A dual system platform for drug metabolism: Nalbuphine as a model compound

Journal Pre-proof

A dual system platform for drug metabolism: Nalbuphine as a model compound Ren-Jong Liang , Yin-Ning Shih , Yen-Lun Chen , Wei-Yang Liu , Wan-Ling Yang , Shih-Yu Lee , Hong-Jaan Wang PII: DOI: Reference:

S0928-0987(19)30366-5 https://doi.org/10.1016/j.ejps.2019.105093 PHASCI 105093

To appear in:

European Journal of Pharmaceutical Sciences

Received date: Revised date: Accepted date:

4 July 2019 26 August 2019 28 September 2019

Please cite this article as: Ren-Jong Liang , Yin-Ning Shih , Yen-Lun Chen , Wei-Yang Liu , Wan-Ling Yang , Shih-Yu Lee , Hong-Jaan Wang , A dual system platform for drug metabolism: Nalbuphine as a model compound, European Journal of Pharmaceutical Sciences (2019), doi: https://doi.org/10.1016/j.ejps.2019.105093

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Elsevier B.V. All rights reserved.

Original Research Article

A dual system platform for drug metabolism: Nalbuphine as a model compound Ren-Jong Liang1,†, Yin-Ning Shih2,†, Yen-Lun Chen3, Wei-Yang Liu4, Wan-Ling Yang4, Shih-Yu Lee5, Hong-Jaan Wang2,3,4* 1

Medical Supplies and Maintenance, Hualien Armed Forces General Hospital, Hualien, Taiwan, Republic of China (R.O.C.) 2 Graduate Institute of Pharmacy, National Defense Medical Center, Taipei, Taiwan, Republic of China (R.O.C.) 3 Graduate Institute of Life Science, National Defense Medical Center, Taipei, Taiwan, Republic of China (R.O.C.) School of Pharmacy, National Defense Medical Center, Taipei, Taiwan, Republic of China (R.O.C.) 5 Graduate Institute of Aerospace and Undersea Medicine, National Defense Medical Center, Taipei, Taiwan, Republic of China (R.O.C.) 4

†

These two authors contributed equally to this work.

*

Corresponding author Address correspondence to: Hong-Jaan Wang, Ph.D. School of Pharmacy, National Defense Medical Center R9304, No.161, Sec. 6, Min-Chuan E. Rd., Neihu District, Taipei 114, Taiwan, R.O.C. Tel: +886-2-87923100 ext. 18849 Fax: +886-2-87924859 E-mail: [email protected]

Running Title Page Running title: A dual system platform for drug metabolism study Number of figures: 6 Number of tables: 3 Number of references: 35 Number of words in Abstract: 243 Number of words in Introduction: 746 Number of words in Discussion: 1564

Nonstandard abbreviations: CYP, cytochrome P450; fm, fraction metabolized; G6P, glucose-6-phosphate; G6PD, glucose-6-phosphate dehydrogenase; HLMs, human liver microsomes; HSA, human serum albumin; IVIVC, in vitro-in vivo correlation; LC/MS/MS, liquid chromatography/tandem mass spectrometry; NAL, nalbuphine; N3G, nalbuphine-3-glucuronide; N6G, nalbuphine-6-glucuronide; 3′-OH NAL, 3′-hydroxylnalbuphine; 4′-OH NAL, 4′-hydroxylnalbuphine; UDPGA, uridine diphosphate glucuronic acid; UGT, UDP-glucuronosyltransferase

Abstract Reaction phenotyping is a method commonly used for characterizing drug metabolism. It determines the drug metabolic pathways and ratios by measuring the metabolized fractions of individual enzymes. However, currently published results have focused on cytochrome P450 (CYP), while not considering phase II metabolism. Therefore, the morphinan analgesic, nalbuphine, primarily metabolized in the liver via CYPs and UDP-glucuronosyltransferases (UGTs), was selected as a model drug to establish a dual-phase platform to elucidate its comprehensive metabolic pathway. Enzyme kinetics was studied using 8 common recombinant (r)CYPs,10 rUGTs, and pooled human liver microsomes. The overall fraction of nalbuphine metabolized by each isozyme was evaluated by determining parent drug depletion. Finally, in vitro-in vivo correlation was validated in animal studies. Fluconazole, a specific UGT2B7 inhibitor, was administered orally to rats to determine the pharmacokinetic effects on nalbuphine and nalbuphine metabolites. Seventy-five percent and 25% of nalbuphine was metabolized by UGTs and CYPs, respectively. UGT2B7, UGT1A3, and UGT1A9 were primarily responsible for nalbuphine glucuronidation; only UGT2B7 produced nalbuphine-6-glucuronide. CYP2C9 and CYP2C19 were the two CYP isozymes that produced 3′-hydroxylnalbuphine and 4′-hydroxylnalbuphine. In vivo, the maximum serum concentration (Cmax) and area under the curve (AUC) of nalbuphine increased 12.4-fold and 13.2-fold, respectively, with fluconazole co-administration. A dual system platform for drug metabolism was successfully established in this study and was used to generate a complete

metabolic scheme for nalbuphine. This platform has been verified by in vivo evaluations and can be utilized to study drugs that undergo multisystem metabolism.

Keywords: reaction phenotyping, fraction metabolized, CYP, UGT, nalbuphine, N6G, metabolism

Introduction Reaction phenotyping is a method commonly used for characterizing drug metabolism. It determines drug biotransformation from individual catabolic enzymes and elucidates their metabolic ratios by measuring the fraction metabolized (fm). This allows drug-drug interactions to be predicted clinically (Di, 2017). In the early stages of drug development, once a drug is confirmed to be metabolized primarily by the liver, its in vivo metabolism is generally evaluated using an in vitro human liver-derived system (Zientek and Youdim, 2015). Phenotyping using human liver microsomes (HLMs) and specific inhibitors for cytochrome P450s (CYPs) is currently one of the most common methods for this purpose (Harper and Brassil, 2008; Lu et al., 2003; Zhang et al., 2007). Although an increasing number of drugs are known to be metabolized in vivo through phase II conjugation, particularly via UDP-glucuronosyltransferases (UGTs), the aforementioned methods are rarely used to identify the reaction phenotype of UGTs. Currently, most published results were obtained from CYP-based systems regardless of whether the drug’s metabolic fraction was calculated from correlation factors or parent drug depletion (Zientek and Youdim, 2015). This has occurred primarily because CYP reaction conditions are well-established, but the substrates, inhibitors, tissue contents, and optimum reaction conditions of phase II enzymes are still being discovered (Miners et al.,

1

2010; Walsky et al., 2012). Therefore, reaction phenotyping currently does not assess phase II metabolism, which leads to a lack of critical information, especially when the roles of UGTs in drug metabolism need to be investigated. Considering this, the model drug nalbuphine (NAL), which is metabolized by both CYPs and UGTs, was selected in this study, and a reaction phenotyping platform, which can consider dual-phase metabolism, was established as a reference for future research. NAL is a semisynthetic opioid analgesic with equal potency to morphine in pain treatment. It has agonist-antagonist effects on the κ-receptor and the μ-receptor and does not cause severe respiratory depression or drug addiction (Schmidt et al., 1985). Due to the “opioid crisis” resulting from the abuse of μ-receptor agonists, the advantages of mixed-type opioids in pain treatment have been broadly discussed (Liang et al., 2019). However, to date, the details of its mechanism in humans are still unclear. Our previous results confirmed that the metabolism of NAL involves phase I oxidation-reduction and phase II glucuronidation. 3′-Hydroxylnalbuphine (3′-OH NAL) and 4′-hydroxylnalbuphine (4'-OH NAL) are generated via CYP-mediated oxidative metabolism. Nalbuphine-3-glucuronide (N3G) and nalbuphine-6-glucuronide (N6G) are two conjugation metabolites produced by UGT activity (Liang et al., 2019; Wang et al., 2014a). Human in vivo experiments have also shown that the ratio of metabolite production via CYPs and

2

UGTs is approximately 23:77 (Wang et al., 2014a). Like morphine, NAL undergoes extensive first-pass metabolism in the intestine and the liver with an oral bioavailability of only ~15% (Wang et al., 2014b). Some prior reports have claimed that the main metabolic pathway of this drug in the human body is the oxidation of the cyclobutyl ring (Harrelson and Wong, 1988; Jaillon et al., 1989). However, considering the structural similarity to morphine, the main pathway of NAL metabolism should involve glucuronic acid conjugation (Errick and Heel, 1983; Schmidt et al., 1985). It is known that 90% of morphine is metabolized in the liver, of which 70% is transformed into morphine glucuronide (Sverrisdottir et al., 2015), but only ~5% is converted to normorphone by CYP3A4 and CYP2C8 (Lalovic et al., 2006; Trescot et al., 2008). From the fate of morphine, NAL was therefore anticipated to be highly metabolized via UGTs. When co-administering NAL, the risk of drug-drug interactions may occur with therapeutic agents that are also metabolized by UGTs. In addition, NAL metabolism is mediated by CYPs, but the proportion of each isozyme within the CYPs is still unknown. Accordingly, the metabolic ratio of each enzyme can be determined through reaction phenotyping and fm calculation, and a comprehensive metabolic pathway of NAL can be generated. In this study, we first proposed a dual-phase system to assess the overall fraction metabolized (fm, overall) by each enzyme in NAL metabolism. This method clearly defines the

3

extents to which different enzyme systems participate in the drug’s metabolism, and the overall metabolic assessment does not lose accuracy due to the limitations of a single enzyme system. Finally, after reaction phenotyping, animal studies were conducted to validate in vitro-in vivo correlation (IVIVC) (Jaiswal et al., 2014). Our study results can serve as a reference for future clinical use of NAL. Moreover, the dual-phase metabolic platform we developed can be applied to future multisystem drug metabolism studies.

Materials and methods 1.

Materials N3G, N6G, 3′-OH NAL, and 4′-OH NAL were synthesized chemically or biologically

by our laboratory (Liang et al., 2019; Wang et al., 2014a). Quinidine, niflumic acid, ticlopidine hydrochloride, sulfaphenazole, glucose-6-phosphate (G6P), glucose-6-phosphate dehydrogenase (G6PD), uridine diphosphate glucuronic acid (UDPGA), and nicotinamide adenine dinucleotide phosphate (NADP+) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Fluconazole was obtained from ACROS (Geel, Belgium). Buprenorphine hydrochloride was a gift from Lotus Pharmaceutical Co. Ltd. (Taipei, Taiwan, ROC). Recombinant human CYP and UGT SupersomesTM including CYP1A2, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP2E1, CYP3A4, CYP3A5, UGT1A1, UGT1A3, UGT1A4, UGT1A6, UGT1A9, UGT1A10, UGT2B4, UGT2B7, UGT2B15, and UGT2B17 were 4

purchased from BD Gentest (Woburn, MA, USA), and pooled human liver microsomes (HLMs) were from Thermo Fisher Scientific (Waltham, MA, USA). Nalbuphine hydrochloride was obtained from PhytoHealth Corp., (Taipei, Taiwan, ROC). Nalbuphine-D3 that served as an internal standard in the analyses was from Cerilliant (Round Rock, TX, USA). All other chemicals used were of reagent grade or higher.

2. Animals Male Sprague-Dawley rats (280 to 330 g) were purchased from BioLASCO Taiwan Co., Ltd. (Taipei, Taiwan, ROC). Experimental protocols involving animals were reviewed and approved by the Institutional Animal Care and Use Committee of the National Defense Medical Center (IACUC Approval Number: 19–128). Before study, the rats were allowed at least a one-week acclimation period at the climate-controlled animal quarter (22 ± 3 °C, 12-h light/dark cycle) with free access to food and water except for fasting 12 h before pharmacokinetic experiments.

3.

Reaction phenotyping: Metabolic reactions of NAL with recombinant CYPs and with recombinant UGTs Long-chain unsaturated fatty acids produced by microsomal membranes have been

reported to inhibit enzyme activity and increase KM, which in turn leads to decreased

5

intrinsic clearance of drugs by metabolic enzymes (Gill et al., 2012; Rowland et al., 2008). Therefore, the addition of bovine or human serum albumin (HSA) during the reaction can reduce the inhibitory activity of fatty acids. In this study, 0.5% (w/v) HSA was added to in vitro incubations to optimize enzyme reaction conditions. Eight, high-purity cDNA-expressed CYPs (rCYP1A2, rCYP2C8, rCYP2C9, rCYP2C19, rCYP2D6, rCYP2E1, rCYP3A4, and rCYP3A5) were used to determine the oxidative metabolism of NAL. The reactions were performed at two NAL concentrations (30 μM and 3 mM) in 0.1 M Tris-HCl buffer (Tris-HCL, pH 7.4). A total volume of 500 μl containing 20 pmol/ml SupersomesTM, 10 mM MgCl2, 0.5% HSA, 4 U/ml G6PD, and 20 mM G6P was employed in the enzyme incubations (Table 1). After preheating at 37°C for 5 min, the reaction was initiated by adding NADP+ (2 mM). After 1 h of incubation, the reaction was terminated by adding 1 ml of ice-cold methanol. Next, 50 μl of a 200 ng/ml NAL-D3 standard was added, and the mixture was transferred to a 2 ml microcentrifuge tube and centrifuged at 16,200 g for 20 min at 4°C. One milliliter of supernatant was collected and dried under a stream of nitrogen at 50°C and then reconstituted with 100 μl of the mobile phase. After centrifugation, the supernatant was collected for quantitative analysis of metabolites using liquid chromatography-tandem mass spectrometry (LC-MS/MS). After determining which enzymes were involved in the reaction,

6

metabolic reactions were performed at 8 NAL concentrations (10-3000 μM), and metabolite concentrations were similarly measured for enzyme kinetics analysis. The reaction rate was adjusted by linear optimization between enzyme concentration and reaction time. With respect to phase II metabolism, 10 high-purity cDNA-expressed UGTs (rUGT1A1, rUGT1A3, rUGT1A4, rUGT1A6, rUGT1A9, rUGT1A10, rUGT2B4, rUGT2B7, rUGT2B15, and rUGT2B17) were used to determine the glucuronidation of NAL. The metabolic reactions were performed at two NAL concentrations (30 μM and 3 mM) in the same manner as described above. A total reaction volume of 500 μl containing 0.1 mg/ml SupersomesTM, 10 mM MgCl2, 0.5% HSA, and 25 μg/ml alamethicin was used. The UGTs were pretreated with alamethicin for 30 min on ice. After preheating at 37°C for 5 min, the reaction was initiated by adding UDPGA (5 mM), and the remaining procedure including the calculation of reaction rate was performed as described above.

4.

