Discovery of N-quinazolinone-4-hydroxy-2-quinolone-3-carboxamides as DNA gyrase B-targeted antibacterial agents

Abstract Emerging drug resistance is generating an urgent need for novel and effective antibiotics. A promising target that has not yet been addressed by approved antibiotics is the bacterial DNA gyrase subunit B (GyrB), and GyrB inhibitors could be effective against drug-resistant bacteria, such as methicillin-resistant S. aureus (MRSA). Here, we used the 4-hydroxy-2-quinolone fragment to search the Specs database of purchasable compounds for potential inhibitors of GyrB and identified AG-690/11765367, or f1, as a novel and potent inhibitor of the target protein (IC50: 1.21 µM). Structural modification was used to further identify two more potent GyrB inhibitors: f4 (IC50: 0.31 µM) and f14 (IC50: 0.28 µM). Additional experiments indicated that compound f1 is more potent than the others in terms of antibacterial activity against MRSA (MICs: 4–8 µg/mL), non-toxic to HUVEC and HepG2 (CC50: approximately 50 µM), and metabolically stable (t1/2: > 372.8 min for plasma; 24.5 min for liver microsomes). In summary, this study showed that the discovered N-quinazolinone-4-hydroxy-2-quinolone-3-carboxamides are novel GyrB-targeted antibacterial agents; compound f1 is promising for further development.


Introduction
Antibiotic resistance poses a significant threat to global public health. The number of individuals who die from infections caused by antibiotic-resistant bacteria is projected to rise to 10 million by 2050. 1 In response to antibiotic resistance, the WHO published a priority list of pathogens in 2017, for which new antibiotics are urgently needed. In the category of Gram-positive bacteria, methicillin-resistant S. aureus (MRSA) was designated as "high-priority status". 2 The antibiotics currently available to treat infectious diseases caused by MRSA are vancomycin, daptomycin, and linezolid. Unfortunately, their use has been limited in clinical practice. The first concern is their safety, e.g. the nephrotoxicity of vancomycin, the unknown proper dose of daptomycin, as well as the potential risk of thrombocytopenia caused by the high plasma levels of linezolid. 3 The second is the antibiotic resistance, which makes them lose antibacterial activity; strains of resistant S. aureus have been isolated in the clinic. 4 Emerging antibacterial compounds in late stage clinical trials often have favourable toxicity profiles, but they are still in the same class as existing drugs and thus may still fail to treat resistant strains. [5][6][7] Therefore, new classes of antibacterial agents for MRSA are urgently needed.
Bacterial DNA gyrase B subunit (GyrB) is a promising target for discovery and development of a new class of antibiotics. 8 As an indispensable component of DNA gyrase (A 2 B 2 ), GyrB binds ATP at the ATPase domain and catalyses ATP hydrolysis; it provides energy for DNA supercoiling. 9 When the GyrB inhibitor novobiocin was approved for clinical use (cf. Figure 1), antibiotics with the same mode of action were considered as promising therapeutics for the treatment of bacterial infections. 7 Since the decline of novobiocin in 1960s due to its toxicity and low efficacy, several diverse GyrB inhibitors have been discovered, e.g. ethyl ureas, 10 pyrazolopyridones, 11 pyrrolamides, 12 and quercetin diacylglycosides. 13 Unfortunately, none of these have been approved. Two compounds, i.e. SPR720 (ethyl ureas) 14 and DS-2969b (pyrrolamides), 15 are in phase I clinical trials (cf. Figure 1), but the clinical outcomes of these chemotypes are also unpredictable. Therefore, the identification of diverse structures as GyrB inhibitors is still necessary.
Here, we used the chemical information of the N-thiadiazole-4hydroxy-2-quinolone-3-carboxamides to identify new GyrB inhibitors. With the essential fragment for GyrB inhibition as the substructure, we searched the Specs database of purchasable compounds for related substances. By testing the compounds in S. aureus GyrB inhibition assay, we discovered that the 4-hydroxy-2-quinolone-3-carboxamide derivative that has an N-quinazolinone moiety inhibits GyrB. To understand the preliminary structure-activity relationship (SAR), we synthesised derivatives and evaluated their activities against GyrB. Representative GyrB inhibitors were submitted for in vitro evaluation of the antibacterial activity against a panel of S. aureus strains. Finally, we studied the cytotoxicity, ADMET profile, and important physicochemical properties of the most active antibacterial agent.

