Carbazole-based semicarbazones and hydrazones as multifunctional anti-Alzheimer agents

Abstract With the aim to combat a multi-faceted neurodegenerative Alzheimer’s disease (AD), a series of carbazole-based semicarbazide and hydrazide derivatives were designed, synthesized and assessed for their cholinesterase (ChE) inhibitory, antioxidant and biometal chelating activity. Among them, (E)-2-((9-ethyl-9H-carbazol-3-yl)methylene)-N-(pyridin-2-yl)hydrazinecarbothioamide (62) and (E)-2-((9-ethyl-9H-carbazol-3-yl)methylene)-N-(5-chloropyridin-2-yl)hydrazinecarbothioamide (63) emerged as the premier candidates with good ChE inhibitory activities (IC50 values of 1.37 µM and 1.18 µM for hAChE, IC50 values of 2.69 µM and 3.31 µM for EqBuChE, respectively). All the test compounds displayed excellent antioxidant activity (reduction percentage of DPPH values for compounds (62) and (63) were 85.67% and 84.49%, respectively at 100 µM concentration). Compounds (62) and (63) conferred specific copper ion chelating property in metal chelation study. Molecular docking studies of compounds (62) and (63) indicate strong interactions within the active sites of both the ChE enzymes. Besides that, these compounds also exhibited significant in silico drug-like pharmacokinetic properties. Thus, taken together, they can serve as a starting point in the designing of multifunctional ligands in pursuit of potential anti-AD agents that might further prevent the progression of ADs. Communicated by Ramaswamy H. Sarma


Introduction
Alzheimer's disease (AD) is a neurodegenerative disease, causing severe neuronal disruption in the hippocampus and cerebral cortex regions of the brain. This is recognized by impairment of the cognitive functions, usually leading to deterioration in emotional control, social behavior or motivation (Goedert & Spillantini, 2006). Globally, over 50 million people are currently affected by AD and the toll is expected to rise upto 150 million by 2050 World Alzheimer Report, 2019).
The neuropathological hallmarks of the AD brain are low levels of acetylcholine (ACh) (Talesa, 2001), diffused extracellular amyloid plaques caused by self or biometal-mediated b-amyloid (Ab) aggregation, which are frequently surrounded by damaged neurons (Selkoe, 2003), and hyperphosphorylated intracellular neurofibrillary tangles (Maccioni et al., 2010). These well-recognized pathological hallmarks are often accompanied by high levels of oxidative stress generated by reactive oxygen species (ROS) and biometal dyshomeostasis (Greenough et al., 2013). Therefore, to combat a multifactorial disease like AD, development of multi-target-directed ligands (MTDLs) is considered a promising drug discovery approach (Bajda et al., 2011;Oset-Gasque & Marco-Contelles, 2018;Patel et al., 2019;Wang et al., 2019). Currently, there are only four FDA-approved medications available to treat AD, of which three are acetylcholinesterase (AChE) inhibitors (donepezil, galantamine and rivastigmine) and the fourth one is N-methyl-D-aspartate (NMDA) receptor antagonist (memantine). These drugs can relieve the symptoms of AD, but they are not able to reverse the progression of the illness.
Deficits in the cholinergic neurons and transmission located in the basal forebrain could severely affect many facets of cognition and behavior, including cortical and hippocampal memory processing. The low levels of ACh are due to its excessive hydrolysis by cholinesterases (ChEs) present in the brain. Therefore, inhibition of ChEs is considered to be an effective strategy to increase ACh levels within the brain (Greig et al., 2002;Hartmann et al., 2007;Jing et al., 2019). Recent studies have indicated that AChE could play a key role in deposition of amyloid plaques. However, in later stages of AD, diminished levels of AChE and upsurged or unaltered levels of BuChE have been reported (Holzgrabe et al., 2007). This suggests that if AD patients are treated with dual AChE and BuChE inhibitors it may lead to more assured results.
Under normal physiological conditions, the production of ROS and reactive nitrogen species (RNS) is counter balanced by the antioxidant activities of various defensive mechanisms that include several antioxidants and enzymes. The imbalance between the ROS generation or reduction in its metabolic destruction will shift the equilibrium of the cellular redox balance to an oxidative imbalance, leading thereby to the overproduction of ROS (Poljuha et al., 2015). The change in the redox transition metal balance, particularly of copper, iron and zinc ions are crucial. Binding of these metal ions to Ab causes dyshomeostasis that leads to an apparent state of Cu deficiency in the intracellular region and an extracellular rise in Cu and Zn levels. The metal dyshomeostasis causes production of oxidative stress by dysfunction of various important cuproenzymes and antioxidants such as superoxide dismutase 1 (SOD1) resulting in the dysfunction of neuronal cells, which in turn reduces the synaptic transmission.
Based on these reports and extending our ongoing research work, we report herein designing, synthesis and biological evaluation (ChE inhibition, antioxidant and metal chelation studies) of a series of novel carbazole-based substituted thiosemicarbazide and hydrazide derivatives as potential anti-AD agents. The binding modes and interactions of the compounds with the target proteins were determined by performing molecular docking and molecular dynamics simulation studies. The most promising compounds were then screened in silico for their virtual ADMET properties.

