SID7969543 was purchased from Tocris Bioscience (Bristol, United Kingdom). 2-Cl-ATP was procured from Santa-Cruz Biotechnology (Texas, USA). 2-CADO and cladribine were purchased from Sigma Aldrich (New South Wales, Australia). For the generation of lipid-based nanoparticles used to transfect leukemia cells with siRNA, cholesterol and the lipids 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC), 1,2-dimyristoyl-rac-glycero-3-methoxypolyethylene glycol-2000 (DMG-PEG), and 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-[lissamine rhodamine B sulfonyl] (DOPE-Rho) were obtained through Merck & Co., Inc. D-Lin-MC3-DMA was purchased from Assay Matrix Pty Ltd. siGENOME Human NR5A1 siRNA (M-003429-00-0010) targeting NR5A1 and On-TARGETplus control pool non-targeting scrambled siRNA (D-001810-10-20) were purchased from Millennium Science Pty Ltd.

All cell lines used in this study were mycoplasma-free and have been authenticated using STR profiling. Cells were cultured as previously described (7, 8). Characteristics of the leukemia cell lines used in this study are described in Supplementary Table 1 . Peripheral blood mononuclear cells from healthy donors were purchased from Australian Red Cross.

The following chemical libraries were used in the screening as previously described (9): Prestwick Chemical Library (Prestwick Chemical PC SAS, France), LOPAC^®1280 library (Sigma-Aldrich, Australia), Tocriscreen Plus library (Tocris Bioscience, UK) and Selleck Inhibitor Library (Selleck Chemicals, USA), together combining 3707 compounds. PER-485 (KMT2A-r) and CCRF-CEM (CEM, KMT2A-wt) leukemia cells were seeded into 384-well culture plates. Test compounds were added to the assay plates for a final concentration of 5 μM (in DMSO stock). Plates were incubated for 72 hours at 37°C, 5% CO₂. Resazurin was added and plates were incubated for 7 hours. The difference in relative fluorescence units at time zero and 7 hours of resazurin incubation was calculated for each well. The percentage cell viability for each test compound was calculated relative to the cells treated with vehicle (100% viability).

The cytotoxicity of drugs was assessed through resazurin-based assays and the inhibitory concentration resulting in 50% reduction of cell survival relative to control (IC₅₀) was calculated as previously described (7). In synergy studies, cells were exposed to a dilution series of compounds, as single agents or in combination, in resazurin-based assays (72 hours) as described (8–10). The occurrence of synergy was determined with the Bliss Independence model as previously described (8–10). Bliss Prediction curves indicated the predicted percentage viability of the cells when exposed to the combination of compounds if both compounds work additively together. Synergy was visualised as the presence of a lower cell viability upon combination of two compounds compared to the viability predicted based on the presence of an additive effect of the compounds (i.e., the viability curve of the combination runs below the Bliss Prediction curve). The percentage of apoptotic cells was assessed by annexin V and 7-Aminoactinomycin D (7-AAD) (BD Biosciences, Australia) flow cytometry as previously described (7).

Western blotting experiments were performed as previously described (7). Antibodies used are listed in Supplementary Table 2 . Actin was used as a loading control.

Total RNA was isolated using a Qiagen RNeasy kit and cDNA synthesized using the iScript cDNA Synthesis Kit (BioRad). Quantitative PCR was run and analyzed on Applied Biosystem’s QuantStudio5 Real-Time PCR Detection System, using Taqman gene expression assays for genes NR5A1 (Hs00610436_m1), GUSB (Hs99999908_m1) and HPRT (Hs99999909_m1) Relative expressions were determined as previously described (7).

Compound half-life was estimated by mouse liver microsomal stability assays as described (10).

Lipid-based nanoparticles encapsulating siRNA targeting NR5A1 and scrambled siRNA (LNP-siRNA) were prepared using a NanoAssemblr™ Spark^® instrument (Precision Nanosystems) as previously described (11). Prior to addition to cell cultures, LNP-siRNA suspensions were further diluted in PBS (pH 7.4) or culture media to lessen traces of ethanol to non-toxic levels (<0.5% vol/vol). Dynamic light scattering using a Malvern Zetasizer Ultra (Malvern Ltd) was used to determine the LNP-siRNA size (~67 nm) and polydispersity index (<0.2). siRNA encapsulation (%) in LNPs was determined by Quant-iT™ RiboGreen™ RNA Assay (ThermoFisher Scientific) in the presence and absence of Triton X-100 (0.5% vol/vol).

