Visual Overview
Abstract
This study aimed to explore the cytochrome P450 (CYP) metabolic and inhibitory profile of hydroxychloroquine (HCQ). Hydroxychloroquine metabolism was studied using human liver microsomes (HLMs) and recombinant CYP enzymes. The inhibitory effects of HCQ and its metabolites on nine CYPs were also determined in HLMs, using an automated substrate cocktail method. Our metabolism data indicated that CYP3A4, CYP2D6, and CYP2C8 are the key enzymes involved in HCQ metabolism. All three CYPs formed the primary metabolites desethylchloroquine (DCQ) and desethylhydroxychloroquine (DHCQ) to various degrees. Although the intrinsic clearance (CLint) value of HCQ depletion by recombinant CYP2D6 was > 10-fold higher than that by CYP3A4 (0.87 versus 0.075 µl/min/pmol), scaling of recombinant CYP CLint to HLM level resulted in almost equal HLM CLint values for CYP2D6 and CYP3A4 (11 and 14 µl/min/mg, respectively). The scaled HLM CLint of CYP2C8 was 5.7 µl/min/mg. Data from HLM experiments with CYP-selective inhibitors also suggested relatively equal roles for CYP2D6 and CYP3A4 in HCQ metabolism, with a smaller contribution by CYP2C8. In CYP inhibition experiments, HCQ, DCQ, DHCQ, and the secondary metabolite didesethylchloroquine were direct CYP2D6 inhibitors, with 50% inhibitory concentration (IC50) values between 18 and 135 µM. HCQ did not inhibit other CYPs. Furthermore, all metabolites were time-dependent CYP3A inhibitors (IC50 shift 2.2–3.4). To conclude, HCQ is metabolized by CYP3A4, CYP2D6, and CYP2C8 in vitro. HCQ and its metabolites are reversible CYP2D6 inhibitors, and HCQ metabolites are time-dependent CYP3A inhibitors. These data can be used to improve physiologically-based pharmacokinetic models and update drug–drug interaction risk estimations for HCQ.
SIGNIFICANCE STATEMENT While CYP2D6, CYP3A4, and CYP2C8 have been shown to mediate chloroquine biotransformation, it appears that the role of CYP enzymes in hydroxychloroquine (HCQ) metabolism has not been studied. In addition, little is known about the CYP inhibitory effects of HCQ. Here, we demonstrate that CYP2D6, CYP3A4, and CYP2C8 are the key enzymes involved in HCQ metabolism. Furthermore, our findings show that HCQ and its metabolites are inhibitors of CYP2D6, which likely explains the previously observed interaction between HCQ and metoprolol.
Introduction
The 4-aminoquinoline hydroxychloroquine (HCQ), an old antimalarial drug, is regarded as a safe and reasonably effective treatment of rheumatoid arthritis and systemic lupus erythematosus (Munster et al., 2002; Rainsford et al., 2015). Beyond its approved indications, HCQ repurposing for the prevention and treatment of various diseases, including diabetes, myocardial infarction, and various cancers, is currently assessed in several clinical trials, due to its anti-inflammatory/immunomodulating, anti-thrombotic, and anti-autophagic properties (Plantone and Koudriavtseva, 2018; Ulander et al., 2021). HCQ is preferred over chloroquine because of lower incidence of cardiac, gastrointestinal, and ocular adverse reactions. During the first stages of the Coronavirus Disease of 2019 (COVID-19) pandemic, HCQ was among the most used repurposed therapeutic agents. Although the potential benefits of systemically administered HCQ for the management of COVID-19 are no longer considered to outweigh its potential risks (FDA, 2020; Horby et al., 2020; Skipper et al., 2020; Pan et al., 2021), investigation of its therapeutic and prophylactic use against COVID-19 still continues (47 recruiting and not yet recruiting clinical studies as of June 15, 2022, https://clinicaltrials.gov/).
HCQ has been in clinical use for more than 60 years, but its clinical pharmacology is not well understood (White et al., 2020). HCQ has a complex pharmacokinetic profile, displaying a high degree of variability in its concentrations and unclear pharmacokinetic–pharmacodynamic relationships in terms of therapeutic and adverse effects (Rainsford et al., 2015). Following oral administration, approximately 70–80% of HCQ is absorbed (Tett et al., 1989). In the body, HCQ distributes extensively to aqueous cellular and intercellular compartments. As a weak base, it accumulates in acidic organelles, such as lysosomes and endosomes (Tett et al., 1993; Schrezenmeier and Dorner, 2020). Thereby, HCQ has an enormous distribution volume (700 l/kg based on plasma data) (Tett et al., 1988). HCQ also concentrates in platelets and leukocytes, leading to approximately 7-fold higher concentrations in blood than in plasma (Tett et al., 1988; Tett et al., 1989; Brocks et al., 1994). Accordingly, whole blood is frequently used as the matrix in pharmacokinetic studies (White et al., 2020). HCQ is eliminated through both metabolism and renal excretion and has a long terminal elimination half-life of 26–53 days (Tett et al., 1988; Tett et al., 1989). The full mass balance profile of HCQ is unclear, but renal elimination is estimated to account for 20–55% of the total clearance (Tett et al., 1988; Tett et al., 1989; White et al., 2020).