Rationale of designing
The mono-targeted anti-AD agents, even though having high binding affinity and selectivity for a specific target, might not be that much effective to combat AD having a complex etiology. In contrast to a mono-targeted ligand, an MTDL with well-balanced affinities for multiple targets can offer additional advantages. So, development of MTDLs for AD is considered as the most promising drug discovery approach.
Thiosemicarbazides and acyl hydrazides have received great attention among medicinal chemists as important chemical functionalities due to their wide spectrum of biological activity. Various substituted thiosemicarbazide and acyl hydrazide derivatives have been identified to possess interesting biological activities (Gomes et al., 2014;Mathew et al., 2018;€ Onkol et al., 2013;Prathima et al., 2012;Sens et al., 2018) (Figure 1(B)). Thiosemicarbazone group is reported to be an indispensable pharmacophore due to its antioxidant and metal chelating properties.
The interesting biological activities associated with the carbazole scaffold, and thiosemicarbazone and acyl hydrazide moieties motivated us to merge these biologically active pharmacophoric groups into a single entity to obtain molecules with multiple beneficial effects for the treatment of AD ( Figure 2). Herein, we report the synthesis of carbazole-based thiosemicarbazone and acyl hydrazone derivatives and their biological assessment for possible anti-AD activities.

Chemistry
The designed carbazole-based thiosemicarbazone and acyl hydrazone derivatives were divided into a set of two series. In series-I, the carbazole scaffold was linked to the substituted thiosemicarbazide moiety (compounds 54-65) and semicarbazide moiety (compound 66), while in series-II, it was linked to the acyl hydrazide moiety (compounds 67-73). Syntheses of these designed compounds are depicted in Schemes 1-3.
The required semicarbazide and thiosemicarbazide intermediates were synthesized as shown in Scheme 1. Substituted benzylamines (11-15) were reacted with carbon disulfide and methyl iodide in the presence of triethylamine to obtain methyl carbamodithioate intermediates, which were reacted further with hydrazine hydrate by using methanol as the solvent to give the desired thiosemicarbazide intermediates (31-35) (He et al., 2019;Pervez et al., 2017). Similarly, substituted anilines (16-22) were reacted with carbon disulfide in the presence of N,N-dimethylformamide (DMF) and sodium hydroxide, followed by reaction with hydrazine hydrate to offer the desired thisemicarbazides (36-42). The semicarbazide intermediate (43) was synthesized from 4-chloroaniline (23) in two-step procedure including acylation reaction followed by hydrazinolysis. 4-Chloroaniline (23) was acylated by phenyl chloroformate in the presence of sodium carbonate in acetone, followed by reaction with hydrazine hydrate to offer the desired semicarbazide intermediate (43).
The required acyl hydrazide intermediates (44-50) were synthesized as depicted in Scheme 2. Substituted carboxylic acids (24-30) were converted to their respective methyl esters by the Fischer esterification method, and then these esters were converted to the corresponding acyl hydrazides (44-50) by refluxing them in methanol with excess hydrazine hydrate.