Computer-aided hit identification
We previously 16 proposed the likely binding mode of the N-thiadiazole-4-hydroxy-2-quinolone-3-carboxamides bearing heteroaromatic rings to the ATP binding site of S. aureus GyrB by molecular docking. Here, we utilised LigPlotþ 17 to generate a 2D diagram of the representative GyrB inhibitor g37. Figure 2(a) clearly shows that the 4-hydroxy-2-quinolone fragment plays the most important role in the binding to GyrB: its carbanyl group forms hydrogen bonds with Arg144, and its 4-hydroxyl group is involved in the formation of hydrogen bonds with Glu58 and Arg84. We performed a substructure search of the Specs database with the 4-hydroxy-2-quinolone fragment and identified 272 matches (in approximately 210,000 compounds). Aided by a clustering analysis and visual inspection of the molecular structures, we selected 14 potential GyrB inhibitors for experimental validation. Figure 2(b) shows that all compounds except for AE-406/ 41056087 and AE-406/41056637 are based on the chemotype of 4-hydroxy-2-quinolone-3-carboxamides and have diverse substituents attached to the amide nitrogen atom, i.e. thiazole, pyridine, 4-oxoquinazolin, phenyl, alkyl, oxazole, biphenyl, phenacetylamino, alkyl amide, benzamide, pyridine acetamide, and benzsulfamide.
In the S. aureus GyrB assay, AG-690/11765367 (also named as f1) reduced the activity of S. aureus GyrB by 72% at a concentration of 10 lM (cf. Figure 2(b)). The IC 50 value was determined to be 1.21 ± 0.13 lM based on the dose-response curve (cf. Figure  2(c)). Interestingly, it contained a 4-oxoquinazolin fragment that was different from the previously reported thiadiazole. The identification of such a novel GyrB inhibitor confirms the capacity of the computer-aided strategy. Notably, the other 13 compounds could not inhibit S. aureus GyrB by 50% at 10 lM (cf. Figure 2(b)), thus indicating that the type of substituents at position N of the amide is essential to GyrB inhibition. We performed molecular docking with OEDocking version 3.0.1 (OpenEye Scientific Software, Inc., Santa Fe, NM) 18 to derive a plausible binding mode of AG-690/11765367 to the ATP binding site of GyrB. In general, the binding mode of AG-690/11765367 is predicted to be similar to that of compound g37. 16 More specifically, the 4-hydroxy-2quinolone-3-carboxamide is expected to form hydrogen bonds with Arg84 and Arg144, electrostatic interactions with Arg84, and hydrophobic interactions with Pro87. The 4-oxoquinazolin fragment is predicted to be located in a sub-pocket surrounded by the hydrophobic residues including Asn54 and Ile86. The carbonyl group of the 4-oxoquinazolin fragment is expected to be uniquely involved in water-mediated interactions with Thr173, Asp81, and Gly85 (cf. Figure 2(d)).