Cells were seeded at 0.5 x 10⁶ cells/mL in 12 wells culture plates and incubated at 37°C, 5% CO2, with LNPs encapsulating anti-NR5A1 siRNA or scrambled siRNA at 250 nM siRNA concentration for up to 72 hours.

To identify novel compounds targeting KMT2A-r leukemia, a phenotypic cell-based viability screen on a library containing 3707 compounds performed previously was re-analyzed (9). The library, consisting of the Prestwick (n=1200), Tocris (n=1119), LOPAC (n=1280) and Selleck (n=108) libraries, comprised a mixture of drugs approved by the FDA and other agencies, such as the European Medicines Agency, and biologically active compounds that cover a wide range of targets including kinases, neurotransmitter receptors and G-protein-coupled receptors. The library was screened against a KMT2A-r leukemia cell line, PER-485, harboring the most common KMT2A translocation found in infants and pediatric patients [t(4;11)], and a KMT2A-wt leukemia cell line CCRF-CEM (9, 12, 13). Both lines were isolated from children with relapsed leukemia, providing excellent in vitro models for high-risk disease. The viability of cells treated with a fixed concentration (5 µM) of the library compounds was assessed in resazurin-based cytotoxicity assays after a 72-hour incubation to allow selection of drugs acting rapidly on cell viability as previously described (9). To discover potent compounds with selective cytotoxicity towards the KMT2A-r leukemia cells, we defined two stringent criteria to select hit compounds: 1) decreasing cell viability of PER-485 cells to 10% or less, and 2) producing a greater growth inhibition of PER-485 KMT2A-r leukemia cells compared to CEM KMT2A-wt leukemia cells (%viability of CEM – %viability of PER-485 ≥ 40%). One hundred ninety-seven compounds decreased the viability of the PER-485 cells to 10% or less after a 72-hour incubation. One hundred and forty agents showed a viability difference of ≥ 40% in CEM versus PER-485 cells ( Figures 1A, B ). A total of six compounds fell into both categories and were subsequently retested in a secondary validation screen in triplicate ( Figure 1A ). Four of these agents were confirmed to preferentially target KMT2A-r leukemia cells, including the adenosine analog 2-chloroadenosine triphosphate (2-Cl-ATP), SID7969543, an inhibitor of NR5A1, β-carboline-3-carboxylate (β-CCB), an endogenous proconvulsant and anxiogenic benzodiazepine receptor ligand, and oxethazaine, a Na⁺ channel blocker ( Figure 1C ).

To investigate the selectivity of the four short-listed hit compounds towards KMT2A-r leukemia, cytotoxicity assays were performed in a leukemia cell line panel (n=8) that included the PER-485 and CEM cells used in the original screen, as well as four additional KMT2A-r cell lines (both ALL and acute myeloid leukemia (AML)) with either t(4;11) (n=2) or t(9;11) (n=2) KMT2A translocations, two KMT2A-wt leukemia cell lines and a non-malignant cell line MRC-5 ( Figure 1C and Supplementary Table 1 ). While β-CCB displayed strong cytotoxic activity with a concentration of 2.6 µM needed to achieve 50% inhibition of cell survival (IC₅₀) in PER-485 cells, confirming our results from the original screen, it was less potent against the other KMT2A-r leukemia cell lines tested ( Figure 1C ). Oxethazaine demonstrated cytotoxicity against several KMT2A-wt and KMT2A-r leukemia cell lines with limited selectivity towards KMT2A-r leukemia cells. However, the other two compounds, 2-Cl-ATP and SID7969543 demonstrated selective cytotoxic activity against a subset of KMT2A-r leukemia cells without inhibiting the KMT2A-wt leukemia cells or the non-malignant cell line, MRC-5 ( Figure 1C ). 2-Cl-ATP and SID7969543 were therefore selected for further evaluation.

