one BCL2-driven DLBCL cell line treated with increasing dosages of single compounds and in combination using a variety of targeted therapies including the BCL-XL inhibitor A1331825, the PI3K inhibitor bimiralisib (PCR309), the STAT3 inhibitor C-188-9, the BMI-1 inhibitor PTC596, the MCL1 inhibitor S63845, and the MEK inhibitor trametinib. Drug concentrations in the combination studies were chosen to correspond to minimally effective concentrations in single compound assays determined in initial titration. Five AML cell lines and OCI-Ly1 were susceptible to 100 nM venetoclax and multiple combination treatments, while PL-21 and MOLM-16 cells were resistant to 100 nM venetoclax and most combination treatments, with the exception of venetoclax and A1331825 ( Figure 2). Across the panel of AML cell lines, cytotoxic effects were enhanced in the combination treatments, for venetoclax and bimiralisib (p = 0.002), trametinib (p = 0.003), S63845 (p = 0.005), A1331825 (p = 0.01), PTC596 (p = 0.01), or C-188-9 (p = 0.04), Combination indexes were calculated according to Chou Talalay [38]. Synergistic effects on cell viability were calculated to be moderate to strong for venetoclax combined with S63845, A1331825, trametinib or bimiralisib, and mild to moderate for venetoclax combined with PTC596 or C-188-9 (Table 3).

Altered Susceptibility to Targeted Therapies in AML Cells Grown in the Presence of Bone Marrow Stroma
To investigate the efficacy of venetoclax combination treatments in the bone marrow environment, cell viability was determined in the susceptible AML cell lines MOLM-13, ML-2, SKM-1, and OCI-AML3, grown in the absence or presence of bone marrow stroma cells. Stroma cells secrete granulocyte and macrophage colony-stimulating factors (G-CSF, GM-CSF, M-CSF) and a variety of cytokines, which can induce STAT signaling in leukemic cells [39][40][41][42]. AML cells grown in the presence of bone marrow stroma were generally less affected by the combination treatment than AML cells grown in the absence of stroma, indicating a protective effect of the bone marrow environment on AML cells ( Figure 3). We observed that only the combination of venetoclax and MCL1 inhibitor S63845 induced cell death with equal efficacy in AML cells grown in the absence or presence of bone marrow stroma. MOLM-13 cells were protected toward venetoclax and combination treatments when grown on stroma ( Figure 3A), ML-2 cells appeared to be protected toward trametinib when grown on stroma ( Figure 3B). OCI-AML3 cells were more susceptible to venetoclax when grown on stroma, but protected toward all venetoclax combination treatments ( Figure 3C). SKM-1 cells were more susceptible to A1331825, and protected toward bimiralisib and trametinib, when grown on stroma ( Figure 3D). row stroma. MOLM-13 cells were protected toward venetoclax and combination treatments when grown on stroma ( Figure 3A), ML-2 cells appeared to be protected toward trametinib when grown on stroma ( Figure 3B). OCI-AML3 cells were more susceptible to venetoclax when grown on stroma, but protected toward all venetoclax combination treatments ( Figure 3C). SKM-1 cells were more susceptible to A1331825, and protected toward bimiralisib and trametinib, when grown on stroma ( Figure 3D). Concentrations of inhibitors were 100 nM for venetoclax, S63845, PTC596 and trametinib, 1 μM for bimiralisib, C-188-9, and A133825. Significance was calculated in a graph pad prism using grouped analysis with multiple unpaired t-test comparing cell viabilities of treated cells grown in the absence or presence of HS-5 stroma. Significance denoted for p < 0.05 (*); p < 0.005 (**); p < 0.0005 (***); no significance denoted for p > 0.05 (ns). A significance level of 0.05 indicates a 5% risk of concluding that a difference exists when there is no actual difference.   Depending on DAPI staining intensity cells were classified as subG1, G0/G1, S phase, or G2 phase (A). Treatment-induced cell death (subG1 fraction) (B), and G1 cell cycle arrest (C). Depending on Annexin V and PI staining intensity, cells were classified as vital (Ann lo, PI lo), early apoptotic (Ann hi, PI lo), late apoptotic (Ann hi, PI hi) or necrotic (Ann lo, PI hi) (D). Treatment-induced loss of vital cells (E) and amount of apoptotic cells (F) were significanty enhanced in the combination treatment. Significance of differences in median values was calculated by the Mann-Whitney test. Significance denoted for p < 0.05 (*); p < 0.005 (**); no significance denoted for p > 0.05.

