Biological features, drug-likeness, pharmacokinetic properties, and docking of 2-arylidenehydrazinyl-4-arylthiazole analogues

Thiazoles are an important class of heterocyclic compounds that possess a sulfur and nitrogen containing five-membered ring, which acts as a pharmacophore, and show a wide range of complex biological activities. A series of sixteen 2-arylidenehydrazinyl-4-arylthiazole analogues (3a–p) were evaluated for cytotoxic activity against brine shrimp (Artemia salina) nauplii and their minimum inhibitory concentrations were determined against two Gram-positive (Listeriamonocytogenes and Enterococcus faecalis) and two Gram-negative bacterial strains (C.sakazakii and E. coli). Of the tested compounds, 3g demonstrated highest cytotoxicity with a LC50 value of 54 ppm followed by compound 3h(LC50 = 85 ppm), in a short-term bioassay using A. salina, whereas compound 3i exhibited the most potent antibacterial activities against L.monocytogenes, E. faecalis, and C.sakazakii with MIC values ranging from 50 to 100 μg mL−1. Compound 3g showed highest antibacterial activity against E. coli (MIC = 50 μg mL−1). In silico drug-likeness, pharmacokinetic (ADME) properties, toxicity effects, and drug scores were also evaluated, and none of the sixteen compounds were found to violate Lipiniski’s rule of five or Veber’s rule, indicating potential for development as oral drug candidates. In addition, a docking study of compound 3i into the active site of E. coli FabH receptor, an attractive target for the development of new antibacterial agents, showed it has good binding properties.

Due to the emergence of bacterial resistance to almost all commercial antibiotics, the treatment of infectious disease has become a challenging problem in hospitals and communities, and thus, the identification of new antibacterial agents with novel targets is of considerable importance. In particular, fatty acid biosynthesis (FAB) is a promising target for new antibacterial agents (Price et al. 2001), because it is essentially required for bacterial cell viability and growth. b-Ketoacyl-acyl carrier protein (ACP) synthase III, which is also known as FabH or KAS III, plays an important role in bacterial FAB (Khandekar et al. 2003), and thus, bacterial FabH is an attractive target for the design of new antibiotics. A literature survey revealed some thiazoles that potently inhibit FabH possess an amide (-NHCO-) or azomethine amine (-C=N-NH-) moiety at the 2-position of the thiazole ring exhibit antibacterial activity (Lv et al. 2009;Sarojini et al. 2010;Lee et al. 2012;Yang et al. 2012;Li et al. 2014). Furthermore, these structural motifs are also present in many pharmaceutically active compounds. Figure 1 shows structures of several reported antibacterial FabH inhibitors that possess a thiazole scaffold. In our previous study, we reported the synthesis and antimicrobial and antioxidant activities of a series of 2-arylidenehydrazinyl-4-arylthiazole analogues (Alam et al. 2014a). In the present study, we describe the in vivo cytotoxic activities and minimum inhibitory concentration (MIC) values of sixteen 2-arylidenehydrazinyl-4-arylthiazole analogues (3a-p) against two Gram-positive (Listeria monocytogenes and Enterococcus faecalis) and two Gram-negative bacterial strains (Cronobacter sakazakii and E. coli). The synthesis of compounds 3a-p have been demonstrated in our previous study (Alam et al. 2014a); however, in vivo cytotoxic actions and MIC values of compounds 3a-p, have not yet been reported. In addition, an in silico study was performed to predict the drug-likeness, pharmacokinetic properties, e.g., absorption, distribution, metabolism, and elimination (ADME), MTIR (mutagenic, tumorigenic, irritant, and reproductive) toxicity profiles, and drug scores in an effort to identify molecular features responsible for the cytotoxic properties of these compounds. Furthermore, docking simulations were performed using the X-ray crystallography determined structure of E. coli FabH to investigate binding modes at its active site. Molecular docking is a computer-assisted drug design method that is usually used to predict the favorable binding conformations of ligands at active sites of target receptors. It also provides information regarding the natures of bonding interactions, which contribute to binding affinities between ligands and receptors  (Sarojini et al. 2010). Therefore, molecular docking is considered an important in silico tool for pharmaceutical lead discovery (Shoichet et al. 2002).

