Synthesis of 2,4-Diaminopyrimidine Core-Based Derivatives and Biological Evaluation of Their Anti-Tubercular Activities

Tuberculosis (TB) is a chronic, potentially fatal disease caused by Mycobacterium tuberculosis (Mtb). The dihyrofolate reductase in Mtb (mt-DHFR) is believed to be an important drug target in anti-TB drug development. This enzyme contains a glycerol (GOL) binding site, which is assumed to be a useful site to improve the selectivity towards human dihyrofolate reductase (h-DHFR). There have been previous attempts to design drugs targeting the GOL binding site, but the designed compounds contain a hydrophilic group, which may prevent the compounds from crossing the cell wall of Mtb to function at the whole cell level. In the current study, we designed and synthesized a series of mt-DHFR inhibitors that contain a 2,4-diaminopyrimidine core with side chains to occupy the glycerol binding site with proper hydrophilicity for cell entry, and tested their anti-tubercular activity against Mtb H37Ra. Among them, compound 16l showed a good anti-TB activity (MIC = 6.25 μg/mL) with a significant selectivity against vero cells. In the molecular simulations performed to understand the binding poses of the compounds, it was noticed that only side chains of a certain size can occupy the glycerol binding site. In summary, the novel synthesized compounds with appropriate side chains, hydrophobicity and selectivity could be important lead compounds for future optimization towards the development of future anti-TB drugs that can be used as monotherapy or in combination with other anti-TB drugs or antibiotics. These compounds can also provide much information for further studies on mt-DHFR. However, the enzyme target of the compounds still needs to be confirmed by pure mt-DHFR binding assays.


Introduction
There is an urgent need to develop new drugs for the treatment of tuberculosis (TB), a chronic disabling infection caused by Mycobacterium tuberculosis (Mtb). This pathogen has developed resistance to standard first-and second-line anti-TB drugs, leaving very few options for effective therapy. para-Aminosalicylic acid (PAS) is a key anti-TB drug that has been in use for over 60 years.
Its anti-mycobacterial mechanism was not clearly understood until recently, when it was reported to be the pro-drug of an inhibitor of the Mtb dihyrofolate reductase (mt-DHFR) [1]. mt-DHFR catalyzes the reduction of dihydrofolate to tetrahydrolate in the folate metabolic pathway that leads to the synthesis of purines, pyrimidines and other proteins. Inhibition of the enzyme would cause cell death via the inhibition of DNA synthesis. This association of PAS with mt-DHFR inhibition encouraged scientists to focus once again on mt-DHFR as a potential target for anti-TB drugs.
According to their chemical structures, DHFR inhibitors can be divided into "classical" and "non-classical" types [2,3]. The structures of classical inhibitors are similar to that of folate, as in methotrexate (Figure 1), which is a commonly used anticancer drug [4][5][6]. The core structure of nonclassical inhibitors is 2,4-diaminopyrimidine, as in trimethoprim (Figure 1), an anti-bacterial drug [7][8][9]. The crystal structures of mt-DHFR (PDB ID: 1DF7, Figure 2) and human DHFR (h-DHFR, PDB ID: 1OHJ), show a glycerol (GOL) binding site in mt-DHFR that does not exist in h-DHFR. To take advantage of this difference, Threadgill evaluated a group of compounds containing glycerol-like side chains, among which one compound, El-7a ( Figure 1) showed notable selectivity for mt-DHFR inhibition over h-DHFR [10]. However, this evaluation was conducted using TB5 Saccharomyces cerevisiae carrying mt-DHFR and h-DHFR genes, therefore, there is no direct evidence to show that El-7a can inhibit the growth of Mtb. Furthermore, the inhibition of Mtb may require an appropriate lipophilicity in the compound [11]. Even if El-7a were able to selectively inhibit the mt-DHFR, its low hydrophobicity may prevent it from passing through the Mtb cell wall. Hence, we designed and synthesized a series of compounds containing more hydrophobic groups on the 6-position of 2,4diaminopyrimidine to evaluate the ability of these compounds to inhibit Mtb cells directly.   (GOL) in mt-DHFR, in which MTX is represented as a sticks model, GOL as a ball-stick model, and protein as a molecular surface; Right: the designed molecule is predicted to be able to occupy the GOL binding site, in which the molecule is represented as sticks and protein as a molecular surface. Its anti-mycobacterial mechanism was not clearly understood until recently, when it was reported to be the pro-drug of an inhibitor of the Mtb dihyrofolate reductase (mt-DHFR) [1]. mt-DHFR catalyzes the reduction of dihydrofolate to tetrahydrolate in the folate metabolic pathway that leads to the synthesis of purines, pyrimidines and other proteins. Inhibition of the enzyme would cause cell death via the inhibition of DNA synthesis. This association of PAS with mt-DHFR inhibition encouraged scientists to focus once again on mt-DHFR as a potential target for anti-TB drugs. According to their chemical structures, DHFR inhibitors can be divided into "classical" and "non-classical" types [2,3]. The structures of classical inhibitors are similar to that of folate, as in methotrexate (Figure 1), which is a commonly used anticancer drug [4][5][6]. The core structure of nonclassical inhibitors is 2,4-diaminopyrimidine, as in trimethoprim (Figure 1), an anti-bacterial drug [7][8][9]. The crystal structures of mt-DHFR (PDB ID: 1DF7, Figure 2) and human DHFR (h-DHFR, PDB ID: 1OHJ), show a glycerol (GOL) binding site in mt-DHFR that does not exist in h-DHFR. To take advantage of this difference, Threadgill evaluated a group of compounds containing glycerol-like side chains, among which one compound, El-7a ( Figure 1) showed notable selectivity for mt-DHFR inhibition over h-DHFR [10]. However, this evaluation was conducted using TB5 Saccharomyces cerevisiae carrying mt-DHFR and h-DHFR genes, therefore, there is no direct evidence to show that El-7a can inhibit the growth of Mtb. Furthermore, the inhibition of Mtb may require an appropriate lipophilicity in the compound [11]. Even if El-7a were able to selectively inhibit the mt-DHFR, its low hydrophobicity may prevent it from passing through the Mtb cell wall. Hence, we designed and synthesized a series of compounds containing more hydrophobic groups on the 6-position of 2,4diaminopyrimidine to evaluate the ability of these compounds to inhibit Mtb cells directly.   (GOL) in mt-DHFR, in which MTX is represented as a sticks model, GOL as a ball-stick model, and protein as a molecular surface; Right: the designed molecule is predicted to be able to occupy the GOL binding site, in which the molecule is represented as sticks and protein as a molecular surface.  (GOL) in mt-DHFR, in which MTX is represented as a sticks model, GOL as a ball-stick model, and protein as a molecular surface; Right: the designed molecule is predicted to be able to occupy the GOL binding site, in which the molecule is represented as sticks and protein as a molecular surface.