Molecular design and chemical synthesis
To preliminarily study the SAR of f1, we designed two series of derivatives by (i) introducing the electron-donating groups or electron-withdrawing substituents to the benzene ring of the newly identified 4-oxoquinazolin fragment (R 1 in Scheme 1, f2-f12) and (ii) replacing the pentyl group with other alkyl groups at position 2 of the 4-oxoquinazolin fragment (R 2 in Scheme 1, f13-f16).
The synthetic route of the above-mentioned derivatives was planned as the condensation reactions between the 4hydroxy-2-oxo-1,2-dihydroquinoline-3-carboxylic acid and different substituted 3-aminoquinazolin-4(3H)-ones according to published methods. 19 In practice, the synthesis of the key intermediate c1, namely 1ethyl-4-hydroxy-2-quinolone-3-carboxylic acid, was composed of three consecutive steps (cf. Scheme 1): first, isatoic anhydride was ethylated in the presence of N,N-diisopropylethylamine and iodoethane. This reaction introduced the ethyl group to the heterocyclic nitrogen of isatoic anhydride. Second, the ethylated anhydride was treated with diethyl malonate and sodium hydride to afford the intermediate b1. Third, b1 was converted to the acid by hydrolysis under the condition of 12N hydrochloric acid and the refluxing methanol. The key amine intermediates e1-e16 were prepared by conversion of different substituted methyl 2-aminobenzoates d1-d16 into amides under mild condition and with triethylamine as base, followed by annulation with hydrazine hydrate in boiling ethanol. The target molecules f1-f16 were obtained by coupling the above-mentioned acid c1 with the corresponding amines (e1-e16). All synthesised compounds were characterised by melting points, 1 H NMR, 13 C NMR, and HRMS. The details of the chemical synthesis and structural characterisation are described in the experimental section.
The 1 H NMR spectra of f1-f16 revealed the appearance of a methylene signal from 4.46 to 4.07 (-CH 2 -N) and a methyl signal from 1.

Inhibitory activity against S. aureus Gyrase B
The inhibitory activities of all the derivatives for S. aureus GyrB in terms of IC 50 are reported in Table 1. In general, the IC 50 values of most derivatives were close to that of compound f1 (1.21 lM). Among the derivatives with R 1 substituents at different positions of the benzene ring, the most potent one was compound f4 with a 6-methoxyl substituent (IC 50 : 0.31 lM); the weakest was compound f7 with a 7-methyl group (24.40 lM). Other potent compounds include f3 (0.83 lM), f5 (0.81 lM), f6 (0.83 lM), f11 (0.77 lM), and f12 (0.88 lM). The introduction of electron-donating groups to position 6/8 or electron-withdrawing groups to position 7 is favourable for GyrB inhibition. For instance, f4 (6-MeO, 0.31 lM) showed greater inhibition against GyrB than f6 (6-CF 3 , 0.81 lM). Compound f12 (8-MeO, 0.878 lM) was more potent than compound f10 (8-Cl, 9.70 lM). In contrast, f8 substituted by a 7-methoxyl group (7.90 lM) was much less potent than f6 with a 7-chlorine moiety (0.83 lM). Derivatives with different alkyl substituents at position 2 of the 4-oxoquinazolin fragment (compounds f13-f16) have IC 50 values between 0.28 and 11.9 lM (cf, Table 1). These values are close to that of compound f1 (1.21 lM), and the replacement of the pentyl group with other alkyl groups only led to a small change in GyrB inhibition.
The most potent derivatives in two series, i.e. f4 and f14, were docked against the ATPase domain of GyrB. Their predicted binding modes are shown in Figure 3. As both f4 and f14 are based on a 4-hydroxy-2-quinolone-3-carboxamide scaffold and the 4oxoquinazolin fragment, the predicted binding poses were similar to that of f1. First, they are predicted to bind to the same position of the ATP binding site and are superimposed well with the predicted binding mode of f1 (cf. Figure 3(a)). Second, the predicted key interactions are identical including the hydrogen bonds with Arg84 and Arg144, electrostatic interactions with Arg84, hydrophobic interactions with Pro87, and water-mediated interactions with Thr173, Asp81, and Gly85. Uniquely, f4, with a methoxy substituent at position 6, seems to better occupy the hydrophobic region defined by Asn54, Ser55, and Thr173 (cf. Figure 3(b)). The pentyl group of compound f1 is outside the pocket (cf. Figure  2(d)), while the ethyl group fits the ligand binding pocket well and is positioned close to the edge of the surface (cf. Figure 3(c)).