Synthesis
The 2-arylidenehydrazinyl-4-arylthiazole analogues (3ap) used in the present study were prepared as we previously described (Alam et al. 2014a) from their corresponding arylidenethiosemicarbazones (2a-j) as presented in Scheme 1. Briefly, to a stirred solution of thiosemicarbazide (1 mmol) in an ethanol-water mixture (1:1), an ethanolic solution of a substituted benzaldehyde (1 mmol) was added slowly and refluxed for 10-20 min. After cooling the reaction mixture to ambient temperature, the mixtures obtained were filtered to provide solid crude products, which were crystallized from ethanol to furnish pure compounds 2a-j at yields of 83-94 %. The arylidenethiosemicarbazones (1 mmol) so obtained were refluxed with 2-bromoacetophenone or 2,4 0 -dibromoacetophenone (1 mmol) in ethanol for 30-60 min and then cooled to ambient temperature. The resulting precipitates were filtered and washed with water to give crude products, which were purified by crystallization from DMF-EtOH (1:1) to afford the pure 1,3-thiazole derivatives (3a-p) at yields of 72-95 %.

Cytotoxicity assay
Brine shrimp nauplii (A. salina) were used for the in vivo cytotoxicity assay, which was performed as described by Mayer et al. (1982), with some modifications. Briefly, brine shrimp nauplii were hatched in a small tank containing artificial seawater (3.8 % NaCl). The tank was partially exposed to incandescent light to attract the nauplii. The assay was performed 24 h after hatching; no food was provided during hatching or the experimental periods. Test samples (3 mg) were dissolved in 0.6 mL of DMSO to obtain stock solutions of 5 mg mL -1 . Different concentrations of test samples (5 mL made up artificial seawater) were then placed in separate vials. Twenty brine shrimp nauplii were then placed in each vial. The negative control was prepared in the same manner, except the sample was omitted. Gallic acid was used as the positive control. After for incubation 24 h, the vials were observed using a magnifying glass and the numbers of survivors were counted. Tests were performed in triplicate, and LC 50 values in ppm were obtained from resulting data using Probit analysis software (Finney 1978).

Docking studies
The molecular geometries of compounds 3c, 3i, and 3j were investigated using standard bond lengths and angles using the ChemBio3D Ultra 14.0 molecular modeling program (CambridgeSoft Corporation, Cambridge, MA 02140 USA). The molecular energies were minimized using a semi-empirical molecular orbital AM1 method (Dewar et al. 1985) using GAMESS Interface in the ChemBio3D Ultra Ver. 14.0. The crystal structure of E. coli FabH-CoA complex was obtained from Protein Data Bank (PDB code: 1HNJ) for docking studies. All bound water and co-crystallized ligand were eliminated and polar hydrogen atoms and Kollman-united charges were added to molecules. The pdb and pdbqt files of ligands and the receptor were prepared using AutoDock 4.2 software. Free rotation was allowed about single bonds during docking. Docking was conducted using the standard protocol implemented by AutoDock Vina in PyRx 0.8 software and the geometries of resulting complexes were investigated using Discovery Studio 4.0 (Accelrys, Inc. San Diego, CA 92121 USA) and Molegro Molecular viewer 2.5 (CLC bio company, Waltham, MA 02451 USA).

Molecular descriptors and field points
Calculations of geometrical descriptors and field point maps of compounds 3c, 3i, and 3j were performed to investigate molecular surface properties using ChemAxon (ChemAxon LLC, Cambridge, MA 02142 USA) and Tor-chLite software (Cresset BioMolecular Discovery Ltd., Cambridgeshire, SG8 0SS, UK). Representative field points are presented using colors and sizes, and larger field points indicate stronger potential interactions.

