Synthesis and Structural Optimization of 2,7,9-Trisubstituted purin-8-ones as FLT3-ITD Inhibitors

Therapy of FLT3-positive acute myeloid leukemia still remains complicated, despite the availability of newly approved kinase inhibitors. Various strategies to avoid the reduced efficacy of therapy have been explored, including the development of dual targeting compounds, which inhibit FLT3 and another kinase necessary for the survival and proliferation of AML cells. We have designed new 2,7,9-trisubstituted 8-oxopurines as FLT3 inhibitors and report here the structure-activity relationship studies. We demonstrated that substituents at positions 7 and 9 modulate activity between CDK4 and FLT3 kinase, and the isopropyl group at position 7 substantially increased the selectivity toward FLT3 kinase, which led to the discovery of compound 15a (9-cyclopentyl-7-isopropyl-2-((4-(piperazin-1-yl)phenyl)amino)-7,9-dihydro-8H-purin-8-one). Cellular analyses in MV4-11 cells revealed inhibition of autophosphorylation of FLT3 kinase in nanomolar doses, including the suppression of downstream STAT5 and ERK1/2 phosphorylation. We also describe mechanistic studies in cell lines and activity in a mouse xenograft model in vivo.


Introduction
Acute myeloid leukemia (AML) is the most common type of acute leukemia in adults who have unfortunately poor prognosis and low survival rates. Approximately 30% of patients bear mutations in FMS-like tyrosine kinase 3 (FLT3), a transmembrane receptor kinase, which plays a key role in normal hematopoietic cell maturation, but its oncogenic activation results in the uncontrolled expansion of myeloid precursors. FLT3 can be activated by the internal tandem duplications (ITD) of the juxtamembrane domain or by point mutations in the kinase domain [1]. FLT3 mutation status relates to a prognosis, but it has also been validated as a drug target in AML and hematooncologists can now treat patients with newly approved drugs [2], including those based on small molecule kinase inhibitors, such as sorafenib, midostaurin, gilteritinib, crenolanib, or quizartinib ( Figure 1) [1,3]. eral FLT3/CDK4 dual inhibitors (including clinical candidate AMG925) have been reported [17][18][19][20]. These compounds usually retain carbonyl function, which serves as a key hydrogen bond acceptor from D158 in CDK4. Carbonyl function in a corresponding position is also present in pteridin-7(8H)-one FLT3 inhibitor ( Figure 1) [18]. In the current work, we have designed analogous 2,7,9-trisubstituted 8-oxopurines as FLT3 inhibitors and report here the structure-activity relationship (SAR) studies, which led to the discovery of compound 15a. We also describe mechanistic studies in cell lines and activity in a mouse xenograft model in vivo.

Chemistry
The synthetic route to various purines 7-10 is described in Scheme 1. Reductive amination was used as a convenient method for the synthesis of derivatives 2 and 3 from commercially available pyrimidine 1. Thus, we changed the typical reaction sequence to N-methyl pyrimidine 2, where alkylation usually precedes to chlorination step [21]. In the case of 3, we optimized reaction conditions using 2,2-dimethoxypropane and sodium triacetoxyborohydride [22].
The further diverse step was the aromatic substitution of chloride at position 4 with various amines using N,N-diisopropylethylamine in butanol. Amines were subjected under our modified conditions [23], but some aromatic amines were more problematic. Despite available targeted drugs, the therapy for FLT3-positive AML remains complicated due to the high rate of resistance to FLT3 inhibitors, often caused by changes in the kinase domain [4]. Various strategies to avoid reduced efficacy of therapy are explored, including combinations with conventional cytotoxics [2] or modern targeted drugs, such as BH3 mimetics [5,6], azacytidine [7], or kinase inhibitors [8][9][10]. Other approaches focus directly on the kinase realm by generating kinase inhibitors targeting resistance-causing FLT3 point mutations [11], or by developing dual targeting compounds, which inhibit FLT3 and another kinase necessary for survival and proliferation of AML cells, for example, JAK2, MEK, MER, or CDK4 [12].
One of these additional targets upregulated in AML is CDK4 [13], which is a key cell cycle regulating enzyme functioning downstream of FLT3. AML cells are sensitive to CDK4/6 inhibition [14,15]. In addition, the approved CDK4 inhibitor, trilaciclib, (Figure 1) also displays low nanomolar affinity to oncogenic variants of FLT3 [16]. In addition, several FLT3/CDK4 dual inhibitors (including clinical candidate AMG925) have been reported [17][18][19][20]. These compounds usually retain carbonyl function, which serves as a key hydrogen bond acceptor from D158 in CDK4. Carbonyl function in a corresponding position is also present in pteridin-7(8H)-one FLT3 inhibitor ( Figure 1) [18]. In the current work, we have designed analogous 2,7,9-trisubstituted 8-oxopurines as FLT3 inhibitors and report here the structure-activity relationship (SAR) studies, which led to the discovery of compound 15a. We also describe mechanistic studies in cell lines and activity in a mouse xenograft model in vivo.

