The multi-kinase inhibitor CG-806 exerts anti-cancer activity against acute myeloid leukemia by co-targeting FLT3, BTK, and Aurora kinases

Background: Despite the development of several FLT3 inhibitors that have improved outcomes in patients with FLT3-mutant acute myeloid leukemias (AML), drug resistance is frequently observed, which may be associated with the activation of additional pro-survival pathways such as those regulated by BTK, aurora kinases, and potentially others in addition to acquired tyrosine kinase domains (TKD) mutations of FLT3 gene. FLT3may not always be a driver mutation. Objective: To evaluate the anti-leukemia efficacy of the novel multi-kinase inhibitor CG-806, which targets FLT3 and other kinases, in order to circumvent drug resistance and target FLT3 wild-type (WT) cells. Methods: The anti-leukemia activity of CG-806 was investigated by measuring apoptosis induction and analyzing cell cycle with flow cytometry in vitro, and its anti-leukemia Results: CG-806 demonstrated superior anti-leukemia efficacy compared to commercially available FLT3 inhibitors, both in vitro and in vivo, regardless of FLT3 mutational status. The mechanism of action of CG-806 may involve its broad inhibitory profile of FLT3, BTK, and aurora kinases. InFLT3 mutant cells, CG-806 induced G1 phase blockage, while in FLT3WT cells, it resulted in G2/M arrest. Targeting FLT3 and Bcl-2 and/or Mcl-1 simultaneously resulted in a synergistic pro-apoptotic effect in FLT3mutant leukemia cells. Conclusion: The results of this study suggest that CG-806 is a promising multi-kinase inhibitor with anti-leukemia efficacy, regardless of FLT3 mutational status. A phase 1 clinical trial of CG-806 for the treatment of AML has been initiated (NCT04477291).


Introduction
Fms-like tyrosine kinase 3 (FLT3)-internal tandem duplication (ITD) mutations occur in approximately 30% of adult acute myeloid leukemia (AML) patients and are associated with poor clinical outcomes (1), particularly in patients with higher mutant-to-wild type (WT) allelic ratios (2). Several putative FLT3 inhibitors (FLT3is) such as sorafenib, midostaurin, quizartinib, crenolanib, and gilteritinib have been investigated alone and in combination with chemotherapeutic drugs or other targeted agents in preclinically and clinical trials (3,4). However, as monotherapies, these inhibitors tend to have limited effectiveness and resistance to the drugs often develops quickly (4)(5)(6)(7)(8)(9). The acquisition of secondary mutations in the tyrosine kinase domain (TKD) of FLT3 is one mechanism of such resistance (6, 10), and has been identi ed in AML patients who relapsed or developed resistance to type-2 FLT3is, such as sorafenib and quizartinib (6, 11).
Moreover, aberrant activation of a number of other receptor and non-receptor tyrosine kinases has been linked to resistance to FLT3 inhibitors, including mitogen-activated protein kinases (MAPKs), signal transducer and activator of transcription 5 (STAT5), cellular myelocytomatosis oncogene (c-Myc) protein and Bruton's tyrosine kinase (BTK) (12)(13)(14)(15). Over-expression of c-Myc has been observed in most human hematopoietic malignancies and is associated with poor prognosis (16). The phosphorylation of c-Myc is regulated by the FLT3 downstream MAPK and PI3K/AKT signaling pathways (1). BTK is also part of the FLT3-ITD signalosome and is activated in an FLT3-ITD-dependent manner to induce proliferation in AML cells (17). In fact, targeting BTK with ibrutinib shows anti-proliferative effects in AML by mediating suppression of FLT3 downstream signaling MAPK, AKT, and NF-κB; and combined inhibition of FLT3 and BTK reportedly has additive anti-leukemia effects (17,18).