In vitro ChE inhibition studies
The in vitro ChEs inhibitory activity of the synthesized compounds was determined by Ellman's assay, and protocol of the study was same as described in our previous reports Sinha et al., 2015). Donepezil and tacrine were taken as the standards for Ellman's assay. The calculated IC 50 values and selectivity indices of the test compounds for ChEs are reported in Tables 1 and 2. In series-I, thiosemicarbazones (54-65) showed good ChEs inhibitory activity in micromolar ranges.
The benzyl derivative (54) exhibited balanced inhibitory activity on both the ChEs (IC 50 values of 2.21 and 2.33 mM for hAChE and EqBuChE, respectively). Introduction of small lipophilic electron-withdrawing fluoro group (in compound 57) and electron-donating methyl group (in compound 58) led to a moderate increment in hAChE (1.64-and 1.38-fold, respectively) and EqBuChE (1.31-fold for compound 58) inhibitory activities, whereas introduction of methoxy group (compound 55) and chloro group (compound 56) at the para position of the benzyl ring resulted into a modest reduction in the inhibitory activities. Replacement of the benzyl ring with phenyl ring (compound 59) led to a 2.27fold improvement in the inhibitory activity of hAChE (IC 50 value of 0.97 mM) and a 1.20-fold increase in the inhibitory activity of EqBuChE (IC 50 value of 1.93 mM). Substitution of the phenyl ring with a methoxyl group at its para position (compound 60) or replacing the phenyl ring with other heterocyclic rings i.e. pyridine (compound 61), substituted pyridines (compounds 62 and 63), pyrimidine (compound 64) or pyrazine (compound 65) had a detrimental effect, leading to moderate decrease in ChEs inhibitory activities. Substituting the thiosemicarbazone linker with semicarbazone (compound  66) offered good hAChE and moderate EqBuChE inhibitory activities (IC 50 value of 1.64 mM and 5.57 mM, respectively).
In series-II, acyl hydrazone derivatives (67-73) demonstrated good hAChE inhibitory and moderate EqBuChE inhibitory activities in micromolar ranges ( Table 2). The phenylacetic acid hydrazone derivative (67) showed good ChEs inhibitory activity (IC 50 values of 1.0 mM for hAChE and 2.07 mM for EqBuChE). Introduction of a 4-methoxyl group (compound 68) or 4-chloro group (compound 69) on the phenyl ring led to a drastic decrease in EqBuChE inhibitory activity. Replacement of the phenylacetic hydrazide moiety with benzhydrazide moiety (compound 70) did not affect the ChEs inhibitory activities. Incorporation of substituted benzhydrazide moiety (compounds 71 and 72) and picolinoyl hydrazide moiety (compound 73) had a detrimental effect, leading to a decrease in ChEs inhibitory activity.

DPPH radical scavenging activity
The DPPH (1,1-diphenyl-2-picrylhydrazyl) radical scavenging assay was used to evaluate the antioxidant activity of the synthesized compounds (Kedare & Singh, 2011). It was determined by the ability of the selected compounds to quench the purple-colored DPPHÁ radical to a yellow colored stable DPPHH molecule and the corresponding free radical-scavenging potential of the test compounds was calculated by the decline in absorbance of the test solutions at 517 nm. The free radical scavenging activity was calculated as a reduction percentage (RP) of DPPH. Among the tested compounds, compounds (61 and 62) exhibited excellent free radical scavenging activity with IC 50 values of 9.91 and 10.52 mM, respectively, that were near to that of ascorbic acid used as a reference compound (IC 50 value of 9.01 mM). Donepezil and tacrine were observed to be devoid of any notable radical scavenging activity with IC 50 values greater than 500 lM (Table 1).

Metal chelation study
Dyshomeostasis of redox transition metals, particularly copper, iron and zinc, and their excessive deposition in the brain have been correlated with AD etiology (Savelieff et al., 2013). Thus, the use of metal ion chelators has been proposed as a possible therapy for a disease like AD. The potential of the test compounds to form chelates with biometals was determined by means of UV-vis spectroscopy (Wang et al., 2015).
Compounds (61 and 62) exhibited maximum absorption peaks at 342 and 359 nm respectively, in methanol (as shown in Figure 3). When Cu 2þ ion solution was added to the solution of the test compounds, a sharp decrease in the heights of the absorbance peaks were observed, indicative of the interactions between the metal ions and the test compounds to form ligand-Cu þ2 complexes. Insignificant changes in the position and height of absorption peaks were observed when AlCl 3 , CaCl 2 , FeCl 3 or FeSO 4 were added into the test compounds' solutions. This suggests that the test compounds had poor chelating abilities for these ions (Al þ3 , Ca 2þ , Fe 2þ and Fe 3þ ). The test compounds were also evaluated for their binding potential to other key biometal ions, such as Zn 2þ and Mg 2þ where these compounds (61 and 62) showed negligible binding affinities for these metals. 8-Hydroxyquinoline (8-HQ) was used as a reference compound Scheme 3. Synthetic route for the designed compounds (54-73). Reagents and conditions: (i) Ethyl iodide, sodium hydroxide, dimethyl sulfoxide, RT, 4-5 h; (ii) phosphoryl chloride, DMF, chloroform, reflux, approx. 12 h; (iii) acetic acid, methanol, reflux, 4-5 h. for validation of the assay protocol. A significant alteration in the height and position of absorption peaks was observed when the solutions of the metal ions and the HQ solution were mixed indicating the ability of HQ to form metal-HQ chelates non-selectively. The above finding asserted the selective Cu 2þ ion chelation potential of the test compounds (61 and 62). Selectivity of these compounds for the copper ions is a crucial factor in the designing of biometal chelating agents to escape binding with other important biometal ions, the imbalance of which could drive to undesired adverse effects.