Chemistry
The synthetic route to various purines 7-10 is described in Scheme 1. Reductive amination was used as a convenient method for the synthesis of derivatives 2 and 3 from commercially available pyrimidine 1. Thus, we changed the typical reaction sequence to N-methyl pyrimidine 2, where alkylation usually precedes to chlorination step [21]. In the case of 3, we optimized reaction conditions using 2,2-dimethoxypropane and sodium triacetoxyborohydride [22]. aromatic nucleophilic substitution compared to an N-isopropyl one (3) or non-substituted amine 1. In these cases, a longer reaction time or the addition of a base was needed.
The further diverse step was the aromatic substitution of chloride at position 4 with various amines using N,N-diisopropylethylamine in butanol. Amines were subjected under our modified conditions [23], but some aromatic amines were more problematic. Noteworthy, only the desired position was involved in the substitution and no side products or hydrolysis were observed. N-methyl intermediate 2 was the most reactive toward aromatic nucleophilic substitution compared to an N-isopropyl one (3) or non-substituted amine 1. In these cases, a longer reaction time or the addition of a base was needed.
Cyclization to purines 7 proceeded with trifluoroacetic anhydride and purinones 8-10 were isolated after reaction with a phosgene solution. Several aromatic and one styryl substituents were introduced using the Chan-Lam cross-coupling reaction using conditions from the literature [24], since the conventional method of aromatic nucleophilic substitution failed. 2-Methoxy, 4-methoxyphenyl, 2-tolyl, naphthyl, and styryl boronic acid were used as coupling partners with purinone intermediates 9 with the NH group at position 9. This intermediate was prepared easily after the deprotection of 2,4-dimethoxybenzyl group with TFA.
The Buchwald-Hartwig amination was used with piperazineaniline 11 to substitute the chlorine atom at position 2 (Scheme 2) since no general and high-yielded method for conventional aromatic nucleophilic substitution was found. The reaction was efficiently performed with precatalyst XPhos Pd G2 under microwave irradiation [25][26][27]. Only derivatives 13 with R 1 = H were more problematic. Prolongation of the reaction time from 1-2 to 6-9 h and an additional amount of the precatalyst solved the problem; however, deprotected purine 9 (R 2 = H) was still resistant to the coupling reaction. Finally, the Boc protecting group was removed to give a library of purines 12-15; only a cleavage of the 2,4-DMB group required harsh conditions to get purine 14a. Int

Kinase Inhibitory Activities-The SAR Study
Newly prepared compounds 12-15 were assayed for inhibition of FLT3-ITD kinase as well as CDK4 and CDK2, known off-targets of structurally related compounds. We observed that CDK4 was sensitive to nearly all compounds with IC50 values in mid-nanomolar values. Slight improvement in the anti-kinase potencies was observed with the increasing size of cycloalkyl moieties at position 9 (R2 substituent), namely from cyclopentyl

