Metabolic clearance of select opioids and opioid antagonists using hepatic spheroids and recombinant cytochrome P450 enzymes

Abstract The opioid crisis is a pressing public health issue, exacerbated by the emergence of more potent synthetic opioids, particularly fentanyl and its analogs. While competitive antagonists exist, their efficacy against synthetic opioids is largely unknown. Furthermore, due to the short durations of action of current antagonists, renarcotization remains a concern. In this study, metabolic activity was characterized for fentanyl‐class opioids and common opioid antagonists using multiple in vitro systems, namely, cytochrome P450 (CYP) enzymes and hepatic spheroids, after which an in vitro‐in vivo correlation was applied to convert in vitro metabolic activity to predictive in vivo intrinsic clearance. For all substrates, intrinsic hepatic metabolism was higher than the composite of CYP activities, due to fundamental differences between whole cells and single enzymatic reactions. Of the CYP isozymes investigated, 3A4 yielded the highest absolute and relative metabolism across all substrates, with largely negligible contributions from 2D6 and 2C19. Comparative analysis highlighted elevated lipophilicity and diminished CYP3A4 activity as potential considerations for the development of more efficacious opioid antagonists. Finally, antagonists with a high degree of molecular similarity exhibited comparable clearance, providing a basis for structure‐metabolism relationships. Together, these results provide multiple screening criteria for early stage drug discovery involving opioid countermeasures.

approximately 20% of all opioid-related deaths in the U.S. during 2016-2017. 6,7 Use of FEN-class opioids as a potential lethal agent was also underscored by the 2002 Moscow theater incident, where an aerosol mixture of carfentanil (CRF) and remifentanil (RMF) was introduced into the ventilation system in an attempt to manage a hostage situation, resulting instead in 125 deaths. 8,9 These concerning trends highlight the severity of the opioid crisis and emphasize the need to better understand the pharmacokinetics of opioids and potential therapeutics.
Currently, competitive antagonists that are effective against morphine and FEN exist. The most extensively studied of these include naloxone (NX), naltrexone (NTX), and nalmefene (NMF), all of which have been shown to reverse opioid-induced respiratory depression in humans and non-human species. [10][11][12][13][14] However, the therapeutic potential of current antagonists against more potent synthetic opioids remains largely unknown. [15][16][17][18] Renarcotization, wherein opioid-induced effects reappear following countermeasure administration and apparent recovery, represents an ongoing concern, particularly for more potent opioids, as current approved antagonists are often characterized by relatively short durations of action. 19,20 Notably, episodes of renarcotization have been reported for NTX, even though it is inherently longer-acting and more potent than NX. 21 This possibility of renarcotization necessitates a more comprehensive understanding of the metabolic stability of opioids and opioid antagonists, especially in relation to each other.

Metabolism of drugs can occur via a variety of biotransformation
processes (e.g., oxidation, hydrolysis, conjugation), although the ultimate goal is always to increase hydrophilicity and facilitate excretion from the body. 22 In general, drug metabolism can be further differentiated into two phases, phase I modification and phase II conjugation, with both phases primarily localized in the liver. 23 The majority of opioids are metabolized extensively via phase I oxidation by cytochrome P450 enzymes (CYPs), a superfamily of enzymes involved in approximately 75% of all marketed drug metabolism. 24,25 Current opioid antagonists, on the other hand, are thought to undergo phase II glucuronidation more readily, likely due to their structural similarity to morphine. 25,26 These differences are important distinctions for metabolic investigations as commonly used in vitro models can contain diverse amalgamations of metabolic pathways, and not all models include relevant transport phenomena that influence drug uptake. [27][28][29] In theory, intrinsic clearances should be equivalent after physiological scaling; however, several fold differences have been observed between intrinsic clearances as determined using different in vitro systems or different donors within a single in vitro system. [30][31][32] As such, while metabolic clearance has been studied for specific opioids and opioid antagonists, multiple in vitro systems were employed, which hinders the ability to directly compare between studies. [32][33][34] In the present study, metabolic clearance was characterized for FEN-class opioids (FEN, RMF, CRF) and common opioid antagonists (NX, NTX, NMF). Individual recombinant CYPs and hepatic spheroids were investigated to determine contributions of individual metabolic routes and total metabolism, respectively. Substrate consumption was quantified using liquid chromatography with tandem mass spectrometry (LC-MS/MS), and Michaelis-Menten parameters were derived from initial reaction velocities. Finally, an in vitro-in vivo correlation (IVIVC) was applied to convert in vitro metabolic activity to in vivo hepatic clearance for comparison and applications in future models.

| Metabolism: recombinant CYPs
Individual CYPs with substantial predicted metabolic activity (>20% of total metabolism) were chosen for each test article based on predictions generated using ADMET Predictor (v. 10.0.0.11; Simulations Plus, Inc.). 35 Total predicted metabolic activity and percent contributions for individual CYPs are given in Table S1. To summarize, all test articles were predicted to be metabolized via CYP3A4, while a select few were projected to undergo metabolism via multiple CYP isozymes (2C19 for NX and 2D6 for FEN and CRF).

| Metabolism: hepatic spheroids
Spheroid-qualified human hepatocytes were obtained from and 200 rpm. Following trypsinization, manual agitation (i.e., vigorous pipetting for 1 min) was applied to fully dissociate spheroids, confirmed by light microscopy. Once a single cell suspension was attained, trypsin action was terminated with the addition of an equal volume of saline solution containing 10% FBS. Cell count was then determined using flow cytometry (Attune™ NxT acoustic focusing cytometer; ThermoFisher) with a 50 μl total acquisition volume per sample. Cell debris was excluded using a forward-and side-scattered light gate (Figures S1-S3), from which cell count was calculated using the gated cell density and total cell suspension volume.

| Sample preparation
Prior to sample extraction, samples were diluted 5-to 200-fold in water to produce final analyte concentrations in the range of  Analytes were separated using gradient programs optimized for each target (see Tables S2-S4) with a flow rate of 500 μl/min, an injection volume of 5 μl, and mobile phases composed of (A) water with 0.2% formic acid and (B) methanol with 0.2% formic acid. Observed retention times (t R ) are given in Table 1 for all targets and internal standards.

| LC-MS/MS
Analytes were detected with tandem mass spectrometry using a Sciex 6500 QTRAP triple quadrupole mass spectrometer (Sciex) operated in positive electrospray ionization mode with multiple reaction monitoring (MRM) and the following ion source parameters: curtain gas, 35 psi; ion spray voltage, 5500 V; ion source temperature 550°C; nebulizer gas; 70 psi; and heater gas, 60 psi. For each target analyte, multiple transitions were monitored, with one transition selected for quantification, and the remaining transitions serving as qualifiers to confirm analyte identity (Table 1). Analyte-specific parameters, specifically collision energy (CE) and collision exit potential (CXP), were optimized for maximum signal and are given in Table 1. Declustering potential, entrance potential, and collisionassisted dissociation gas did not exhibit analyte-specific effects and were set at 50, 10 V, and low, respectively, for all analytes.

| Data analysis
Cell counts were analyzed using Attune™ NxT Software (v. 3.1.2; ThermoFisher) to determine cell density and exclude cell debris.
Analyte peak areas for quantifiers and qualifiers were integrated using Analyst Instrument Control and Data Processing Software (v. Intrinsic clearances (CL int ) were calculated for a 70 kg human (see Table S5 for liver properties) using experimentally determined

Michaelis-Menten parameters and the correlation between hepatic
CL int and enzyme kinetics as derived in Choi et al. 37 where V max is the maximal rate of reaction, K m is the Michaelis constant, and A is the amount of CYPs or hepatocytes in the liver tissue.
The liver's contribution to systemic clearance, also termed hepatic clearance (CL H ), was calculated using a well-stirred clearance model: where Q H is the hepatic blood flow, f u,p is the fraction unbound (Table S6), and CL int is the intrinsic clearance. In this model, the liver is approximated as a well-mixed compartment with a fixed drug concentration. 37

Metabolism of opioids and opioid antagonists via individual CYPs
with considerable predicted metabolic activity (>20% of total metabolism) is shown in Figure 1. Rates of substrate consumption were evaluated for multiple substrate concentrations and  Table S7 for each substrate-CYP pair. CL int were calculated from in vitro parameters using an in vitro-in vivo correlation and are shown in Figure 3 in red (see Table S8 for numerical values and individual CYP contributions). CYP3A4 metabolism was characterized for all

| Metabolism via hepatic spheroids
Metabolism of opioids and opioid antagonists was characterized in hepatic spheroids and is shown in Figure 2, with fitted V max and K m values given in Table S9. Although all substrates exhibited typical Michaelis-Menten kinetics, data spread was considerably greater when compared with those observed in CYP metabolism. CL int were derived analogously using Michaelis-Menten parameters and an in vitro-in vivo correlation and are shown in Figure 3 in black (see