Docking studies of compounds (61 and 62) with ChEs
To gain insight into the interactions of the reference compounds and test compounds within the active sites of ChEs, docking studies were performed using Schrodinger Suite (Glide version 5.5 (2009)   of AChE consists of an oxyanion hole, an acyl-binding pocket, catalytic active site (CAS) and peripheral anionic site (PAS).
The key amino acid residues responsible for binding are Trp84 and Phe330 in the CAS, and Trp279 in the PAS. The docking study shows that compound (61) is fitting intimately inside the groove formed by Trp84, Tyr121 and Phe330 amino acid residues. It was positioned along the active site gorge similar to donepezil, starting from the active site amino acid residue Trp84 to the peripheral site amino acid residue Trp279. The pyridine ring of compound (61) was sandwiched between Trp84 and Phe330 residues of CAS. It showed p-p stacking interactions with Trp84 (3.9 Å) and Phe330 (3.6 Å). The tricyclic carbazole ring was positioned near the Trp279 residue of CAS. Binding mode of the compound (61) within the active site of BuChE is depicted in Figure 4(C,D). Precise analysis of interaction of 61 in the active site of BuChE showed that the key hydrogen bonding interaction observed experimentally for the tacrine moiety with His438 residue was maintained for compound (61). In addition, nitrogen atom of the pyridine ring formed a hydrogen bond with Trp-82 (2.8 Å). The -NH of semicarbazone group showed hydrogen bonding with His438 (2.0 Å). The tricyclic carbazole ring showed T-shape edge-to-edge p-p stacking interactions with Trp231 (4.8 and 4.9 Å). In the structure of BuChE, Trp82 amino acid residue is conserved, but Phe330 present in AChE is replaced by Ala328 in BuChE. Absence of phenylalanine residue influences the affinity of compounds for BuChE. Similarly, the interaction views of 62 within the active sites of AChE and BuChE are represented in Figure 5. Compound (62) fits compactly inside the groove formed by amino acid residues Trp84, Try121 and Phe330. 5-Chloropyridine ring of compound (62) was oriented near Trp84 and Phe330 residues of CAS. The tricyclic carbazole ring was positioned near Trp279 residue of CAS. In case of BuChE, the tricyclic carbazole ring of compound (62) showed T-shaped edge-to-edge p-p stacking interactions with Trp231 (4.8 and 4.9 Å) and Phe329 (5.3 and 5.5 Å). 5-Chloropyridine ring was oriented parallel to Trp82 residue in CAS. The -NH of semicarbazone group was observed to form a hydrogen bond with His438 (2.4 Å).

