Targeting of SFK and G6PD Contributes to Anti-Leukemic Activity of Novel FLT3 Trisubstituted Purine Inhibitors: Effective Therapeutic Approach for AML with FLT3 Mutations

We have recently developed compound LGR3922, a potent inhibitor of FLT3 receptor tyrosine kinase, which displays selectivity and potency in in vitro and in vivo AML models (compound 7d, Gucky T et al, J Med Chem. 2018;61(9):3855). In addition, LGR3922 showed strong inhibitory activity against Src family kinases (SFK) in the in vitro settings (Fig. 1A; ASH meeting abstract, Radosova Muchova et al, Blood. 2016;128(22):1569). Here, we present additional characterization of the lead compound LGR3922 and demonstrate its potent antileukemic activity in MLL/FLT3-ITD cell lines, murine xenografts and ex vivo primary AML samples. LGR3922 exhibited strong anti-proliferative effect against MV4-11 and MOLM-13 cells. Contrary to quizartinib (quiz), LGR3922 simultaneously reduced phosphorylation levels of FLT3-SFK and their downstream molecules STAT5 and ERK in both AML cell lines. Interestingly, the inhibitory effect of LGR3922 on SFK, STAT5 and ERK increased over time, especially in the MOLM-13 cells. Similar results were only seen for the combination treatment with quiz and dasatinib (das). Dual-targeting of FLT3-SFK by LGR3922 was further confirmed on MV4-11 xenografts. In the long-term in vivo experiment with repeated doses of the tested inhibitors, LGR3922 showed similar trends in initial treatment response as the combination treatment with quiz and das; quiz exhibited less efficient inhibitory effects at this stage. The tumor reduction was most significant (p ≤ 0.0001) in LGR3922 group; 3 out of 6 tumors were completely eradicated during the treatment period (Fig. 1B). Complete tumor eradication (in 1 out of 5 animals) over the treatment period was then seen only in combination quiz and das (Fig. 1B). No adverse effects on blood count parameters or body weight were associated with LGR3922 compared to combination treatment with quiz and das, where temporary body weight loss and only limited weight gain were observed. Next, we searched for additional target of the FLT3/SFK inhibitor that would have a synthetic lethal effect for the kinase pair. Besides the kinases that belong to other druggable nodes of AML-associated networks (Fig. 1A), we also tested antioxidant defense enzymes, previously validated as being synthetic lethal with FLT3 inhibitor therapy (Gregory MA et al. Proc Natl Acad Sci U S A. 2016;113(43):E6669). Indeed, we observed that LGR3922 inhibited G6PD activity, already after one-hour exposure (with a dose response over the concentration range), in contrast to quiz and das. We hypothesize that this activity modulating leukemia cell redox metabolism, cooperating with FLT3/SFK inhibition, may elicit more effective eradication of AML cells than quiz and das or quiz/das combination. We then evaluated toxicity of LGR3922 on normal hematopoiesis using colony forming unit (CFU) assays with progenitor cells from several healthy donor controls. LGR3922 inhibited colony formation in a dose-dependent manner; maximal tolerated concentration (MTC) for all healthy samples was 100 nM, in some samples the reduced granulocyte-macrophage (CFU-GM) and burst forming unit-erythroid (BFU-E) colony forming ability was not observed up to a concentration of 200 nM. Quiz and das exposure showed MTCs at 20 nM and 5 nM, respectively; in quiz/das combination at 10 nM/0.5 nM; 10 nM/5 nM was already toxic. Then, we performed CFU assays on 4 primary human AML samples with FLT3-ITD mutations (n=3) or FLT3-TKD+KMT2A::MLLT3 (n=1). Cultures with LGR3922 at 100 nM reduced the number of colonies by ∼70% in all FLT3-ITD patient samples. The inhibitory effects were comparable to or a little more potent than MTCs of quiz or quiz/das combination. Das culture had significantly lower effect on colony formation for the tested concentrations. In the FLT3-TKD+KMT2A::MLLT3 positive sample, fusion transcript was quantified in harvested colonies. While LGR3922 significantly reduced KMT2A::MLLT3 fusion transcript, no effects were observed for MTCs of single agent cultures with quiz or das. Only a combination of quiz/das exceeding MTC (10 nM/5 nM) resulted in a slight decrease of KMT2A::MLLT3 fusion transcript levels. Taken together, these data suggest effectiveness of our new LGR3922 compound and provide a rationale for further development and testing of this molecule for MLL/FLT3-ITD AMLs, and, potentially, other types of AMLs. Grant support: Program EXCELES, ID Project No. LX22NPO5102; GA19-09086S; MH CZ-DRO (UHKT, 00023736). Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal