Discovery of a Highly Selective FLT3 Inhibitor with Specific Proliferation Inhibition Against AML Cells Harboring FLT3-ITD Mutation

Abstract

FLT3 is frequently over-expressed in acute myeloid leukemia (AML), and FLT3 mutants, particularly FLT3-ITD, are closely related to poor prognosis in AML patients. Therefore, FLT3 has become an attractive target for AML therapy. A range of FLT3 inhibitors have undergone evaluation in clinical trials, with one recently approved for AML treatment. However, current FLT3 inhibitors face challenges with kinase selectivity and drug resistance, especially due to concurrent FLT3-ITD-TKD mutations. In this work, we identified a new FLT3 inhibitor (compound 1) with a simple structure through virtual screening of an extensive in-house molecule database containing numerous kinase inhibitors. Compound 1 was discovered to have potent inhibitory activity against several FLT3 mutants and exhibited high selectivity for FLT3 over other kinases. Furthermore, its anti-proliferative effects on tumor cells in vitro were dependent on the presence of the FLT3-ITD mutation, displaying little cytotoxicity to MV4-11 and human normal cells. Mechanistic studies showed that compound 1 blocked the FLT3 pathway, induced cell cycle arrest, and triggered apoptosis in MV4-11 cells. Molecular dynamics simulations were employed to investigate the binding mode of compound 1, providing insight into key intermolecular binding residues and strategies for further structural optimization. Thus, compound 1 represents a promising starting point for research on FLT3 inhibitors with improved kinase selectivity and potential to overcome drug resistance.

Introduction

Acute myeloid leukemia (AML) is a hematopoietic malignancy characterized by the dysfunction of bone marrow hematopoietic systems, abnormal proliferation of hematopoietic stem cells, and deficiencies in cell differentiation and apoptosis. Multiple factors contribute to AML incidence, including gene mutations, abnormal gene or non-coding RNA expression, and commonly acquired mutations. The “two-hit” hypothesis describes AML pathogenesis as requiring at least two sequential gene mutations. For instance, initial driver mutations in genes such as NPM1, DNMT3A, IDH2, and TET2 can extend the survival of hematopoietic stem and progenitor cells. Additional mutations then complete AML development, among which internal tandem duplication of the Fms-like tyrosine kinase 3 (FLT3-ITD) is the most common and challenging. FLT3-ITD and other FLT3 mutations occur in over 30% of AML patients. FLT3-ITD mutations drive abnormal hematopoietic cell proliferation and hinder normal differentiation by continuously activating the PI3K/AKT/STAT5 pathway and the MAPK pathway. Patients with increased FLT3-ITD to wild-type ratio display higher sensitivity to FLT3-targeted therapies, highlighting the significance of FLT3-ITD in AML pathology. In contrast, point mutations in FLT3′s kinase domain (FLT3-TKD) are less frequent in early AML, and their significance for prognosis is complex. Concurrent FLT3-ITD and -TKD mutations can drive resistance to FLT3 inhibitors, such as the resistance of FLT3-ITD-F691L to quizartinib and of FLT3-ITD-D835V to PLX3397. Therefore, FLT3, particularly FLT3-ITD, is a highly attractive target for anti-AML drug research.

FLT3 belongs to the type III receptor tyrosine kinase (RTK) family and shares high homology with other transmembrane kinases, including c-KIT, FMS, and platelet-derived growth factor receptor (PDGFR). As a transmembrane kinase, FLT3 contains four domains: an extracellular domain, a single transmembrane domain, a juxtamembrane domain (JM), and a typical kinase domain composed of two lobes linked by a hinge region. The JM domain regulates FLT3 through auto-inhibition by closely contacting the kinase domain and preventing activation. However, ITD mutations in the JM domain lead to the loss of auto-inhibition, causing ligand-independent and constitutive autophosphorylation of FLT3.

Numerous FLT3 inhibitors have been described in published literature and patents. Several molecules have entered clinical trials for AML, including crenolanib, tandutinib, linifanib, sunitinib, midostaurin, lestaurtinib, sorafenib, pacritinib, PLX3397, and quizartinib. FLT3 inhibitors are generally classified as type I or type II based on their mechanisms of interaction with FLT3. While some favorable clinical results have been achieved, rapid development of resistance, often driven by secondary FLT3-TKD mutations, is a significant challenge. Recently, midostaurin was approved for AML treatment in combination with chemotherapy, confirming FLT3 as a druggable target for AML. Another persistent problem is selectivity: many FLT3 inhibitors exhibit multi-kinase activity, as in the case of midostaurin. Additionally, resistance frequently occurs for type II inhibitors due to FLT3 D835Y/V/I/F and F691L mutations. Inspired by crenolanib’s improved selectivity and resistance profile, it is believed that type I inhibitors may be preferable, as they do not interact with the D835 residue in the activation loop and may not be extensively affected by the F691 mutation.