Antibacterial activity against a panel of S. aureus strains
To identify GyrB inhibitors with potent antibacterial activity, we next tested the hit compound f1 and the two highly potent derivatives, f4 and f14, against five isolates of methicillin-sensitive S. aureus (MSSA; ATCC 29213, 15, 18-3, BAA976, and BAA1708) in a broth microdilution assay. 20 The MIC (minimal inhibitory concentration) values of f4 and f14 were 32 mg/mL, 64 mg/mL, or greater than 64 mg/mL for different isolates (Table 2); the MICs for f1 ranged from 4 to 8 mg/mL. Compounds f4 and f14 showed much weaker activity against S. aureus though they inhibited S. aureus GyrB more potently. To explain this inconsistency, we calculated the logP values of the three compounds with ChemDraw Ultra version 14.0 (Cambridge Scientific Computing, Inc., Cambridge, MA), i.e. 3.13 for compound f1, 3.01 for compound f4, and 1.88 for compound f14. Accordingly, we postulate that high hydrophobicity may favour the antibacterial activity of GyrB inhibitors.
The MIC value of compound f1 (4-8 mg/mL) was close to that of vancomycin (0.5-1 mg/mL), and thus, we further tested f1 against a panel of MRSA strains (ATCC 33591, ATCC 43300, 18-2) and vancomycin-intermediate-resistant S. aureus (VISA) strains  (ATCC 700699, HIP 5836, and HIP 5827). Likewise, the antibacterial activities of compound f1 for MRSA and VISA were similar to those of vancomycin (MICs: 4-16 vs. 0.5-4 mg/mL). The data listed in Table 2 demonstrate that the GyrB inhibitor f1 was the most potent antibacterial agent.

Drug-likeness of compound f1
To see whether f1 is a promising lead compound for further optimisation, we first investigated its cytotoxicity against human umbilical vein endothelial cells (HUVECs) and human hepatocellular liver carcinoma cells (HepG2) in the sulforhodamine B (SRB) assay. 21 The CC 50 values of this compound were 49.6 ± 0.2 mM for HepG2 and 51.5 ± 4.5 mM for HUVECs ( Table 3). The MIC values for the tested MRSA/VISA strains were between 4 and 16 mg/mL (i.e. 8.97 and 35.87 mM). These data indicate that compound f1 is not toxic to mammalian cells at concentrations at which bacterial growth is inhibited.
To gain insights into the pharmacokinetic profile of f1, the compound was first tested in vitro for metabolic stability by incubating it with the mouse plasma and liver microsomes. The data indicate that f1 is quite stable in mouse plasma, with a half-life (t 1/2 ) value greater than 372.8 min. Compound f1 was also metabolically stable in mouse liver microsomesthe t 1/2 value was 24.5 min. These results are consistent with our predictions with FAME 3a model for the prediction of sites of metabolism for phase 1 and 2 metabolic enzymes. 22,23 FAME 3 flagged only the ethyl side chain of f1 with a moderate likelihood of being a site of metabolism (cf. Figure S3).
Apart from in vitro cytotoxicity and metabolic stability, we used a free web service, SwissADME (http://www.swissadme.ch/), 24 to predict further key properties related to drug-likeness. Compound f1 was predicted to have high gastrointestinal (GI) absorption and did not penetrate the brain-blood barrier (BBB). Also, the physicochemical properties do not violate Lipinski's Rule-of-Five. Nevertheless, aqueous solubility may be an issue that should be addressed. Full details on all predictions are provided in Figure S4.