Molecular dynamics simulation studies
To determine the conformational stability and promising molecular interactions of compounds (61) and (62) in the active sites of ChEs, time-dependent molecular dynamics (MD) stability analysis was performed. The MD simulations were performed for a period of 50 ns using GROMACS2020.1. The study was performed using the docked pose of the ligand-receptor complex as the reference frame. The conformational changes and dynamic behavior in complexes were analyzed by various computational parameters like root-mean-square deviation (RMSD)-P, RMSD-L, root-meansquare fluctuation (RMSF)-P (P ¼ protein, L ¼ ligand) and solvent accessible surface area (SASA) for protein while having ligand in the receptor active site are shown in Figures 6-9.
The protein RMSD is determined mainly to understand the extent of movement of receptor atoms or groups while having the ligand inside the receptor active site. This type of analysis is indicative of the stability of conformation of the receptor structure and deviation over a period of time. The average RMSD-P for AChE in complexation with the docked compound (61) was 0.1386 nm (range of 0.08-0.253) ( Figure  6(A)). This suggests that over a period of time the protein having compound (61) in the active site was quite stable. The average RMSD-L was 0.3497 nm (range of 0.2-0.48 nm) which strongly indicated the stability of the protein-ligand complex without a major deviation in orientation of the compound in the receptor active site over the period of MD analysis (Figure 6(B)). RMSF indicates the integrity of the structure and mobility of the residues in the protein structure. It shows the flexibility of each amino acid residue of the protein upon binding to the ligand. Here, all the residues of the protein structure including loop and terminal residues, with compound (61) docked in the receptor active site, showed RMSF-P within 0.35 nm (Figure 6(C)). Apart from these observations, SASA analysis also supported the stability of the protein-ligand complex (Figure 6(D)). The short-range electrostatic (Coul-SR) and van der Waals/hydrophobic (LJ-SR) interaction energies of the protein-ligand complex suggested potential electrostatic and hydrophobic interactions.
The average values of Coul-SR and LJ-SR were observed to be -57.14 ± 1.90 kJ/mol and -182.43 ± 2.10 kJ/mol respectively, suggesting that the contribution of hydrophobic interactions was higher than the electrostatic interactions in stabilizing the protein-ligand complex.
In a similar way, the MD analysis of compound (61) docked in the BuChE active site was also carried out. The average RMSD-P was 0.156 nm (range of 0.10-0.218 nm) (Figure 7(A)) and the RMSD-L was found in the range of 0.14-0.43 nm (Figure 7(B)). Including terminal and loop residues, the observed RMSF-P for all the residues was below 0.25 nm (Figure 7(C)). SASA analysis also supported the stability of the protein-ligand complex (Figure 7(D)).The shortrange electrostatic (Coul-SR energy value of -76.14 ± 2.9 kJ/ mol) and van der Waals/hydrophobic (LJ-SR energy value of -172.43 ± 2.7 kJ/mol) interaction energies explained promising electrostatic and hydrophobic interactions between the protein and the ligand. The obtained values suggested that   the role of hydrophobic interactions was much more than the electrostatic interactions in stabilizing the ligand within the protein.
Similar MD simulation studies for ChEs with compound (62) were also performed. For compound (62) with AChE, the average RMSD-P of 0.151 nm (range of 0.08-0.223 nm) suggested high stability of the receptor with compound (62) in the active site (Figure 8(A)). RMSD-L was observed to be 0.134 nm with a range of 0.08-0.214 nm (Figure 8(B)). Here, the observed overall RMSF-P value was below 0.45 nm supporting the stability of the protein while having the ligand in the receptor active cavity (Figure 8(C)). Apart from these  observations, SASA analysis also supported the stability of the protein-ligand complex (Figure 8(D)). Here again, the short-range electrostatic (Coul-SR energy value of -61.34 ± 1.6 kJ/mol) and van der Waals/hydrophobic (LJ-SR energy of -206.73 ± 2.4 kJ/mol) interaction energies explained strong interactions between the protein and the ligand.
The RMSD-P value for BuChE-compound (62) complex was 0.148 nm (range 0.12-0.18 nm) explaining high stability of the protein (Figure 9(A)). The average RMSD-L was 0.237 nm (range of 0.1-0.46 nm) (Figure 9(B)) and RMSF-P was less than 0.35 nm including loop and terminal residues (Figure 9(C)). Here also the hydrophobic interaction contribution was found to be higher than the electrostatic interactions (found Coul-SR energy and LJ-SR energy values were -66.14 ± 2 kJ/mol and -192.46 ± 1.8 kJ/mol, respectively).

In silico prediction of physicochemical and pharmacokinetics parameters
Determination of pharmacokinetics properties of the lead molecules is an important step in the drug discovery process to prevent the drug attrition in the later stages of clinical trials due to inadequate absorption, distribution, metabolism, excretion, and toxicity (ADMET) profile. Application of computer-based tools in prediction of ADMET indicators is attracting considerable attention to screen out such outliers in the very early stages of the drug discovery process. Here, the ADMET parameters for the most promising compounds (61 and 62) along with the standard drugs donepezil and tacrine were assessed using QikProp module of Schrodinger Suite (Table 3), and pkCSM web server (http://structure.bioc. cam.ac.uk/pkcsm) ( Table 4). Table 3, compounds (61 and 62) obeyed all of the Lipinski's rule-of-five parameters in the given permissible ranges. Veber et al. introduced two additional key parameters viz. number of rotatable bonds (NRB) and topological polar surface area (TPSA) (Veber et al., 2002). NRB describes the conformational flexibility of the molecules. For better oral bioavailability and drug-likeness, a molecule should have lesser than 8 rotatable bonds or 7 atoms in a linear chain outside the ring. TPSA is a measure of polarity. It is a major determinant of passive diffusion of a molecule through biological membranes and thus, it enables the prediction of drug absorption, including bioavailability, blood-brain barrier (BBB) permeation and intestinal absorption. Lenz and coworker have reported an average TPSA value to be less than 90 Å 2 for a central nervous system (CNS) active drug (Pajouhesh & Lenz, 2005). The average value for marketed CNS drugs is 40.5 Å 2 (range 4.63-108 Å 2 ). Both the compounds (61 and 62) have NRB values of six and TPSA values of $56.5 Å 2 . QPCaco-2 value correlates the oral absorption of a drug predicting its gut-blood barrier permeability. Values higher than 500 predict significant oral absorption, which was attained by both the test compounds. Significant quantum of oral bioavailability of the test compounds is further supported by the predicted human oral absorption percent (% HOA) values. The n-octanol-water partition coefficient (QPlogP o/w ), brain/blood partition coefficient (QPlogBB), CNS and apparent MDCK (Madin-Darby canine kidney) cell permeability (QPPMDCK) collectively predict the potential of a ligand to cross the BBB. Compounds (61 and 62) are predicted to have borderline CNS penetration as they possess CNS values of zero, and QPlogBB values of -0.276 and -0.113, respectively. QPPMDCK value is a good criterion for the prediction of BBB permeation. A QPPMDCK value greater than 500 is considered as good, and the test compounds displayed significantly high values for this parameter. QPlogK hsa predicts the binding of test molecules to the human serum albumin. The test compounds (61 and 62) show adherence to the recommended values. The descriptor '#star' annotates the number of parameters having values falling outside the 95% range of the same values for the reported drugs. A larger number of #star suggests that a molecule is less druglike than molecules with lesser #stars.