Overexpression of aurora kinases (AURK), a family of serine/threonine kinases, has also been consistently demonstrated in a variety of leukemia cell lines and primary AML samples (19). The aberrant expression of AURK may be associated with poor-risk cytogenetic abnormalities and high blood cell counts in patients with AML (20). It has been reported that these CD34+/CD38-cells aberrantly expressed elevated levels of AURK that are required for cell cycle progression (21,22), and targeting AURK demonstrated in vitro e cacy against human Myc-overexpressing AML cells (23). Concomitantly targeting AURK and FLT3 exhibited potent cytotoxicity with lower half-maximal inhibitory concentrations (IC 50 s) values against FLT3-ITD-mutant MV4-11, MOLM13 and MOLM13-resistant AML cells. The latter harbor ITD and D835Y dual mutations (24,25). Interestingly, Druker's group recently identi ed aurora kinase B as an early resistance factor to FLT3 inhibition (26). Therefore, co-targeting FLT3/AURK may be another potential therapeutic strategy in AML treatment. Furthermore, a therapeutic kinase inhibitor directed against all three novel targets, FLT3, BTK, and AURK, may be able to achieve much enhanced anti-leukemia e cacy in the targeted therapy of AML. Furthermore, FLT3 mutations may be acquired late in leukemogenesis and therefore may not be critical driver mutations. Activity against non-FLT3 mutant AML will then present a distinct advantage by targeting several subclones, FLT3 mutant and non-mutant.
The small molecule drug CG-806 (luxeptinib; Aptose, San Diego, CA) is a multi-kinase inhibitor with high activity against FLT3 mutations (27). Here, we evaluated the anti-leukemia activity of CG-806 by investigating its anti-proliferation and survival effects in AML in vitro as well as in AML xenograft models. We found that CG-806 has superior anti-leukemia effects against both FLT3 WT and mutant AML, especially in AML harboring FLT3 and TKD point mutations, which is a potential mechanism of secondary resistance to FLT3i. CG-806 profoundly inhibited phosphorylated (p)-BTK and p-AURK in addition to its FLT3 inhibition in vitro and signi cantly extended survival in vivo AML murine models. Our ndings suggest that co-targeting FLT3, BTK, and AURK with a multi-kinase inhibitor CG-806 may be potent against AML regardless of FLT3 mutational status. c-Myc knockdown with siRNA transfection The indicated siRNAs and mock control (scramble) siRNAs were purchased from Dharmacon Research, Inc. (Lafayette, CO). Transfections of MOLM14 leukemia cells were carried out by electroporation using the Amaxa Nucleofector system (solution V, program O-017; Lonza, Basel, Switzerland) following the manufacturer's instructions. The nal concentration of siRNA was 300 nM. After 24 h of transfection, the indicated concentrations of CG-806 were added to the cells for an additional 24 h of culture.
Animal studies An AML model was established by intravenously xenografting GFP-tagged Ba/F3-FLT3-ITD cells (0.5 x 10 6 cells/mouse), infected with lentivirus expressing re y luciferase, into NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ mice (eight-week-old female) (The Jackson Laboratory, Bar Harbor, ME). The mice were treated with 10 mg/kg or 100 mg/kg CG-806 (oral gavage daily, 10 mice/group) starting on day 4 after leukemia cell injection, when an unambiguous luciferase signal was recorded. Control mice (n=10) received vehicle once daily for 5 weeks. Mice were noninvasively imaged using an Xenogen-200 in vivo bioluminescence imaging system (Xenogen, Hopkinton, MA) after being injected with 4 mg of luciferin substrate (D-luciferin, GoldBoi, St. Louis, MO). Bioluminescence images were obtained and quantitated as described previously (4). Three mice per group were humanely killed on day 11 after leukemia cell injection, and peripheral blood (PB) and bone marrow (BM) samples were collected to assess leukemia cell engraftment by measuring GFP positivity with ow cytometry. The in vivo studies were performed under approved animal care protocol and the standards of the Association for Assessment and Accreditation of Laboratory Animal Care.

Statistical analyses
The Student t-test was used to analyze immunoblotting and cell apoptosis data. P-values ≤ 0.05 were considered statistically signi cant. All statistical tests were two-sided, and the results are expressed as the means of triplicate samples/experiments ± standard deviations or as means with 95% con dence intervals. Survival was estimated using the Kaplan-Meier method (29), and log-rank statistics was used to assess differences in survival between groups.