Reaction phenotyping: Metabolic reactions of NAL with HLMs Eight reaction concentrations of NAL (10-3000 µM) were added to HLMs (0.4

mg/ml), and metabolic reactions were carried out at 37°C for 60 min. The reaction volume, composition, and conditions were as described above (Table 1). In addition, the coenzyme of the dual-phase enzyme system was added and the HLMs were pretreated with alamethicin (25 μg/ml) on ice for 30 min. After preheating at 37°C for 5 min, NADP+ (2 7

mM) and UDPGA (5 mM) were added to initiate the reaction, and the subsequent procedure was as described above. HLM inhibition experiments were carried out under separate reaction conditions for the phase I and phase II systems. Under the conditions for initiating the UGT system, 5 µM NAL was used as the reactant concentration, and the UGT-specific inhibitors buprenorphine (for UGT1A3, 100 μM), niflumic acid (for UGT1A9, 2.5 μM), and fluconazole (for UGT2B7, 2.5 mM) were used. After combining the reactants and inhibitors, HLMs (1 mg/ml) were added, and metabolic reactions were incubated at 37°C for 30, 45, and 60 min (Table 1). The conditions for initiating the CYP system were a reactant concentration of 1 µM NAL and the CYP-specific inhibitors sulfaphenazole (for CYP2C9, 2.5 μM), ticlopidine (for CYP2C19, 3 μM), and quinidine (for CYP2D6, 1 μM). After HLMs (1 mg/ml) were added, the metabolic reactions were carried out at 37°C for 45, 60, and 75 min (Table 1).

5.

Analytical process NAL, NAL metabolites, and the internal standard NAL-D3 were determined using

LC-MS/MS. The LC-MS/MS consisted of an Agilent Technologies 1200 series high performance liquid chromatograph (Boeblingen, Germany) coupled to a Biosystems-Sciex API 3000 series triple-quadrupole mass spectrometer (Foster City, CA, USA) with a 8

chromatographic C18 column (Waters Symmetry C18, 4.6 × 100 mm, 3.5 μm). The mobile phase was composed of solvent A (2.5 mM ammonium formate with 0.1% formic acid in water) and solvent B (0.1% formic acid in methanol) with a gradient elution program as follows: 0–10 min 90% (A), 10–10.10 min 90%–10% (A), 10.10–15 min 90% (A). The flow rate was 0.4 ml/min and the total run time was 15 min. Analyte retention times were 5.04 min for 4′-OH NAL, 6.25 min for 3′-OH NAL, 6.66 min for N3G, 7.29 min for NAL-D3, 7.31 min for NAL, and 7.73 min for N6G. The standard curves of all analytes (25-800 ng/ml) were linear with the coefficients of determination (R2) greater than 0.99.

6.

Enzyme kinetic assays The recombinant CYPs and UGTs, as well as pooled HLMs were used for in vitro

enzyme kinetics assays. Eight NAL concentrations ranging from 10 μM to 3000 μM were used for the metabolic reactions. LC-MS/MS with non-linear fitting was used to measure metabolite production in the enzyme kinetics analysis. The Michaelis-Menten plot (x-axis is the substrate concentration [S], y-axis is the metabolite production rate ν) was created using software (version 6.01; GraphPad Prism Software Inc., San Diego, CA, USA), linear transformation was performed, and the Eadie-Hofstee plot was generated with ν/[S] as the x-axis and ν as the y-axis for kinetics model interpretation. The optimal model for curve fitting and data calculation was found, the kinetic parameters of the corresponding enzymes 9

were obtained, and the following two equations were used to fit the enzyme kinetics models in this study. Michaelis-Menten kinetics equation: Vmax ×[S] KM +[S]

Eq. 1

Substrate inhibition kinetics equation:

ν

Vmax KM [S] 1+ [S] +

Eq. 2

where Ki is the constant for describing the substrate inhibition reaction.

7.

Determination of fm The NAL metabolic fraction for each enzyme was determined by enzyme activity

inhibition experiments using HLMs. The initial rate of disappearance of the parent drug is expressed as ke. Natural logarithm plots in either the presence or absence of inhibitor were made with the x-axis as time and the y-axis as the % disappearance of the parent drug (Yang et al., 2016). The fm of the enzyme in each phase (CYP or UGT system) was obtained from the following calculations, and the overall metabolic fraction (fm, overall) of the enzyme was calculated from the ratio of metabolite production in the in vitro tests.

slope × 2.303

Eq 3 10

inhibition

e, inh free

e, with inh

Eq. 4

e, inh free

sum of totoal

verall

8.

inhibition inhibition across CYP or UGT isozymes

× metabolite formation ratio in

system of overall

Eq. 5

system

Eq. 6

Pharmacokinetics study The UGT2B7 specific inhibitor fluconazole was selected to determine the effects on

the oral bioavailability of NAL in rat and to support the concept derived from the in vitro experiments. The animals were divided into two groups. The rats (n = 8) in the study group received fluconazole (50 mg/kg p.o. in saline containing 1% DMSO) 1.5 h before NAL (20 mg/kg) administration, while the rats (n = 8) in the control group received vehicle instead. Ten blood samples were taken from the caudal vein during the pharmacokinetic evaluation including the blank and at 5, 10, 20, 40, 60, 120, 180, 240, and 360 min after NAL administration. Whole blood samples (300 μl) combined with 20 μl of heparin (50 IU/ml) were centrifuged (4°C, 16,200 g, 10 min), and 150 μl of supernatant was transferred to a 1.5 ml microtube. The plasma samples were stored at -80°C until analysis. The methods used in the animal study, including blood sample preparation and analysis, were performed according to our previous reports with slight modifications (Chen et al., 2019; Wang et al., 2014a). The plasma levels of NAL and its four metabolites were determined by a Waters 11

AcquityTM ultrahigh-performance liquid chromatograph (Milford, MA, USA) coupled to a Biosystems-Sciex API 3000 series triple-quadrupole mass spectrometer (Foster City, CA, USA). A chromatographic C18 column (Waters Acquity UPLC BEH C18, 2.1 × 100 mm, 1.7 μm) was applied for analyte separations; the retentions were 1.12 min for 4′-OH NAL, 1.46 min for 3′-OH NAL, 1.60 min for N3G, 3.02 min for N6G, 3.26 min for NAL-D3, and 3.29 min for NAL. The linearity, precision, and accuracy of calibrations were used to evaluate the concentrations of all analytes in the rat plasma. The standard curves of all analytes (5-1000 ng/ml) were linear with R2 values greater than 0.9930. The precision and accuracy were determined by analyzing quality control (QC) samples in terms of relative standard deviation and relative error in three batches on separate days. Except for the lower limit of quantification, which was set at ≤ 20 , the acceptable precision and accuracy of all QC samples were set at ≤ 15% according to the USFDA bioanalytical method validation (USFDA, 2018, Guidance for Industry).

9.

Statistical analyses The pharmacokinetic parameters of the animal study were analyzed by

non-compartment model using WinNonlin version 5.3 software (Scientific Consulting Inc., North Carolina, USA). For NAL and each metabolite, statistical analyses of the pharmacokinetic data obtained for each group were compared by independent sample t test 12

at the 0.05 significance level using SPSS v.17 (SPSS Inc., Chicago, IL, USA).

Results 1. Metabolic reaction between NAL and cDNA-expressed CYPs The metabolic reaction of 20 pmol/ml rCYPs with NAL (30 μM and 3 mM) for 1 h at 37°C was used to evaluate whether the eight CYP isozymes (i.e. CYP1A2, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP2E1, CYP3A4, and CYP3A5) were involved in oxidative metabolic reactions of NAL. Analysis of C3′-hydroxylation showed that 3′-OH NAL was primarily produced by rCYP2C9 and rCYP2C19 (Fig. 1A), and the enzymatic activity of rCYP2C9 was greater than that of rCYP2C19 (Table 1). In addition, rCYP2D6 only had a low level of metabolic activity. At high substrate concentration (3 mM), the reactivity of the rCYP2C9 and rCYP2C19 isozymes decreased. Similarly, rCYP2C9 and rCYP2C19 are responsible for NAL C4′-hydroxylation (Fig. 1B). The activity of rCYP2C19 activity was far greater than that of rCYP2C9 (Table 1). The activity of both isozymes also decreased at high substrate concentrations, while rCYP3A4 possessed little metabolic activity under these conditions.

