Synthesis, In Vitro and In Silico Anticancer Activity of New 4-Methylbenzamide Derivatives Containing 2,6-Substituted Purines as Potential Protein Kinases Inhibitors

A novel class of potential protein kinase inhibitors 7–16 was synthesized in high yields using various substituted purines. The most promising compounds, 7 and 10, exhibited inhibitory activity against seven cancer cell lines. The IC50 values for compounds 7 and 10 were 2.27 and 2.53 μM for K562 cells, 1.42 and 1.52 μM for HL-60 cells, and 4.56 and 24.77 μM for OKP-GS cells, respectively. In addition, compounds 7 and 10 dose-dependently induced the apoptosis and cell cycle arrest at G2/M phase, preventing the cell division of OKP-GS cells. Compounds 7, 9, and 10 showed 36–45% inhibitory activity against PDGFRα and PDGFRβ at the concentration of 1 μM. Molecular modeling experiments showed that obtained compounds could bind to PDGFRα as either type 1 (compound 7, ATP-competitive) or type 2 (compound 10, allosteric) inhibitors, depending on the substituent in the amide part of the molecule.


Introduction
The discovery of imatinib, a selective inhibitor of Bcr-Abl protein kinase, in the late 1990s, was an important step not only in revealing the relationship between the Philadelphia chromosome and the onset of chronic myeloid leukemia, but also in the development of targeted cancer therapy in general [1,2]. The success of imatinib showed how effective a strategy of attacking a specific biological target in cancer cells can be and, no less important, drew attention to protein kinases as a key element of the cell signaling system, to which the action of targeted drugs can be directed [3,4]. More than 20 years have passed since the registration of imatinib, and every year the number of protein kinase inhibitors used for the treatment of cancer is only increasing [5,6]. In addition, the number of protein kinase targets for approved inhibitors is constantly expanding as well [7][8][9][10][11]. Recent progress also shows that protein kinase inhibitors can be used to treat non-oncological diseases [12,13]. It is important to emphasize that the known kinase targets constitute only a small part of the human kinome.
It should be noted, that, in most cases, the use of small-molecule kinase inhibitors (SMKIs) still does not provide as favorable a prognosis as it does in the case of imatinib and other Abl inhibitors. However, their relatively high efficacy, combined with good tolerance, makes them, in most cases, the first line of drug treatment [14]. The common causes of the primary or subsequent resistance to SMKIs are the mutations of their protein targets due to the genomic instability of cancer cells or the specificity of a particular patient [15]. Mutations change the structure of the binding site of a kinase, reducing the affinity of the inhibitor. Some of the most problematic mutations (for example, T315I for Abl kinase) make the binding completely impossible [16][17][18][19][20]. It should be noted that all kinases use the same substrate (ATP) and in general share many similarities in their structure. In most 2 of 23 cases, an inhibitor targets several kinase enzymes at once. Such multi-targeting could be a positive moment for the cases when cancer cells over-express several protein kinases. However, blocking non-oncogenic proteins could lead to side effects [21]. Modulating drug selectivity is a major challenge in SMKI development.
X-ray structural data make it possible to distinguish a number of features of SMKI binding in a protein active site [22]. Most of the kinase inhibitors discovered to date are ATPcompetitive and are classified as type 1 inhibitors. This type of inhibitor binds to the active conformation of a kinase. Type 1 inhibitors usually consist of a heterocyclic ring system that occupies the purine binding site, where it serves as a framework for its substituents to occupy adjacent hydrophobic regions I and II [23]. At the same time, a significant portion of inhibitors, including imatinib, bind to a kinase that is in a biologically inactive conformation (type 2 inhibitors), in which the flexible kinase activation loop opens an additional, allosteric pocket. This pocket can be used by an inhibitor to form additional interactions. Common pharmacophore features of type 2 inhibitors are: the presence of a specific heterocyclic base creating interactions in the ATP pocket, and the amide bond, usually with a 3-trifluoromethylaniline substituent, which provides strong interactions in the allosteric site. The linker that connects these two fragments is, as a rule, represented by a benzene ring with substituents at different positions [24].
In our previous work, we suggested to use a flexible 4-methylbenzamide linker to obtain new chemical compounds capable of inhibiting protein kinases. A number of the obtained derivatives of 4-methylbenzamide showed high biological activity in vitro and in silico [25,26].
In this study, the design, synthesis and study of the in vitro biological activity of new 4-methylbenazmide derivatives were carried out. The key idea of the design of new compounds is to preserve the N-(3-trifluoromethyl-phenyl)-4-methylbenzamide backbone, and to use various purine derivatives as substituents of the methyl group of 4-methylbenzamide. The N-9 atom of a purine was used for this connection. On the one hand, flexible linker and purines as a heterocyclic base provide the pharmacophore similarity of the target compounds with known type 2 SMKIs. On the other hand, purine derivatives can imitate the natural ATP substrate, being potential type 1 kinase inhibitors.
In this investigation, an attempt is made to trace the evolutionary pathway of purinecontaining analogs by a systematic search for structural and electronic similarities, using an in silico method for screening the obtained compounds with a number of different protein kinases and to assess the contribution of modified purine bases and imidazole groups to binding to the active site of various protein kinases.