Chemistry
The synthesis of the designed compounds 10a-q or 11a-q was carried out in five steps-chlorination, nucleophilic substitution, iodination, Suzuki reaction and deprotection-according to our patented method [12], with 2,4-diamino-6-hydroxypyrimidine (1) as the starting material (Scheme 1). Initially, 2,4-diamino-6-chloropyrimidine (2) was generated from 1 by treatment with phosphorus oxychloride. After the reaction was quenched with ice water, the solution was hydrolyzed at 90 • C to obtain a good yield (85%) of the target intermediate [13]. This procedure yields pure 2 without the need for chromatography.

Determination of In Vitro Anti-Tubercular Activity
Based on the different R 2 substituents on the 2,4-diamino-5-aryl-6-substituted pyrimidine derivatives, the compounds can be divided into four types: (1) the R 2 substituents bearing hydroxy groups (10a-q, 11a-q); (2) the R 2 substituents bearing alkoxy groups (16a-g); (3) the R 2 substituents bearing thiazole groups (16h-l); (4) the R 2 substituents bearing phenyl substituted triazole groups (16m-p). Only compounds containing the thiazole group act as Mtb inhibitors. Among this group of compounds, five compounds (16h-l) showed potentially useful inhibitory effects, with 16l showing the lowest MIC (6.25 µg/mL or 12.45 µM) and MBC (12.5 µg/mL) ( Table 2). In order to see the selectivity of 16l against mammalian cells, the MTT assay was performed on vero cells, and the IC 50 on cells viability was found to be around 50.22 µM. The selectivity ratio of 16l on H37Ra vs vero cells is around 4-fold.