Kinase Inhibitory Activities-The SAR Study
Newly prepared compounds 12-15 were assayed for inhibition of FLT3-ITD kinase as well as CDK4 and CDK2, known off-targets of structurally related compounds. We observed that CDK4 was sensitive to nearly all compounds with IC 50 values in mid-nanomolar values. Slight improvement in the anti-kinase potencies was observed with the increasing size of cycloalkyl moieties at position 9 (R2 substituent), namely from cyclopentyl to cycloheptyl. The substitution of the oxo for trifluoromethyl group at position 8 (derivatives 12a-12d) dramatically improved inhibitory activities to all tested kinases to nanomolar values with slight preferences for CDK4 and FLT3. Compounds 13a-13d having no substitution at position 7 of the purine ring did not display potency to the tested kinases.
Finally, we explored the activities of four compounds that possess an isopropyl substituent at position 7 and cyclopentyl, cyclohexyl, 4-methylcyclohexyl, or cycloheptyl groups (15a-15d) at position 9 of the purine ring. All these derivatives displayed interesting activities for FLT3 kinase (IC 50 = 30-70 nM) and selectivity over tested CDKs. Comparison with counterparts differing in the methyl and isopropyl moieties (pairs 15d and 14h, 15a and 14d, 15c and 14g, and 15b and 14e) indicates that isopropyl markedly contributes to selectivity against FLT3.

Kinase Selectivity
Three representative compounds (14d, 14e, and 15a) were profiled in a 1 µM concentration against 30 kinases that were selected as known off-targets of approved CDK4/6 inhibitors [28]. Compounds 14d and 14e, differing only in the cyclopentyl and cyclohexyl moiety at position 9 of the purine ring, were confirmed to be potent CDK4 and FLT3 inhibitors ( Figure S1). Additional sensitive kinases include STK16, EphA1, MEKK2, SRC, and MER (≤10% of residual enzyme activities), which are not targets of the structurally similar CDK4/6 inhibitor, trilaciclib. We also screened compound 15a, an analogue of 14d, bearing isopropyl at position 7 of the purine instead of the methyl group. This change rapidly diminished the potency against CDK4 (dropped to 90% of residual enzyme activity), but the preference for FLT3 remained (1st rank) ( Figure 2). Further, four kinases, namely Src, TTK, Lyn, and Abl were also potently inhibited (≤10% of residual activities). moiety at position 9 of the purine ring, were confirmed to be potent CDK4 and FLT3 inhibitors ( Figure S1). Additional sensitive kinases include STK16, EphA1, MEKK2, SRC, and MER (≤10% of residual enzyme activities), which are not targets of the structurally similar CDK4/6 inhibitor, trilaciclib. We also screened compound 15a, an analogue of 14d, bearing isopropyl at position 7 of the purine instead of the methyl group. This change rapidly diminished the potency against CDK4 (dropped to 90% of residual enzyme activity), but the preference for FLT3 remained (1st rank) ( Figure 2). Further, four kinases, namely Src, TTK, Lyn, and Abl were also potently inhibited (≤10% of residual activities).

Binding of 15a in the FLT3 Active Site
Molecular docking of 15a into the FLT3 in active conformation (DFG-in) [29] proposed plausible binding in the active site in a manner corresponding to known inhibitors of CDK4, CDK6, and FLT3 [30,31] (Figure 3). The orientation of the compound is consistent with type I binding. In the hinge region, two putative conserved hydrogen bonds are formed between the backbone of Cys694 and 15a. The isopropyl in position 7 of purine core can make extensive hydrophobic interactions with Val624, Lys644, Val675, Leu 767, and the gatekeeper, Phe691. The cyclopentyl ring forms hydrophobic contacts with Asp698 and Leu767.

Binding of 15a in the FLT3 Active Site
Molecular docking of 15a into the FLT3 in active conformation (DFG-in) [29] proposed plausible binding in the active site in a manner corresponding to known inhibitors of CDK4, CDK6, and FLT3 [30,31] (Figure 3). The orientation of the compound is consistent with type I binding. In the hinge region, two putative conserved hydrogen bonds are formed between the backbone of Cys694 and 15a. The isopropyl in position 7 of purine core can make extensive hydrophobic interactions with Val624, Lys644, Val675, Leu 767, and the gatekeeper, Phe691. The cyclopentyl ring forms hydrophobic contacts with Asp698 and Leu767.