Cytotoxic activities
The in vivo cytotoxic activities of compounds (3a-p) were screened using an Artemia salina (brine shrimp) lethality bioassay as previously described (Alam et al. 2014a, b). This bioassay is an excellent tool for the preliminary screening of bioactive compounds (Mayer et al. 1982;Hartl and Humpf 2000). Lethal doses (LC 50 ) obtained using the brine shrimp assay can be used to determine more specific activities (Weidenbörner and Chandra Jha 1993;Alam et al. 2014a). The results obtained for 2-arylidenehydrazinyl-4-arylthiazole analogues (3a-p) are presented in Table 1. Gallic acid was also tested under identical conditions as a positive control. Of the tested compounds, 3g showed greatest brine shrimp toxicity with a LC 50 of 54 ppm, followed by compounds 3h (LC 50 = 85 ppm) and 3e (LC 50 = 113 ppm). However, all compounds were less cytotoxic than gallic acid (LC 50 = 12 ppm). Compounds 3f, 3k, and 3l showed moderate cytotoxicity with LC 50 values of 319, 522, and 848 ppm, respectively, whereas compounds 3a and 3p exhibited low activities with LC 50 values of 1933 and 1794 ppm, respectively. On the other hand, compounds 3b-3d, 3m, and 3n demonstrated similar weak activities (LC 50 [ 2500 ppm) and compounds 3i, 3j, and 3o were not cytotoxic to A. salina.
In most compounds, the substituent types and positions play significant role to exert cytotoxic activity. Structureactivity relationships (SAR) may be explained briefly as follows: introduction of a moderate electron donating group (OMe) at the R 1 and R 3 or R 2 and R 3 positions favored activity (e.g., 3 g or 3e), while the presence of a more polar group (OH) at the same positions caused reduced activity (e.g., 3a and 3c). Substitution of a bromine atom (a bulky and moderate ring deactivating group) at R 5 position greatly decreased cytotoxicity (e.g., 3 g [ 3 h or 3e [ 3f).

Antibacterial activities
The MICs (minimum inhibitory concentrations) of compounds 3a-p were evaluated using two gram-positive bacterial strains, L. monocytogenes ATCC 43256 and E. faecalis CARS 2011-012, and two gram-negative bacterial strains, C. sakazakii CARS 2012-J-F and E. coli CARS 2011-016 using the serial dilution method. MIC values are presented in Table 2. The antibacterial nalidixic acid was used as a positive control. All synthesized compounds showed antibacterial activity and MIC values ranged from 50 to 300 lg mL -1 . Compounds 3c, 3g, 3i, 3j, and 3m exhibited significant antibacterial activity. Summarizing, compounds 3c, 3i, and 3j had lowest MIC values (*50 lg mL -1 ) against L. monocytogenes, whereas 3i and 3 m had similar MIC values (50 lg mL -1 ) against E. faecalis. Compounds 3b, 3c, and 3i had the lowest MIC values (*100 lg mL -1 ) against C. sakazakii, and compound 3 g had the lowest MIC value (50 lg mL -1 ) against E. coli. However, nalidixic acid had the strongest effect with MIC values ranging from 12.5 to 25 lg mL -1 .
The results presented in Table 2 indicate that structureantimicrobial activity relationships did not appear to play a significant role as all compounds exhibited similar MIC values ranging from 100 to 200 lg mL -1 against all bacterial strains with some exceptions of compounds 3c, 3g, 3i, 3j, and 3 m against some selected bacterial strains (MIC = 50 lg mL -1 ). This result led us to speculate that the mode of action of these compounds is microorganism dependent. However, the presence of an electron donating group (OH or OMe) at the ortho-and para-positions favored activity, e.g., 3c and 3i against L. monocytogenes, 3g against E. coli, 3i and 3m against E. faecalis, while the presence of these groups at the meta-position greatly reduced activity (e.g., 3a and 3k). Introduction of a  bromine atom (a bulky as well as moderate ring deactivating group) at R 5 position greatly reduced antibacterial efficacy (e.g., 3i [ 3j or 3c [ 3d), which would be due to the size effect or steric effect.