Molecular Design
Based on the structural analysis of type 2 SMKI, we began the development of new potential Bcr-Abl inhibitors using the combination of bioisosteric substitution and conformational restriction. Nilotinib, a second-generation Bcr-Abl inhibitor, was used as base structure. Structural modifications were focused on the replacement of the phenylaminopyrimidine, which forms hydrogen bonds with the amino acids Met-318 and Thr-315 in the adenine pocket, with 2,6-substituted purines ( Figure 1). As a linker between the adenine pocket and the DFG-out pocket, we used 4-(aminomethyl)benzamide, the derivatives of which showed a significant in vitro and in silico anti-kinase activity in previous studies [24]. 3-(4-Methyl-1H-imidazol-1-yl)-5-(trifluoromethyl)aniline was chosen as the terminal fragment considering the additional interactions that are formed due to the presence of a trifluoromethyl substituent in the phenyl ring. This allow nilotinib to penetrate deeper into the allosteric pocket, enhancing hydrophobic interactions around the phenyl group and the imidazole fragment, which is in close contact with the Phe-359 residue in the C-lobe [27]. To clarify the importance of these interactions, the imidazole fragment was replaced in some structures by a hydrogen atom (Figure 1).
Based on these design principles, ten new analogs have been developed, synthesized, and studied as potential inhibitors of cell proliferation and protein kinase activity.

Chemistry
The synthesis of starting compounds 2-4 was carried out according to previously reported methods [28,29]. Compounds 5 and 6 were obtained by reacting chlorobenzoate 4 with 3-(trifluoromethyl) aniline or 3-(4-methyl-1H-imidazol-1-yl)-5-(trifluoromethyl) aniline, followed by isolation by column chromatography as described in [25]. Compounds 7-12 were synthesized by the condensation of the corresponding chlorine derivative 5 and 6 with the potassium salt of the purine analogue c-e, obtained by treating the corresponding base with potassium carbonate in dimethylformamide at reflux, in high yields (Scheme 1). 3-(4-Methyl-1H-imidazol-1-yl)-5-(trifluoromethyl)aniline was chosen as the terminal fragment considering the additional interactions that are formed due to the presence of a trifluoromethyl substituent in the phenyl ring. This allow nilotinib to penetrate deeper into the allosteric pocket, enhancing hydrophobic interactions around the phenyl group and the imidazole fragment, which is in close contact with the Phe-359 residue in the C-lobe [27]. To clarify the importance of these interactions, the imidazole fragment was replaced in some structures by a hydrogen atom (Figure 1).
Based on these design principles, ten new analogs have been developed, synthesized, and studied as potential inhibitors of cell proliferation and protein kinase activity.

Chemistry
The synthesis of starting compounds 2-4 was carried out according to previously reported methods [28,29]. Compounds 5 and 6 were obtained by reacting chlorobenzoate 4 with 3-(trifluoromethyl) aniline or 3-(4-methyl-1H-imidazol-1-yl)-5-(trifluoromethyl) aniline, followed by isolation by column chromatography as described in [25]. Compounds 7-12 were synthesized by the condensation of the corresponding chlorine derivative 5 and 6 with the potassium salt of the purine analogue c-e, obtained by treating the corresponding base with potassium carbonate in dimethylformamide at reflux, in high yields (Scheme 1).
The treatment of 2,6-dichloropurine derivative 7 or 10 with potassium carbonate in dry methanol led to the formation of purine analogs 13 and 14 with a yield of 69% and 45%, respectively (Scheme 2). Analogs 15 and 16 were obtained from the corresponding compounds 7 or 10 by treatment with 25% aq. NH 4  The treatment of 2,6-dichloropurine derivative 7 or 10 with potassium carbonate in dry methanol led to the formation of purine analogs 13 and 14 with a yield of 69% and 45%, respectively (Scheme 2). Analogs 15 and 16 were obtained from the corresponding compounds 7 or 10 by treatment with 25% aq. NH4OH in high yields. The structures of synthetic intermediates and products were established by 1 H, 13 C, and 19 F NMR spectroscopy and high-resolution mass spectrometry (HRMS). In the 1 H NMR spectra of compounds 12-16, there are chemical shifts in the range 5.45-5.62 ppm, which are typical proton signals of the methylene group connected to the N-9 nitrogen atom of the imidazole ring of purine. In comparison, for related compounds containing a CH2 group attached not to a purine, but to a secondary amine, the proton signal is shifted to a strong field and is located in the range of 4.54-4.69 ppm [25]. In the case of analogs containing a CH2 group at the piperazine ring, an even stronger shift of proton signals in a strong field up to 3.62 ppm is observed [25]. In the 19 F NMR spectra of compounds 8 and 11, an additional signal appears at −52.1 ppm, related to the fluorine atom of the purine moiety besides the CF3 signal at −61.3 ppm. Graphical NMR spectra of target compounds can be found in the Supplementary Materials.

Anti-Proliferation Activity
To determine the selectivity of the synthesized compounds 7-16 toward various cancer cells and normal human cells, the anti-proliferative activity against K562 (chronic myelogenous leukemia), HL-60 (promyelocytic leukemia), MCF-7 (breast adenocarcinoma), HeLa (cervix carcinoma), HepG2 (human liver cancer), A549 (lung carcinoma), The treatment of 2,6-dichloropurine derivative 7 or 10 with potassium carbonate in dry methanol led to the formation of purine analogs 13 and 14 with a yield of 69% and 45%, respectively (Scheme 2). Analogs 15 and 16 were obtained from the corresponding compounds 7 or 10 by treatment with 25% aq. NH4OH in high yields. The structures of synthetic intermediates and products were established by 1 H, 13 C, and 19 F NMR spectroscopy and high-resolution mass spectrometry (HRMS). In the 1 H NMR spectra of compounds 12-16, there are chemical shifts in the range 5.45-5.62 ppm, which are typical proton signals of the methylene group connected to the N-9 nitrogen atom of the imidazole ring of purine. In comparison, for related compounds containing a CH2 group attached not to a purine, but to a secondary amine, the proton signal is shifted to a strong field and is located in the range of 4.54-4.69 ppm [25]. In the case of analogs containing a CH2 group at the piperazine ring, an even stronger shift of proton signals in a strong field up to 3.62 ppm is observed [25]. In the 19 F NMR spectra of compounds 8 and 11, an additional signal appears at −52.1 ppm, related to the fluorine atom of the purine moiety besides the CF3 signal at −61.3 ppm. Graphical NMR spectra of target compounds can be found in the Supplementary Materials.

Anti-Proliferation Activity
To determine the selectivity of the synthesized compounds 7-16 toward various cancer cells and normal human cells, the anti-proliferative activity against K562 (chronic myelogenous leukemia), HL-60 (promyelocytic leukemia), MCF-7 (breast adenocarcinoma), HeLa (cervix carcinoma), HepG2 (human liver cancer), A549 (lung carcinoma), The structures of synthetic intermediates and products were established by 1 H, 13 C, and 19 F NMR spectroscopy and high-resolution mass spectrometry (HRMS). In the 1 H NMR spectra of compounds 12-16, there are chemical shifts in the range 5.45-5.62 ppm, which are typical proton signals of the methylene group connected to the N-9 nitrogen atom of the imidazole ring of purine. In comparison, for related compounds containing a CH 2 group attached not to a purine, but to a secondary amine, the proton signal is shifted to a strong field and is located in the range of 4.54-4.69 ppm [25]. In the case of analogs containing a CH 2 group at the piperazine ring, an even stronger shift of proton signals in a strong field up to 3.62 ppm is observed [25]. In the 19 F NMR spectra of compounds 8 and 11, an additional signal appears at −52.1 ppm, related to the fluorine atom of the purine moiety besides the CF 3 signal at −61.3 ppm. Graphical NMR spectra of target compounds can be found in the Supplementary Materials.
The data obtained for compounds 7-16 are provided in Table 1. Known protein kinase inhibitors were used as reference compounds: imatinib, sorafenib and nilotinib. For imatinib, the IC 50 value for the chronic myeloid leukemia cell line (K562) turned out to be less than 1 µM, which is consistent with the data hereof [30]. OKP-GS (renal cell carcinoma), and normal cells RPMI 1788 (B lymphocyte, the human cell line) was investigated.
The data obtained for compounds 7-16 are provided in Table 1. Known protein kinase inhibitors were used as reference compounds: imatinib, sorafenib and nilotinib. For imatinib, the IC50 value for the chronic myeloid leukemia cell line (K562) turned out to be less than 1 μM, which is consistent with the data hereof [30]. The compounds 7 and 10, containing two chlorine atoms at the C-2 and C-6 positions of the purine heterocycle, demonstrated high activity against all studied cell lines, comparable to the data for sorafenib. The highest activity was observed against the leukemic cell line K562 with IC50 values of 2.27 and 2.53 μM, and IC50 values for HL-60 equal to 1.42 and 1.52 μM, respectively.
The compound 7 showed high activity against the OKP-GS renal carcinoma cell line with an IC50 value of 4.56 μM. Additionally, a significant inhibitory ability against OKP-GS was found in compounds 13 and 14 containing a chlorine atom and a methoxy group at the C-2 and C-6 positions of the purine fragment, respectively. It should be noted that the absence of a 4-methyl-imidazole fragment in structures 7 and 13 had no noticeable effect on their antiproliferative activity against the studied cell lines as compared with analogs 10 and 14.
According to the data in Table 1, the synthesized compounds were toxic to the normal cell line RPMI 1788, comparable to the results for reference compounds that inhibited the growth of normal human dermal fibroblasts (NHDF) in experiments in vitro [31,32].

12
NH 2 H f n/a n/a n/a n/a n/a n/a n/a n/a 13 OMe NH 2 Cl f n/a n/a n/a n/a n/a n/a n/a 5.91 The compounds 7 and 10, containing two chlorine atoms at the C-2 and C-6 positions of the purine heterocycle, demonstrated high activity against all studied cell lines, comparable to the data for sorafenib. The highest activity was observed against the leukemic cell line K562 with IC 50 values of 2.27 and 2.53 µM, and IC 50 values for HL-60 equal to 1.42 and 1.52 µM, respectively.

Reference Compounds
The compound 7 showed high activity against the OKP-GS renal carcinoma cell line with an IC 50 value of 4.56 µM. Additionally, a significant inhibitory ability against OKP-GS was found in compounds 13 and 14 containing a chlorine atom and a methoxy group at the C-2 and C-6 positions of the purine fragment, respectively. It should be noted that the absence of a 4-methyl-imidazole fragment in structures 7 and 13 had no noticeable effect on their antiproliferative activity against the studied cell lines as compared with analogs 10 and 14.
According to the data in Table 1, the synthesized compounds were toxic to the normal cell line RPMI 1788, comparable to the results for reference compounds that inhibited the growth of normal human dermal fibroblasts (NHDF) in experiments in vitro [31,32].

Cell Apoptosis Assay of Compounds 7 and 10
Apoptosis is a versatile and extremely effective pathway for cell death. This phenomenon can be caused by internal signals, such as genotoxic stress, or external signals, such as the binding of ligands to death receptors on the cell surface [33]. The induction of apoptosis was studied by measuring the translocation of phosphatidylserine from the cytoplasmic to the extracellular side of the plasma membrane. OKP-GS cells were treated with compound 7 or 10 at 5, 10 and 20 µM for 72 h, after which annexin-V/propidium iodide (PI) binding was measured by flow cytometry (Figure 2).  The data obtained show that analogs 7 and 10 at a concentration of 10 μM have a significant effect on the induction of apoptosis, comparable to sorafnib at a concentration of 20 μM. At a concentration of 20 μM, compounds 7 or 10 exhibited a more than two-fold higher level of apoptosis compared with sorafnib. Compounds 7 and 10 were found to promote apoptosis in a dose-dependent manner ( Figure 3). The number of necrotic cells increased almost linearly and did not exceed 7.3%, which suggests that after 72 h cell death is mainly due to apoptosis, not necrosis.

In Vitro Cell Cycle Effects of Compounds 7 and 10
Cell cycle experiments were performed with OKP-GS cells treated with 5.0 μM, 10 μM and 20 μM compounds 7, 10 and 20 μM sorafenib, respectively (Table 4, Figure 4). To The data obtained show that analogs 7 and 10 at a concentration of 10 µM have a significant effect on the induction of apoptosis, comparable to sorafnib at a concentration of 20 µM. At a concentration of 20 µM, compounds 7 or 10 exhibited a more than two-fold higher level of apoptosis compared with sorafnib. Compounds 7 and 10 were found to promote apoptosis in a dose-dependent manner ( Figure 3). The number of necrotic cells increased almost linearly and did not exceed 7.3%, which suggests that after 72 h cell death is mainly due to apoptosis, not necrosis.  The data obtained show that analogs 7 and 10 at a concentration of 10 μM have a significant effect on the induction of apoptosis, comparable to sorafnib at a concentration of 20 μM. At a concentration of 20 μM, compounds 7 or 10 exhibited a more than two-fold higher level of apoptosis compared with sorafnib. Compounds 7 and 10 were found to promote apoptosis in a dose-dependent manner ( Figure 3). The number of necrotic cells increased almost linearly and did not exceed 7.3%, which suggests that after 72 h cell death is mainly due to apoptosis, not necrosis.

In Vitro Cell Cycle Effects of Compounds 7 and 10
Cell cycle experiments were performed with OKP-GS cells treated with 5.0 μM, 10 μM and 20 μM compounds 7, 10 and 20 μM sorafenib, respectively (Table 4, Figure 4). To     Table 4). For all studied substances the number of cells in the S phase was slightly increased in comparison with the control when treated at a concentration of 10 μM and 20 μM. The obtained result for compounds 7 and 10 showed that cells were able to progress from the G1 phase to the S phase, being then blocked in the G2/M phase. That caused a decrease in the number of cells in the G1 phase and an increase in the G2/M phase with an almost constant number of cells in the S-phase. Based on the results, we came to the conclusion that compound 7 dose-dependently blocks cells in the G2/M phase.
Compound 10 suppressed the development of the cell cycle in a similar way, although in a less pronounced form. At the same time, the effect of compounds 7 and 10 on cell cycle progression, leading to a decrease in OKP-GS cell proliferation, was more significant than that of sorafenib at 20 μM. The cell cycle analysis of OKP-GS cells is shown in Figure 4. The results for compounds 7 and 10 are summarized in Table 4.

Docking and Molecular Dynamics
The anticancer activity of the obtained compounds was evaluated in silico using Autodock Vina [36] as the most popular open source docking software. As receptors for  Table 4). For all studied substances the number of cells in the S phase was slightly increased in comparison with the control when treated at a concentration of 10 µM and 20 µM. The obtained result for compounds 7 and 10 showed that cells were able to progress from the G1 phase to the S phase, being then blocked in the G2/M phase. That caused a decrease in the number of cells in the G1 phase and an increase in the G2/M phase with an almost constant number of cells in the S-phase. Based on the results, we came to the conclusion that compound 7 dose-dependently blocks cells in the G2/M phase.
Compound 10 suppressed the development of the cell cycle in a similar way, although in a less pronounced form. At the same time, the effect of compounds 7 and 10 on cell cycle progression, leading to a decrease in OKP-GS cell proliferation, was more significant than that of sorafenib at 20 µM. The cell cycle analysis of OKP-GS cells is shown in Figure 4. The results for compounds 7 and 10 are summarized in Table 4.

Docking and Molecular Dynamics
The anticancer activity of the obtained compounds was evaluated in silico using Autodock Vina [36] as the most popular open source docking software. As receptors for docking, we used 33 experimental structures of cancer-related protein targets, most of which were protein kinases (Abl, Src, Aurora). Two receptors were poly (ADP-ribose) polymerases (PARP), since this class of proteins is a promising target in the treatment of cancer. Three-dimensional models of these proteins in complexes with known inhibitors were taken from the Protein Data Bank (PDB) [37].
The potential inhibitory activity of the synthesized derivatives of 4-methylbenazmide was evaluated based on the comparison of their docking scores with the docking scores of known inhibitors from PDB complexes. Structures of native inhibitors were docked twice. Initially, the original three-dimensional structure of the original ligand from the PDB file was used. In the second experiment, all original ligand structures were preliminarily minimized using a GAFF force field, which aims to simulate constructing them from scratch. For a number of receptors, docking scores for native ligands were taken from our previous works [25,26]. The correctness of the reproduction of the experimental docking poses for original ligands was assessed visually by the comparison with the ligands contained in the PDB complexes.
The values of the obtained docking scores for original ligands and studied structures are provided in Table 5. For the known ligands, the difference in binding energies for minimized and non-minimized structures was found to be negligible. In addition, for correctly reproduced poses, the docking scores were significantly better than for poses that did not correspond to the real binding model.    It should be noted that for the complexes of type 2 inhibitors, in which the ligand occupies the ATP pocket, allosteric pocket, and linker cavity, Autodock Vina made it possible to correctly reproduce the position of the inhibitor in all cases. Good results were also obtained for PARP receptors. For complexes of known inhibitors of type 1 or mixed type, which do not occupy the allosteric pocket, the docking error was quite high. The arrangement of complex aromatic systems often coincided with the experiment (PDB: 3ZBF, 4C2W, 5EW9, 6NEC), but the flexible parts of the structures were not docked properly. For example, for the complex of Aurora kinase with ATP (PDB: 4C2W), AutoDock Vina correctly identified the location of the purine and sugar fragments, but triphosphate residue did not match the original ligand. The strongest discrepancies between the predicted docking pose and experiment were observed for complexes 2BFY, 2VRX, 4B8M, and 4G5P.
The best docking scores were obtained for complexes of 7-9, 13 and 15, which do not contain a 4-methyl-imidazole fragment in their structure, with protein kinases of the epidermal growth factor family, HER2, ErbB4, VEGFR1, and EGFR T790M, as well as with kinase TRKb of the neurotrophin receptor group (Trk), which suggests the presence of inhibitory activity against these enzymes (Table 5).
For protein kinases Aurora A and Aurora B, the docking energies were from −8.7 to −10.4 kcal/mol, which corresponded to or exceeded the original ligand.
To refine the binding energy, a number of molecular dynamics experiments were carried out to calculate the binding energy and its components for complexes of the studied structures and some known inhibitors as reference. The binding energy calculations were performed using the MM-PBSA method [38] on the frames of the trajectory obtained after 2 ns molecular dynamics simulation, which was carried out using GROMACS [39]. The obtained binding energies are provided in Table 6. The studied compounds demonstrated, in general, lower binding energies, in comparison with the results of known inhibitors obtained in this work and in previous studies [25]. Complexes of compounds 10-12 with Abl kinase (PDB: 3CS9) showed high values of VDW component in binding energy, but the energy of electrostatic interactions turned out to be low. High values of electrostatic interactions were found for complexes of 8, 9 and 15 with BRAF kinase (PDB: 5HI2).

Molecular Modeling Study
To better understand the results of molecular docking and data on kinase inhibition for compounds 7-16, several complexes of target compounds with various proteins were visualized using the Chimera software [41].
Most of the known type 1 kinase inhibitors form one to three hydrogen bonds with amino acids located in the hinge region of a kinase, thereby mimicking the hydrogen bonds that are usually formed by the adenine ring of ATP [47,48]. They generally do not use a ribose-binding pocket or triphosphate-binding pocket.
The studied compounds demonstrated, in general, lower binding energies, in comparison with the results of known inhibitors obtained in this work and in previous studies [25]. Complexes of compounds 10-12 with Abl kinase (PDB: 3CS9) showed high values of VDW component in binding energy, but the energy of electrostatic interactions turned out to be low. High values of electrostatic interactions were found for complexes of 8, 9 and 15 with BRAF kinase (PDB: 5HI2).

Molecular Modeling Study
To better understand the results of molecular docking and data on kinase inhibition for compounds 7-16, several complexes of target compounds with various proteins were visualized using the Chimera software [41].
Most of the known type 1 kinase inhibitors form one to three hydrogen bonds with amino acids located in the hinge region of a kinase, thereby mimicking the hydrogen bonds that are usually formed by the adenine ring of ATP [47,48]. They generally do not use a ribose-binding pocket or triphosphate-binding pocket.
As shown in Figure 5A, compound 7 occupies the PDGFRα kinase ATP binding site (PDB: 6JOL). In this binding model, Arg-52 forms two hydrogen bonds with the N-3 purine atom (distances: 2.47 and 2.08 Å). In addition, the NH2 hydrogen of the Cys-132 group serves as an H-bond donor for the amide group of 4-metylbenzamide (distance: 1,71 Å).

Glu-99
Asp-219 In the case of dichloro compound 10, the structure behaves as a type 2 inhibitor, with a purine fragment being located in the adenine pocket, and the allosteric region is occupied by the trifluoromethyl and imidazole substituents of the phenyl ring. The amide group forms two hydrogen bonds with Glu-99 (distance: 1.97 Å) and Asp-219 (distance: 2.10 Å), as shown in Figure 5B. Compound 10 is characterized by a higher calculated binding energy (-133.473 kJ/mol) to receptor tyrosine kinase PDGFRα (PDB: 6JOL) than compound 7 (−109.837 kJ/mol), as shown in Table 6. Figure 5C shows differences in type 1 (lapatinib) and type 2 (nilotinib) inhibitors binding. Lapatinib binds to the ATP-binding site, not touching allosteric and linker pockets. Nilotinib occupies all three binding site regions: pyrimidine imitates ATP, 3-(4-methyl-1H-imidazol-1-yl)-5-(trifluoromethyl)aniline rests in the allosteric pocket, central benzamide ring implements proper orientation for these two fragments also forming hydrogen bonds with aspartic and glutamic residues. In fact, these two hydrogen bonds can be seen in the majority of type 2 inhibitor complexes.
Comparing the binding models of dichloro derivative 10 and purine analogue 12 to Abl-kinase (PDB: 3CS9), it should be noted that their positions are typical for type 2 inhibitors. The presence of the adenine ring in compound 12 leads to the formation of three hydrogen bonds in the adenine pocket. The amino group of purine (distance: 2.24 Å) and In the case of dichloro compound 10, the structure behaves as a type 2 inhibitor, with a purine fragment being located in the adenine pocket, and the allosteric region is occupied by the trifluoromethyl and imidazole substituents of the phenyl ring. The amide group forms two hydrogen bonds with Glu-99 (distance: 1.97 Å) and Asp-219 (distance: 2.10 Å), as shown in Figure 5B. Compound 10 is characterized by a higher calculated binding energy (-133.473 kJ/mol) to receptor tyrosine kinase PDGFRα (PDB: 6JOL) than compound 7 (−109.837 kJ/mol), as shown in Table 6. Figure 5C shows differences in type 1 (lapatinib) and type 2 (nilotinib) inhibitors binding. Lapatinib binds to the ATP-binding site, not touching allosteric and linker pockets. Nilotinib occupies all three binding site regions: pyrimidine imitates ATP, 3-(4-methyl-1H-imidazol-1-yl)-5-(trifluoromethyl)aniline rests in the allosteric pocket, central benzamide ring implements proper orientation for these two fragments also forming hydrogen bonds with aspartic and glutamic residues. In fact, these two hydrogen bonds can be seen in the majority of type 2 inhibitor complexes.
Comparing the binding models of dichloro derivative 10 and purine analogue 12 to Abl-kinase (PDB: 3CS9), it should be noted that their positions are typical for type 2 inhibitors. The presence of the adenine ring in compound 12 leads to the formation of three hydrogen bonds in the adenine pocket. The amino group of purine (distance: 2.24 Å) and the N-1 purine atom (distance: 2.27 Å) interact with the Met-95 residue, the N-3 atom of purine forms hydrogen bond with Thr-92 (distance: 2.13 Å), as shown in Figure 6B. At the same time, the presence of two chlorine atoms in analogue 10 does not allow the formation of hydrogen bonds in the hinge (Hinge) region, but the amide group forms two hydrogen bonds with residues Glu-63 (distance: 2.26 Å) and Asp-258 (distance: 1.86 Å), as shown in Figure 6A. However, compound 10 has a higher calculated binding energy (−144.216 kJ/mol) to Abl tyrosine kinase (PDB: 3CS9) than 12 (−137.285 kJ/mol), and a high antiproliferative activity against a number of cell lines (Tables 1 and 6). the N-1 purine atom (distance: 2.27 Å) interact with the Met-95 residue, the N-3 atom of purine forms hydrogen bond with Thr-92 (distance: 2.13 Å), as shown in Figure 6B. At the same time, the presence of two chlorine atoms in analogue 10 does not allow the formation of hydrogen bonds in the hinge (Hinge) region, but the amide group forms two hydrogen bonds with residues Glu-63 (distance: 2.26 Å) and Asp-258 (distance: 1.86 Å), as shown in Figure 6A. However, compound 10 has a higher calculated binding energy (−144.216 kJ/mol) to Abl tyrosine kinase (PDB: 3CS9) than 12 (−137.285 kJ/mol), and a high antiproliferative activity against a number of cell lines (Tables 1 and 6). The position of the 2-fluoroadenine derivative 8 in the active site of the BRAF (PDB: 5HI2) kinase is more consistent with type 1 inhibitors ( Figure 7A). Binding of compound 8 revealed four hydrogen bonds: the proton of the N-6 amino group as an H-donor (distance 1.91 Å), and the N-7 adenine atom as an acceptor (distance 2.28 Å) binds to the Cys-92 residue, while the amide group forms third and fourth H-bonds with Glu-61 (distance: 2.29 Å) and Asp-154 (distance: 1.76 Å). Despite the formation of four hydrogen bonds, compound 8 has a low calculated binding energy (−95.507 kJ/mol) with respect to BRAF kinase and did not show significant activity against the studied cell cultures and cancer-related kinases (Tables 1-3).

Met-95
Thr-92 Met-95 The position of the 2-fluoroadenine derivative 8 in the active site of the BRAF (PDB: 5HI2) kinase is more consistent with type 1 inhibitors ( Figure 7A). Binding of compound 8 revealed four hydrogen bonds: the proton of the N-6 amino group as an H-donor (distance 1.91 Å), and the N-7 adenine atom as an acceptor (distance 2.28 Å) binds to the Cys-92 residue, while the amide group forms third and fourth H-bonds with Glu-61 (distance: 2.29 Å) and Asp-154 (distance: 1.76 Å). Despite the formation of four hydrogen bonds, compound 8 has a low calculated binding energy (−95.507 kJ/mol) with respect to BRAF kinase and did not show significant activity against the studied cell cultures and cancer-related kinases (Tables 1-3 The BRAF-binding domain in Figure 7B demonstrates that the amide group of adenine derivative 9 participates in two H-bonds with side chains residues Glu-61 and Asp-154 (distance: 1.90 and 2.14 Å, respectively). The adenine fragment does not participate in the formation of hydrogen bonds, unlike 2-fluoroanalog 8. Purine analog 9 showed good inhibitory activity against PDGFRα PDGFRβ receptor kinases, 45% and 36%, respectively, at a concentration of 1 μM ( Table 3).
The binding model of compound 10 to Aurora B kinase (PDB: 2BFY) is mediated by two hydrogen bonds, as shown in Figure 8A. H-bond interactions can be seen between the hydrogen of the NH amide group and the main chain residue Glu-100 (distance: 2.49 Å). Another H-bond is formed by Tyr-102, which acts as a hydrogen bond acceptor, with the N-3 atom of the 4-methylimidazole ring (distance: 2.30 Å). The BRAF-binding domain in Figure 7B demonstrates that the amide group of adenine derivative 9 participates in two H-bonds with side chains residues Glu-61 and Asp-154 (distance: 1.90 and 2.14 Å, respectively). The adenine fragment does not participate in the formation of hydrogen bonds, unlike 2-fluoroanalog 8. Purine analog 9 showed good inhibitory activity against PDGFRα PDGFRβ receptor kinases, 45% and 36%, respectively, at a concentration of 1 µM ( Table 3).
The binding model of compound 10 to Aurora B kinase (PDB: 2BFY) is mediated by two hydrogen bonds, as shown in Figure 8A. H-bond interactions can be seen between the hydrogen of the NH amide group and the main chain residue Glu-100 (distance: 2.49 Å).
Another H-bond is formed by Tyr-102, which acts as a hydrogen bond acceptor, with the N-3 atom of the 4-methylimidazole ring (distance: 2.30 Å). In the case of 6-ОMе-2-Cl-purine derivative 14, a similar arrangement of the ligand in the adenine pocket of the Aurora-B kinase with the formation of two hydrogen bonds is observed ( Figure 8B). As in the case of compound 10, one H-bond is a donor-acceptor interaction between the hydrogen of the amide NH group and Glu-100 (distance: 1.86 Å). The N-3 atom of the 4-methylimidazole ring forms a hydrogen bond with Lys-26 (distance: 2.22 Å).
Obtained binding models, analysis of hydrogen bonds and binding energy calculations suggest that compounds 10 and 14 could bind to the active site of Aurora-B as type 1 inhibitors, despite the presence of a bulky N-(3-(4-methyl-1H-imidazol-1-yl)-5-(trifluoromethyl)phenyl) fragment.