To further characterize the cytotoxic profile of the adenosine analog 2-Cl-ATP, the compound was evaluated in an expanded leukemia cell line panel comprising six additional KMT2A-r leukemia cell lines, including five derived from infants with KMT2A-r ALL and two CALM-AF10 translocated leukemias, which are KMT2A-wt but represent an aggressive leukemia subtype that shares underlying molecular etiological pathways with KMT2A-r leukemias, such as their dependency on DOT1L histone-lysine methyltransferase and an upregulation of HOXA cluster genes (14–16). In addition, to determine the specificity of the compound towards leukemia cells, we widened our cell line panel to incorporate non-leukemia cell lines, including seven solid tumor cell lines (neuroblastoma, breast cancer and lung carcinoma) and another non-malignant cell line, totaling 25 cell lines (KMT2A-r n=11, KMT2A-wt n=5, solid tumors n=7, non-malignant n=2) ( Supplementary Table 1 ). 2-Cl-ATP showed inhibitory activity, with IC₅₀ values below 10 µM, towards 9 out of 11 (82%) KMT2A-r leukemia cells with IC₅₀ values ranging from 2.5 to 5.9 µM ( Figure 2A and Supplementary Table 1 ). The compound also exerted significant inhibitory effects against both CALM-AF10 translocated leukemia cell lines when compared to other KMT2A-wt leukemia cell lines, the solid cancer cell lines and non-malignant cells ( Figure 2A and Supplementary Table 1 ). 2-Cl-ATP inhibited KMT2A-r leukemia cells by inducing apoptosis ( Supplementary Figure 1A ).

2-Cl-ATP is a phosphorylated derivative of the adenosine analog 2-chloroadenosine (2-CADO). Corroborating our observations, previous studies have reported sensitivity of KMT2A-r ALL cells to nucleoside analogs (17, 18). To further confirm that 2-Cl-ATP exerts a similar KMT2A-r selective action as other nucleoside analogs such as cladribine and that KMT2A-r leukemia cells are indeed susceptible to nucleoside analogs, we evaluated whether 2-Cl-ATP and 2-CADO had a similar cytotoxicity profile as cladribine in resazurin-based viability assays with a panel of KMT2A-r and KMT2A-wt leukemia cell lines ( Supplementary Figure 1B ). Indeed, a highly similar cell line selectivity was observed for all three drugs, with 2-Cl-ATP-resistant cell lines THP-1 and REH also being less responsive to 2-CADO and cladribine than the 2-Cl-ATP-sensitive KMT2A-r cells confirming our observed sensitivity of KMT2A-r leukemia cells to adenosine agonists.

The selectivity of SID7969543 towards KMT2A-r leukemia cells was also determined in the expanded cell line panel. Similar to 2-Cl-ATP, all cell lines that were sensitive towards SID7969543 (IC₅₀ < 10 µM) were leukemias with either a KMT2A or CALM-AF10 translocation ( Figure 2B and Supplementary Figure 2 ). SID7969543 decreased the viability of a subset (7/11) of KMT2A-r cells as well as both CALM-AF10 leukemia cells, with IC₅₀ values ranging from 1 µM to 5 µM, while not affecting the other KMT2A-wt leukemia cells, solid tumors or non-malignant cells ( Figure 2B and Supplementary Figure 2 ; Supplementary Table 1 ). Five of the seven sensitive KMT2A-r leukemia cell lines were derived from infants. No significant association was found between the occurrence of a certain KMT2A translocation and sensitivity to SID7969543. Both of the t(9;11), four out of six t(4;11) and one t(1;11) translocated KMT2A-r leukemia cell were sensitive to SID7969543, whereas the remaining two t(11;19) KMT2A-r leukemia cells did not display sensitivity ( Supplementary Table 1 ). No significant associations were observed between sensitivity to SID7969543 and the presence of a lymphoid (ALL) or myeloid (AML) leukemia phenotype ( Supplementary Figure 3 ). Remarkably, SID7969543 did not significantly impact the viability of peripheral blood mononuclear cells (PBMC) isolated from healthy donors at doses at which it significantly inhibited KMT2A-r leukemia cells ( Figure 2C ). This is in contrast to the inhibitory effects of 2-Cl-ATP against these PBMCs with IC₅₀ values closer to those achieved against KMT2A-r leukemia cells ( Figure 2D ).