2. Metabolic reaction between NAL and cDNA-expressed UGTs

13

The metabolic reactions of the rUGTs at 0.1 mg/ml with two substrate concentrations of NAL (30 µM and 3 mM) for 1 h at 37°C were evaluated to determine whether the ten UGTs (i.e., UGT1A1, UGT1A3, UGT1A4, UGT1A6, UGT1A9, UGT1A10, UGT2B4, UGT2B7, UGT2B15, and UGT2B17) participate in NAL-conjugating reactions. The C3-glucuronidation of NAL was primarily mediated by rUGT1A3 and rUGT2B7, and the metabolic activity of rUGT2B7 was higher than that of rUGT1A3 (Fig. 1C). Although rUGT2B7 was twice as active as rUGT1A3 at low concentrations (30 μM), the metabolic activity of rUGT1A3 was higher than that of rUGT2B7 at high concentrations (3 mM). However, the activity of both isozymes decreased at high substrate concentrations. In addition, rUGT1A9 exhibited slight metabolic activity at both low and high substrate concentrations, but its metabolic activity was significantly lower than that of either rUGT1A3 or rUGT2B7 (Table 2). The other enzymes rUGT1A1, rUGT1A4, rUGT1A6, rUGT1A10, rUGT2B4, rUGT2B15, and rUGT2B17 also had minor metabolic activities at high substrate concentrations, but their product concentrations were below the quantitative range of the assay and were omitted. With respect to C6-glucuronidation of NAL, only rUGT2B7 produced N6G at low substrate concentration (30 μM), whereas at a high concentration (3 mM), rUGT1A1, rUGT1A4, rUGT2B7, and rUGT2B17 generated N6G (Fig. 1D). However, except for

14

rUGT2B7, the product concentrations for the enzymes were below the quantitative range of the assay; thus, rUGT2B7 is considered primarily responsible for NAL C6-glucuronidation. Furthermore, the C6-glucuronidation activity of NAL was less than that of C3-glucuronidation with respect to the amount of metabolite produced.

3. Analysis of NAL enzyme kinetics The reactions of rCYPs with different concentrations of NAL (10–3000 μM) are shown in Fig. 2A–B. The substrate inhibition model was applicable to the metabolic conversion of NAL by all rCYPs. The results of the metabolic reaction between rUGTs and NAL (10–3000 μM) are shown in Fig. 2C–D. The substrate inhibition model was also applicable to the C3 and C4-glucuronidation of NAL, which was catalyzed by rUGT1A3, rUGT1A9, and rUGT2B7. All kinetic parameters, namely Vmax, KM, and Ki are detailed in Table 2. The four metabolites obtained from the metabolic reactions of NAL using HLMs were identified by LC-MS/MS, of which two were the oxidative metabolites 3′-OH NAL and 4′-OH NAL and two were the glucuronidation metabolites N3G and N6G (retention times of 6.25, 5.04, 6.66, and 7.73 min, respectively) (Fig. S1). All four major metabolic pathways of NAL were interpreted as having substrate inhibition mechanisms based on Eadie-Hofstee plots; the fitted plots are shown in Fig. 2. Notably, the proportion of the four 15

metabolites in the dual-phase enzyme system changes with substrate concentrations in the HLM incubations (Fig. 3).

4. Metabolic reactions of NAL with HLMs and CYP-specific inhibitors The inhibitory effect of specific CYP inhibitors was determined in reactions using pooled HLMs. The effects of the specific CYP inhibitors sulphaphenazole (CYP2C9 specific), ticlopidine (CYP2C19 specific), and quinidine (CYP2D6 specific) on the enzyme activity at three time points (45, 60, and 75 min) are shown in Fig. 4A. Sulphaphenazole had a strong inhibitory effect (43%, 40%, and 49%, respectively), indicating that CYP2C9 is the major enzyme among CYPs in NAL metabolism. In contrast, ticlopidine (10%, 4%, and 17%, respectively) did not significantly alter metabolic reactions of NAL, indicating that CYP2C19 contributes lesser to NAL hydroxylation. Quinidine did not produce an obvious inhibitory effect (<1%) at any of the three time points, showing that CYP2D6 has a negligible effect on NAL metabolism. When all three inhibitors were added, the inhibitory effects were 99%, 100%, and 72%, respectively. In addition, the inhibition of metabolite production was similar to that of parent drug depletion. The production of 3′-OH NAL was primarily affected by sulphaphenazole (65%, 63%, and 62%, respectively), whereas ticlopidine (5%, 0%, and 5%, respectively), and quinidine (1%, 1%, and 4%, respectively) had almost no inhibitory effect (Fig. 4B). The production of 4′-OH NAL was primarily 16

inhibited by sulphaphenazole (49%, 44%, and 39%, respectively), followed by ticlopidine (26%, 16%, and 30%, respectively), and quinidine (3%, 0%, and 8%, respectively) (Fig. 4C). The fm values of CYP2C9, CYP2C19, and CYP2D6 were calculated using Eq. (5) after determining the excretion rate ke from the curve slope. The fm values were 0.76, 0.24, and ~0, respectively (Fig. 4A).

5. Metabolic reactions of NAL with HLMs and specific UGT inhibitors The inhibitory effect of UGT-specific inhibitors was determined in reactions using pooled HLMs. The inhibitory effects of the specific UGT inhibitors buprenorphine (UGT1A3 specific), niflumic acid (UGT1A9 specific), and fluconazole (UGT2B7 specific) on the enzyme activity at three time points (30, 45, 60 min) are shown in Fig. 4D. Buprenorphine (84%, 73%, and 68%, respectively) and fluconazole (82%, 80%, and 78%, respectively) showed extremely strong inhibitory effects, indicating that UGT1A3 and UGT2B7 are the major metabolic enzymes in the UGT system for NAL conjugation. In contrast, niflumic acid had a weaker inhibitory effect (19%, 11%, and 4%, respectively), indicating that UGT1A9 makes a smaller contribution to the glucuronidation of NAL. When all three inhibitors were added, the metabolism of NAL was almost completely blocked, and metabolite production was similar to that of parent drug depletion. The production of N3G was primarily inhibited by buprenorphine (87%, 79%, and 68%, respectively) and 17

fluconazole (82%, 80%, and 78%, respectively), whereas the effect of niflumic acid (8%, 12%, 11%) was less pronounced. The production of N6G was primarily inhibited by buprenorphine (74%, 61%, and 56%, respectively) and fluconazole (69%, 68%, and 65%, respectively), whereas niflumic acid (2%, 0%, 4%) had almost no inhibitory effect. The fm values of UGT2B7, UGT1A3, and UGT1A9 were calculated by Eq. (5) after determining the excretion rate ke from the curve slope. The fm values were 0.50, 0.46, and 0.04, respectively. Based on the formation ratio (CYP: UGT = 25:75) observed at high reactant concentrations (Table S1), the fm,overall of each CYP and UGT isozyme calculated by Eq. (6) are shown in Fig. 5 along with the proposed NAL metabolic scheme.