Chemistry
Organic solvents were purified and dried by standard methods before usage in the synthesis. TLC analysis was carried out using TLC Silica gel 60 F254 (aluminium-backed sheets, Merck). The following solvent systems were used: CHCl3: MeOH, 7: 1, v/v or In the case of 6-OMe-2-Cl-purine derivative 14, a similar arrangement of the ligand in the adenine pocket of the Aurora-B kinase with the formation of two hydrogen bonds is observed ( Figure 8B). As in the case of compound 10, one H-bond is a donor-acceptor interaction between the hydrogen of the amide NH group and Glu-100 (distance: 1.86 Å). The N-3 atom of the 4-methylimidazole ring forms a hydrogen bond with Lys-26 (distance: 2.22 Å).
Obtained binding models, analysis of hydrogen bonds and binding energy calculations suggest that compounds 10 and 14 could bind to the active site of Aurora-B as type 1 inhibitors, despite the presence of a bulky N-(3-(4-methyl-1H-imidazol-1-yl)-5-(trifluoromethyl)phenyl) fragment.

Chemistry
Organic solvents were purified and dried by standard methods before usage in the synthesis. TLC analysis was carried out using TLC Silica gel 60 F 254 (aluminium-backed sheets, Merck). The following solvent systems were used: CHCl 3 : MeOH, 7:1, v/v or EtOAc, isocratic. Preparative column chromatography was performed on silica gel Merck 60 (70-230 mesh). NMR spectra were registered on a Bruker Avance 500 MHz spectrometer at 500 MHz ( 1 H), 125 MHz ( 13 C) and 470 MHz ( 19 F). Chemical shifts (δ) are given in ppm related to internal SiMe 4 and coupling constants (J) in Hz. The signals were designated as follows: s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet. High-resolution mass spectra (HRMS) were recorded on an Agilent 1290 Accurate-Mass 6500 Series Q-TOF using ESI (electrospray ionization). The purity of the synthesized compounds was determined by high-performance liquid chromatography (HPLC) (Waters, 996, Milford, MA, USA) with an EC 250/4,6 NUCLEODUR 100-5 C18ec column using KH 2 PO 4 (0.02 M, pH = 6.8) with a formic acid/acetonitrile 45:55 mobile phase (1.0 mL/min). Melting points (m.p.) were determined on an electrically heated melting point apparatus and were uncorrected. The synthesis of starting compounds 2-4 was carried out according to the previously described methods [28,29].
A solution of amine 3-(trifluoromethyl)aniline or 3-(4-methyl-1H-imidazol-1-yl)-5-(trifluoromethyl)aniline (one equivalent) and N,N-dimethylformamide (DMF, 1.1 equivalents) in dry CHCl 3 cooled to 0 • C was added dropwise to a solution of 4-(chloromethyl)benzoyl chloride (4) (one equivalent) in dry CHCl 3 . The reaction mixture was stirred at room temperature. The progress was monitored by TLC. Cold water was added to the reaction mixture. The organic layer was separated, and the water layer was extracted by CHCl 3 three times. The combined organic layers were dried over Na 2 SO 4 , filtered and concentrated under vacuum. The product was purified by column chromatography on silica gel. The fractions containing the products 5 or 6 were collected and evaporated to dryness under vacuum. The residue was crystallized.  2, 141.7, 139.4, 138.4, 135.5, 134.4, 129.4, 128.6, 115.4, 114.7, 112.2, 45.8, 14.0.

General Method for the Synthesis of Key Compounds 7-12
To a solution of purine base (c/d/e) (1.1 equivalents) in dry DMF was added K 2 CO 3 . To this mixture, chloride 5 or 6 (one equivalent) in DMF was added under stirring. Then, the reaction mixture was stirred at 80-90 • C under nitrogen. The progress of the reaction was monitored by TLC. Then, the reaction mixture was brought to room temperature and diluted with CHCl 3 . The organic phase was washed with water, dried over Na 2 SO 4 and evaporated to dryness. The product was purified by column chromatography on silica gel. The fractions containing the products 7-12 were collected and evaporated to dryness under vacuum. The residue was crystallized.
degrees of inhibition of tumor cells (%). The average of three parallel measurements was calculated for each concentration of test compound. The degree of cell growth suppression was plotted as a function of the logarithm of the concentration. The half maximal inhibitory concentration (IC 50 ) was determined graphically for active compounds.

Abl1 Kinase Inhibitory Assay
The Abl1 inhibitory activity assay was performed using Abl1 Kinase Enzyme System (Promega, Madison, WI, USA) and ADP-Glo™ Kinase assay kit (Promega, Madison, WI, USA) according to the Technical Manual. The general procedure was as follows: kinase (2.5 ng/reaction) was incubated with substrate (1 µg), compound (0.5 and 10µM) and ATP (50 µM) in a commercial buffer solution with a reaction volume of 5 µL. In every experiment, no-enzyme and no-compound control reactions were included to represent background luminescence (0% activity) and uninhibited kinase activity (100% activity), respectively. The assay 384-well plate was incubated at room temperature for 1 h. Afterwards, 5µL of ADP-Glo reagent was added into each well to stop the reaction and consume the remaining ADP within 40 min. At the end, 10 µL of kinase detection reagent was added into the well and incubated for 30 min to produce a luminescence signal. Kinase activity assays were performed in triplicate at each inhibitor concentration. The luminescence was measured with the help of a Tecan Infinite M200 Plate Reader. The luminescence data were analyzed using the computer software, Magellan™. Percent kinase activity (KA) was calculated by subtracting the average no-enzyme control luminescence values from all kinase-containing reactions with or without compound, then converting these net luminescence values to percent activity based on the no-compound control reactions representing 100% kinase activity. Inhibition (I, %) was calculated as 100%-KA.