Venetoclax Combination Treatments with Differential Efficacy in Subsets of AML Patients
After initial studies in AML cell lines, the treatment combinations of venetoclax with A1331825, S63845, bimiralisib, trametinib, C-188-9, or PTC596 were applied to patientderived mononuclear cells isolated from peripheral blood (PBMC) or bone marrow (BMMC) ( Table 4). A total of 26 AML, one CML, two NHL, as well as PBMCs of four healthy donors (HD) were subjected to single compound and combination treatments. The tested combination treatments induced minor reduction of cell viabilities in mononuclear cells isolated from healthy donors, and substantial reduction of cell viability in 50-70% of AML samples treated with venetoclax monotherapy ( Figure 5A) or in combination with A1331825, S63845 or bimiralisib ( Figure 5B-D). Venetoclax in combination with trametinib, C-188-9 or PTC-596 was less effective with substantial reduction of cell viability in 20-40% of AML samples ( Figure 5E,F). The patient samples were divided in two subgroups of similar size, one with major (strong) response (SR), and one with minor (normal) response (NR). The median cell viabilities in the SR groups were 70% in 100 nM venetoclax treatment ( Figure 5A), 50% in the venetoclax combination with 1 µM A1331825 or 1 µM bimiralisib ( Figure 5B,C), 43% in combination with S63845 ( Figure 5D), 55% in combination with 100 nM trametinib ( Figure 5E), 60% in combination with 1 µM C-188-9 ( Figure 5F), and 62% in combination with 200 nM PTC596 ( Figure 5G).  Potential response markers were deduced from the correlation analysis of cell viabilities grouped according to diagnostic parameters including gene mutation status, peripheral blood and bone marrow blast cells percentage, and CD34 positivity. In venetoclaxtreated AML, the presence of IDH2 mutation, as well as elevated blast cell percentage in peripheral blood or bone marrow associated with response, presence of PTPN11, TET2 or  Potential response markers were deduced from the correlation analysis of cell viabilities grouped according to diagnostic parameters including gene mutation status, peripheral blood and bone marrow blast cells percentage, and CD34 positivity. In venetoclax-treated AML, the presence of IDH2 mutation, as well as elevated blast cell percentage in peripheral blood or bone marrow associated with response, presence of PTPN11, TET2 or ASXL1 mutation indicated lack of response, while FLT3, NPM1, RUNX1, and TP53 status as well as CD34 levels were not associated with treatment response (Figure 6). In the venetoclax and A1331825 combination treated AML gene mutation status of TET2, FLT3, and TP53 as well as CD34, levels were inconsequential, IDH2 mutation and elevated blast cell percentage were associated with response, while the presence of PTPN11 or ASXL1 mutation indicated lack of response (Figure 7). In the venetoclax and bimiralisib combination treatment, the presence of IDH2 or FLT3 mutation and elevated blast cell counts were indicators of response, ASXL1 and TP53 status was inconsequential, while presence of PTPN11 or TET2 mutation indicated a lack of response ( Figure 8). In the combination treatment of venetoclax and S63845, the presence of IDH2 mutation and elevated blast counts were associated with response, presence of TET2 mutation indicated lack of response, while NPM1, FLT3, ASXL1, PTPN11, and TP53 status were inconsequential (Figure 9). Notably, AML patient samples with TET2 mutations carried additional mutations in PTPN11, KRAS, FLT3, or TP53 genes, all of which may be associated with venetoclax resistance (Table 4). In the combination treatments of venetoclax and trametinib, C-188-9 or PTC596, the number of samples with reduced cell viability was small, and response markers were not identified ( Figure S9).