6. Pharmacokinetic evaluations in rats Linearity (Table S2), accuracy, and precision (Table S3) all met the minimal requirements in accordance with bioanalytical guidelines; thus, the determination of plasma samples was reliable. In order to achieve the same concentration of fluconazole in vivo as in in vitro (2.5 mM ≈ 765 μg/ml) to inhibit UGT2B7 appreciably, previous reports have recommended an oral pretreatment of fluconazole (50 mg/kg) 1.5 h before NAL administration in vivo (Azeredo et al., 2012). As shown in Fig. 6, pretreatment with fluconazole significantly increased NAL plasma concentration, with Cmax increaseing from 38.2 ng/ml to 475.5 ng/ml, and AUCt increasing from 60.5 h·ng/ml to 797.8 h·ng/ml. 18

Compared to that of the control group, the Cmax and AUCt increased by 12.4-fold and 13.2-fold, respectively (Table 3). With respect to the metabolites, the Cmax of 3′-OH NAL increased from 436.3 ng/ml to 1103.3 ng/ml, the Cmax of 4′-OH NAL increasing from 65.3 ng/ml to 299.9 ng/ml, and AUCt increased by 4.5-fold and 5.7-fold, respectively. The effect on the UGT system was smaller, with the Cmax of N3G increasing from 168.3 ng/ml to 283.6 ng/ml (experimental group) and AUCt increasing by 1.33-fold. Compared to those of the control group, the metabolic ratios (ratio of parent drug to metabolite, MR) of NAL relative to the three metabolites 3′-OH NAL, 4′-OH NAL, and N3G were MRAUCt (0.49 vs 0.15, 0.68 vs 0.26, and 2.18 vs 0.20, respectively) and MRCmax (0.44 vs 0.08, 1.59 vs 0.51, and 1.70 vs 0.22, respectively). The differences in MRAUCt and MRCmax between the two groups were statistically significant (p < 0.001). These data reveal that fluconazole significantly inhibits NAL metabolism in rats (Fig. S2).

Discussion The present investigation is the first comprehensive mechanistic study of NAL metabolism, and its complete metabolic pathway was constructed through a dual-phase catalytical platform. The CYP and UGT isozymes in NAL metabolism were clearly delineated through reaction phenotyping, and a cross-system metabolic ratio was proposed for estimating their fm,overall (Eq. (6)). 19

Reaction phenotyping revealed that the UGT responsible for NAL metabolism was primarily UGT2B7 (fm = 0.50), followed closely by UGT1A3 (fm = 0.46). Enzyme kinetic assays indicated that the metabolic activity (V max) of UGT1A3 for NAL was no less than that of UGT2B7 (Table 2), suggesting that UGT1A3 also plays an important role in the metabolism of morphine-like drugs. However, it must be noted that the degree of NAL metabolism by UGT2B7 may be underestimated due to insufficient specificity of fluconazole and buprenorphine for the inhibition of UGTs. Previous reports have shown that buprenorphine, a specific inhibitor of UGT1A3, is also metabolized by UGT2B7 (Rouguieg et al., 2010). Due to its inhibition of UGT2B7, buprenorphine could likely overestimate the capacity of UGT1A3 for NAL metabolism. According to our reaction phenotyping results, N6G is only produced from UGT2B7 (Fig. 1D). However, N6G formation was significantly inhibited by buprenorphine at all time points (Fig. 4F), indicating that buprenorphine not only reduced UGT1A3 activity but also interfered with UGT2B7 simultaneously during the reactions. Therefore, the lack of specificity inevitably leads to overestimate the fm of UGT1A3, as well as underestimate the fm of UGT2B7 (Eq. (5)). As for the CYP system, CYP2C19 was much more active than CYP2C9 in NAL C4′-hydroxylation (Table 2), but the enzyme inhibition study demonstrated that CYP2C9 plays a more important role in NAL metabolism (fm,overall = 0.19). It is believed that CYP2C9 is much more abundant in the liver

20

than CYP2C19 (Zhang et al., 2016), making the former more important in overall metabolism. NAL undergoes both phase I and phase II metabolic reactions in vivo. Since the published calculation of fm depends only on a single enzyme system, there was no calculation method that encompassed dual-phase metabolism. Therefore, to calculate the fraction metabolized concurrently in different systems and reflect in vivo drug metabolism, we developed fm,overall which considers the overall metabolic ratio. However, fm,overall was estimated from the metabolic ratios obtained at different reactant concentrations (10–3000 μM), and it was found that the metabolite formation ratio shifted as the substrate concentration changed. NAL is highly metabolized in the liver and the extraction ratio (ER) is 0.7 (Hawi et al., 2015; Jaillon et al., 1989). Our previous reports have indicated that the maximal plasma concentration of NAL (represented as Cout in the following formula) after 66 mg p.o. administration would be 20.6 ± 6.5 ng/ml in human body (Wang et al., 2014b). Assuming that the metabolism pattern of NAL in the liver was following the well-stirred model, the concentration entry into the liver (Cin) could be calculated by the equation ER = (Cin−Cout)/Cin. Thus, the Cin of NAL was roughly estimated as 100 ng/ml, which would be 0.25 µM. The in vitro metabolism performed with HLMs was therefore carried out using 0.25 µM NAL as a reference. However, the dual-phase metabolic ratio (CYP:UGT = 36:64)

21

obtained from the reaction was very different from the previously accepted metabolic mechanism of morphinan derivatives. At this concentration of NAL, its metabolism by CYPs was nearly 40%, whereas the known high-capacity UGTs accounted for only ~60%. However, at the high reactant concentrations of NAL (> 500 µM) with HLMs, the dual-phase metabolite production (CYP:UGT = 25:75) approached the metabolite ratio in the human body (CYP:UGT = 23:77, estimated by AUCinf) (Wang et al., 2014a). Notably, when using in vitro experiments to predict in vivo metabolism, estimates are generally made from first-order reactions at concentrations approaching those in actual organisms (Harper and Brassil, 2008). Therefore, in the absence of human data, CYPs can be mistakenly assumed to contribute almost 40% to the overall metabolism of NAL. Consequently, the choice of substrate concentration in in vitro tests may be a critical factor in IVIVC. However, regardless of the substrate concentration used in the reaction, the proportion of NAL metabolism via CYPs was estimated to be > 25%. This means that phase I metabolism is indispensable for NAL depletion in the human body and it reduces the fraction metabolized by UGTs. The result differs completely from previous predictions, suggesting that NAL is similar to morphine and has an extremely low metabolic ratio (~5%) in the CYP system. Therefore, phase II conjugation is not unique in the metabolism of opioid analgesics, but the importance of CYPs also cannot be ignored.

22

UGT2B7, UGT1A3, CYP2C9, and CYP2C19 were confirmed as critical enzymes of NAL metabolism in HLMs, but extrapolation of these in vitro results to in vivo experiments is needed for further corroboration. Drugs with high fm (e.g. midazolam > 0.85) combined with enzyme-specific inhibitors increase the AUC ratio significantly, making it easy to observe the occurrence of drug interactions (Di, 2017). However, the fm,overall of UGT2B7 in the present study was approximately 0.3 (Fig. 5), which is quite different from previously predicted values, suggesting that NAL was metabolized primarily by UGT2B7. Fluconazole is known to block UGT2B7 effectively, but also to inhibit CYP2C9 and CYP2C19 (Niwa et al., 2005). It was expected to have inhibitory effects on all metabolic enzymes, thereby blocking the in vivo NAL metabolic pathway. In rats, CYP2C11 and UGT2B1 are major enzymes in the phase I and II systems and are homologous to human CYP2C9 and UGT2B7, respectively (Chen et al., 2016; King et al., 2001). Due to this homology with the human enzymes, we speculated that rat UGT2B1 and CYP2C11 also play the major roles in NAL metabolism. To justify the in vitro results, fluconazole, a moderate to strong repressor of both UGT2B1 and CYP2C (Mano et al., 2007; Scheer et al., 2012) was selected for co-administration with NAL in our animal studies. Although the fm of various enzymes involved in NAL metabolism were small, the non-selective inhibitory effects of fluconazole blocked the drug’s metabolic pathways and was expected to result in significant alterations

23

of NAL pharmacokinetic properties in rats. The Cmax and AUC of NAL in rats were greatly increased, indicating that the four enzymes UGT2B7, UGT1A3, CYP2C9, and CYP2C19, which were confirmed by reaction phenotyping, are truly important in NAL metabolism. However, the plasma concentrations of 3′-OH NAL, 4′-OH NAL, and N3G increased as the metabolism of the parent drug decreased. This result is counterintuitive, since blocking metabolism of the parent drug should reduce metabolite production and lead to decreased metabolite concentration in the blood. However, the experimental observations contradict this assumption. A proposed hypothesis is that fluconazole also blocks the secondary metabolism of “metabolites” in rats, which leads to accumulation of 3′-OH NAL, 4′-OH NAL, and N3G in the body and increased plasma concentrations. Additionally, enzyme-transporter interplay is a possible mechanism to explain some of this controversy (Lau et al., 2007) and it makes the situation much more complicated, particularly when the drug is subjected to multi-metabolic pathways. Lacking sufficient information about other potential effects, the reason why the plasma levels of NAL metabolites were elevated remains unclear. Nevertheless, MR calculations have indicated that fluconazole effectively inhibits the metabolic pathways mediated by UGT2B7, UGT1A3, CYPC2C9, and CYP2C19 (van der Weide et al., 2005; Wang et al., 2019). The MR values of NAL relative to the three metabolites show that

24

MRAUCt and MRCmax were significantly increased in the experimental group compared to the control group values (Fig. S2), which also indicates that the metabolic activity of UGT2B7 (UGT2B1 in rats) and CYPC2C9 (CYP2C11 in rats) is indeed reduced by fluconazole, thus reflecting the in vitro outcomes. The accuracy of IVIVC as applied to drug metabolism has always been criticized and is open to discussion (Jaiswal et al., 2014). Using NAL as the drug tested in vitro with HLMs in the present study, only the dual system metabolic ratio obtained at high substrate concentrations (> 500 µM) matched the results observed in the clinical trial. Thus, the prediction of in vivo metabolism by reaction phenotyping somehow requires consideration of other correction factors. In addition, unlike drugs that undergo single-phase metabolism, the dual-phase metabolism of NAL requires simultaneous initiation of both UGTs and CYPs, which makes calculating the fm of individual enzymes practically impossible. Since the UGT inhibitors selected for the assays are not as specific as the CYP inhibitors, these literature-recommended UGT inhibitors frequently regulate CYPs as well. For example, the UGT2B7 inhibitor fluconazole is also a potent inhibitor of CYP2C9 and CYP2C19 (Niwa et al., 2005). Therefore, when calculating the fm of UGT2B7, the metabolic ratio of UGT2B7 to NAL is often overestimated. Considering this, the dual-phase enzymes were initiated separately, and the metabolic ratios of UGTs and CYPs in the isolated system were

25

evaluated, which is beneficial to accurately calculate the fm of individual isozymes. Reaction phenotyping, together with in vivo testing, demonstrated that NAL possesses a unique metabolism. The role of CYPs in this metabolism cannot be overlooked. Among them, CYP2C9, and not CYP3A4, is the most important. With respect to phase II conjugation, UGT1A3 also plays an important role in addition to UGT2B7. Therefore, NAL metabolism is different than previously believed and is not necessarily the same or similar to that of morphine. Even if two compounds have similar structures, they may have completely different fates in the human body.

Conclusion The results of the present study can be used as a reference for the clinical application of NAL. Using cDNA-expressed enzymes, UGT1A3, UGT1A9, UGT2B7, CYP2C9, CYP2C19, and CYP2D6 were identified to participate in NAL metabolism, and studies of their respective enzyme kinetics were performed. A detailed NAL metabolic pathway was constructed via a dual-phase metabolism platform. This platform can be further applied to the study of multisystem drug metabolism. The fm,overall of UGT2B7 is higher than that of other isozymes, and it is the primary enzyme for NAL metabolism. Unlike morphine, CYP2C9 and CYP2C19 account for ~25% of the overall NAL metabolism. Clinically relevant drug interactions may need to be re-evaluated, especially when azole-derivative 26

antifungals or the anticoagulant warfarin are used during analgesic treatments. Finally, it should be noted that species differences are always an inevitable risk and cannot be ignored when predicting outcomes of in vivo systems from in vitro models. Further human clinical verification is still required in predicting an accurate IVIVC.

Acknowledgments This

work

was

supported

by

grants

(MOST106-2320-B-016-001

and

MOST103-2320-B-016-007-MY3) from the Ministry of Science and Technology, Taipei, Taiwan (R.O.C.), and a grant (805-C108-13) from the Hualien Armed Forces General Hospital, Hualien, Taiwan (R.O.C.). We express our gratitude to Dr. Li-Heng Pao (Chang Gung University of Science and Technology) and Dr. Cheng-Huei Hsiong (National Defense Medical Center) for their valuable advices in data interpretation. Special thanks to Dr. Yune-Fang Ueng (National Research Institute of Chinese Medicine) for her technical assistance in dealing with enzyme kinetic assays.

Conflicts of Interest The authors declare no conflicts of interest.

27

Reference Azeredo, F.J., de Araujo, B.V., Haas, S.E., Torres, B., Pigatto, M., de Andrade, C., Dalla Costa, T., 2012. Comparison of fluconazole renal penetration levels in healthy and Candida albicans-infected Wistar rats. Antimicrob. Agents Chemother. 56, 5852-5857. Chen, A., Zhou, X., Tang, S., Liu, M., Wang, X., 2016. Evaluation of the inhibition potential of plumbagin against cytochrome P450 using LC-MS/MS and cocktail approach. Sci. Rep. 6, 28482. Chen, W.C., Huang, P.W., Yang, W.L., Chen, Y.L., Shih, Y.N., Wang, H.J., 2019. Fundamentals of Pharmacokinetics to Assess the Correlation Between Plasma Drug Concentrations and Different Blood Sampling Methods. Pharm. Res. 36, 32. Di, L., 2017. Reaction phenotyping to assess victim drug-drug interaction risks. Opin. Drug Discov. 12, 1105-1115. Errick, J.K., Heel, R.C., 1983. Nalbuphine. A preliminary review of its pharmacological properties and therapeutic efficacy. Drugs. 26, 191-211. Gill, K.L., Houston, J.B., Galetin, A., 2012. Characterization of in vitro glucuronidation clearance of a range of drugs in human kidney microsomes: comparison with liver and intestinal glucuronidation and impact of albumin. Drug Metab. Dispos. 40, 825-835. Harper, T.W., Brassil, P.J., 2008. Reaction Phenotyping: Current Industry Efforts to Identify Enzymes Responsible for Metabolizing Drug Candidates. AAPS J. 10, 200-207. Harrelson, J.C., Wong, Y.J., 1988. Species variation in the disposition of nalbuphine and its acetylsalicylate ester analogue. Xenobiotica. 18, 1239-1247. Hawi, A., Alcorn, H., Jr., Berg, J., Hines, C., Hait, H., Sciascia, T., 2015. Pharmacokinetics of nalbuphine hydrochloride extended release tablets in hemodialysis patients with exploratory effect on pruritus. BMC Nephrol. 16, 47-47. Jaillon, P., Gardin, M.E., Lecocq, B., Richard, M.O., Meignan, S., Blondel, Y., Grippat, J.C., 28