| DISCUSS ION
Opioid abuse is an ongoing public health concern, exacerbated by the emergence of more potent synthetic opioids, specifically FEN-class opioids. 1,2,4,5 Existing competitive antagonists (e.g., NX, NTX, NMF) have been proven effective against morphine and FEN-induced respiratory depression; however, efficacy against more potent opioids is not well-established. 10,[12][13][14][15]18 In particular, renarcotization remains a relevant issue due to the relatively short half-lives of available opioid antagonists. 19 To characterize the stability of antagonists For both in vitro systems investigated, metabolism followed typical Michaelis-Menten kinetics for all substrates (Figures 1 and 2).
Unsurprisingly, a larger data spread was observed in hepatocytes, Although only a single enzymatic reaction is represented, CYPs play a crucial role in drug and xenobiotic metabolism, with five isozymes (1A2, 2C9, 2C19, 2D6, and 3A4) responsible for the oxidation and metabolism of more than half of all marketed drugs. 24,40 Metabolic simulations for opioids and opioid antagonists identified CYP3A4, CYP2D6, and CYP2C19 as major contributors to overall metabolism (Table S1). Of these, only the 3A4 isozyme was predicted and experimentally confirmed to significantly process all substrates of interest (Table S8). This is anticipated for opioids, which are known to be predominantly metabolized via CYP-mediated oxidation. 25 On the other hand, even though opioid antagonists undergo glucuronidation more readily, substantial CYP3A4 activity was observed, albeit to a lesser degree than that measured for opioids. This additional metabolic activity from phase I CYP-mediated oxidation may culminate in the short half-lives of opioid antagonists relative to FEN-class opioids, which are not known to undergo glucuronidation to any significant degree. 25 Such consequences may present an important consideration for novel antagonist design, as the apparent ubiquity of CYP metabolism may hinder antagonist action if peripheral metabolic pathways are simultaneously present.
Select opioids and opioid antagonists were projected to undergo metabolism via multiple CYP isozymes, in this case, 2C19 for NX and 2D6 for FEN and CRF, in conjunction with 3A4 (Table S1). For

F I G U R E 2
Michaelis-Menten kinetics for the metabolism of opioids and opioid antagonists in spheroid-qualified human hepatocytes. Cells were seeded at a density of 1500 viable cells per well and exposed on the seventh day, following 5 days of undisturbed spheroid aggregation and an additional 2 days in serum-free maintenance media. Reactions were initiated with the addition of substrate, incubated at 37°C and 5% CO 2 , and terminated after 0, 1, 2, 4, and 24 h. Initial velocities were determined using an exponential plateau fit and are expressed in units of substrate consumed per hour per 10 6 cells. Substrate consumption was determined using LC-MS/MS. Cell concentrations were determined using flow cytometry following spheroid dissociation at the end of each time point. n = 2 for each time point at each concentration these substrates, measured CYP3A4 metabolic rates were considerably dominant in terms of intrinsic enzymatic activity, with CL int,3A4 at least an order of magnitude greater than that of other isozymes.
Moreover, relative contributions of individual CYPs were markedly different from simulations, with far greater metabolic activity via CYP3A4 than predicted (Table S8). In fact, oxidation mediated by either CYP2D6 or CYP2C19 was practically negligible (<5% and <1% of CYP3A4 activity, respectively, Contrary to the distinct separation between opioids and antagonists observed for CYP metabolism, intrinsic metabolic activities as measured in hepatic spheroids were largely comparable for all substrates ( Figure 3). Only minute differences were discernable in terms of the spans of CL int calculated for each class, with antagonists characterized by a considerably narrower range. This consistency in measured metabolic activity may be due to fundamental similarities in chemical structure. While agonists investigated share a common backbone, the overall degree of molecular similarity, as often quantified by the Tanimoto coefficient (T C ), is far greater amongst the morphine-based antagonists (T c >0.9 for antagonists vs. T c = 0.6-0.9 for agonists, Tables S10 and S11). 41 Given the nature of enzyme-substrate specificity, it may not be uncommon for structurally analogous molecules to share metabolic pathways, undergo similar biotransformations, and ultimately exhibit comparable metabolic rates. 42 The largely similar CL int observed for highly structurally analogous antagonists supports this theory and may provide the un- Given this relationship, piecewise structural modifications could be implemented to improve the lipophilicity of potential opioid countermeasures in an effort to extend antagonist effect. In theory, these physicochemical considerations, along with the desired metabolic features discussed previously (e.g., low CYP3A4 activity), should produce antagonists that are effective even against more potent opioids and should be further investigated to confirm the validity of utilizing these elements as a foundation for future drug development studies.
The results presented herein represent a comprehensive metabolic study utilizing the same in vitro system across multiple opioids and opioid antagonists, with hopes to facilitate the development of antagonists that are efficacious against potent synthetic opioids.
While metabolism via the CYP system and hepatic spheroids were explored, metabolism via other major metabolizing pathways (e.g. glucuronidation) or other in vitro systems (e.g. human liver microsomes) were not investigated in this study. Future studies incorporating the impact of glucuronidation and other metabolic pathways, such as metabolism via enzymes in plasma, and in either well established models, such as liver microsomes, or advanced in vitro systems (e.g. "liver-on-a-chip" platforms) may produce more robust data

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.

D I SCLOS U R E
The authors declare no competing interest.

E TH I C S S TATEM ENT
The views expressed herein are those of the authors and do not reflect the official policy of the Department of Army, Department of Defense, or the U.S. Government.