Discussion
Treatment response to the BCL2 inhibitor venetoclax together with hypomethylating agents may be short-lived with leukemia relapse as the major cause of treatment failure. Multiple studies have indicated that the upregulation of other anti-apoptotic proteins of the B-cell lymphoma 2 (BCL2) family and the activation of intracellular signaling pathways were the major factors leading to resistance to venetoclax [4,5,7]. Accordingly, targeting anti-apoptotic proteins BCL-XL and MCL-1 as well as targeting signaling pathways leading to the induction of BCL-XL and MCL-1 may enhance and prolong treatment response to the BCL-2 inhibitor venetoclax.
In this study, we describe a panel of venetoclax combination treatments with enhanced cytotoxic effects on AML cells grown in the absence or presence of bone marrow stroma, including the BCL-XL inhibitor A133825, the MCL1 inhibitor S63845, the BMI1 inhibitor PTC596, the dual PI3K-mTOR inhibitor bimiralisib, the STAT3 inhibitor C-188-9, and the MEK inhibitor trametinib. The in vitro IC50 concentrations of the tested inhibitors were determined to be in the range of physiologically relevant concentrations for all compounds, except C-188-9, where in vitro IC50 levels exceeded the plasma concentrations present in PDX mice. AML cells grown in the presence of bone marrow stroma were generally less affected by the combination treatments than AML cells grown in the absence of stroma, indicating a protective effect of the bone marrow environment on AML cells. MOLM-13 cells were protected toward venetoclax and combination treatments when grown on stroma, ML-2 cells were protected toward trametinib, OCI-AML3 cells were protected toward all venetoclax combination treatments, and SKM-1 cells were more susceptible to venetoclax and A1331825, and protected toward bimiralisib and trametinib, when grown on stroma. SKM-1 cells have been found to be more susceptible to venetoclax and to the STAT5 inhibitor AC-4-130 when grown on stroma [15]. A cell-type-specific dependence on STAT5 signaling may cause elevated susceptibility to the BCL-2-and BCL-XL inhibitors and a reduced response to PI3K-and MEK inhibitors, in the bone marrow environment. Various cellular components, cytokines, and chemokines present in the bone marrow may impact AML initiation and therapy resistance at the cellular and molecular level [43,44]. We found that only the combination of venetoclax and MCL1 inhibitor S63845 induced cell death with equal efficacy in AML cells grown in the absence or presence of bone marrow stroma, indicating a potential advantage of applying this combination in the treatment of AML, as this may eradicate leukemic stem cells in the bone marrow. A synergistic effect of S63845 toward venetoclax-mediated apoptosis of AML cells in the context of interaction with the BM microenvironment that intrinsically mediates resistance to BCL2 inhibition has been previously described [45]. Targeting MCL-1 may dysregulate the cellular metabolism and leukemia-stroma interactions and re-sensitize acute myeloid leukemia to BCL-2 inhibition [46].
To validate the findings in a translational setting, venetoclax combination treatments were applied to mononuclear cells isolated from the peripheral blood or bone marrow of primary AML patients. The addition of the BCL-XL inhibitor A133825, the MCL1 inhibitor S63845 or the PI3K inhibitor bimiralisib to venetoclax induced substantial reductions of cell viability in 50-70% of the tested AML samples, while the addition of the MEK inhibitor trametinib, the BMI-1 inhibitor PTC596 or the STAT3 inhibitor C-188-9 to venetoclax was less effective, with substantial reductions of cell viability in 20-40% of tested AML samples. In order to define the patient subgroups who may profit from novel targeted combination therapies, potential response markers were deduced from the correlation analysis of cell viabilities grouped according to diagnostic parameters, including gene mutation status of prevalent tumor suppressors and oncogenes, peripheral blood and bone marrow blast cell percentage, and levels of CD34 positive cells. A significant association between venetoclax response and elevated blast cell percentage has been previously described [47], and was reproduced in our study with a boundary value of 60% peripheral blood blast percentage, and 70% bone marrow blast cell infiltration. Mononuclear cells isolated from AML patients with elevated blast cell percentage were more susceptible to venetoclax and multiple venetoclax combinations in vitro. IDH2, NPM1, FLT3, DNMT3A, PTPN11, ASXL1, TET2, KRAS, RUNX1, and TP53 genes have been described as response markers to venetoclax treatment [4,5,7,8]. In our study, IDH2 mutation was the single most significant and consistent biomarker associated with response to venetoclax and multiple venetoclax combination treatments, while TP53 mutation was not associated with response. Notably, the TP53 double mutant cell line OCI-Ly1 was most sensitive to venetoclax, while the TP53 double mutant cell line MOLM-16 was resistant.