Bergnieres, J., Vergnoux, O., 1989. Pharmacokinetics of nalbuphine in infants, young healthy volunteers, and elderly patients. Clin Pharmacol Ther 46, 226-233. Jaiswal, S., Sharma, A., Shukla, M., Vaghasiya, K., Rangaraj, N., Lal, J., 2014. Novel pre-clinical methodologies for pharmacokinetic drug-drug interaction studies: spotlight on "humanized" animal models. Drug Metab. Rev. 46, 475-493. King, C., Tang, W., Ngui, J., Tephly, T., Braun, M., 2001. Characterization of rat and human UDP-glucuronosyltransferases responsible for the in vitro glucuronidation of diclofenac. Toxicol. Sci. 61, 49-53. Lalovic, B., Kharasch, E., Hoffer, C., Risler, L., Liu-Chen, L.Y., Shen, D.D., 2006. Pharmacokinetics and pharmacodynamics of oral oxycodone in healthy human subjects: role of circulating active metabolites. Clin. Pharmacol. Ther. 79, 461-479. Lau, Y.Y., Huang, Y., Frassetto, L., Benet, L.Z., 2007. effect of OATP1B transporter inhibition on the pharmacokinetics of atorvastatin in healthy volunteers. Clin. Pharmacol. Ther. 81, 194-204. Liang, R.J., Lai, Y.H., Kao, Y.T., Yang, T.H., Chen, Y.L., Wang, H.J., 2019. A novel finding of nalbuphine-6-glucuronide, an active opiate metabolite, possessing potent antinociceptive effects: Synthesis and biological evaluation. Eur. J. Med. Chem. 178, 544-551. Lu, A.Y., Wang, R.W., Lin, J.H., 2003. Cytochrome P450 in vitro reaction phenotyping: a re-evaluation of approaches used for P450 isoform identification. Drug Metab. Dispos. 31, 345-350. Mano, Y., Usui, T., Kamimura, H., 2007. Comparison of inhibition potentials of drugs against zidovudine glucuronidation in rat hepatocytes and liver microsomes. Drug Metab. Dispos. 35, 602-606. Miners, J.O., Mackenzie, P.I., Knights, K.M., 2010. The prediction of drug-glucuronidation parameters in humans: UDP-glucuronosyltransferase enzyme-selective substrate and

29

inhibitor probes for reaction phenotyping and in vitro-in vivo extrapolation of drug clearance and drug-drug interaction potential. Drug Metab. Rev. 42, 196-208. Niwa, T., Shiraga, T., Takagi, A., 2005. Effect of antifungal drugs on cytochrome P450 (CYP) 2C9, CYP2C19, and CYP3A4 activities in human liver microsomes. Biol. Pharm. Bull. 28, 1805-1808. Rouguieg, K., Picard, N., Sauvage, F.L., Gaulier, J.M., Marquet, P., 2010. Contribution of the different UDP-glucuronosyltransferase (UGT) isoforms to buprenorphine and norbuprenorphine metabolism and relationship with the main UGT polymorphisms in a bank of human liver microsomes. Drug Metab. Dispos. 38, 40-45. Rowland, A., Knights, K.M., Mackenzie, P.I., Miners, J.O., 2008. The "albumin effect" and drug glucuronidation: bovine serum albumin and fatty acid-free human serum albumin enhance the glucuronidation of UDP-glucuronosyltransferase (UGT) 1A9 substrates but not UGT1A1 and UGT1A6 activities. Drug Metab. Dispos. 36, 1056-1062. Scheer, N., Kapelyukh, Y., Chatham, L., Rode, A., Buechel, S., Wolf, C.R., 2012. Generation and characterization of novel cytochrome P450 Cyp2c gene cluster knockout and CYP2C9 humanized mouse lines. Mol. Pharmacol. 82, 1022-1029. Schmidt, W.K., Tam, S.W., Shotzberger, G.S., Smith, D.H., Clark, R., Vernier, V.G., 1985. Nalbuphine. Drug Alcohol Depend. 14, 339-362. Sverrisdottir, E., Lund, T.M., Olesen, A.E., Drewes, A.M., Christrup, L.L., Kreilgaard, M., 2015. A review of morphine and morphine-6-glucuronide's pharmacokinetic-pharmacodynamic relationships in experimental and clinical pain. Eur. J. Pharm. Sci. 74, 45-62. Trescot, A.M., Datta, S., Lee, M., Hansen, H., 2008. Opioid pharmacology. Pain Physician. 11, S133-153. U.S. Food and Drug Administration. Bioanalytical Method Validation Guidance for Industry.

30

2018. van der Weide, J., van Baalen-Benedek, E.H., Kootstra-Ros, J.E., 2005. Metabolic ratios of psychotropics as indication of cytochrome P450 2D6/2C19 genotype. Ther. Drug Monit. 27, 478-483. Walsky, R.L., Bauman, J.N., Bourcier, K., Giddens, G., Lapham, K., Negahban, A., Ryder, T.F., Obach, R.S., Hyland, R., Goosen, T.C., 2012. Optimized assays for human UDP-glucuronosyltransferase (UGT) activities: altered alamethicin concentration and utility to screen for UGT inhibitors. Drug Metab. Dispos. 40, 1051-1065. Wang, H.J., Hsiong, C.H., Pao, L.H., Chang, W.L., Zhang, L.J., Lin, M.J., Ho, S.T., Huang, P.W., Hu, O.Y.P., 2014a. New finding of nalbuphine metabolites in men: NMR spectroscopy and UPLC–MS/MS spectrometry assays in a pilot human study. Metabolomics. 10, 709-718. Wang, H.J., Hsiong, C.H., Ho, S.T., Lin, M.J., Shih, T.Y., Huang, P.W., Hu, O.Y.P., 2014b. Commonly used excipients modulate UDP-glucuronosyltransferase 2b7 activity to improve nalbuphine oral bioavailability in humans. Pharm. Res. 31, 1676-1688. Wang, Z., Sun, W., Lin, Z.-F., Sun, R., Huang, C.-K., Ye, W.-J., Dong, Y.-Y., Zhang, X.-D., Chen, R.-J., 2019. A UHPLC-MS/MS method coupled with liquid-liquid extraction for the quantitation of phenacetin, omeprazole, metoprolol, midazolam and their metabolites in rat plasma and its application to the study of four CYP450 activities. J. Pharm. Biomed. Anal. 163, 204-210. Yang, X., Atkinson, K., Di, L., 2016. Novel Cytochrome P450 Reaction Phenotyping for Low-Clearance Compounds Using the Hepatocyte Relay Method. Drug Metab. Dispos. 44, 460-465. Zhang, H., Davis, C.D., Sinz, M.W., Rodrigues, A.D., 2007. Cytochrome P450 reaction-phenotyping: an industrial perspective. Expert Opin. Drug Metab. Toxicol. 3,

31

667-687. Zhang, H.F., Wang, H.H., Gao, N., Wei, J.Y., Tian, X., Zhao, Y., Fang, Y., Zhou, J., Wen, Q., Gao, J., Zhang, Y.J., Qian, X.H., Qiao, H.L., 2016. Physiological Content and Intrinsic Activities of 10 Cytochrome P450 Isoforms in Human Normal Liver Microsomes. J. Pharmacol. Exp. Ther. 358, 83-93. Zientek, M.A., Youdim, K., 2015. Reaction phenotyping: advances in the experimental strategies used to characterize the contribution of drug-metabolizing enzymes. Drug Metab. Dispos. 43, 163-181.

