Preclinical Disposition (In Vitro) of Novel μ-Opioid Receptor Selective Antagonists

Abstract

Recently, two novel N-heterocyclic derivatives of naltrexone [designated 17-cyclopropylmethyl-3,14β-dihydroxy-4,5α-epoxy-6β-[(4′-pyridyl)acetamido]morphinan (NAP) and 17-cyclopropylmethyl-3,14β-dihydroxy-4,5α-epoxy-6α-[(3′-isoquinolyl) acetamido]morphinan (NAQ)] have been proposed as μ-opioid receptor (MOR) selective antagonists. The goal of this study was to examine their absorption and metabolism. The bidirectional transport of NAP and NAQ was determined in Caco-2 and MDCKII-MDR1 cells, and the permeability directional ratio (PDR) was estimated (PDR = P_{app, B-A}/P_{app, A-B}, where P_app is the apparent permeability, A is apical, and B is basolateral). Oxidative metabolism of NAQ (0.5–80 μM) and NAP (0.5–30 μM) was determined in pooled human liver microsomes. The reaction monitored the disappearance of NAQ/NAP. NAP and NAQ were quantitated by high-performance liquid chromatography-UV at 270 or 232 nm, respectively. The permeability of NAQ or NAP was similar to that of naltrexone or paracellular markers, respectively. NAP also exhibited a high PDR and was determined to be a P-glycoprotein (P-gp) substrate. Unbound fractions in human plasma for NAQ and NAP were 0.026 ± 0.019 and 0.85 ± 0.12, respectively. The metabolic oxidative reaction rates, fitted to a Michaelis-Menten model, yielded K_m and V_max values of 15.8 ± 5.5 μM and 192 ± 24 pmol/min for NAQ and 1.8 ± 1.5 μM and 8.1 ± 1.4 pmol/min for NAP. Intrinsic hepatic clearance was estimated to be 13 and 5 ml · min⁻¹ · kg⁻¹ for NAQ and NAP, respectively. Neither NAQ nor NAP underwent detectable glucuronidation. Thus, NAP was a P-gp substrate with low apparent permeability, whereas NAQ was not a P-gp substrate and showed better permeability. Therefore, in contrast to NAP, NAQ would be more suitable for oral absorption and penetration of the blood-brain barrier, yielding potential pharmacokinetic and pharmacodynamic advantages over naltrexone.

Introduction

Chronic opioid treatment is associated with undesirable effects such as tolerance, addiction, respiratory depression, and constipation. These side effects of opioid treatment, especially respiratory depression, are more prevalent at the μ-opioid receptor, although they are mediated by all three opioid receptors (μ, δ, and κ) (Le Merrer et al., 2009). Thus, a selective μ-opioid receptor antagonist may be useful for treatment of opioid overdose.

The μ-opioid receptor (MOR) antagonists currently marketed for the treatment of excessive opiate-mediated side effects (especially respiratory depression) exhibit minimal selectivity for the MOR over the κ- and δ-opioid receptors. In addition to their antianalgesic effects, MOR antagonists can also improve drug dependence of other dependence-producing substances such as nicotine and alcohol (Bidlack and Mathews, 2009). Naltrexone is one of the early opioid antagonists that is devoid of agonist activity at the MOR. It has recently been demonstrated that two novel N-heterocyclic derivatives of naltrexone, 17-cyclopropylmethyl-3,14β-dihydroxy-4,5α-epoxy-6β-[(4′-pyridyl)acetamido]morphinan (NAP) and 17-cyclopropylmethyl-3,14β-dihydroxy-4,5α-epoxy-6α-[(3′-isoquinolyl) acetamido]morphinan (NAQ) (Fig. 1), exhibited high selectivity for the MOR relative to naltrexone. Both compounds also exhibited comparable antagonistic activity at the MOR and insignificant agonistic activity (Li et al., 2009). These results piqued interest in the potential clinical use of NAP and NAQ.

Fig. 1.

Chemical structures of naltrexone, NAP, and NAQ.

Naltrexone and naloxone are currently used opioid antagonists bearing structural similarity to NAP/NAQ. In naloxone, the cyclopropyl group of naltrexone is substituted by an allyl group. Naltrexone is used orally as maintenance therapy for treating alcohol addiction and opioid addiction. In contrast to naltrexone, naloxone is primarily used as an emergency measure for parenterally treating opioid overdose. It is also used in combination with opioid agonists such as buprenorphine or oxycodone for treating opioid-induced constipation. Both naltrexone and naloxone exhibit high first-pass metabolism with bioavailability being ∼5 to 40% and ∼3%, respectively (Fishman et al., 1973; Weinstein et al., 1973; Wall et al., 1981). The primary metabolite of naltrexone is 6β-naltrexol with minor metabolites being 2-hydroxy-3-O-methyl-6β-naltrexol and 6β-naltrexone glucuronide (Wall et al., 1981). In both NAP and NAQ, the metabolically labile ketone group at position 6 of naltrexone has been substituted by an amide bond (Fig. 1). Because amides are hydrolytically and metabolically stable, the hypothesis is that NAP and NAQ will undergo less metabolism than naltrexone, at least at position 6. However, the prediction of metabolic stability is not necessarily straightforward, because both NAP and NAQ contain additional aromatic rings, which may make them more susceptible to aromatic hydroxylation.

The overall aim of this study was to obtain a preliminary assessment of the gastrointestinal absorption and hepatic metabolism of the two novel compounds, NAP and NAQ, with the ultimate goal of determining their potential as therapeutic agents. This screen is important because naltrexone exhibits high hepatic first-pass (oxidative and conjugative) metabolism, resulting in low oral bioavailability and a short systemic half-life. A more selective MOR antagonist with higher oral bioavailability and a longer half-life would have significant therapeutic advantages over currently available therapies such as naltrexone.

Materials and Methods

Materials.

NAP and NAQ were synthesized as described previously (Li et al., 2009). The Caco-2 cells were obtained from American Type Culture Collection (Manassas, VA). The MDCKII-MDR1 cell line (Madin-Darby canine kidney cells transfected with MDR1) was a generous donation from Dr. Piet Borst (Netherlands Cancer Institute, Amsterdam, The Netherlands). Human liver microsomes (HLM) (pooled from 200 patients, mixed gender) were obtained from Xenotech, LLC (Lenexa, KS). Saccharo-1,4-lactone monohydrate was purchased from Calbiochem (San Diego, CA). N-(4-[2-(1,2,3,4-Tetrahydro-6,7-dimethoxy-2-isoquinolinyl)ethyl]-phenyl)-9,10-dihydro-5-methoxy-9-oxo-4-acridine carboxamide (GF120918; elacridar) was purchased from Toronto Research Chemicals (Toronto, ON, Canada). All other reagents and supplies were obtained commercially.

Cell Culture and Bidirectional Transport.

Caco-2 and MDCKII-MDR1 cells were cultured as described previously (Evers et al., 2000; Young et al., 2006). Caco-2 cells were subcultured within 6 to 7 days and MDCKII-MDR1 cells within 3 to 4 days.

Caco-2 cells (passage numbers 39–45) and MDCKII-MDR1 cells (passage numbers 3–5) were used to study the directionality of NAP and NAQ transport. Cells were seeded onto 12-well Transwell inserts (0.4 μM pore size) at a density of 80,000 cells/cm² for the Caco-2 cells and 50,000 cells/cm² for the MDCKII-MDR1 cells, respectively. Bidirectional transport studies were conducted between days 20 to 25 for the Caco-2 cells and on day 6 for the MDCKII-MDR1 cells.

Transport studies were carried out in Hanks' balanced salt solution buffered with 10 mM HEPES (pH 7.4). Monolayers were incubated in transport buffer at 37°C (50 rpm) for 20 min (pre-equilibration period). The solutions were then aspirated off, and the cells were treated with NAP/NAQ (10 μM) in the apical (A) chamber for apical-to-basolateral studies and in the basolateral (B) chamber for basolateral-to-apical studies. All solutions were prepared in Hanks' balanced salt solution and the vehicle concentration was <0.5% v/v. Experiments in which the effect of the P-glycoprotein (P-gp) inhibitor GF120918 (2 μM) were tested were set up in the same way, except that the inhibitor was added to both chambers, during both pre-equilibration and experimental periods. Aliquots (200 μl) were removed from the receiver chambers at predetermined time points (up to 2 h), and replaced with an equal volume of transport buffer (37°C). Acetonitrile (ACN; 25 μl) was added to the withdrawn samples and centrifuged at 2500g for 10 min at 4°C. A portion of the supernatant was used for analysis by HPLC-UV.

The integrity of the monolayer was assessed by determining the permeability of lucifer yellow (100 μM) for 30 min after transport of NAP/NAQ (Inokuchi et al., 2009). Results were used only from monolayers for which the apparent permeability coefficient of lucifer yellow was less than 1.0 × 10⁻⁶ cm/s.

After lucifer yellow transport, cells were lysed to determine intracellular concentrations of NAP/NAQ. Cells were washed once in transport medium, and ice-cold methanol (500 μl) was added to each Transwell insert. Transwell plates were stored on ice for 15 min. Cells were scraped and sonicated for 10 min for further disruption. After centrifugation at 2500g for 10 min, a portion of the supernatant was evaporated to dryness, reconstituted in 1:8 acetonitrile-transport buffer, and analyzed by HPLC-UV.

Metabolic Stability of NAP and NAQ.

Oxidation.

Because metabolites of NAP/NAQ were unknown at this stage, disappearance of the substrates was monitored by HPLC-UV (232 nm). Reactions were initially optimized so that they were linear with respect to protein concentration and time. NAQ (0.5–80 μM) and NAP (0.5–30 μM) were added to 50 mM potassium phosphate buffer (pH 7.4) containing 3 mM MgCl₂, and the mixture was stored on ice. An NADPH-generating system (5 mM glucose 6-phosphate, 1 mM nicotinamide adenine dinucleotide phosphate, and 1 U/ml glucose-6-phosphate dehydrogenase in 50 mM phosphate buffer, pH 7.4) was added to this mixture, and the components were warmed at 37°C for 2 min. The reactions were initiated by addition of HLM to a final concentration of 0.5 mg/ml (total reaction volume 400 μl) for NAQ and 1 mg/ml for NAP (total reaction volume 250 μl). Reactions were performed at 37°C in a shaking water bath (90 rpm), with tubes open to the humidified atmosphere. Aliquots (50 μl) were withdrawn at predetermined time points (0, 15, 30, 45, 60, and 120 min for NAQ and 0, 15, 30, and 45 min for NAP) into an equal volume of ice-cold acetonitrile. The mixture was centrifuged at 2500g for 10 min at 4°C. A portion of the supernatant was withdrawn and was evaporated to dryness under vacuum. The samples were reconstituted in 90:10 0.05% trifluoroacetic acid in water-acetonitrile. Phenacetin (80 μM) deethylation was used as a positive control for phase 1 oxidative enzyme activity in the assays (0.5 mg/ml HLM). Control experiments contained dimethyl sulfoxide in place of NAP/NAQ.

Glucuronidation.

Because glucuronidation is known to be a major metabolic pathway for opioid compounds including naltrexone (Wall et al., 1981; Lötsch, 2005), we resolved to study the extent to which NAP and NAQ may be glucuronidated. Glucuronidation of NAP and NAQ (0.5–100 and 1–100 μM, respectively) were examined in human liver microsomes in the presence of UDP-glucuronic acid (3 mM), 0.5 mg/ml microsomal protein, alamethicin (50 μg/ml microsomal protein), saccharolactone (6 mM), magnesium chloride (10 mM), and Tris · HCl buffer, pH 7.4 (50 mM). Reactions were initiated by the addition of UDP-glucuronic acid, and were performed at 37°C in a shaking water bath (90 rpm). 17β-Estradiol (50 μM) was used as the positive control for glucuronidation activity. The total reaction volume was 300 μl. Aliquots of 50 μl were withdrawn at 0, 0.5, 1, and 2 h into equal volumes of 6% trifluoroacetic acid. The tubes were centrifuged at 2500g for 10 min at 4°C. Aliquots of the supernatant were withdrawn and analyzed by HPLC-UV for NAP/NAQ, HPLC-UV for naltrexone, and HPLC fluorescence for estradiol, as outlined below. Control experiments contained dimethyl sulfoxide in place of NAP/NAQ.

Identification of the Oxidative Metabolite of NAQ.

The analytical method was successful in monitoring the time-dependent appearance of an oxidative metabolite of NAQ (M1). To identify this metabolite, several reaction mixtures were set up, each containing NAQ at a final concentration of 1 μM (HLM at 0.5 mg/ml, total reaction volume 400 μl). Tubes were incubated under conditions that facilitated oxidative metabolism as outlined earlier and prepared for HPLC-UV analysis as outlined below. Fractions containing M1 were collected, evaporated to dryness, and reconstituted in 50:50 water containing 0.1% formic acid-ACN. NAQ (50 μg/ml) was prepared fresh in 50:50 water containing 0.1% formic acid-ACN.

Standards and HPLC fractions were analyzed under the same conditions. They were directly infused onto a Micromass ZMD single quadrupole mass spectrometer and analyzed by electrospray ionization (positive mode). Ionization conditions were as follows: capillary voltage, 4.27 V; cone voltage, 20 V; source block temperature, 120°C; and desolvation temperature, 150°C. Signals were compared with analogous infusions with the same solvent system to determine unique ions associated with NAP, NAQ, and their metabolites.

HPLC Analysis.

Samples were maintained at 4°C during analysis. The analytical column was a C18 column (Alltima HP C18, 4.6 mm × 100 mm, 3 μm; Alltech/Grace Davison, Deerfield, IL). The mobile phase consisted of 0.05% trifluoroacetic acid in water (A) and acetonitrile (B) and was delivered at a flow rate of 1 ml/min.

For transport assay samples, the initial mobile phase composition was 90:10 A/B and was held at this composition for 1 min. The mobile phase composition was ramped to 50:50 over the next 6 min, changed to 90:10 A/B over the next minute, and held at 90:10 A/B for 3 min. NAP and NAQ were detected at 232 nm, whereas naltrexone was detected at 281 nm. Analytes were quantified from standard curves prepared in transport buffer-ACN (8:1). Calibration curves for NAP, NAQ, and naltrexone were all linear in the range 0.05 to 5 μM (R² = 0.999). Samples from the oxidative metabolism studies of NAP and NAQ, were analyzed using the same method as above, except that the standards were prepared in 90:10 A/B.

Analysis of samples from the glucuronidation studies of NAP, NAQ, naltrexone, and estradiol were performed as outlined below. The initial mobile phase composition was 99:1 A/B. It was held at this composition for 1 min, ramped to 91:9 and 45:55 over the next 2 and 6 min. The composition was changed to 99:1 over 3 min and allowed to equilibrate at 99:1 for another 3.5 min. Standards were prepared in deionized water. NAP, NAQ, and naltrexone were detected by UV at 232, 232, and 281 nm, respectively, in accordance with their UV absorbance peaks. Estradiol, estradiol 3-glucuronide, and estradiol 17-glucuronide were detected by fluorescence (275 nm/315 nm). Calibration curves for NAP, NAQ, naltrexone, and estradiol were linear in the range of 0.5 to 100 μM (R² = 0.999), 0.1 to 50 μM (R² = 0.999), 5 to 5000 μM (R² = 1.000), and 0.5 to 25 μM (R² = 0.997).

Acetaminophen and phenacetin were detected at 240 nm. The analytical method for separating phenacetin and acetaminophen used the same mobile phase solvents as follows: 95:5 A/B for 1 min, ramped up to 70:30 over 4 min, ramped down to 95:5 over 4 min, and held at 95:5 for 3 min. Calibration curves for acetaminophen and phenacetin were linear in the range of 0.05 to 5 μM (R² = 0.999) and 0.25 to 25 μM (R² = 0.999).

Human Plasma Protein Binding.

Plasma samples from four healthy tobacco-free human subjects (three male and one female) were obtained from BioChemed Services (Winchester, VA). Plasma samples were spiked with NAQ (10 μM), NAP (10 μM), [³H]gentamicin (0.3 μCi, as a low-binding control), or [³H]2-methoxyestradiol (0.05 μCi, as a high-binding control) and dialyzed against phosphate-buffered saline (pH 7.4) at 37°C using the Rapid Equilibrium Device (Thermo Fisher Scientific, Waltham, MA) as described previously (Gulati et al., 2009) for 20 h to ensure equilibrium. Plasma and buffer samples (200 μl) were mixed with 200 μl of buffer or plasma (respectively), and 600 μl of acetonitrile containing 0.01 N HCl, vortexed, chilled, centrifuged, evaporated, reconstituted with 60 μl of the mobile phase, and centrifuged again. Then 30 μl of the supernatant were injected into the HPLC system described above but using an Agilent Microsorb-MV 3μ C18 100 × 4.6 mm column. NAP was eluted at 40°C using 25% methanol 75% aqueous (20 mM trifluoroacetic acid and 10 mM triethylamine) with detection at 270 nm. NAQ was eluted at 40°C using 55% methanol 75% aqueous with detection at 232 nm. Standard curves were linear (r² > 0.99) between 0.08 and 20 μM or 0.04 and 20 μM for NAP and NAQ, respectively.

Data Analysis.

Permeability calculations.

The following equation was used to calculate apparent permeability in either the apical-to-basolateral direction (P_{app, A-B}) or basolateral-to-apical direction (P_{app, B-A}) (eq. 1):

where dC/dt is the change in concentration in the receiver chamber with respect to time, V_r is the volume of the receiver chamber, A is the growth surface area, and C_d is the dosing concentration of the drug in the donor compartment.

The permeability directional ratio (PDR) was calculated as follows (Young et al., 2006) (eq. 2):

Metabolism calculations.

The oxidation and glucuronidation rates of NAP and NAQ were determined by the method of initial rates (Copeland, 2000). Plots of initial reaction rates (picomoles per minute per milligram of protein) versus substrate concentration (micromolar) were fitted by the Michaelis-Menten equation (GraphPad Prism, version 5; GraphPad Software Inc., San Diego, CA) to estimate K_m and V_max, where V_max is the maximal rate of oxidation/glucuronidation and is expressed in picomoles per minute per milligram of protein and K_m is the substrate concentration at the half-maximal reaction rate. The Michaelis-Menten parameter estimates were used to estimate the intrinsic clearance Cl_int (Cl_int = V_max/K_m). Cl_int was extrapolated using scaling methods to the in vivo clearance [CL_int^total (IVIVE)] expected in a healthy 75-kg individual, with the help of eq. 3 (Naritomi et al., 2001):

CL_int (IVIVE) is presented as milliliters per minute per kilogram. With the values obtained from eq. 3, the in vivo hepatic clearance (CL_hep) was estimated as in eq. 4 (Naritomi et al., 2001):

where f_u is unbound fraction of the compound in plasma (determined as above) and Q_hep is the hepatic blood flow rate (1500 ml/min).

Results

Bidirectional Transport of NAP and NAQ in Caco-2 Cells.

The apical-to-basolateral and basolateral-to-apical transport of NAP and NAQ were examined in Caco-2 cells to obtain an estimate of the human gastrointestinal absorption potential of these two compounds and to assess whether or not they are subject to efflux transporters. Appearance of NAP, NAQ, and naltrexone in the receiver chamber as a function of time was linear up to 2 h, the final time point of this study (Supplemental Fig. S1). The integrity of the Caco-2 monolayers was validated by estimating the apparent permeability coefficient (P_app) of lucifer yellow in the apical-to-basolateral and basolateral-to-apical directions after transport of the compound of interest. Lucifer yellow is a well known endothelial cell permeability marker but is also used as a paracellular permeability marker in epithelial cells (Young et al., 2006). The cutoff for an intact monolayer was set at an apparent permeability coefficient of (1 × 10⁻⁶ cm/s) for lucifer yellow. Apparent permeability constants are reported in Table 1, including only data from monolayers judged to be intact from lucifer yellow permeability (<1 × 10⁻⁶ cm/s). PDR values ≥2.0 indicate net apical secretion (“efflux” transport).

TABLE 1.

Bidirectional transport in Caco-2 cells

P_app values of substrates in the apical-to-basolateral (A-B) and basolateral-to-apical (B-A) directions were determined by incubating Caco-2 monolayers with substrate at 37°C for 2 h. PDR = P_{app, B-A}/P_{app, A-B}. Data represent means ± S.D.

	Papp		PDR
	A-B	B-A
	× 106 cm/s
NAQ (10 μM)	2.85 ± 0.66	4.24 ± 1.04	1.5
NAP (10 μM)	0.6 ± 0.17	7.8 ± 1.0	13.1
NAP (10 μM) + GF120918 (2 μM)	0.8 ± 0.2	1.32 ± 0.13	1.6
Naltrexone (10 μM)	4.39 ± 1.22	3.7 ± 0.96	0.8
Caffeine (100 μM)	10.2 ± 3.2	11.4 ± 2.6	1.1
Mannitol (3 μM)	0.51 ± 0.05	0.67 ± 0.05	1.3
Digoxin (0.10 μM)	1.05 ± 0.17	5.3 ± 0.38	5.1
Digoxin + GF120918 (2 μM)	1.62 ± 0.29	1.89 ± 0.09	1.2

As shown in Table 1, the absorptive (apical to basolateral) permeability of NAQ was approximately 4- to 5-fold lower than that of caffeine, which is a highly permeable compound and undergoes complete absorption in humans (Yazdanian et al., 1998). This result is in contrast to NAP, which exhibits an apparent absorptive permeability coefficient similar to that of mannitol, which is a marker of low permeability or paracellular permeability. The permeability of NAQ in the absorptive direction was approximately 4.5-fold greater than that of NAP and was closer to that of naltrexone.

NAQ also bears other similarities to naltrexone in that for both compounds the PDR was less than 2, suggesting that neither is an efflux transporter substrate (Giacomini et al., 2010). In contrast, NAP exhibited a PDR substantially greater than 2, indicating efflux transport. Efflux transporters expressed on the apical membrane of Caco-2 cells include P-gp, breast cancer resistance protein (BCRP), and multidrug resistance-associated protein 2 (Englund et al., 2006; Seithel et al., 2006). GF120918 is a commonly used efflux transporter inhibitor that abrogates efflux mediated by P-gp as well as by BCRP (IC₅₀ = 0.3 and 20 μM, respectively) (Maliepaard et al., 2001; Matsson et al., 2009). GF120918 reduced the PDR of digoxin, a well known P-gp substrate, from 5.1 to 1.2, confirming its inhibitory potential in our experimental setup (Table 1). Used at the same concentration (2 μM), GF120918 decreased the PDR of NAP to <2, suggesting that the high PDR, and thus the efflux transport, of NAP is mainly due to P-gp, although contributions from other apical transporters such as BCRP (or a basolateral transporter) cannot yet be ruled out.

Bidirectional Transport of NAP in MDCK-MDR1 Cells.

To confirm that NAP is a P-gp substrate, its bidirectional transport was estimated in MDCKII-MDR1 cells overexpressing human P-gp. In these cells, NAP exhibited a very high PDR, which was abolished in the presence of GF120918 (Table 2). This confirmed that NAP is a P-gp substrate. In contrast to the results from Caco-2 cells, the PDR reduction in the presence of GF120918 resulted from a much greater increase in P_app in the apical-to-basolateral direction compared with the decrease observed in the basolateral-to-apical direction.

TABLE 2.

Bidirectional transport of NAP in MDCK-MDR1 cells

P_app values of NAP were determined in the apical-to-basolateral (A-B) and basolateral-to-apical (B-A) directions by incubating MDCK-MDR1 monolayers with NAP (10 μM) at 37°C for 2 h. PDR = P_{app, B-A}/P_{app, A-B}. Data represent means ± S.D.

	Papp		PDR
	A-B	B-A
	× 106 cm/s
NAP (10 μM)	0.037 ± 0.024	12.5 ± 1.9	341
NAP (10 μM) + GF120918 (2 μM)	1.5 ± 0.5	3.3 ± 0.12	2.1

Plasma Protein Binding.

Mass balance of both compounds between plasma and dialysate was not significantly different from 100%, indicating good stability. Fraction unbound estimates for gentamicin and 2-methoxyestradiol, as low- and high-binding controls, were similar to previously reported values (Lakhani et al., 2006; Lexi-Comp, 2011). The unbound fraction in human plasma for NAP was 0.85 ± 0.12, whereas that of NAQ was 0.026 ± 0.012. Thus, these data indicate that NAP is 15% bound, whereas NAQ is 97% bound.

Oxidative Metabolism of NAQ and NAP.

Phenacetin is a substrate of CYP1A2 and undergoes deethylation to form acetaminophen (Venkatakrishnan et al., 1998; Walsky and Obach, 2004). When HLM were incubated with phenacetin (80 μM), the analytical method was successfully able to monitor the appearance of acetaminophen (Supplemental Fig. S2), indicating that the experimental setup facilitated oxidative metabolism. The pooled human liver microsomes had already been certified by the vendor to be active for the major drug-metabolizing enzymes (CYP1A2, CYP2C9, CYP2D6, and CYP3A4). CYP1A2 was favored as an in-house positive control over the other enzymes because of the ease of availability of acetaminophen, the O-deethylation product of phenacetin. Because the metabolites of NAP and NAQ were unknown at this stage, we monitored the disappearance of NAP or NAQ, rather than appearance of metabolites. Figure 2, A and B, shows representative curves obtained at 0.5 and 80 μM NAQ, the lowest and highest concentrations of NAQ used. Regression analysis of the linear portion of the curves (120 min at 80 μM and 45 min at all other concentrations) returned the initial reaction velocities at each of these concentrations. Likewise, oxidation of NAP was examined in the concentration range of 0.5 to 30 μM (Fig. 3, A and B), and regression analysis was performed on the linear portion of the curves (45 min) to estimate initial reaction velocities. The initial reaction velocities were plotted versus initial substrate concentration to obtain the kinetic constants K_m and V_max for NAQ (15.8 ± 5.5 μM and 192 ± 24 pmol · min⁻¹ · mg protein⁻¹, respectively) (Fig. 2C) and NAP (1.8 ± 1.5 μM and 8.1 ± 1.4 · min⁻¹ · mg protein⁻¹, respectively) (Fig. 3C). Higher concentrations of NAP were also tested for their susceptibility to oxidative metabolism, but presumably because of sensitivity issues, loss of NAP was not observed at these high concentrations. NAP has a higher apparent affinity (∼10-fold lower K_m) for microsomal oxidative enzymes compared with NAQ, but the V_max of oxidation of NAQ is much higher than that of NAP. With use of the methods described above (eqs. 3 and 4), intrinsic hepatic clearances were predicted to be 13 and 5 ml · min⁻¹ · kg⁻¹ for NAQ and NAP, respectively, thus predicting that of NAQ to be 2.5-fold greater than that of NAP, mainly due to the greater V_max for NAQ. At all concentrations, the amount of substrate lost is more for NAQ than NAP (Supplemental Table S1). Combining these estimates with the experimental values for fraction unbound in human plasma (as described above), we obtained estimates of hepatic clearance for NAQ of 0.33 ml · min⁻¹ · kg⁻¹ and for NAP of 3.5 ml · min⁻¹ · kg⁻¹, suggesting low hepatic extraction for both compounds. By comparison, the systemic clearance of naltrexone (presumably mainly due to hepatic extraction) is approximately 17 ml · min⁻¹ · kg⁻¹ (McEnvoy, 1999).

Fig. 2.

Metabolism of NAQ in human liver microsomes. To estimate the initial reaction velocities, HLM (0.5 mg/ml, total reaction volume of 400 μl) were incubated with NAQ (0.5–80 μM), and loss of NAQ was monitored as a function of time. A and B, reaction profiles observed at 0.5 and 80 μM, respectively. C, concentration-dependent oxidative metabolism of NAQ in HLM. HLM were incubated with NAQ under conditions that facilitated oxidative metabolism. The initial reaction velocities at each concentration were plotted versus the concentration. The curve was fitted by the Michaelis-Menten equation to estimate the kinetic constants K_m and V_max (see text). Data are plotted as mean ± S.D. (n = 3).

Fig. 3.

Determination of initial reaction velocities from oxidative metabolism studies of NAP in HLM. To estimate the initial reaction velocities, HLM (1 mg/ml, total reaction volume 250 μl) were incubated with NAP (0.5–30 μM), and loss of NAP was monitored as a function of time. A and B, reaction profiles observed at 0.5 and 30 μM, respectively. C, concentration-dependent oxidative metabolism of NAP in HLM. HLM were incubated with NAP under conditions that facilitated oxidative metabolism. The initial reaction velocities at each concentration were plotted versus the concentration. The curve was fitted by the Michaelis-Menten equation to estimate the kinetic constants K_m and V_max (see text). Data are plotted as mean ± S.D. (n = 3).

For NAQ, we observed the appearance of a new peak concomitantly with the disappearance of NAQ. The peak area of this presumed metabolite of NAQ (M1) increased linearly with incubation time (data not shown). The apparent K_m for the formation of M1 was 77.9 ± 10.3 μM. This is in the range of the K_m value obtained from the disappearance of NAQ (15.8 ± 5.5 μM), and consistent with M1 being the primary metabolite of NAQ. In contrast, whereas the analytical method was able to capture the loss of NAP, we were not able to monitor the appearance of any new metabolite peak by UV absorbance at either 232 or 270 nm.

Glucuronidation of NAP and NAQ.

One of the major known circulating metabolites of naltrexone is its phenolic glucuronide conjugate (glucuronidated at position 3; Fig. 1) (Ventura et al., 1988). Based on their structural similarity to naltrexone, NAP and NAQ were hypothesized to undergo glucuronidation at the aromatic hydroxyl group. Therefore, we exposed NAP and NAQ to glucuronidation reactions, as described above. Because of the unavailability of metabolite standards, we monitored the loss in peak areas of NAP and NAQ. Contrary to the hypothesis, neither NAP (1–100 μM) nor NAQ (0.5–100 μM) underwent any detectable glucuronidation. Glucuronidation of naltrexone was evaluated over a wider concentration range (1–2500 μM). Naltrexone glucuronidation was observed in the concentration range 50 to 200 μM, ranging from 0.1 to 5.8 · min⁻¹ · mg protein⁻¹. Under the same conditions, the analytical method was successfully able to monitor loss of estradiol (10 μM) (Supplemental Fig. S3) and the appearance of the 3- and 17-glucuronic acid conjugates of estradiol (data not shown).

Identification of the Oxidative Metabolite of NAQ.

For NAQ standard, the peak of highest intensity was observed at m/z of 498.73, which corresponds to the [M + H]⁺ peak of NAQ. In the HPLC eluent fractions containing pure NAQ, highest peak intensities were observed at m/z 249.74 and 498.86. These correspond, respectively, to the double and single positively charged peaks of NAQ. For the metabolite fraction, the peak of highest m/z was 444.66 and corresponds to the single positively charged N-dealkylation product of NAQ. To facilitate fragmentation, the cone voltage was increased to 106 V. Fragmentation patterns of the peaks obtained at 498.86 and 444.66 were similar: for both, peaks were observed that corresponded to loss of 18, 22, and 38 m/z units. This result indicates that the 444.66 molecular ion retains the same molecular structure as NAQ (except for the N-dealkylation), suggesting that the primary metabolite of NAQ is the N-dealkylated product.

Discussion

The results suggest that between NAP and NAQ, NAQ is expected to have superior absorption across the gastrointestinal epithelia and holds more promise as an oral drug candidate, assuming similar gastrointestinal solubility. If permeability were to be the rate-determining factor driving oral absorption, NAQ can be expected to have absorption similar to naltrexone. The permeability of NAP is not only lower than that of naltrexone but is also similar to that of paracellular permeability markers. This could be a major obstacle for further development of NAP as a drug candidate intended for oral absorption. The comparatively lower permeability of NAP does not necessarily eliminate its promise as a therapeutic agent. Methylnaltrexone, a quaternary amine opioid antagonist, has poor permeation across the blood-brain barrier. It is used clinically for reversing opioid-induced peripheral side effects without causing loss of analgesia or without precipitation of withdrawal symptoms. NAP has the potential to have similar clinical usage.

Of interest, NAP was determined to be a P-gp substrate, whereas NAQ was not. In mice, NAP and NAQ were able to antagonize the effects of morphine, which is an agonist at the MOR (Li et al., 2009). The dose at which NAP produced 50% of this effect was 10-fold higher than that of NAQ. Opioid receptors are prevalent in the brain (Le Merrer et al., 2009). It is likely that P-gp, which is expressed and is functional at the blood-brain barrier (Giacomini et al., 2010), limits distribution of NAP to the brain. In Caco-2 cells, naltrexone did not undergo efflux, as reported previously (Kanaan et al., 2009). Compared with naltrexone, NAP carries a substituted pyridinyl side chain at the 6-position through an amide linkage, replacing the ketone group on naltrexone. This structural difference between naltrexone and NAP may explain the difference in their efflux profiles, because structure-activity relationships of P-gp substrates indicate that increasing the number of hydrogen bond acceptors on a molecule increases its chances of being recognized as a substrate by P-gp (Hochman, 2006). More interesting are the different efflux patterns of NAP and NAQ: NAQ differs from NAP by the presence of an additional aromatic ring and by the placement of the aromatic nitrogen atom, as well as the stereochemistry of the side chain. Although it is not possible to identify the exact reason for the difference in the efflux patterns of NAP and NAQ without molecular modeling experiments, possible reasons could be suboptimal distances between the hydrogen bond acceptors and/or steric hindrance because of the additional aromatic ring (Hochman, 2006).

P-gp-mediated efflux is expected to be an additional impediment to the development of NAP as a drug candidate. Inhibition experiments in Caco-2 cells suggest that the large PDR results from P-gp facilitating secretory transport rather than inhibiting absorptive transport; the inhibitor enhanced the absorptive transport only to a modest extent. This asymmetrical effect of P-gp has also been observed previously with several other substrates (Troutman and Thakker, 2003). Thus, efflux by P-gp may not additionally hamper the intestinal transport of NAP beyond its already low diffusional permeability. However, the intended pharmacological uses of NAP and NAQ are as selective antagonists at the central μ-opioid receptors, which necessitates their crossing the blood-brain barrier where P-gp is functional. It has been observed that for compounds that exhibit a high P-gp efflux ratio in vitro, brain penetration is a greater hurdle than intestinal permeation (Hochman, 2006). Thus, although NAQ may easily cross the blood-brain barrier (based on its observed in vitro permeability), it may be a more daunting task for NAP. Furthermore, in MDCKII-MDR1 cells, NAP had a very high PDR, demonstrating its activity as a good P-gp substrate.