Inhibitory Assay on the Panel of Tyrosine Kinases
The single-dose screening of kinase selectivity was performed using Kinase Selectivity Profiling System TK-1 (Promega, Madison, WI, USA) and ADP-Glo™ Kinase assay kit (Promega, Madison, WI, USA) according to Technical Manual. Kinase Selectivity Profiling System TK-1 includes a set of kinases from the tyrosine kinase family (EGFR, HER2, HER4, IGF1R, InsR, KDR, PDGFα, PDGFβ) in 8-tube strip format. The general procedure was as follow: kinase working stock (2 µL/reaction) was incubated with ATP/substrate working stock (1 µL) and compound (1 µM) in a commercial buffer solution with a reaction volume of 5 µL. The reaction conditions and inhibition (%) calculation were the same as in Section 3.3.1. Kinase activity assays were performed in duplicate at each inhibitor.

Cell Apoptosis Assay
Tumor cells were seeded on a 6-well plate at a density of 200.000 per well for HL-60 and 100.000 per well for OKP-GS. Test compounds 7 and 10 were added at concentrations of 5, 10, and 20 µM for 72 h. Then, the cells were washed from the culture medium by centrifugation at 1000 rpm for 5 min, washed again with PBS, then fixed with ice-cold ethanol during constant shaking and incubated at −20 • C for 24 h. After that, we washed it from alcohol by centrifugation, washed it again with PBS and added a solution containing 100 µg/mL RNase A and 10 µg/mL propidium iodide (PI), incubated for 40 min in the dark and measure it on a Cytomics FC500 Beckman Coulter flow cytometer (FL4 channel for PI). Received data were analyzed with the help of Kaluza 2.0 Software (Beckman Coulter, Brea, CA, USA).

In Vitro Cell Cycle Effects
Tumor cells were plated on a 6-well plate at a density of 100.000 per well for OKP-GS. Test compounds 7 and 10 were added at concentrations of 5, 10, and 20 µM for 72 h. Then, the cells were washed from the culture medium by centrifugation at 1000 rpm for 5 min, washed again with phosphate buffer, then the Annexin V-FITC and PI dyes from the Annexin 5A-FITC Kit (Beckman Coulter, Brea, CA, USA) were added, according to the manufacturer's instructions. Colored suspensions were incubated for 15 min in the dark on ice and measured on a Cytomics FC500 Beckman Coulter flow cytometer (channels FL1 for Annexin V-FITC and FL4 for PI). Data were analyzed with Kaluza 2.0 Software (Beckman Coulter, Brea, CA, USA).

Molecular Docking
AutoDock Vina was used for docking studies [24]. Two-dimensional structures were created using Marvin Sketch [33]. Three-dimensional structures were generated using molconvert [34]. Chemical file format conversions were performed with Open Babel [36]. Missing protein residues were restored with MODELLER [37]. Preparation of ligands and receptors for docking was carried out in MGL Tools [38]. Visualization of docking results, H-bond search and structure minimizations were made using Chimera [39]. Twodimensional interactions maps were generated by PoseView [49]. The most informative snapshots of molecular dynamics trajectory were used for H-bond location and interaction map creation.

Molecular Dynamics Simulations
Molecular dynamics were carried out using GROMACS package using AMBER FF99SB-ILDN force field. All ligands were parametrized with ACPYPE [38]. Simulation workflow was as follows: solvation, neutralizing and adding NaCl ions up to concentration of 0.15 mol, energy minimization, NPT and NVT equilibration steps of 200 ps each and a final 2 ns production run at 300 K. A dodecahedron box of 1.2 nm and periodic bounding conditions were used. A Berendsen thermostat was used for equilibration. Long electrostatic coupling was treated according to PME method. G_mmpbsa tool was used for MM-PBSA binding energy calculations [26]. To perform energy calculations, every twentieth frame of the final molecular dynamics trajectory was extracted, skipping the first 100 frames.

Conclusions
As part of our study, a series of new (arylaminomethyl) benzamide derivatives containing substituted purines as promising fragments to enforce the binding to a kinase's hinge region were developed and studied for their antitumor activity.
Compounds 7 and 10, containing chlorine atoms in the C-2 and C-6 positions of the purine heterocycle, demonstrated the highest activity against all studied cancer cell lines, comparable to the data for sorafenib. The highest activity was observed against the leukemic cell line K562 with IC 50 values of 2.27 µM and 2.53 µM, and IC 50 values for HL-60 equal to 1.42 and 1.52 µM, respectively. The absence of a 4-methyl-imidazole fragment in structures 7 and 13 did not have a noticeable effect on their antiproliferative activity in comparison with analogs 10 and 14. Several synthesized purine derivatives were also able to inhibit platelet-derived growth factor (PDGF) in vitro at a concentration of 1 µM with a best value of 45% for compound 9.
Compounds 7 and 10 were found to induce apoptosis induction in a dose-dependent manner and, at a concentration of 20 µM, the compounds exhibited a more than twofold higher level of apoptosis compared with sorafnib. In cell cycle progression studies, compounds 7 and 10 decreased the proliferation of OKP-GS cells. The effect was more significant than that of sorafenib at 20 µM. Compound 7 dose-dependently delayed cells in the G2/M phase.
The results of docking and molecular dynamics show that the presence of a 4methylimidazole fragment in the structure of titled compounds promotes their binding to protein kinases predominantly as type 2 inhibitors. However, in the case of Aurora kinases, one type of binding prevailed, regardless of the presence of a 4-methylimidazole fragment.
Thus, substituted purines in the form of a hinge-binding moiety in combination with a flexible linker, showing promising enzymatic inhibition as well as antiproliferative activity, can be considered a valuable starting point for further studies.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/ijms222312738/s1. Author Contributions: E.K. coordinated the project, was responsible for general supervision and participated in manuscript writing; A.F. performed the molecular modeling study, reviewed and edited the paper; T.B. performed part of the chemical synthesis and the kinase inhibitory assay, reviewed and edited the paper; A.P. participated in the cellular assay and interpretation of biological activities. All authors have read and agreed to the published version of the manuscript.