Pharmacokinetic and drug-likeness properties
Absorption, distribution, metabolism, and elimination (ADME) are crucial potential drug parameters. To assess drug-likeness of a compound, Lipiniski's rule of five (ROF) and Veber's rule are usually used. These rules consider the following parameters: topological polar surface area (TPSA), octanol-water partition coefficient (logP), NROTB, HBD and acceptors (HBA), and molecular weight (MW). logP and PSA have been shown to be excellent descriptors of drug absorption (including intestinal absorption), bioavailability, Caco-2 permeability, and blood-brain barrier penetration. Therefore, we studied the synthesized compounds (3a-p) in silico to assess ADME and drug-likeness properties and then compared these with nalidixic acid and gallic acid controls. The descriptors of Lipinski's ROF and Veber's rule (TPSA, logP, NROTB, HBD, HBA, and MW) were calculated using the Molinspiration online property toolkit (Lipinski et al. 2001), and results obtained are presented in Table 3. TPSA was calculated using the method developed by Ertl et al. (2000), which considers the sum of fragment contributions, and Oand N-centered polar fragments. The results obtained showed that none of the sixteen synthesized compounds violated Lipiniski's or Veber's rule, suggesting good druglike properties. The four major developmental criteria of an orally active drug candidate are logP B 5, MW B 500, HBA B 10, and HBD B 5 (Ertl et al. 2000). The calculated logP and logS values of the of the synthesized compounds were within the ranges of 3.198 to 5.44 and -4.07 to -6.23, respectively; these values are -2.0 to 6.5 and 0.5  (Zhang and Wilkinson 2007). Cytochrome P450 is an important enzyme and is responsible for many ADME problems. Its inhibition or the generation of redundant metabolites by cytochrome P450 can result in many adverse drug side effects. The prediction of ADME-toxicity is important prior to structural drug design and many potential drugs fail to reach the clinical stage because of ADME-Tox issues. Hence, we examined the mutagenicities, tumorigenicities, irritancies, and reproductive toxicities and drug scores (Sadowski and Kubinyi 1998) of the sixteen synthesized compounds and compared them with those of the nalidixic and gallic acid standards using the Osiris program. Results are presented in Fig. 2. Osiris predicted that none of the sixteen compounds would have irritant, mutagenic, or reproductive toxicity, but all were predicted to have high tumorigenic toxicity. Interestingly, nalidixic acid was predicted to have high mutagenic, tumorigenic, and reproductive effects, whereas gallic acid was predicted to have high mutagenic and reproductive effects. Most of the sixteen compounds (except compounds 3h, 3o, and 3p) had a higher drug  scores than nalidixic acid, whereas only compounds 3a, 3c, and 3i had a higher drug score than gallic acid. The above in silico ADME results suggest 2-arylidenehydrazinyl-4arylthiazole analogues (3a-p) have good drug-likeness properties and that they represent pharmacologically active base worthy of further study.