Conclusion
S. aureus has developed resistance against multiple antibiotics in clinical use including even the recently introduced daptomycin and linezolid. 4 Thus, antibacterial agents based on novel chemotypes or modes of actions are urgently needed to tackle antibiotic resistance. GyrB is a promising target for compounds breaking the antimicrobial resistance of S. aureus. Unfortunately, there is no GyrB inhibitor approved for clinical use, and the outcome of GyrB inhibitors in clinical trials is unpredictable at this stage. Here, we combined computer-aided hit identification, chemical synthesis, and in vitro biological evaluation to identify diverse GyrB-targeted antibacterial agents.
We concluded that the 4-hydroxy-2-quinolone fragment is essential to GyrB inhibition. Structural searches of the Specs compound library and experimental testing helped identify f1 (AG-690/11765367) as a novel, moderate inhibitor of S. aureus GyrB (IC 50: 1.21 mM). This hit compound is of great interest because it contains a 4-oxoquinazolin moiety instead of the previously identified thiadiazole. According to the predicted binding mode of f1, the water-mediated interactions that involve the carbonyl group of the 4-oxoquinazolin moiety seem to be of relevance to bioactivity.
We also performed a preliminary SAR study by synthesising 15 new derivatives and evaluating their GyrB inhibitory activities in S. aureus Gyrase ATPase inhibition assays. This led to eight derivatives that were more potent than f1. Compounds f4 and f14 were the two most potent S. aureus GyrB inhibitors, with IC 50 values of 0.31 and 0.28 mM, respectively. We compared these two derivatives with f1 in terms of antibacterial activity against a panel of S. aureus strains and decided to select the initial hit compound f1 for further testing in light of its better anti-MRSA activity (MICs: 4-8 mg/mL).
The in vitro cytotoxicity assay and metabolic stability assay (mouse) indicated that compound f1 does not exhibit significant cytotoxicity against HUVECs and HepG2 cells and is metabolically stable. The compound has favourable physicochemical properties, although its aqueous solubility may require optimisation.
In summary, this study reports on the computer-aided discovery of N-quinazolinone-4-hydroxy-2-quinolone-3-carboxamides as new GyrB inhibitors. The GyrB inhibitor f1 is a good starting structure for the development of new antibacterial agents.

Substructure search
The protocol for "substructure search" implemented in Pipeline Pilot version 16.2.0.58; (Dassault Syst emes Biovia Corp., San Diego, CA) was used for our purpose. Here, the 4-hydroxy-2-quinolone scaffold (cf. Figure 2) was set as the substructure, while the Specs chemical library version 2015 (accessed at http://www.specs.net) that contains more than 210,000 compounds was selected as the screening database. The outputs from the protocol, i.e. the compounds with the 4-hydroxy-2-quinolone scaffold from the Specs chemical library, were further assigned to 30 subsets by clustering based on MACCS fingerprints. Clustering was performed with the "Cluster Ligands" protocol of Discovery Studio version 16.1.0; (Dassault Syst emes Biovia Corp., San Diego, CA) Finally, the list of compounds for further bioassays was determined according to the structural diversity shown by the clustering, commercial availability, and synthetic feasibility.

Molecular docking
The structural model of S. aureus GyrB that we previously generated 16 was used for molecular docking of the new inhibitors. This model was derived from the crystal structure of S. aureus GyrB in complex with Novobiocin (PDB code 4URO; http://www.rcsb.org/). The binding site of this model was defined by Novobiocin in the crystal structure. Here, the conserved co-crystallised water (i.e. wat46 in the publication 25 ) was retained, and the hydrogen bond acceptor on the guanidine of Arg144 was defined as a docking restraint. These settings aimed to facilitate the generation of binding poses that may form hydrogen bonds with the conserved water and the guanidine of Arg144.
Prior to molecular docking, a maximum of 200 ligand conformers were generated with OMEGA version 2.5.1.4 (OpenEye Scientific Software, Inc., Santa Fe, NM). 26 These conformers were subsequently placed into the binding site of the protein model with OEDocking version 3.0.1 (OpenEye Scientific Software, Inc., Santa Fe, NM). 18 Finally, the docking poses were visually inspected, and the most plausible binding pose was selected.