Antiproliferative Properties of Prepared Compounds
In parallel to kinase assays, we measured the effect of compounds on the viability of cancer cell lines MV4-11 and K562 using a resazurin-based assay. MV4-11 is an AML cell line dependent on a constitutively activated FLT3-ITD kinase with high sensitivity to FLT3 inhibition, K562 cells (chronic myelogenous leukemia) are independent on FLT3. As shown in Table 1, most of the compounds have GI50 values in a mid-nanomolar range in MV4-11 cells, while K562 cells were substantially less sensitive (most of GI50 > 1 µM). Eleven compounds displayed GI50 below 200 nM in MV4-11 and from these, compounds 14i and 14e exerted a selectivity index > 20 (ratio for GI50′s of K562 and MV4-11 cells). Compound 15a was the most potent in MV4-11 cells expressing GI50 = 50 nM.
FLT3 inhibition causes G1-cell cycle arrest in FLT3-dependent cell lines such as MV4-11; we, therefore, performed flow cytometric analysis to evaluate cell cycle changes in MV4-11 and K562 cell lines treated with 200 nM compounds for 24 h. The data showed that most compounds block the cell cycle in the G1 phase in MV4-11 cells. Specifically, 18 compounds arrested >85% of cells in the G1 phase, which is >30% cells in the G1 phase of the cell cycle than in control cells ( Figure 4A). The observed results confirmed the data obtained from the viability assays because all non-toxic compounds (e.g., 14a, 14j, 14l, and 14n) did not significantly affect the cell cycle in MV4-11 cells. No changes in cell cycle distribution upon treatment were observed in K562 cells, confirming the specificity of the

Antiproliferative Properties of Prepared Compounds
In parallel to kinase assays, we measured the effect of compounds on the viability of cancer cell lines MV4-11 and K562 using a resazurin-based assay. MV4-11 is an AML cell line dependent on a constitutively activated FLT3-ITD kinase with high sensitivity to FLT3 inhibition, K562 cells (chronic myelogenous leukemia) are independent on FLT3. As shown in Table 1, most of the compounds have GI 50 values in a mid-nanomolar range in MV4-11 cells, while K562 cells were substantially less sensitive (most of GI 50 > 1 µM). Eleven compounds displayed GI 50 below 200 nM in MV4-11 and from these, compounds 14i and 14e exerted a selectivity index > 20 (ratio for GI 50 s of K562 and MV4-11 cells). Compound 15a was the most potent in MV4-11 cells expressing GI 50 = 50 nM.  FLT3 inhibition causes G1-cell cycle arrest in FLT3-dependent cell lines such as MV4-11; we, therefore, performed flow cytometric analysis to evaluate cell cycle changes in MV4-11 and K562 cell lines treated with 200 nM compounds for 24 h. The data showed that most compounds block the cell cycle in the G1 phase in MV4-11 cells. Specifically, 18 compounds arrested >85% of cells in the G1 phase, which is >30% cells in the G1 phase of the cell cycle than in control cells ( Figure 4A). The observed results confirmed the data obtained from the viability assays because all non-toxic compounds (e. g., 14a, 14j, 14l, and 14n) did not significantly affect the cell cycle in MV4-11 cells. No changes in cell cycle distribution upon treatment were observed in K562 cells, confirming the specificity of the compounds ( Figure 4B). Only a few compounds, such as 12c, 14e, and 14g-14i, caused moderate G1 block in K562 that probably stems from the potent inhibition of CDK4 (see Table 1). The control experiments were performed with palbociclib and quizartinib; palbociclib could significantly arrest both cell lines in the G1 phase, while quizartinib arrested selectively only MV4-11. for 24 h ( Figure 5). Increasing doses of 15a arrested cells in the G1 phase of the cell cycle up to 90% which was also observed for the FLT3 inhibitor, quizartinib. Oppositely, cell line K562 did not show any changes upon treatment with 15a and quizartinib.

Inhibition of FLT3 Signaling In Vivo
Next, we investigated the compounds for their ability to inhibit FLT3 and its downstream signaling in the MV4-11 cell line. Western blot analysis after 1 h treatment of MV4-11 cells revealed that 14e inhibits autophosphorylation of FLT3 kinase at Y589/591 and subsequently diminishes the phosphorylation of its downstream proteins, STAT5 and ERK1/2, in nanomolar doses ( Figure 6). A similar inhibitory pattern was observed in quizartinib-treated cells.