Determination of In Vitro Anti-Tubercular Activity
Based on the different R 2 substituents on the 2,4-diamino-5-aryl-6-substituted pyrimidine derivatives, the compounds can be divided into four types: (1) the R 2 substituents bearing hydroxy groups (10a-q, 11a-q); (2) the R 2 substituents bearing alkoxy groups (16a-g); (3) the R 2 substituents bearing thiazole groups (16h-l); (4) the R 2 substituents bearing phenyl substituted triazole groups (16m-p). Only compounds containing the thiazole group act as Mtb inhibitors. Among this group of compounds, five compounds (16h-l) showed potentially useful inhibitory effects, with 16l showing the lowest MIC (6.25 μg/mL or 12.45 μM) and MBC (12.5 μg/mL) ( Table 2). In order to see the selectivity of 16l against mammalian cells, the MTT assay was performed on vero cells, and the IC50 on cells viability was found to be around 50.22 μM. The selectivity ratio of 16l on H37Ra vs vero cells is around 4-fold.

Molecular Docking and Simulation
Through the structural analysis of the compounds, the Clog P of El-7a was noticed to be −0.17, which showed the compound El-7a to be very hydrophilic, and led us to assume it would not be able to cross the Mtb cell wall. This assumption was indirectly confirmed by the observation that the hydrophilic compounds 10a-q and 11a-q (analogs of El-7a, with Clog P around −1 to 2), could not inhibit the growth of Mtb. Based on the above assumption, more hydrophobic compounds were analyzed by using molecular docking and molecular dynamic simulations, and based on the size of the substituents on the 6-position of 2,4-diaminopyrimidine, they were divided into three groups, which are large side chain groups (compounds 16m-p), medium side chain groups (compounds 16h-l) and small side chain groups (compounds 16a-g). With molecular docking, it was noticed that the large side chain group, which contains the 1-benzyl-1H-1,2,3-triazole-4-methoxy group on the 6-position, cannot fit into the GOL binding site (Figure 3a), and this could be the reason why this group of compounds did not show any inhibition effects on Mtb. Although the small side chain group derivatives (compounds 16a-g), which contain the methoxyethoxy or methoxypropoxy group on the 6-positions, can fit into the GOL binding site (Figure 3b,c), they cannot form strong interactions or fully fill the GOL binding site.

Determination of In Vitro Anti-Tubercular Activity
Based on the different R 2 substituents on the 2,4-diamino-5-aryl-6-substituted pyrimidine derivatives, the compounds can be divided into four types: (1) the R 2 substituents bearing hydroxy groups (10a-q, 11a-q); (2) the R 2 substituents bearing alkoxy groups (16a-g); (3) the R 2 substituents bearing thiazole groups (16h-l); (4) the R 2 substituents bearing phenyl substituted triazole groups (16m-p). Only compounds containing the thiazole group act as Mtb inhibitors. Among this group of compounds, five compounds (16h-l) showed potentially useful inhibitory effects, with 16l showing the lowest MIC (6.25 μg/mL or 12.45 μM) and MBC (12.5 μg/mL) ( Table 2). In order to see the selectivity of 16l against mammalian cells, the MTT assay was performed on vero cells, and the IC50 on cells viability was found to be around 50.22 μM. The selectivity ratio of 16l on H37Ra vs vero cells is around 4-fold.

Molecular Docking and Simulation
Through the structural analysis of the compounds, the Clog P of El-7a was noticed to be −0.17, which showed the compound El-7a to be very hydrophilic, and led us to assume it would not be able to cross the Mtb cell wall. This assumption was indirectly confirmed by the observation that the hydrophilic compounds 10a-q and 11a-q (analogs of El-7a, with Clog P around −1 to 2), could not inhibit the growth of Mtb. Based on the above assumption, more hydrophobic compounds were analyzed by using molecular docking and molecular dynamic simulations, and based on the size of the substituents on the 6-position of 2,4-diaminopyrimidine, they were divided into three groups, which are large side chain groups (compounds 16m-p), medium side chain groups (compounds 16h-l) and small side chain groups (compounds 16a-g). With molecular docking, it was noticed that the large side chain group, which contains the 1-benzyl-1H-1,2,3-triazole-4-methoxy group on the 6-position, cannot fit into the GOL binding site (Figure 3a), and this could be the reason why this group of compounds did not show any inhibition effects on Mtb. Although the small side chain group derivatives (compounds 16a-g), which contain the methoxyethoxy or methoxypropoxy group on the 6-positions, can fit into the GOL binding site (Figure 3b,c), they cannot form strong interactions or fully fill the GOL binding site.
The molecular docking showed that the medium side chain group (compounds 16h-l), which contain a (thiazol-5-yl)methoxy on the 6-position, can fit into the GOL binding site properly, and a molecular dynamics simulation was performed to understand the binding of compound 16l to mt-DHFR. During 100 ns simulations, 16l was stable in the binding site, and the side chain of 6.25/12.5 Rifampicin 0.313/0.313