As shown in
Compounds (61 and 62) showed values of #star within the limit, thus they could be taken as drug-like compounds.
pkCSM tool was employed to predict some additional ADMET properties. Some of these descriptors are mentioned in Table 4. Compounds (61 and 62) are predicted to have good absorption values suggesting high chances of crossing the gut by passive diffusion. The water solubility, Caco2 permeability, human intestinal absorption and skin permeability values for the test compounds were observed to be in acceptable ranges. The test ligands (61 and 62) are predicted to have high VDss  The results indicated that these compounds were likely to cross these barriers. Metabolism of any drug molecule depends mainly on the action of the cytochrome (CYP) class of enzymes. The lead molecules (61 and 62) were predicted to be the substrates for CYP3A4 suggesting possible metabolism of these compounds with the action of CYP3A4. They were also predicted to be the inhibitors of CYP1A2, CYP2C19, CYP2C9 and CYP3A4 which could cause interference with the metabolism of other drugs metabolized by these isoforms. The predicted excretion values also supported drug-like behavior of the test compounds based on the total clearance and interaction with renal organic cation transporter 2 (OCT2). The test ligands were found to be non-substrates of OCT2 predicting their proper renal clearance. Compounds (61 and 62) were predicted to have maximum tolerated doses (human) of 0.638 and 0.762 mg/kg/day, respectively. The AMES test predicts the non-mutagenic and noncarcinogenic nature of compound (61), whereas compound (62) was predicted to have possible mutagenic and/or carcinogenic properties. Furthermore, the lead molecules were found to be non-inhibitors of human Ether-a-Go-go Related Gene-I (hERG-I) and hERG-II channels. The minnow toxicity was found to be higher than -0.3 log mM which predicted that the test ligands were acutely nontoxic. All these results together indicate that the lead molecules (61 and 62) have drug-like pharmacokinetics profiles, which would increase their biological significance.

Site of metabolism prediction by SMARTCyp
The test compounds (61 and 62) were predicted to be metabolized by CYP3A4 isoform. Common sites of metabolism (SOM) of the test ligands were predicted using the SMARTCyp web server (https://smartcyp.sund.ku.dk/mol_ to_som). For the validation of SMARTCyp predictions, verapamil, a substrate of CYP3A4 was used as its metabolic products, was well reported. The predicted SOM of verapamil by SMARTCyp was in accordance with the literature (Sun et al., 2004). As predicted by SMARTCyp (Figure 10), the primary SOM in compounds (61 and 62) is S.1 of thiocarbonyl group of semicarbazone as it ranked first with the least scores and energies. The secondary predicted SOM are C.20 and C.21 sites on the carbazole ring of compound (61 and 62) respectively, while the tertiary predicted SOM are the C.23 and C.24 methylene carbons of ethyl group attached to the carbazole ring of compounds (61) and (62), respectively.

Conclusion
In conclusion, a series of carbazole-based thiosemicarbazone, semicarbazone and hydrazone derivatives were designed, synthesized and biologically evaluated for their ChE inhibitory, antioxidant and biometal chelating properties. In vitro study showed that the synthesized compounds possessed ChE inhibitory activity with IC 50 values ranging between 0.9 and 2.6 mM for hAChE and 1.6 and 23.32 mM for EqBuChE.  (61 and 62) were predicted to have in silico drug-like ADMET properties. Findings of the present study suggest compounds (61 and 62) as promising leads for further development as balanced MTDLs to be used as anti-AD agents.

Experimental
General: All the required reagents and solvents for the experiments were procured from commercial suppliers such as Sigma Aldrich and Spectrochem. The solvents and chemicals were purified whenever required by basic laboratory techniques prior to use. The silica gel 60 GF 254 (Merck) TLC plates having a thickness of 0.25 mm were used to observe the progress of the reaction by visualizing in a UV (254 nm) chamber and/or exposure to various staining reagents. Melting points of the synthesized intermediates and final compounds were recorded in a glass capillary using a VEEGO melting point apparatus having a silicone oil bath and a temperature resolution of 0.1 C. The IR spectra were obtained using FTIR spectrophotometer [Bruker ALPHA-T (Germany)] using KBr pellets. The 1 H NMR spectra were acquired on a Bruker 400 MHz spectrometer in deuterated solvents. The chemical shift (d) values are expressed in parts per million (ppm) in relation to tetramethylsilane (TMS) as internal standard. The pattern of the peaks is expressed as s (singlet), d (doublet), dd (doublet of doublet), t (triplet), q (quartet) and m (multiplet). The mass spectrometer of Thermo Fisher with electrospray ionization (ESI) ion source was used for the recording of mass spectra.