This study presents the discovery and validation of a selective type I FLT3 and FLT3-mutant inhibitor by virtual screening of a kinase inhibitor-focused in-house database, with the aim of developing a compound that could serve as a promising starting point for AML chemotherapy.

Results and Discussion

Hit Compound Identification by In Silico Screening

Virtual screening was performed using a structure-based approach with Glide on an in-house molecule database focused on kinase inhibitors. All compounds were docked into the catalytic site of activated FLT3, the target of type I inhibitors. As no suitable crystal structure was available at the start of the study, a three-dimensional FLT3 structure was constructed by homology modeling and optimized via loop refinement, energy minimization, and molecular dynamics simulations. Docking results were filtered by docking score, and compounds with scores below -5.0 were selected for further evaluation. Preliminary FLT3 inhibition assays identified five potential FLT3 inhibitors with IC50 values below 1 μM. Compound 1, originally synthesized as a PLK1 inhibitor, was further studied due to its potent FLT3 inhibitory activity (IC50 = 3.03 nM).

Chemistry

Compound 1 was re-synthesized through a multi-step process, beginning with substitution of a 4-bromo group to obtain intermediate A. Reaction with acetonitrile generated the 1,3-bifunctional intermediate B, followed by cyclization to yield the substituted pyrazole amine (compound C). Salt E was prepared via Buchwald-Hartwig coupling and alkaline hydrolysis. The final compound 1 was produced by coupling intermediates C and E.

Biological Validation of Compound 1

To confirm the inhibitory activity of compound 1 against FLT3, in vitro enzyme assays were conducted. Compound 1 displayed strong potency against wild-type FLT3 and two common mutants (FLT3-D835Y and FLT3-ITD), with IC50 values of 3.03, 0.735, and 0.94 nM, respectively. As secondary TKD mutations in FLT3-ITD can cause resistance to current inhibitors, compound 1 was also tested against the FLT3-ITD-D835V and FLT3-ITD-F691L mutations, and still showed moderate potency, indicating potential to overcome certain types of resistance.

To evaluate kinase selectivity, compound 1′s enzyme activity was tested against 18 kinases and further measured by IC50 values. It exhibited significantly greater potency against FLT3 compared to other kinases, including FLT3 homologues FMS and c-Kit, with selectivity over two orders of magnitude in most cases.

As type II FLT3 inhibitors are vulnerable to D835Y mutations, and compound 1 showed consistent activity against FLT3-D835Y and FLT3-ITD, it was further characterized as a type I inhibitor. Analysis of its affinity for various phosphorylated and non-phosphorylated forms of kinases showed no significant discrepancy, supporting its classification as a type I inhibitor and confirming its selectivity.

Cell Growth-Inhibition Potency and Cell Toxicity

Compound 1’s anti-proliferative activity was assessed in AML cell lines. It suppressed the proliferation of MOLM13 and MV4-11 cell lines, validating its anti-AML efficacy. In contrast, leukemia cell lines expressing wild-type FLT3 or lacking FLT3 expression were resistant to compound 1, as were breast cancer and hepatoma cell lines. The results confirm that the compound’s activity is dependent on the presence of the FLT3-ITD mutation.

MTT assays demonstrated that compound 1 was not overtly cytotoxic to MV4-11 cells, even at high concentrations (up to 1 μM), matching the performance of sorafenib. Furthermore, it showed no toxicity to human peripheral blood mononuclear cells at 1 μM.

Cellular Mechanism

The effects of compound 1 on the FLT3 pathway in MV4-11 cells were examined. Western blotting revealed dose-dependent inhibition of the phosphorylation of FLT3 and its downstream effector STAT5, while non-phosphorylated forms remained unaffected, highlighting specific inhibition of activated FLT3. Flow cytometry analysis confirmed that the compound caused cell cycle arrest in the G2/M phase and induced dose-dependent apoptosis, both effects comparable to those of sorafenib.

In Vitro Metabolic Stability

Compound 1′s metabolic stability was evaluated in liver microsomes. It displayed greater stability in human liver microsomes compared to rat microsomes, consistent with acceptable stability for a hit compound.

Molecular Simulation Study

The binding mode of compound 1 with FLT3 was investigated through molecular docking and optimized via molecular dynamics simulation over 30 nanoseconds. Considering the tautomerization of the pyrazole amine moiety, both tautomers were analyzed. The complex with tautomer A reached equilibrium faster during simulation, and MM-GB/PBSA binding free energy calculations showed that tautomer A possessed slightly stronger binding affinity.

Further decomposition of binding free energies on a per-residue basis demonstrated that residues Cys694, Tyr693, and Leu616 in the hinge region played critical roles in binding. Hydrogen bond analysis showed stable interactions between the NH group of Cys694 and the nitrogen atom of the pyrazole ring in compound 1, supporting its key role in molecular recognition. Non-polar interactions, such as pi-pi stacking with the phenyl ring of Tyr693 and hydrophobic interactions with Leu616, further stabilized binding. The molecular insights suggest that optimizing the cyano-thiophene moiety may improve interactions in the back pocket and enhance potency.