To identify how SID7969543 limited the viability of KMT2A-r leukemia cells, the percentage of cells expressing surface annexin V was quantified to evaluate induction of apoptosis. SID7969543 significantly increased the percentage of annexin V-positive KMT2A-r cells as early as 6 hours post-exposure ( Figure 3A ). This was associated with increased cleavage of PARP and caspase-3 in sensitive KMT2A-r, but not in KMT2A-wt leukemia cells ( Figure 3B and Supplementary Figure 4 ), indicating that the compound rapidly kills sensitive KMT2A-r leukemia cells through caspase-dependent apoptosis.

To elucidate whether SID7969543 exerted its selective KMT2A-r leukemia cell killing effect through targeting and inhibiting its reported target, the transcription factor NR5A1, we firstly investigated whether the expression of NR5A1 was associated with the occurrence of a KMT2A rearrangement. NR5A1 expression was not significantly different between KMT2A-r and KMT2A-wt infant ALL patient samples ( Supplementary Figure 5A ) nor between KMT2A-r and KMT2A-wt leukemia cell lines ( Supplementary Figure 5B ), and the sensitivity of leukemia cell lines to SID7969543 did not correlate with expression levels of NR5A1 ( Supplementary Figure 5C ). We next assessed whether the drug impacted downstream NR5A1 signaling in treated KMT2A-r leukemia cells. Treatment with SID7969543 significantly decreased the expression of reported NR5A1 target genes involved in metabolism, enolase-1 (ENO1) and pyruvate kinase (PKM) (19) in sensitive KMT2A-r cell lines before the induction of apoptosis, while the expression of genes not regulated by NR5A1 (e.g. the KMT2A target genes HOXA9, HOXA10, MEIS1, and MYB) remained unchanged, demonstrating that SID7969543 hit its reported target ( Figure 3C and Supplementary Figures 6, 7 ). However, siRNA-mediated silencing of NR5A1 expression did not selectively impact the viability of KMT2A-r leukemia cells and thus did not mimic the action of SID7969543 on these cells ( Figures 3D–F and Supplementary Figures 8, 9 ). In addition, another structurally dissimilar inhibitor of NR5A1, the inverse agonist AC-45594, did not display selective cytotoxicity towards KMT2A-r leukemia cells ( Supplementary Figure 10 ) (20). Taken together it is therefore unlikely that the sole inhibition of NR5A1 is causing the observed effects of SID7969543 on KMT2A-r leukemia cell survival.

We next determined whether SID7969543 synergized with conventional chemotherapeutics used in the treatment of high-risk leukemia, namely, cytarabine, daunorubicin, etoposide, mitoxantrone, and topotecan ( Figure 4 and Supplementary Figure 11 ). Two KMT2A-r leukemia cell lines (PER-485, MOLM13) were treated with SID7969543, a conventional chemotherapeutic agent or their combination and potential synergistic effects were determined using the Bliss additivity model. For both cell lines, the viability curves of cells treated with the combination of SID7969543 and cytarabine or daunorubicin tracked below the predicted viability curves by Bliss (dotted lines), demonstrating synergy ( Figure 4 ). For etoposide, synergy with SID7969543 was observed in the PER-485 cells, while the combination was additive in MOLM13 cells ( Figures 4A, B ). The combination of SID7969543 with mitoxantrone or topotecan had additive effects in both cell lines ( Supplementary Figures 11A, B ). No synergy between SID7969543 and cytarabine, daunorubicin or etoposide was found in KMT2A-wt REH cells ( Figure 4C ).

Taken together, these data demonstrate that SID7969543 rapidly kills a subset of KMT2A-r leukemia cells, particularly those arising in infants, via the induction of apoptosis. In addition, the compound potentiates several standard-of-care chemotherapies used for treatment of high-risk leukemia. However, microsomal stability assays showed that SID7969543 has a short half-life and thus limited in vivo stability ( Supplementary Figure 12 ), precluding further testing in animal models in its current form.