Molecular docking studies
In order to predict whether the 2-arylidenehydrazinyl-4arylthiazole analogues could bind to the FabH receptor to exert their antibacterial activity, the most active (3c, 3i, and 3j) compounds were docked into the active site of E. coli FabH (PDB ID: 1HNJ) using AutoDock tools 1.5.6 (Morris et al. 2009) and AutoDock Vina in PyRx 0.8 software (Trott and Olson 2010). E. coli b-ketoacyl-(acyl carrier protein) synthase III, also known as ecKAS III or FabH, plays an essential, regulatory role in the bacterial fatty acid synthesis. The enzyme initiates fatty acid elongation cycles and is involved in the feedback regulation of the biosynthetic pathway (Heath and Rock 1996). Furthermore, the FabH proteins of Gram-positive and Gram-negative bacteria are highly conserved at the sequence and structural level (there are no significant homologous proteins in man).
In addition, the residues that comprise the active site are essentially invariant among Gram-positive and Gram-negative organisms (Nie et al. 2005). Therefore, FabH represents a promising target for the design of novel, selective, nontoxic, broad-spectrum antimicrobial drugs. The crystal structure of E. coli FabH-CoA complex was retrieved from the Protein Data Bank (PDB ID: 1HNJ) and molecular docking of compounds 3c, 3i, and 3j were performed at its active site. The results obtained showed that compounds 3c, 3i, and 3j bind to the same active site of FabH receptor as the endogenous ligand (malonyl CoA) and other reported thiazole FabH inhibitors (Lv et al. 2009;Cheng et al. 2013;Li et al. 2014). The binding model and the different types of interactions found for the most active compound 3i are shown in Fig. 3, and the docking models for compounds 3c and 3j are provided as supporting data (Supplementary Figs. 1, 2, respectively). Since docking scores provide a measure of ligand to receptor binding affinity, we calculated binding scores for compounds 3c, 3i, and 3j using iGEMDOCK software (Yang and Chen 2004) (results are presented in Fig. 4). The binding affinities of compounds 3c, 3i, and 3j with FabH receptor were found to be -69.443, -104.908, and -75.0625 kcal mol -1 with van der Waals contributions of -54.8878, -91.1249, and -52.3552 kcal mol -1 , respectively, and H-bond contributions of -14.5552, -13.7835, and -22.7073 kcal mol -1 ,  groups of the benzylidene phenyl ring. In addition to hydrogen bonding interactions, compounds 3c and 3i exhibited p-sulfur interactions with MET207 (4.10 and 3.89 Å , respectively) amino acid residues, and 3c also showed p-pi interactions with the TRP32 (5.17 Å ) amino acid residue. In addition, p-alkyl interactions with ILE156 (5.38), VAL212 (4.68), and ALA246 (3.98) for 3c, ARG36 (5.21), ILE156 (5.26), VAL212 (4.78), ALA216 (5.21), and ALA246 (4.01 and 4.65) for 3i, and ARG151 (8.82), ILE156 (5.49), MET207 (5.09), and ALA246 (4.26) for 3j were found to involve thiazole, phenyl-, or benzylidenephenyl rings. The following van der Waals' interactions were also observed: between 3c and TRP32, CYS112, ILE250, ASN210, PHE213, ASN274, PHE157, LEU189, PHE304, and GLY305; between 3i and TRP32, CYS112, PHE157, LEU189, ILE250, ASN274, PHE304, GLY305, and GLY306; and between 3j and TRP32, THR37, GLY152, PHE157, LEU189, ILE250, VAL212, and HIS244. The above results indicate compounds 3c, 3i, and 3j bind to the same active site of FabH receptor and suggest that the higher binding affinity of 3i might is caused by strong van der Waals' interactions with amino acid residues at the active site of FabH. The energy maps of receptor ligand binding sites are useful for understanding the different types of molecular interactions in favorable regions. Molecular docking studies are also used to identify pharmacophore, that is, common chemical frameworks responsible for a drug's biological activity. Therefore, the energy maps and pharmacophore models of compounds 3c, 3i, and 3j were also studied at the FabH receptor binding site using Molegro Molecular viewer 2.5 and LigandScout 3.12 software, respectively. Energy maps and the pharmacophore model of compound 3i are shown in Fig. 5, and energy maps for compounds 3c and 3j are provided as supporting data (Supplementary Figs. 3,4,respectively). Referring to Fig. 5A, green regions are energetically favorable for steric interactions, whereas red and blue color regions are energetically favorable for negative-and positive-electrostatic interactions, respectively, in the 3i-FabH receptor complex. Yellow and cyan indicate regions where H-bond donor and acceptor interactions are energetically favorable, respectively, at the 3i binding site in FabH receptor (Fig. 5B). Because compounds 3c, 3i, and 3j bind to the same site, their pharmacophore models were similar (c.f. Supplementary Figs. 3C, 4C, 5C, respectively).