Inhibition of FLT3 Signaling In Vivo
Next, we investigated the compounds for their ability to inhibit FLT3 and its downstream signaling in the MV4-11 cell line. Western blot analysis after 1 h treatment of MV4-11 cells revealed that 14e inhibits autophosphorylation of FLT3 kinase at Y589/591 and subsequently diminishes the phosphorylation of its downstream proteins, STAT5 and ERK1/2, in nanomolar doses ( Figure 6). A similar inhibitory pattern was observed in quizartinib-treated cells.
In parallel, we examined the effect of the most potent and selective FLT3 inhibitor 15a, which showed a similar inhibitory pattern of signalling inhibition. The dephosphorylation of FLT3 was observed already at an 8 nM concentration and nearly complete suppression of ERK1/2 at 40 nM of 15a. Figure 6. Effect of 14e and 15a on phosphorylation of FLT3 and its downstream signaling pathways. MV4-11 cells were treated for 1 h with indicated concentrations of the compounds; quizartinib (Quiz.) was used as a standard. α-tubulin was used for equal protein loading.

In Vivo Activity of 15a in MV4-11 Xenograft Model
Finally, we tested the in vivo activity of 15a in a model of AML based on subcutaneous xenotransplantation of MV4-11 cells in immunodeficient mice. As shown in Figure  7A, daily i.p. treatment with 15a (20 mg/kg) was associated with a significant reduction in the growth of xenografted MV4-11 cells compared to vehicle-dosed control mice. Importantly, no significant reduction in mice weight was observed during the therapy (Figure 7B).  In parallel, we examined the effect of the most potent and selective FLT3 inhibitor 15a, which showed a similar inhibitory pattern of signalling inhibition. The dephosphorylation of FLT3 was observed already at an 8 nM concentration and nearly complete suppression of ERK1/2 at 40 nM of 15a.

In Vivo Activity of 15a in MV4-11 Xenograft Model
Finally, we tested the in vivo activity of 15a in a model of AML based on subcutaneous xenotransplantation of MV4-11 cells in immunodeficient mice. As shown in Figure 7A, daily i.p. treatment with 15a (20 mg/kg) was associated with a significant reduction in the growth of xenografted MV4-11 cells compared to vehicle-dosed control mice. Importantly, no significant reduction in mice weight was observed during the therapy ( Figure 7B).

Inhibition of FLT3 Signaling In Vivo
Next, we investigated the compounds for their ability to inhibit FLT3 and its downstream signaling in the MV4-11 cell line. Western blot analysis after 1 h treatment of MV4-11 cells revealed that 14e inhibits autophosphorylation of FLT3 kinase at Y589/591 and subsequently diminishes the phosphorylation of its downstream proteins, STAT5 and ERK1/2, in nanomolar doses ( Figure 6). A similar inhibitory pattern was observed in quizartinib-treated cells.
In parallel, we examined the effect of the most potent and selective FLT3 inhibitor 15a, which showed a similar inhibitory pattern of signalling inhibition. The dephosphorylation of FLT3 was observed already at an 8 nM concentration and nearly complete suppression of ERK1/2 at 40 nM of 15a. Figure 6. Effect of 14e and 15a on phosphorylation of FLT3 and its downstream signaling pathways. MV4-11 cells were treated for 1 h with indicated concentrations of the compounds; quizartinib (Quiz.) was used as a standard. α-tubulin was used for equal protein loading.

In Vivo Activity of 15a in MV4-11 Xenograft Model
Finally, we tested the in vivo activity of 15a in a model of AML based on subcutaneous xenotransplantation of MV4-11 cells in immunodeficient mice. As shown in Figure  7A, daily i.p. treatment with 15a (20 mg/kg) was associated with a significant reduction in the growth of xenografted MV4-11 cells compared to vehicle-dosed control mice. Importantly, no significant reduction in mice weight was observed during the therapy (Figure 7B).