General methods
All of the reagents were obtained from commercial sources and used without further purification unless stated otherwise. Thinlayer chromatography (TLC) on the silica gel plates GF254 (200-300 mm; Qingdao Haiyang Chemical Co., Ltd., Qingdao, China) with UV light illumination was used to monitor chemical reactions. 1 H NMR (500 MHz) and 13

Preparation of intermediate b1
Intermediate b1 was prepared according to the reported method. 27 A solution of isatoic anhydride (a1, 1 equiv.) in dry DMF (10 mL) was treated with DIPEA (1.3 equiv.) and iodoethane (1.3 equiv.), and the mixture was heated to 45 C and stirred for 10 h. After cooling to room temperature, diethylmalonate and sodium hydride were added to the solution under N 2 atmosphere. The reaction mixture was heated to 70 C and stirred for 8 h. The solution was then poured into 50 mL of cool water, and 4 M HCl was added to make its pH less than 5. The product was obtained after filtering and recrystallisation with EtOAc.

Preparation of intermediate c1
HCl (  Moderately or poorly soluble a Cytotoxicity measured after 72-h treatment with f1. Mean: average of duplicates; SD: standard deviation Paclitaxel was the positive control for the cytotoxicity assay (CC 50 for HepG2 cells: 12.7 ± 2.1 nM; CC 50 for HUVECs: 1.7 ± 0.2 nM). b For mouse plasma and microsomal stability assays, propantheline and dextromethorphan were used as the positive control, and their halflife (t 1/2 ) values were 66.3 and 16.0 min, respectively.
pressure. The residue was then washed with 2-propanol and dried, which afforded intermediate c1.

General procedure a for preparation of intermediates e1-e16
The substituted acyl chlorides (1.2 equiv.) and Et 3 N (1.2 equiv.) were added to a DCM solution of differently substituted methyl 2aminobenzoates (1 equiv.) The reaction mixture was stirred at room temperature for 3 h. The solvent was evaporated under reduced pressure. The residues were used for the next step without further purification.
The residues were then dissolved in ethanol, and hydrazine hydrate (1.5 equiv.) was added. The reaction mixture was stirred at 78 C for 10 h. The residue was then obtained by filtering the mixture and washed with ethanol to afford the intermediates e1-e16.   4.3.5. General procedure B for preparation of target compounds (f1-f16) At 0 C, HATU (1.5 equiv.) was added to the solution of the acid c1 (1.1 equiv.), followed by DIPEA (1.5 equiv.) dissolved in DMF (10 mL). After 10 min, the amines e1-16 (1.0 equiv.) were added, and the reaction mixture was stirred for 48 h at room temperature. The mixture was poured into the cold water. The precipitate was collected and washed with water and dried with anhydrous Na 2 SO 4 . The crude solid was purified through a silica gel column chromatography to afford the target compounds f1-16. The purity of all tested compounds was >95%, as determined by HPLC analysis.  N-(6-chloro-4-oxo-2-pentylquinazolin-3(4H)-yl)-1-ethyl-4hydroxy-2-oxo-1,2-dihydroquinoline-3-carboxamide (f2

Minimal inhibitory concentration (MIC) measurement
All of the bacterial strains used in this study were obtained from the Collection Centre of Pathogen Microorganism of Chinese Academy of Medical Sciences (CAMS-CCPM-A) in China. All of the isolates were stored at À80 C and streaked on tryptic soy agar (TSA) plates to obtain overnight cultures. The MICs of the test compounds were determined by the broth microdilution method in accordance with the Clinical and Laboratory Standards Institute (CLSI) guidelines. 20 Cation-adjusted Mueller-Hinton broth (CAMHB) was used as the growth medium for MIC assays. Briefly, 100 lL of serially diluted compounds (starting concentrations were 64 lg/ mL) in CAMH broth were added to the wells of 96-well microtiter plates. Then, 10 mL of 5 Â 10 6 CFU/mL bacterial culture was added to each well. The plates were incubated at 37 C for 18-20 h prior to the MIC determination. The MIC was defined as the lowest concentration of a compound that inhibited visual growth of the bacteria. Vancomycin was set as the positive control for this assay.