Field points and molecular descriptors
Field points consist of electrostatic, van der Waals, and hydrophobic potentials and molecular descriptors, for example, quantum-chemical, grid-based, volume, and surface descriptors, of a molecule that play significant roles in its chemical and biological activities. The field points of a molecule provide more information on pharmacophore and enable the position and potentials of electrostatic, van der Waals, and hydrophobic interaction points to be directly measured from the physical properties of a molecule. Molecular field patterns, that is, the electrostatic (positive and negative), steric, and hydrophobic properties, of compound 3i are presented in Fig. 6, and those of compounds 3c and 3j are provided as supporting data (Supplementary Figs. 5, 6, respectively). These field point patterns are used to express ligand-receptor interactions simultaneously. The nature of bonding and the abilities of ligands to interact with receptors are categorized using the sizes of field points and larger field points are indicative of strong potential interactions. In addition, geometrical 3D descriptors of the surface interactions of compounds 3c, 3i, and 3j were also analyzed and are presented in Fig. 7 showing their three-dimensional representation in relation to the nature and connectivity of the atoms as well as the overall spatial configurations. The geometrical 3D descriptors are usually use to effective search for the relationships between the molecular structure and biological activities of a molecule. The geometrical 3D descriptors of compounds 3c, 3i, and 3j are as follows: the dreiding energy, which is the energy related to the 3D structure of the molecule using the dreiding force field (Mayo et al. 1990 In conclusion, a series of sixteen 2-arylidenehydrazinyl-4-arylthiazole analogues were evaluated for in vivo cytotoxicity using the brine shrimp (A. salina) and for MICs against two Gram-positive bacterial strains (L. monocytogenes ATCC 43256 and E. faecalis CARS 2011-012) and two Gram-negative bacterial strains (C. sakazakii CARS 2012-J-F and E. coli CARS 2011-016). The data obtained indicated that compound 3g with two methoxyl substituents at ortho-and para-positions of the benzylidene phenyl ring was most cytotoxic to A. salina (LC 50 54 ppm) and that compound 3h with an additional bromo substituent at the para-position of phenyl ring was the second-most cytotoxic (LC 50 = 85 ppm). Compound 3i with three hydroxyl substituents at para-and ortho-positions of the benzylidene phenyl ring showed potent antibacterial activity against L. monocytogenes, E. faecalis, and C. sakazakii (MIC 50-100 lg mL -1 ), and compound 3g exhibited highest antibacterial activity against E. coli (MIC 50 lg mL -1 ). Structure-activity relationships (SAR) studies revealed that moderate electron donating groups (OMe) in the benzylidene phenyl ring are favorable compare to that of more polar groups (OH) to exert brine shrimp cytotoxicity. While structure-antimicrobial activity relationships did not appear to play a significant role. However, the presence of a bromine atom (a bulky and moderate ring deactivating group) at R 5 position greatly reduced both the cytotoxicity and antibacterial activity. Molecular docking studies were performed to investigate interactions between compounds 3c, 3i, and 3j and the active site of E. coli FabH receptor. Compound 3i was found to effectively bind to the active site with a high docking score, indicating it could act as a FabH inhibitor. None of the sixteen analogues violated Lipniski's ROF or Veber's rule, and thus, they exhibited good drug-likenesses with high drug scores. Additionally, the in silico determined ADME properties of 2-arylidenehydrazinyl-4arylthiazole analogues suggested they be viewed as potential oral drug candidates. The details of the ligandreceptor and surface interactions of compounds 3c, 3i, and 3j were also analyzed. Accordingly, we conclude that compounds with thiazole frameworks containing the phenyl azomethine amine (Ar-CH=N-NH-) moiety should be considered a basis for the design and development of a new antibacterial FabH inhibitor.