Results
CG-806 exhibits anti-leukemia activity superior to other FLT3i in samples with FLT3 WT or TKD mutations through the inhibition of FLT3/AURK/BTK Our initial studies indicate that CG-806 has impressive kinase inhibition against WT and mutant FLT3, BTK, and AURK as well at extremely low nanomolar IC 50 s in a cell-free biochemical kinase inhibition assay (Table S2). Therefore, we rst evaluated the anti-leukemia activity of CG-806 in AML cell lines with different FLT3 mutation status. Treatment with extremely low doses (nanomolar to sub-nanomolar) of CG-806 for 72 h profoundly inhibited cell growth via apoptosis induction in both human and murine leukemia cell lines harboring either FLT3-ITD mutations or FLT3-ITD+TKD point mutations (Fig. 1A, B). Cells harboring FLT3 TKD mutations or ITD + TKD dual mutations usually show resistance to most currently available FLT3i in previous studies (6, 10,30). The IC 50 s and EC 50 s of CG-806 against these leukemia cell lines were in the low nanomolar to sub-nanomolar range (Table 1). Interestingly, whereas CG-806 had low IC 50 s (i.e., 4 to 10 nM) in most human and murine FLT3-WT leukemia cell lines, its EC 50 s could not be determined in some human FLT3-WT leukemia cell lines, such as THP-1 and Kasumi-1 (Table 1). Most importantly, CG-806 had profound pro-apoptotic effects in primary AML patient samples irrespective of FLT3 mutation status, but did not induce apoptosis in BM cells from healthy donors (Fig.  1C). This suggested that CG-806 has broad and potent anti-cancer activity against AML cells in addition to a potential therapeutic window with respect to toxicity to normal cells.
To evaluate the anti-leukemia potency of CG-806, we compared the cytotoxic effects of the drug with other currently approved/available FLT3 and multi-kinase inhibitors in FLT3-mutated and FLT3-WT AML cell lines and patient samples. The IC 50 s of CG-806 were much lower than those of other FLT3is particularly in leukemia cells harboring the "gatekeeper" F691 mutation. The IC 50 s were 10.0 nM for CG-806 but 115.3 nM, 98.4 nM, and 257.6 nM for quizartinib, gilteritinib, and crenolenib, respectively ( Table  2). We further compared the apoptogenic effect of CG-806 with that of the other FLT3i in AML patient samples ex vivo. CG-806 demonstrated markedly greater cytotoxicity than quizartinib in primary peripheral blood mononuclear cells with FLT3-ITD mutations or with FLT3-ITD+TKD mutations (Fig 1D,  E).
Immunoblot analyses demonstrated that CG-806 at nanomolar concentrations markedly suppressed phosphorylation levels of FLT3, AURK, and BTK, and their downstream signaling partners p-AKT and p-ERK in the leukemia cell lines and primary AML samples harboring FLT3-ITD mutations and/or FLT3-TKD mutations ( Fig. 2A, B). CG-806 also upregulated the pro-apoptotic protein Bim in FLT3 WT and ITD mutant AML cell lines after 24 h of treatment, and later triggered the cleavage of caspase-3 and PARP in FLT3-ITD-mutated AML cells (Fig. 2C, Supplementary Fig. S2). However, the treatment only marginally affected Bcl-2 and Bcl-xL levels in the leukemia cells, and even upregulated the anti-apoptotic protein Mcl-1 especially in Ba/F3-ITD mutant and Ba/F3-FLT3-WT cell lines (Fig. 2C).

CG-806 blocks leukemia cells in G1 phase in FLT3-ITD-mutated AML cells and triggers G2/M arrest in FLT3-WT AML cells
To further characterize the mechanism(s) underlying the anti-leukemia activity of CG-806, we investigated the impact of CG-806 on cell cycle progression. Results indicated that CG-806 blocked cells in G1 phase in FLT3-ITD-mutated MOLM14 and MV4-11 leukemia cell lines after 24 h of treatment as determined by BrdU incorporation assay (Fig. 3A, B). Immunoblotting analyses showed profound suppression of cell proliferation-related proteins p-mTOR, -S6K, and -RB, upregulation of p27, and reduction of G1 phase checkpoint proteins CDK4, CDK6 and c-Myc as well (Fig. 3C). In terms of cell proliferation, c-Myc has key abilities to control cell cycle progression by promoting transcription of its downstream genes for cell cycle transition from G0/G1 into S phase and antagonizing cell cycle inhibitor activity (31). Therefore, we further determined if c-Myc is critical for CG-806-induced G1 arrest. Knocking down c-Myc with siRNA in MOLM14 cells (Supplementary Fig. S3) triggered more pronounced G1 phase arrest compared to MOLM14 cells without c-Myc knockdown (50.9% vs. 33. 6% in MOLM14-cMyc-siRNA vs. MOLM14-cMycscramble cells, respectively, p < 0.05) (Fig. 3D), implying that c-Myc suppression has a role in CG-806induced inhibition of cell growth through G1 phase arrest in FLT3-mutated leukemia cells.