General Information
Starting materials and reagents were purchased from various commercial sources (VWR, Radnor, PA, USA; Merck, Darmstadt, Germany; Fluorochem, Hadfield, UK; Acros Organics, Waltham, MA, USA) and used as received. All reactions were carried out under air. Reaction workup and column chromatography were performed with commercial grade solvents without further purification. All reactions were monitored using LC/MS analysis or thin-layer chromatography (TLC) using aluminium plates precoated with silica gel (silica gel 60 F254, Merck, US) impregnated with a fluorescent indicator. TLC plates were visualized using exposure to ultraviolet light (λ = 254 nm). Flash chromatography was performed using silica gel (35-70 µm partical size) column chromatography. The cross-coupling reactions under microwave irradiation were performed in a 10 mL glass tube sealed with a PTFE coated septa. Reaction mixture was pre-mixed for 1 min with rapid stirring and then irradiated at a maximum power of 150 W with simultaneous cooling using compressed air (24 psi). When the desired temperature was reached (ramping time ∼1 min), the power was automatically adjusted to maintain the set reaction temperature for a specified time. Finally, the vessel was cooled to approximately 40 • C with compressed air (cooling time ∼1 min) . Intermediates 2, 3, 4a, 4b, 5c, 8a-b, 9c-e, 9j, and 9l, 9p-o have been already reported [21,22,[32][33][34][35], for more information see Supplementary Information.

Instrumentation
The LC−MS analyses were carried out using UPLC Waters Acquity equipped with PDA and QDa detectors. The system comprised XSelect HSS T3 (Waters, Milford, MA, USA) 3 mm × 50 mm C18 reverse phase column XP, 2.5 µm particles. Mobile phases: 10 mM ammonium acetate in HPLC grade water (A) and gradient grade acetonitrile for HPLC (B). A gradient was mainly formed from 20 to 80% of B in 4.5 min, kept for 1 min, with a flow rate of 0.6 mL/min. The MS ESI operated at a 25 V cone voltage, 600 • C probe temperature, and a 120 • C source temperature.
All 1 H and 13 C NMR experiments were performed at magnetic field strengths of 11.75 T (with operating frequencies 500.16 MHz for 1 H and 125.77 MHz for 13 C) and 9.39 T (with operating frequencies 399.78 MHz for 1 H and 100.53 MHz for 13 C) at an ambient temperature (20 • C). 1 H and 13 C spectra were referenced relative to the signal of DMSO-d 6 ( 1 H: δ = 2.50 ppm, 13 C: δ = 39.51 ppm) or CDCl 3 ( 1 H: δ = 7.260 ppm, 13 C: δ = 77.160 ppm).
HRMS analyses were performed using UPLC Dionex Ultimate 3000 equipped with an Orbitrap Elite high-resolution mass spectrometer, Thermo Exactive plus, operating at full scan mode (120,000 FWMH) in the range of 100−1800 m/z. The settings for electrospray ionization were as follows: oven temperature of 150 • C and a source voltage of 3.6 kV. The acquired data were internally calibrated with diisooctyl phthalate as a contaminant in MeOH (m/z 391.2843).
Melting points were determined using VEB Analytik Dresden PHMK 78/1586 apparatus. Microwave experiments were conducted in the CEM Discover SP closed vessel microwave synthesizer.

General Synthetic Procedure 2: Cyclization with TFAA
For compound 4, 0.75 mmol was put into a flask, and toluene (1 mL) and trifluoroacetic anhydride (3 mL) were added. The mixture was heated at 70 • C for 16 h. The crude mixture was diluted with dichloromethane (20 mL), then washed with 10% K 2 CO 3 (30 mL) and distilled water (30 mL), and dried over MgSO 4 . The solvent was evaporated under reduced pressure (rotovap) and the residue was purified using column chromatography (usually 5:1 hexane/EtOAc).

General Synthetic Procedure 3: Cyclization with Phosgene
For compounds 4, 5, or 6, 0.70 mmol was dissolved in anhydrous THF (10 mL) under inert atmosphere of nitrogen and cooled to −10 • C. After the slow addition of 15 wt % COCl 2 in toluene (560 µL, 0.85 mmol), 1 M LiHMDS in hexane (1.4 mL, 1.40 mmol) was added portionwise. After 1 h, the reaction mixture was cooled to ambient temperature and stirred for 20 min. The mixture was then evaporated under reduced pressure and the residue was purified using column chromatography (usually 2:1 EtOAc/hexane).