Comparison with the binding modes of other moderately active and inactive compounds underscored the importance of hydrogen bonding with Cys694 for stable FLT3 inhibition.

Conclusions

FLT3-targeted therapy is validated in AML, with midostaurin as the first approved FLT3 inhibitor for AML monotherapy. Clinical trials continue with quizartinib, crenolanib, and gilteritinib, inspiring efforts to develop new inhibitors with enhanced selectivity and resistance profiles. Through virtual screening, compound 1 was discovered as a type I FLT3 inhibitor with efficacy against both FLT3-ITD and FLT3-TKD mutations, as well as the more challenging concurrent FLT3-ITD-TKD mutations. The compound exhibited strong selectivity for FLT3 and specificity for AML cells with FLT3-ITD expression, causing cell cycle arrest and apoptosis by disruption of FLT3 signaling. Compound 1 also displayed metabolic stability in vitro.

Therefore, compound 1 is a promising hit for further development as a new FLT3 inhibitor with both kinase and cell specificity, as well as activity against various FLT3 mutations. Structural optimization, particularly of the 3-cyanothiophene moiety, may enhance cell potency while maintaining kinase activity. These findings aim to establish compound 1 as a scaffold for future FLT3 inhibitor discovery and AML therapy development.

Experimental Section

Homology Modeling

The human FLT3 sequence was retrieved from the UniProt database and homologous protein templates were identified using the BLAST algorithm. The model was constructed with the Schrödinger 2009 platform using multiple templates from the Protein Data Bank. Model optimization was achieved via steepest descent and Polak-Ribiere conjugate gradient algorithms until the gradient converged to less than 0.05.

Molecular Docking

The FLT3 model was prepared with the Protein Preparation Wizard, and ligands were docked into the ATP binding site using the Glide module in single precision with default parameters. The top-ranked pose was selected for further simulation.

Molecular Dynamics Simulation

Molecular dynamics simulations were performed using the AMBER 12 package. Systems were solvated in a TIP3P water box and neutralized. Energy minimization, heating, and equilibrium simulations were conducted according to standard protocols. The MM-GB/PBSA algorithm was applied to binding free energy calculations for the ligand and decomposed on a per-residue basis.

Kinase Assay

Inhibitory rates and IC50 values against kinases were determined using methods previously reported. Reactions were performed at Reaction Biology Corporation using the HotSpot assay platform. Kinase reactions were initiated by mixing with ATP and radiolabeled ^33P ATP, incubated, and activities detected by a filter-binding method. IC50 values were calculated using Prism software.

Kd values with kinases were measured using the KINOMEscan platform at DiscoveRx Corporation. Magnetic beads liganded with biotinylated small molecules were incubated with kinase and test compound. Following washing and elution, kinase concentration in samples was determined by qPCR and Kd values derived using the Hill equation.

Cell Growth Inhibition Assay

The inhibitory rate or IC50 against human tumor cell lines was evaluated as previously reported. Cells from the ATCC were cultured in IMDM media and exposed to test compounds for 72 hours. Cell viability was measured using the CellTiter-Glo assay according to manufacturer instructions. IC50 values were calculated from three independent experiments.

Cell Viability Assay

MTT assays were performed using established protocols. Tumor cells were cultured in 96-well plates, exposed to compounds for 72 hours, then incubated with MTT for 4 hours. Formazan was dissolved, and absorbance read at 570 nm. Data were normalized to vehicle controls and reported as mean values from six replicates. Human peripheral blood mononuclear cells were isolated and treated similarly, except for cell density.

Western Blotting

MV4-11 cells were seeded in 6-well plates and incubated overnight. Cells were treated with compound solutions, and lysates collected after four hours. Immunoblotting was performed using antibodies against FLT3/phospho-FLT3, STAT5/phospho-STAT5, and GAPDH as controls.

Cell Apoptosis and Cell Cycle Assay

Flow cytometry was performed using a FACScan flow cytometer. MV4-11 cells were exposed to test compounds, then stained with FITC-conjugated Annexin V and propidium iodide for apoptosis analysis or fixed and stained for cell cycle analysis.

In Vitro Liver Microsomal Stability

Compounds (1 μM) were incubated with human or rat liver microsomes at 37°C. Reactions were terminated at specified times, and the percentage of compound remaining determined by LC-MS/MS. Half-life (t1/2) and intrinsic clearance (Clint) were calculated using first-order kinetics.

Chemistry

All commercially available reagents were used without further purification. Melting points and NMR spectra were measured with standard instrumentation. Product purity was HPK1-IN-2 confirmed to be greater than 95% by HPLC.