However, we did not observe G1 arrest in FLT3-WT cells. Conversely, CG-806 inhibited the growth of FLT3-WT THP-1 and OCI/AML3 cells by triggering signi cant G2/M arrest instead (Fig. 4A). Immunoblotting analyses demonstrated that CG-806 profoundly suppressed p-AURK B and C levels and downregulated Polo-like kinase 1 (PLK1), p-CDC25c, and cyclin B1 (Fig. 4B). To con rm that AURK inhibition was associated with G2/M arrest in FLT3-WT cells, we suppressed AURK activity by using an AURK speci c inhibitor SNS-314 (32) in either FLT3-WT or -mutant AML cell lines THP-1 or MOLM14. Results showed a similar G2/M arrest accompanied by p-AURK inhibition in both cell lines (Fig. 4C, D), which suggests that AURK suppression and G2/M arrest are interconnected regardless of FLT3 mutation status in AML cells.
CG-806 has marked anti-leukemia e cacy in murine models of FLT3-mutated leukemia Our preliminary data indicated that mice receiving 100 mg/kg CG-806 had high plasma concentrations 24 h after one dose ( Supplementary Fig. S4). Therefore, we established a leukemia model in NSG mice by xenografting Baf3-FLT3-ITD cells and treated mice with 10 or 100 mg/kg doses of CG-806. CG-806 signi cantly reduced the leukemia burden by 48% (10 mg/kg dose; p < 0.05) and 93% (100 mg/kg dose; p < 0.001) compared to the vehicle group (Fig. 5A, B) and eliminated leukemia-related splenomegaly after 1 week of drug administration (Fig. 5C). CG-806 also eliminated leukemic blasts in both PB and BM in a dose-dependent manner (Fig. 5D, E). In addition, the survival duration of the 10 mg/kg (16 d) and 100 mg/kg CG-806 groups (24 d) was signi cantly longer than that observed in the vehicle group (11 d; p < 0.01) (Fig. 5F). CG-806 at either dose did not affect mouse body weight (data not shown).
Immunoblot analyses showed that targeting Bcl-2 and/or Mcl-1 concomitantly with CG-806 profoundly suppressed Mcl-1, reduced p-FLT3, -BTK, and -AURK, and triggered a marked cleavage of caspase-3 (Fig.  6C), suggesting that the combination regimens trigger potent leukemia cell killing which may translate in bene cial effect in relapsed/refractory AML regardless of FLT3 mutational status.

Discussion
Several FLT3i have been developed over the last two decades including sorafenib (4,5) and FDAapproved gilteritinib and midostaurin (8, 36). A newer generation FLT3i, crenolanib is still under development and showed impressive anti-leukemia effects against resistant AML harboring FLT3 TKD mutations (7). One complicating factor is the oligoclonal nature of AML (37,38). FLT3 mutant clones coexist with FLT3 WT clones or dysregulated signaling, which creates a much more complex scenario. Meanwhile, it also provides a window into therapeutic vulnerabilities. Aberrant expression of survival signaling pathways such as MAPK, BTK and AURK is associated with resistance to FLT3i (12,15,19). CG-806 as a novel multi-kinase inhibitor of targeting FLT3/BTK/AURK might provide a better pharmacological notion for overcoming resistance than that more speci c FLT3i. In the present study, we demonstrated that CG-806 has a superior anti-leukemia e cacy especially against AML harboring "gatekeeper" F691 mutations or FLT3 WT compared to other FLT3i, without detectable toxicity in normal BM samples. Mechanistically, CG-806 profoundly suppresses FLT3, BTK, and AURK activation simultaneously and results in impressive cytotoxicity in these AML cells, particularly against FLT3-WT AML cells which demonstrated abnormally high expression of aurora kinase. Our ndings con rmed that CG-806 is a pan-FLT3 inhibitor that may bene t from the simultaneous suppression of FLT3, BTK and AURK activation, leading to a promising anti-leukemia effect against AML regardless of their FLT3 status.