Cell Cultures and Viability Assay
Human cell lines were obtained from the European Collection of Authenticated Cell Cultures (K562) and Cell lines service (MV4-11) and were cultivated according to the provider's instructions. Briefly, cells were maintained in RPMI-1640 (MV4-11) or DMEM (K562) supplemented with 10% fetal bovine serum, penicillin (100 U/mL), and streptomycin (0.1 mg/mL) at 37 • C in 5% CO 2 . For the viability assays, cells were treated in triplicate with six different doses of each compound for 72 h. After treatments, the resazurin (Sigma Aldrich, St. Louis, MO, USA) solution was added for 4 h, and the fluorescence of resorufin corresponding to live cells was measured at 544/590 nm (excitation/emission) using a Fluoroskan Ascent microplate reader (Labsystems). The GI 50 value, the drug concentration lethal to 50% of the cells, was calculated from the dose-response curves that resulted from the assays.

Kinase Inhibition Assays
CDK/cyclin complexes and FLT3 kinases were assayed as previously described [36,37]. Protein kinase selectivity was evaluated at a single concentration (1 µM) by screening against 30 enzymes at Eurofins Discovery (France).

Flow Cytometry
Asynchronously growing cells were seeded into a 96-well plate and, after a preincubation period, treated with tested compounds for 24 h. The treated cells were stained directly (or after trypsinization for the adherent cells) with the 5× staining solution (17 mM trisodium citrate dihydrate, 0.5% IGEPAL CA-630, 7.5 mM spermine tetrahydrochloride, and 2.5 mM Tris; pH 7.6 containing 50 mg/mL propidium iodide). After the staining, DNA content was analyzed using flow cytometry using a 488 nm laser (BD FACS Verse with software BD FACSuite TM , version 1.0.6). Cell cycle distribution was analyzed using ModFit LT (Verity Software House, version 5.0). Palbociclib, trilaciclib, and quizartinib were purchased from MedChemExpress, Monmouth Junction, NJ, USA.

Molecular Docking
Molecular docking was performed into the previously published model of FLT3 in the active conformation [29]. The 3D structure of 15a and the conformations with the lowest energy were obtained using molecular mechanics with Avogadro 1.90.0. Polar hydrogens were added to ligand and protein with the AutoDock Tools program and rigid docking was performed using AutoDock Vina 1.059 [38]. Interactions between 15a and FLT3 were determined using PLIP (protein-ligand interaction profiler) [39] and the figure was generated using Pymol ver. 2.0.4 (Schrödinger, LLC, New York, NY, USA).

Experimental Therapy
The experimental design was approved by the Institutional Animal Care and Use Committee (MSMT-37334/2020-4). Immunodeficient adult female NOD.Cg-Prkdc scid Il2rg tm1Wjl /SzJ mice (referred to as NSG mice) were purchased from The Jackson Laboratory and preserved in a pathogen-free environment in individually ventilated cages, provided with sterilized food and water. NSG mice were subcutaneously inoculated with 10 × 10 6 leukemia cells. After all mice developed palpable tumors, they were stratified into cohorts with comparable tumor volumes and therapy with 15a (20 mg/kg, daily i.p. dosing, 7 consecutive days) was initiated (=day 1, D1). Each cohort contained 6 animals. Three perpendicular dimensions (in millimeters) were measured daily with a digital caliper and tumor volumes were calculated using the following formula: π/6 × length × width × height. An experiment was terminated (and mice euthanized) when tumors exceeded 20 mm in the largest diameter.

Conclusions
In summary, we prepared a series of 2,7,9-trisubstituted-purin-8-nones to compare structure-activity relationships on CDK2, CDK4, and FLT3 kinases. It was revealed that FLT3 kinase was more tolerant in the structural variety of substituents at positions 7 and 9 than CDK4 kinase. The introduction of the isopropyl group at position 7 was significantly beneficial for selectivity toward FLT3 kinase. The more detailed selectivity study on 30 kinases, known as off-targets of approved CDK4/6 inhibitors, with three representative compounds showed effective inhibition of four additional kinases STK16, TTK, NUAK2, Lyn, and Src. Cellular analyses with MV4-11 cells treated the selected candidates 14e and 15a confirmed the inhibition of autophosphorylation of FLT3 kinase, subsequent dephosphorylation of the downstream proteins STAT5 and ERK1/2 in nanomolar doses, and G1-cell cycle arrest. The pharmacological properties of the model compound studied on Caco-2 cells indicated a substantially higher stability of inhibitor 14e in comparison with trilaciclib. To our delight, a significant reduction in tumor growth was observed during the therapy of the MV4-11 xenograft model.