Cytotoxicity assay
Cytotoxicity in terms of CC 50 was determined by the SRB assay according to the protocol described in our previous publication. 16 Briefly, human HUVECs or HepG2 cells were seeded in the wells of 96-well plates at a concentration of 1 Â 10 5 cells per well and incubated at 37 C for 24 h. The cells were treated with the serially diluted compounds (10 nM and 100 lM) and incubated at 37 C with 5% CO 2 for 48 h. The cells were fixed with 10% trichloroacetic acid (w/v) at 4 C for 1 h, washed five times with distilled water, and stained with 0.4% SRB solution at room temperature for another 20 min. The cells were then washed five times with 1% acetic acid and air-dried. The protein-bound dye was dissolved in 10 mM Tris-base solution and shaken for 5 min. The optical density (OD) was measured at 540 nm using a microplate reader to estimate the cell viability (%). With the viability values (%) of the cells treated by the compound at different concentrations as input, the CC 50 value was determined by the non-linear regression with normalised dose-response fit implemented in GraphPad Prism version 5 software (GraphPad Software Inc., La Jolla, CA). The experiment was performed in duplicate, and paclitaxel was used as the positive drug for this assay.

Mouse plasma stability assay
The mouse plasma was purchased from Charles River and stored at À20 C. The stock solution of the test compound dissolved in DMSO (50 mM) was prepared and diluted to obtain a working solution at the concentration of 0.2 mM. The working solution of the control compound, i.e. propantheline, was similarly prepared (0.2 mM). Terfenadine (5 ng/mL) and tolbutamide (10 ng/mL) dissolved in acetonitrile were used as two quenching solutions.
The mouse plasma was pre-warmed at 37 C for 15 min. The working solution of the test/control compound (2 mL) was then added to the mouse plasma (398 mL) in a 96-well plate. Next, 30 mL of the reaction mixture was sampled and moved to 300 mL of quenching solutions at each time point: 0, 5, 15, 30, 60, and 120 min for the control compound and 0, 15, 30, 60, and 120 min for the test compound. When the sampling was done, each sample was centrifuged at 4000 rpm at 4 C for 15 min. Then, 100 mL of the supernatant was removed and mixed with 100 mL of distilled water for LC-MS/MS analysis with Shimadzu HPLC system and the AB Sciex API 4000 QTRAP instrument. The remaining compound (%) after incubation in plasma versus the incubation time was plotted. The half-life (t 1/2 ) value of the compound was determined via linear regression from the plot.

Mouse microsomal stability assay
The mouse liver microsomes were purchased from BioreclamationIVT and stored at À80 C. The working solutions (0.2 mM) of the control compound, i.e. dextromethorphan, and the test compound dissolved in DMSO were prepared. Terfenadine (5 ng/mL) and tolbutamide (10 ng/mL) dissolved in acetonitrile were used as two quenching solutions. NADPH was dissolved in the phosphate buffer (50 mM K2HPO4, pH 7.4) at 5 mM. The mouse liver microsomes were thawed at 37 C. The concentration of the stock solution was 20 mg/mL. The working solution (0.629 mg/mL) was prepared by dilution in phosphate buffer.
The working solution of the control/test compound (1.5 mL) was mixed with the liver microsome working solution (238.5 mL) in a 96-well plate and pre-incubated at 37 C for 5 min. The 5 mM NADPH working solution (60 mL) was added to the solution to initiate the metabolic reaction. At each time point (0, 5, 15, 30, and 60 for the control/test compound), 30 mL of the reaction mixture was sampled and moved to 300 mL of quenching solutions. When the sampling was done, each sample was centrifuged at 4000 rpm at 4 C for 15 min. The supernatant was analysed with the same HPLC system as the plasma stability assay. The residues (%) of the test/control compound along with the incubation time were calculated based on which the half-life (t 1/2 ) value was determined.