Further, we found that CG-806 achieved its anti-leukemia activity in FLT3-mutated AML and FLT3-WT AML through different mechanisms. Speci cally, CG-806 induced G1 arrest in FLT3-mutated MOLM14 and MV4-11 cells but induced G2/M arrest in FLT3-WT THP-1 cells at IC 50 s of about 1 nM and 5 nM, respectively. In FLT3-mutated cells, G1 cell cycle progression is closely associated with the dominant activation of FLT3 and its downstream effectors, AKT/mTOR/S6K, and MAPK. These signaling cascade proteins are highly expressed and constitutively activated in FLT3-mutated leukemia cells. In addition, BTK, which is expressed in about 80% of human AML, mediates FLT3-ITD-dependent Myc and STAT5 activation (17), and transcriptionally increases the levels of G1 cell cycle checkpoint proteins through c-Myc signaling (39). Thus, by co-targeting FLT3 and BTK, which are dominantly activated in FLT3-mutated AML, CG-806 enhances the downregulation of c-Myc to trigger G1 arrest. In fact, Eriksson et al. reported that targeting FLT3 signaling with the FLT3i AKN-028 or midostaurin for 12-48 h triggered G1 phase arrest in FLT3-mutated MV4-11 AML cells as well (40). In the present study, targeting FLT3 with quizartinib or gilteritinib or targeting BTK with ibrutinib also triggered G1 arrest in FLT3-ITD-mutated AML cells (data not shown). Similarly, c-Myc knockdown elicited G1 arrest in FLT3-mutated MOLM14 cells as well. These ndings strongly imply that the CG-806-induced suppression of FLT3 and BTK and their downstream signaling, and the subsequent downregulation of c-Myc play critical roles in G1 arrest in FLT3-mutated AML cells.
We did not observe c-Myc repression in FLT3-WT AML cells treated with CG-806 although it has a high expression in the FLT3-WT cells. In fact, FLT3-WT cells had higher expression of p-AURK, but lower p-FLT3 or -BTK compared to FLT3-mutated AML cells (Supplementary Fig. S6). An investigation of genes essential for proliferation and survival of cancer cells with CERES (computational method to estimate gene-dependency levels from CRISPR-Cas9 essentiality screens) dependency score (lower score indicates a higher likelihood that the gene of interest is essential in a given cell line) indicated that FLT3-WT leukemia cell lines have low CERES scores of AURK-B (-2.0 and -1.7, respectively, on Kasumi-1 and THP-1 cell lines) and of AURK-A (-1.45 and -1.37, respectively, in Kasumi-1 and OCI/AML3 cell lines); which suggests a higher dependency on AURK with regard to proliferation and survival of FLT3-WT leukemia cell lines (41). Actually, the overexpression of AURK and their downstream PLKs were reported to be associated with tumorigenesis of many human tumors, including leukemias (21,42). AURK signaling plays a fundamental role in regulating cell cycle checkpoints that ensure the timing and order of cell cycle events such as DNA repair, bipolar spindle formation, chromosome segregation and mitotic exit (43)(44)(45). In line with our in vitro results, CG-806 had robust anti-leukemia activity in a murine xenograft model of leukemia created using the aggressive FLT3-ITD-mutated Ba/F3 cells (Fig. 5), and in a PDX AML model as well (50). Compared with vehicle, 100 mg/kg CG-806 signi cantly reduced leukemia burden and the absolute leukemia cell count in PB by up to 17-fold after 1 wk of treatment. CG-806 also conveyed a remarkable survival bene t, as the survival duration of mice that received 10 mg/kg or 100 mg/kg CG-806 (16 and 24 d, respectively) were markedly longer than that of mice that received vehicle (11 d).
Impressively, even at doses of up to 450 mg/kg, the daily oral administration of CG-806 for 14 d elicited no obvious toxicity in Balb/c mice (data not shown). These results suggest that CG-806 could be well-tolerated in a clinical setting as monotherapy or as a combinatorial drug with other targeted agents or chemotherapeutics for AML treatment.
We further observed that combinations of CG-806 with the Bcl-2 antagonist venetoclax and/or the Mcl-1 inhibitor A1210477 had synergistic pro-apoptotic effects not only in FLT3-mutated AML cells (both those with FLT3-ITD mutations and those with FLT3-ITD + "gatekeeper" F691 mutations), but also in FLT3-WT cells, indicating that the combinatorial regimens may have a promising anti-leukemia e cacy in both FLT3-WT and mutant AML. In fact, combinations of venetoclax with various other drugs, including FLT3i, are being actively investigated in AML (35,51,52). The anti-apoptotic protein Mcl-1 is one of the main determinants of venetoclax resistance in AML (53).       (A) FLT3-WT THP-1 and OCI-AML3 cells were exposed to the indicated concentrations of CG-806 for 24 h, and PI staining was used to measure the cell cycle (DNA content) distribution. Data are presented as the