The mutation status of other genes was relevant to response, however, with differential indicators in specific venetoclax combination treatments. FLT3 mutations were associated with response to venetoclax and bimiralisib, but not to venetoclax monotherapy or in combination with A1331825 or S63845, indicating that FLT3-mutated cells may be specifically susceptible to the combined inhibition of BCL2, PI3K, and mTOR. The TET2 gene mutation has been associated with response to venetoclax and HMA combination treatment [5]. In our study, TET2 mutation apparently associated with resistance to venetoclax in combination with A1331825 or bimiralisib, possibly due to presence of concurrent mutations in KRAS, PTPN11, or TP53 genes in the primary AML samples. PTPN11 mutations were associated with lack of response to venetoclax and A1331825 or bimiralisib, but not to venetoclax and S63845, indicating differential target cell specificity and differential mechanisms of action for S63845 in combination with venetoclax. Activating mutations of the SHP2 protein, encoded by the PTPN11 gene, leads to hyper-activation of the downstream RAS-MAPK signaling pathway and confer resistance to venetoclax and multiple venetoclax combinations. The combination of venetoclax and the MCL1 inhibitor AZD5991 was proposed to overcome this resistance [47]. In accordance, the combination treatment with venetoclax and the MCL-1 inhibitor S63845 resulted in reduced cell viabilities in primary AML samples in vitro independent of PTPN11 status. ASXL1 mutations have been associated with distinct epigenomic alterations that lead to sensitivity to venetoclax and azacytidine [48]. In our study, ASXL1 mutation apparently associated with resistance to venetoclax in combination with A1331825, but not in combination with S63845 or bimiralisib, possibly due to the presence of concurrent mutations in KRAS, PTPN11, RUNX1, or TP53 genes in the primary AML samples. Further studies in larger cohorts may be required to validate the relevance of TET2, TP53, and ASXL1 mutation status as response markers to venetoclax combination treatments. Additional biomarkers of response to venetoclax combination treatments may arise from studies on expression levels of pro-apoptotic BCL-2 family proteins and anti-apoptotic proteins (BIM, BAX, and BAK) as well as the components of the upstream signaling pathways in larger cohorts.
In conclusion, different combinations of targeted therapies emerge that are suitable in the treatment of specific subsets of AML patients. In this study, elevated in vitro efficacies were detected in the combination treatments of venetoclax with BH3 mimetics A1331852 and S63845 and the dual PI3K-mTOR inhibitor bimiralisib, in AML with elevated blast cell percentage, at drug concentrations that can be reached in vivo in the plasma, with different response markers in each combination. The combination treatment of venetoclax and A1331825 may be effective in the treatment of IDH2-mutated AML in the absence of PTPN11 mutations, while the combination of venetoclax and S63845 may be effective in AML in the presence of PTPN11 mutations. The combination treatment of venetoclax and bimiralisib may be effective in AML with IDH2, NPM1, and FLT3 mutations. We propose the combination of PI3K/mTOR dual pathway inhibition with bimiralisib and BCL2 inhibition with venetoclax as a candidate treatment in clinical trials for IDH2or FLT3-mutated AML.

Patient Samples
Mononuclear cells of AML patients diagnosed and treated at the University Hospital, Bern, Switzerland, between 2018 and 2022, were included in this study. Informed consent from all patients was obtained according to the Declaration of Helsinki, and the studies were approved by decisions of the local ethics committee of Bern, Switzerland, decision number 221/15. Peripheral blood mononuclear cells (PBMCs) and bone marrow mononuclear cells (BMMCs) were collected at the time of diagnosis before initiation of treatment. The AML cells were analyzed at the central hematology laboratory of the University Hospital Bern according to state of the art techniques [49]. Mutational screening for FLT3, NPM1, TP53, and conventional karyotype analysis of at least 20 metaphases were performed in all samples. In addition, all samples were analyzed by NGS sequencing of the myeloid panel genes. The genes tested within the NGS panel can be categorized into several major functional categories, including the spliceosome (U2AF1, SF3B1, SRSF2, and ZRSR2), epigenetic modifiers (TET2, DNMT3A, BCOR, ASXL1, IDH1, and IDH2), cohesions (STAG2, RAD21, and SMC3), transcription factors (TP53, RUNX1, WT1, and ETV6), signaling molecules (NF1, NRAS, CBL, PTPN11, JAK2, and FLT3), and chromatin modifiers (EZH2 and ASXL1).  (IL-1alpha), IL-1beta, IL-1RA, IL-6, IL-8, IL-11, and leukemia inhibitory factor (LIF) [41,42]. For the co-culture assays HS-5 cells were plated on standard tissue culture plates (REF 83.3920, Sarstedt, Nümbrecht, Germany) on day 1. On day 2, Nunc 0.4 µm cell culture inserts (Thermo Fisher Scientific, Nunc A/S, Roskilde, Denmark) were placed over the HS-5 feeder layer and AML cells were filled into the cell culture inserts. On day 3, AML cells were collected from the six well inserts and replated on tissue culture plates suspension 96 well (REF 83.3924500, Sarstedt, Nünmbrecht, Germany), before addition of compounds. Cytotoxicity assays were performed on day 4.

Cytotoxicity Assays
For assays with AML cell lines, cells were plated at a density of 5 × 10 5 /mL on tissue culture plates suspension 96 well (REF 83.3924500, Sarstedt, Nümbrecht, Germany), and treated with targeted compounds. For assays with patient-derived mononuclear cells, the cells were cultured for 24 h prior to treatment. The BMI1 inhibitor PTC596 (HY-112041), the MCL1 inhibitor S63845 (HY-100741), the MEK inhibitor trametinib (HY-10999), the STAT3 inhibitor C-188-9 (HY-112288) and the PI3K inhibitor PQR309 (HY-12868) were purchased at MedChemExpress (Monmouth Junction, NJ, USA). PQR309 (bimiralisib) is a novel brain-penetrant dual PI3K/mTOR inhibitor with in vitro and in vivo anti-lymphoma activity as single agent and in combination. A stock solution of Venetoclax was prepared by dissolving a tablet in DMSO (Venclexta ® , Abbvie Inc., North Chicago, IL, USA). Cell viability was determined 20 h after the start of treatment using the MTT-based cell proliferation kit I (Ref 11465007001, Roche Diagnostics GmbH, Mannheim, Germany). This time point was selected because the cellular responses were effectual for the calculation of combination indexes after 20 h of treatment with two compounds in leukemic cells. For AML cell lines, four independent assays (biological replicates) with four measurements (technical replicates) per dosage were performed. For hematological patient samples, two independent assays with three technical replicates per dosage were performed. For the calculation of combination indexes, two dosages of venetoclax and two dosages of the other compounds were applied alone and in combination. Combination indexes were calculated on Compusyn software (version 1.0; ComboSyn, Inc., Paramus, NJ, USA). Data are depicted as scatter plots with median values and SD. In grouped analysis, significance of differences in median values was calculated by Mann-Whitney test.

Imaging Cytometry
Imaging cytometry was carried out on the NC-3000 cell analyzer (ChemoMetec, Allerod, Denmark) with reagents supplied by ChemoMetec. To determine the induction of cell death apoptotic cells were stained with AnnexinV-CF488A conjugate (Biotium, Fremont, CA, USA) in AnnexinV buffer and Hoechst 33,342 (10 µg/mL) for 15 min at 37 • C, followed by several washes. Propidium iodide was added shortly before imaging. For cell cycle analysis, cells were incubated in lysis buffer with DAPI (10 µg/mL) for 5 min at 37 • C before imaging on the NC-3000 cell analyzer.