Naltrexone is extensively metabolized after oral administration, oral bioavailability being reported to be 5 to 40%. Major plasma metabolites of naltrexone in man after intravenous administration are 6β-naltrexol, and the conjugates of naltrexone and 6β-naltrexol (Wall et al., 1981), with the highest systemic exposures after oral administration obtained for 6β-naltrexol and its conjugate. After oral administration, the predominant plasma metabolite was the 6β-naltrexol conjugate (Wall et al., 1981). The conjugates of naltrexone and 6β-naltrexol are the glucuronic acid conjugates (Ventura et al., 1988) at position 3 (Fig. 1). Reduction of naltrexone to 6β-naltrexol was observed only in cytosolic fractions but not in microsomal fractions. Enzymes responsible for the metabolism of naltrexone to 6β-naltrexol are aldo-keto reductase isoforms 1C4, 1C1, and 1C2 (Porter et al., 2000; Breyer-Pfaff and Nill, 2004; Tong et al., 2010). Both NAP and NAQ lack the ketone moiety at position 6 of the morphinan ring structure and are hence not expected to be reduced to the corresponding alcohol. Instead, both compounds have substituted amide bonds at that position. Amides are generally stable to hydrolysis. On the basis of this fact, we hypothesized that NAP 3-glucuronide and NAQ 3-glucuronide would be the primary metabolites of NAP and NAQ. Contrary to our expectations, NAP or NAQ was not detectably glucuronidated. We examined the glucuronidation of NAP and NAQ at concentrations much lower than the K_m values reported for naltrexone; on the basis of doses at which NAP/NAQ were administered to mice, their plasma concentration was estimated to be in the nanomolar range (Li et al., 2009).

Both NAP and NAQ underwent modest loss of peak area under conditions that facilitated oxidative metabolism, with the percentage of NAP metabolized being approximately half that of NAQ. This result is in accordance with what has been reported before for other naltrexamine derivatives in human liver S9 fractions, which either did not undergo oxidative metabolism or exhibited half-lives of 100 to 300 min (Ghirmai et al., 2009). The primary oxidative metabolite of NAQ was the N-dealkylated product. Naltrexone itself is not known to undergo N-dealkylation in vivo, but several other opioids such as buprenorphine, morphine, and codeine, which bear structural similarity to naltrexone, do undergo N-dealkylation (Caraco et al., 1996; Kobayashi et al., 1998; Projean et al., 2003), although for most of them glucuronidation is the major route of metabolism. Even though the experimental conditions successfully monitored the loss of NAP, time-dependent appearance of any major metabolite was not observed. A more sensitive detection, such as a tandem mass spectrometry system, could be used to further identify the primary metabolites of NAP.

As mentioned previously, one of the primary circulating metabolites of naltrexone is 6β-naltrexol. The intrinsic clearance (V_max/K_m) for the formation of 6β-naltrexol from naltrexone in human liver cytosolic fractions has been reported to be 12.6 and 5 to 36.7 μl · min⁻¹ · mg protein⁻¹ (Porter et al., 2000; Tong et al., 2010). To obtain an estimate of the hepatic clearance (CL_hep), these values of intrinsic clearance were scaled up using the same method used for NAP and NAQ (metabolism calculations in Data Analysis). The unbound fraction of naltrexone in human plasma is 72 to 79% (McEnvoy, 1999). Therefore, the hepatic clearance (CL_hep) of naltrexone was estimated to be 3.2 to 12.1 ml · min⁻¹ · kg⁻¹. Using this extrapolative scaling approach, we speculate that the hepatic clearance for NAP (3.5 ml · min⁻¹ · kg⁻¹) may be similar to that of naltrexone, whereas that of NAQ (0.33 ml · min⁻¹ · kg⁻¹) is probably lower by an order of magnitude.

In conclusion, both NAP and NAQ have promise as new drug candidates. Selective μ-opioid receptor antagonists NAP and NAQ showed evidence of hepatic oxidative metabolism (NAQ > NAP), but no evidence of hepatic glucuronidation; NAP was found to be a P-gp substrate with low gastrointestinal permeability, whereas NAQ showed better permeability and lacked activity as a P-gp substrate. Therefore, in contrast to NAP, NAQ would be suitable for oral absorption and penetration of the blood-brain barrier. Combining the improved MOR selectivity with its potential pharmacokinetic benefits, NAQ seems to be a good lead compound with advantages over naltrexone as a MOR antagonist.

Authorship Contributions

Participated in research design: Mitra, Venitz, Zhang, and Gerk.

Conducted experiments: Mitra, Yuan, and Gerk.

Contributed new reagents or analytic tools: Mitra, Yuan, and Gerk.

Performed data analysis: Mitra, Venitz, and Gerk.

Wrote or contributed to the writing of the manuscript: Mitra, Venitz, Zhang, and Gerk.