General method for the synthesis of 4-substituted thiosemicarbazide intermediates (31-42)
For intermediates (31-35) (Method A): To a stirred solution of substituted benzylamines (11-15) (1 equiv.) and triethylamine (1.5 equiv.) in methanol (10 mL), a solution of carbon disulfide (1.2 equiv.) in methanol (5 mL) was added gradually. The resulting mixture was allowed to stir for 2-3 h at room temperature (RT). To it, methyl iodide (1 equiv.) in methanol (5 mL) was slowly added keeping temperature of the reaction mixture at -10 C. After complete addition, the reaction mixture was stirred for additional 2 h at RT. After completion of the reaction as checked by TLC, the reaction mixture was concentrated under reduced pressure and partitioned between 30 mL of ether and 0.2 N HCl solution. The ethereal layer was washed with cold water and saturated with NaCl solution, dried using anhydrous MgSO 4 , filtered and evaporated under vacuum to obtain white solid mass. The obtained solid mass was further charged to a solution of hydrazine hydrate (3 equiv.) in methanol (15 mL). The resulting reaction mixture was refluxed for 4 h. After completion of the reaction as confirmed by TLC, the reaction mixture was allowed to attain RT. The solid so precipitated were collected by filtration and washed with cold methanol to obtain the desired intermediates (31-35).
5. 1.1.1. N-Benzylhydrazinecarbothioamide (31). The intermediate (31) was synthesized as per the general Method A from benzylamine (11) (1.0 g, 9.33 mmol) as white solid (He et al., 2019 (32)  For intermediates (36-42) (Method B): To a stirred solution of substituted aniline (16-22) (1.0 equiv.) and NaOH (1.2 equiv.) in DMF (10 mL), a solution of carbon disulfide (1.2 equiv.) in DMF (5 mL) was gradually added and the resulting mixture was stirred for 3 h at RT. After complete consumption of starting material as checked by TLC, hydrazine hydrate (3.0 equiv.) was added to the above reaction mixture and it was heated to 70 C for 1 h. After completion of the reaction as checked by TLC, the reaction mixture was allowed to cool to RT and diluted with cold water. The solid precipitates so obtained were filtered, washed with cold water and dried under reduced pressure to get the intermediate compounds (36-42).

N-(4-Chlorophenyl)hydrazinecarboxamide (43)
To a cold stirring solution of 4-chloroaniline (1.0 equiv.) and potassium carbonate (2.0 equiv.) in acetone (20 mL) at 0-5 C, a solution of phenyl chloroformate (1.1 equiv.) in acetone (10 mL) was added gradually. After complete addition, the resulting reaction mixture was stirred at room temperature for additional 2 h. After completion of the reaction, the reaction mixture was concentrated in vacuo, quenched with 10% NaHCO 3 solution and diluted with cold water. The solid precipitates obtained were filtered and dried. The obtained solids were added to solution of hydrazine hydrate (3 equiv.) in 1,4-dioxane. The reaction mixture was heated to 100 C for 4 h. After completion of the reaction as checked by TLC, the reaction mixture was allowed to cool to RT. The resulting precipitates were filtered and washed with ether to obtain the titled compound as white solid (Beukers et al., 2003). (Yield 75.2%); m.p. 188-190 C (lit m.p. 190-192 C); IR (KBr, cm À1 ): 3334, 3230, 1670, 1142, 815. 5.1.3. General procedure for the synthesis of acyl hydrazide intermediates (44-50) Method C. To a stirred solution of substituted benzoic acid/substituted phenylacetic acids/nicotinic acid (1 equiv.) in methanol (20 mL), conc. sulfuric acid in a catalytic amount was added. The resulting mixture was refluxed for 6 h. After completion of the reaction (confirmed by TLC), the mixture was concentrated to a volume of $10 mL and partitioned between chloroform (30 mL) and water. The organic phase was washed with water, dried over anhydrous sodium sulfate and evaporated under reduced pressure to obtain liquid product. The resulting liquid was further added to a solution of hydrazine hydrate (3 equiv.) in methanol (20 mL). The reaction mixture was heated up to 65 C for 4 h. After completion of the reaction as checked by TLC, the reaction mixture was allowed to cool to RT and the solid precipitates so obtained were filtered and washed with cold methanol to obtain the desired acyl hydrazides (44-50). (44). The intermediate (44) was synthesized as per the general Method C from 2-phenylacetic acid (24) (1.0 g, 7.36 mmol) as white solid (Zimin et al., 2018).  (48) was synthesized as per the general Method C from 4-fluorobenzoic acid (28) (1.0 g, 6.49 mmol) as white solid (Murty et al., 2016 (52) To stirring solution of carbazole (51) (1.0 g, 6 mmol) in DMSO (15 mL), aqueous sodium hydroxide (0.48 g in 2 mL) solution was added, and the resulting reaction mixture was stirred for further 10 min. Ethyl iodide (0.6 mL, 7.06 mmol) was added in a dropwise manner to the above reaction mixture and it was allowed to stir at RT. TLC was used to monitor the progress of the reaction where after completion of the reaction, the reaction mixture was diluted with ice cold water and the solid so precipitated was collected by filtration, rinsed with cold water and dried in vacuum to yield the compound (52) as off white solid (Patel, Patel, Kanhed, Teli, Patel, Joshi, et al., 2020). (Yield: 1.07 g, 94%); m.p. 64-66 C (lit m.p. 64-66 C); IR (KBr, cm À1 ): 3049, 2978, 2869, 1596, 1018MS (m/z): 196.3