Molecular Docking and Simulation
Through the structural analysis of the compounds, the Clog P of El-7a was noticed to be −0.17, which showed the compound El-7a to be very hydrophilic, and led us to assume it would not be able to cross the Mtb cell wall. This assumption was indirectly confirmed by the observation that the hydrophilic compounds 10a-q and 11a-q (analogs of El-7a, with Clog P around −1 to 2), could not inhibit the growth of Mtb. Based on the above assumption, more hydrophobic compounds were analyzed by using molecular docking and molecular dynamic simulations, and based on the size of the substituents on the 6-position of 2,4-diaminopyrimidine, they were divided into three groups, which are large side chain groups (compounds 16m-p), medium side chain groups (compounds 16h-l) and small side chain groups (compounds 16a-g). With molecular docking, it was noticed that the large side chain group, which contains the 1-benzyl-1H-1,2,3-triazole-4-methoxy group on the 6-position, cannot fit into the GOL binding site (Figure 3a), and this could be the reason why this group of compounds did not show any inhibition effects on Mtb. Although the small side chain group derivatives (compounds 16a-g), which contain the methoxyethoxy or methoxypropoxy group on the 6-positions, can fit into the GOL binding site (Figure 3b,c), they cannot form strong interactions or fully fill the GOL binding site. calculation showed that the binding free energy was −3.47 Kcal/mol (Table 3), which indicated that 16l can bind with mt-DHFR tightly. The free energy contributions of each residue was calculated, and contributions greater than −0.5 Kcal/mol were recorded (Ile20, Arg23, Phe31, Leu50, Pro51 and Val54) (Figure 4 Left). Most of these residues showed a strong VDW interaction, except Phe31 and Arg23 which formed a H-bond with 16l. Arg23 could also form strong interactions with the trifluoromethoxy group (Figure 4 Right). Therefore, through the molecular docking and molecular dynamic simulations, we believe that the compounds containing the (thiazol-5-yl)methoxy side chain (medium size group), can fully occupy the GOL binding site, and have reasonable properties. Therefore, such compounds could be used as the lead compounds for further anti-TB drug discovery studies.  The molecular docking showed that the medium side chain group (compounds 16h-l), which contain a (thiazol-5-yl)methoxy on the 6-position, can fit into the GOL binding site properly, and a molecular dynamics simulation was performed to understand the binding of compound 16l to mt-DHFR. During 100 ns simulations, 16l was stable in the binding site, and the side chain of compounds 16h-l occupied the GOL binding site along the full simulation ( Figure 2). The free energy calculation showed that the binding free energy was −3.47 Kcal/mol (Table 3), which indicated that 16l can bind with mt-DHFR tightly. The free energy contributions of each residue was calculated, and contributions greater than −0.5 Kcal/mol were recorded (Ile20, Arg23, Phe31, Leu50, Pro51 and Val54) (Figure 4 Left). Most of these residues showed a strong VDW interaction, except Phe31 and Arg23 which formed a H-bond with 16l. Arg23 could also form strong interactions with the trifluoromethoxy group (Figure 4 Right).  The residues whose binding free energy contributions are greater than −0.5 Kcal/mol; Right: The interactions between key residues and compound 16l.