32

Figure captions

Fig. 1. Eight cDNA-expressed CYP isozymes and ten cDNA-expressed UGT isozymes were screened for NAL metabolism. Each cDNA-expressed enzyme was reacted at substrate concentrations of 30 µM and 3 mM for 1 h at 37°C; enzymatic activity is expressed as the rate of metabolite production: (A) 3′-OH NAL, (B) 4′-OH NAL, (C) N3G, and (D) N6G. Data expressed as means ± SD, n = 3.

33

Fig. 2. Michaelis-Menten plots showing the enzyme kinetics models of NAL hydroxylation and glucuronidation by rCYPs, rUGTs, and HLMs. Insets show Eadie-Hofstee transformations. (A) C3′-hydroxylation using rCYP2C9, rCYP2C19, rCYP2D6, HLMs; (B) C4′-hydroxylation using rCYP2C9, rCYP2C19, HLMs; (C) C3-glucuronidation using

34

rUGT1A3, rUGT1A9, rUGT2B7, HLMs and (D) C6-glucuronidation using rUGT2B7 and HLMs.

Fig. 3. Dual-phase metabolic reaction initiated by HLMs. The ratio of metabolites produced from the CYP and UGT systems varied with different reactant concentrations of NAL.

35

Fig. 4. Single-phase metabolic reactions of NAL (1 µM) and HLMs (1 mg/ml) initiated by the CYP system, as well as NAL (5 µM) and HLMs (1 mg/ml) initiated by the UGT system are respectively plotted in (A) and (D). The inhibition effects of 3′-OH NAL and 4′-OH NAL productions with three CYP-specific inhibitors at different time points (45, 60, and 75 min) are shown in (B) and (C). The inhibition of N3G and N6G formations using three UGT-specific inhibitors at different time points (30, 45, and 60 min) is indicated in (E) and (F). Data expressed as a percentage relative to the control (n = 3 in each group).

36

37

Fig. 5. Based on the dual-phase metabolite formation ratio (25:75) across CYP and UGT systems with HLMs, the overall fraction of NAL metabolized (fm,overall) for each isozyme was calculated and is indicated in the proposed NAL metabolic scheme.

38

Fig. 6. Mean plasma concentration-time curve in rats (n = 8 in each group) after an oral dose of 20 mg/kg NAL with or without fluconazole from time 0 to 6 h. In the control group (○), only NAL was administered; in the experimental group (●), fluconazole (50 mg/kg, p.o.) was pre-administered 1.5 h before NAL administration. Inset plot is depicted in semi-log scale.

Table 1. System components and reaction protocol in NAL reaction phenotyping a. Determination of fm

Enzyme Kinetic Assays Supersomesb

ponent (unit)

HLMsc

HLMsc

rCYPs

rUGTs

CYPs + UGTs

CYPs

UGTs

M)

0.1

0.1

0.1

0.1

0.1

O (mM)

10

10

10

10

10

-

25

25

-

25

20

0.2

0.4

1

1

4

-

4

4

-

20 10–3000

10–3000

20 10–3000

20 2.5 3 1 1

100 2.5 2.5 5

0.5

0.5

0.5

0.5

0.5

-

5

5

-

5

2

-

2

2

-

60

60

60

45, 60, 75

(μg/ml)

d

centration (mg/ml)

l)

azole (CYP2C9) (μM) HCl (CYP2C19) (μM) CYP2D6) (μM) ine (UGT1A3) (μM) id (UGT1A9) (μM) (UGT2B7) (mM) μM)

M)

M)

ime (min)

Total volume: 500 μl a

Numeric values in the table are indicated as the reactant concentrations. Corning® Supersomes: cDNA-encoded recombinant CYP and UGT isozymes. c HLMs: human liver microsomes. d Unit here is pmol/ml. b

Table 2. Kinetic parameters for NAL oxidative metabolism and glucuronidation were obtained by using rCYPs, rUGTs, and HLMs in the presence of HSA 0.5% (w/v) after fitting to a substrate inhibition kinetics model. Pathway/Enzyme

KM

Vmax 39

Ki

30, 45,

(μM)

(pmol/min per unit of isozymea)

(μM)

rCYP2C9

13.0 ± 3.00

1720.0 ± 100.1

1084.0 ± 176.2

rCYP2C19

12.8 ± 1.9

1067.0 ± 32.3

1923.0 ± 235.9

rCYP2D6

9.1± 1.4

457.0 ± 12.8

4909.0 ± 764.6

HLMs

30.3 ± 6.0

275.9 ± 20.5

505.7 ± 77.6

rCYP2C9

15.0 ± 2.7

4362.0 ± 231.4

592.7 ± 72.9

rCYP2C19

10.7 ± 1.5

16516.0 ± 638.9

579.9 ± 54.4

HLMs

25.2 ± 4.0

904.4 ± 54.6

386.8 ± 47.4

rUGT1A3

315.8 ± 50.2

1118.0 ± 107.5

1086.0 ± 176.2

rUGT1A9

32.8 ± 4.7

68.3 ± 2.7

2946.0 ± 403.1

rUGT2B7

46.8 ± 4.9

1066.0 ± 39.3

1537.0 ± 143.7

HLMs

30.7 ± 3.8

1784.0 ± 72.2

1011.0 ± 99.0

rUGT2B7

67.4 ± 4.4

143.2 ± 3.5

2123.0 ± 138.9

HLMs

48.1 ± 5.2

338.5 ± 13.6

1198.0 ± 112.5

C3′-hydroxylation

C4′-hydroxylation

C3-glucuronidation

C6-glucuronidation

a

Unit of each isozyme is: rCYPs (nmol), rUGTs (mg) and HLMs (mg).

Data are means ± SE from triplicate experiments.

Table 3. Pharmacokinetic parameters of oral NAL (20 mg/kg) in rats. Study group a NAL + fluconazole (n = 8)

Control group NAL (n = 8)

Pharmacokinetic parameter NAL

3'-OH NAL

4'-OH NAL

N3G

Cmax (ng/ml)

38.2 ± 436.3 ± 65.3 ± 168.3 ± 36.2 247.9 42.4 37.8

Tmax (h)

0.2 ± 0.1

AUCt (h·ng/ml) T1/2 (h)

NAL

3'-OH NAL

4'-OH NAL

475.5 ±

1103.3 ±

299.9 ±

141.8

***

0.2 ± 0.0 0.3 ± 0.2 0.3 ± 0.0 0.2 ± 0.1

153.8

***

48.7

N3G

***

0.2 ± 0.1 0.8 ± 0.6

*

283.6 ± ** 68.8 0.3 ± *

0.1 ± ± ± 1637.2 1177.5 60.5 ± 362.6 ± 208.0 ± 293.3 ± 797.8 388.8 ± *** *** *** 51.3 181.0 93.9 38.6 133.6 201.3 309.5 159.9 2.0 ± 3.8 ± 7.3 ± 3.1 ± 2.6 ± 1.9 ± 0.6 3.6 ± 2.2 6.1 ± 4.1 0.5 1.1 2.4 1.9 2.8

Statistics method: Independent-Sample T-test. 40

a

Fluconazole (50 mg/kg, p.o.) was pre-dosed prior to NAL. Data are shown as means ± SD. * p< 0.05, ** p < 0.01, *** p < 0.001 compared to control group.

Graphical abstract

41