DPPH radical scavenging activity
The potential of the test compounds to scavenge the DPPH radical was evaluated by DPPH spectrophotometric assay as described in our previous reports (Patel et al., 2019. The quantification of free radical scavenging activity of the test compounds was done in the form of reduction percentage (RP) of DPPH calculated by the equation RP ¼ 100[(A 0 À A C )/A 0 ], where A 0 is the absorbance of DPPH without sample and A C is the absorbance after adding sample with concentration C. Here, ascorbic acid was taken as a positive control (Kedare & Singh, 2011).

Metal chelation study
The tendency of the selected test compounds to form chelates with the biometals was evaluated using UV spectrophotometry as detailed in our earlier report (Patel, Patel, Kanhed, Teli, Patel, Joshi, et al., 2020). The UV absorption spectra of the test compounds (61 and 62) and 8-hydroxyquinoline (reference compound) were recorded alone and in the presence of CuSO 4 , FeCl 3 , FeSO 4 , ZnCl 2 , AlCl 3 , CaCl 2 or MgSO 4 with wavelengths ranging from 200 to 600 nm after incubating the solutions for 30 min. in methanol at room temperature. Final concentration of the test compounds and the reference compound was 25 mM.

Docking studies of compounds (61 and 62) with ChEs
Docking studies were performed using Glide module of Schrodinger Suite (Schr€ odinger 2009, LLC, New York, NY). The protein structures of AChE (PDB code: 2CKM, 1B41) and BuChE (PDB code: 4BDS) were obtained from the RCSB Protein Data Bank and checked for structural correctness with Protein Preparation Wizard of Schrodinger using OPLS force field. The grids were generated on the active sites of the respective protein structures. The generated grids were validated by redocking of the knocked out co-crystallised ligands into the active sites of the grids. The redocked ligands showed similar interactions as shown by the original co-crystallized ligands. The grids were validated by comparing the RMSD values of the redocked ligands with the cocrystallized ligands present in 2CKM and 4BDS and the values obtained were 0.40 and 0.26 A, respectively. The 3D structures of 61 and 62 were built with the Build module within Maestro, and conformational search was carried out using OPLS force field at physiological pH with the LigPrep module of Schr€ odinger. Docking studies were conducted in extra-precision (XP) mode where extensive conformational search is carried out to identify the suitable binding conformation which can bind within the receptor active site.

Molecular dynamics simulation
The MD analyses were carried out for 50 ns time duration by using GROMACS 2020.1 tool (GROMACS 2020). MD protocol was followed as reported previously by our group .

In silico prediction of ADMET properties
Prediction of in silico ADMET properties was performed using the QikProp module of Schrodinger Suite and pkCSM web tool. The ligand structures built for the docking studies were used for the prediction of the different ADMET descriptors. The significant descriptors examined are presented in this report.

Prediction of site of metabolism by SMARTCyp
SMARTCyp, a SOM predicting software released by Rydberg et al. contains a database of compounds metabolized by CYP450 with precalculated density functional theory (DFT) activation energies Rydberg, Gloriam, Zaretzki et al., 2010). To rank the SOM, SMARTCyp uses database having SMiles ARbitrary Target Specification-based fragments in combination with an accessibility descriptor. The test ligands were uploaded as SMILES strings format representing the molecule. The obtained results included the molecular structure and an atom ranking table for each ligand. Three top-ranked atoms were highlighted in the structure as well as in the table, as shown in Figure 5.