General Information
All reagents and solvents were purchased from the suppliers and used directly in the experiments. THF was dried by distillation ver sodium benzophenone. TLC was carried out using silica gel 60 precoated aluminium plates (0.20 mm thickness) from Macherey-Nagel (Darmstadt, Germany) with visualisation by UV light (254 nm). Flash chromatography was performed on silica gel (particle size 40-63 μm). IR spectra were recorded on a Tensor 27 spectrometer (Bruker, Ettlingen, Germany) using KBr discs. 1 H-NMR spectra were obtained from an AVANCE III 400 spectrometer (Bruker, Fällanden Switzerland). The chemical shifts, given as δ values, were quoted in parts per million (ppm); 1 H-NMR chemical shifts were measured relative to internal tetramethylsilane; Apparent coupling constants (absolute values), J, were measured in Hertz and multiplicities quoted as singlet (s), doublet (d), triplet (t), quartet (q) or combinations thereof as appropriate. Mass spectra were obtained from an 6545 Accurate-Mass Q-TOF LC/MS (Agilent Technologies, Santa Clara, CA, USA). Melting points were determined using a WRS-1B melting point measurement instrument (Shanghai, China) and were uncorrected. 2,. 2,4-Diamino-6-hydroxypyrimidine (1) (1.00 g, 7.93 mmol) was added to POCl3 (9 mL), and stirred at 97 °C for 17 h. The reaction solution was added to ice water slowly, and then stirred at 90 °C for 1 h. The pH of this solution was adjusted to 8 with NaOH, and then it was extracted with EtOAC (150 mL × 3). The combined organic layers were dried with Na2SO4, filtered and concentrated to give white solid 0.97 g, yield 85%. m.p. 200. 2-200.4  Under argon, to a solution of (S)-2,3-isopropylideneglycerol or (R)-2,3-isopropylideneglycerol 0.50 mL (4.0 mmol) in dry DMSO (5 mL) was added NaH 0.20 g (60%, 5.0 mmol) and stirred at room temperature for 1 h. 2,4-Diamino-6-chloropyrimidine (2, 0.29 g, 2.0 mmol) was added and stirred at 90 °C for 8 h. The reaction solution was quenched with sat NH4Cl (20 mL) and extracted with EtOAc (30 mL × 3), and the combined organic layers dried with Na2SO4, filtered and concentrated. The residue was purified by column chromatography on silica gel using CH2Cl2/CH3OH (50:1, v/v) as the eluting solvent to give compounds 3 or 4. The residues whose binding free energy contributions are greater than −0.5 Kcal/mol; Right: The interactions between key residues and compound 16l.

ChemistryIt
Therefore, through the molecular docking and molecular dynamic simulations, we believe that the compounds containing the (thiazol-5-yl)methoxy side chain (medium size group), can fully occupy the GOL binding site, and have reasonable properties. Therefore, such compounds could be used as the lead compounds for further anti-TB drug discovery studies.

General Information
All reagents and solvents were purchased from the suppliers and used directly in the experiments. THF was dried by distillation ver sodium benzophenone. TLC was carried out using silica gel 60 pre-coated aluminium plates (0.20 mm thickness) from Macherey-Nagel (Darmstadt, Germany) with visualisation by UV light (254 nm). Flash chromatography was performed on silica gel (particle size 40-63 µm). IR spectra were recorded on a Tensor 27 spectrometer (Bruker, Ettlingen, Germany) using KBr discs. 1 H-NMR spectra were obtained from an AVANCE III 400 spectrometer (Bruker, Fällanden Switzerland). The chemical shifts, given as δ values, were quoted in parts per million (ppm); 1 H-NMR chemical shifts were measured relative to internal tetramethylsilane; Apparent coupling constants (absolute values), J, were measured in Hertz and multiplicities quoted as singlet (s), doublet (d), triplet (t), quartet (q) or combinations thereof as appropriate. Mass spectra were obtained from an 6545 Accurate-Mass Q-TOF LC/MS (Agilent Technologies, Santa Clara, CA, USA). Melting points were determined using a WRS-1B melting point measurement instrument (Shanghai, China) and were uncorrected.
(B) In a pressure tube, to a mixed solution of CH 3 CN/H 2 O (1:1, 40 mL) was added compound 5 or 6 (2.73 mmol), substituted phenylboronic acid 7p or 7q (4.10 mmol), Pd(dbpf)Cl 2 (2.73 × 10 −4 mmol) and K 2 CO 3 (4.10 mmol) consecutively and then stirred at 60 • C for 8 h. The reaction solution was extracted with EtOAc (30 mL × 3), and the combined organic layers were washed by H 2 O and dried with Na 2 SO 4 , filtered and concentrated. The residue was purified by column chromatography on silica gel using CH 2 Cl 2 /CH 3 OH (60:1, v/v) as the eluting solvent to the desired compounds. (R)