HCQ is biotransformed into three active metabolites in humans: the major circulating metabolite desethylhydroxychloroquine (DHCQ), as well as desethylchloroquine (DCQ), and didesethylchloroquine (DDCQ) (Fig. 1) (McChesney, 1983; Tett et al., 1985; Charlier et al., 2018; Shimizu et al., 2022). Although many articles refer to cytochrome P450 (CYP) 2C8, CYP2D6, and CYP3A4 as the key enzymes involved in HCQ metabolism, these statements are based on chloroquine data only (Kim et al., 2003; Projean et al., 2003). In addition, except for a study showing no inhibition of CYP3A4 (Li et al., 2020), little is known about the in vitro inhibitory effects of HCQ on CYP enzymes. In healthy subjects, however, HCQ has increased the plasma exposure of the β1 adrenergic receptor antagonist and CYP2D6 substrate metoprolol by 65% (Somer et al., 2000), suggesting that HCQ is an inhibitor of CYP2D6. These reports implying that HCQ may be both a substrate and an inhibitor of CYP2D6 raise the concern of CYP2D6 autoinhibition and thereby of time-dependent nonlinear pharmacokinetics for HCQ. Due to this concern and these gaps in the knowledge of the metabolism and drug–drug interaction potential of HCQ, this study aimed to comprehensively investigate the in vitro CYP-mediated metabolism of HCQ and the CYP inhibitory effects of HCQ and its three main metabolites.
Hydroxychloroquine main metabolic pathways, with the most important CYP enzymes indicated for each reaction, based upon the present findings. DCQ and DHCQ are primary metabolites of hydroxychloroquine, and DDCQ is formed via subsequent metabolism of either of these primary metabolites.
Materials and Methods
Chemicals and Microsomes
Hydroxychloroquine sulfate was kindly provided by Orion Corporation (Espoo, Finland). Amodiaquine dihydrochloride dihydrate, astemizole, bupropion hydrochloride, desethylchloroquine, desethylhydroxychloroquine, desethylhydroxychloroquine-d4, dextrorphan tartrate, dextrorphan-d3 tartrate, didesethylchloroquine, didesethylchloroquine-d4, hydroxybupropion, hydroxybupropion-d6, hydroxychloroquine-d4, 7-hydroxycoumarin, 7-hydroxycoumarin-d5, ±4-hydroxymephenytoin-d3, 1-hydroxytacrine-d3, hydroxytolbutamide, hydroxytolbutamide-d9, N-desethylamodiaquine hydrochloride, N-desethylamodiaquine-d5, O-desmethylastemizole, S-4-hydroxymephenytoin, S-mephenytoin, montelukast sodium, tacrine hydrochloride dihydrate, and tolbutamide were purchased from Toronto Research Chemicals (Toronto, ON, Canada). Coumarin, dextromethorphan hydrobromide monohydrate, formic acid, α-hydroxymidazolam, α-hydroxymidazolam-d4, β-NADPH tetrasodium, quinidine, and troleandomycin were bought from Sigma-Aldrich (St. Louis, MO, USA). Acetonitrile and methanol were obtained from Honeywell Riedel-de Haën (Charlotte, NC, USA), and midazolam was obtained from Hoffmann-La Roche (Basel, Switzerland). Disodium hydrogen phosphate dihydrate was purchased from Merck (Darmstadt, Germany), sodium dihydrogen phosphate monohydrate from J.T. Baker & Mallinckrodt (Deventer, The Netherlands), 1-hydroxytacrine maleate and gemfibrozil 1-O-β-glucuronide from Santa Cruz Biotechnology (Dallas, TX, USA), O-desmethylastemizole-d4 from Medical Isotopes (Pelham, NH, USA), paroxetine hydrochloride from Synfine Research (Richmond Hill, ON, Canada), and ketoconazole from Janssen Biotech (Olen, Belgium).
Human liver microsomes (HLMs; XTreme 200 pooled, mixed-gender) used in metabolism, inhibition screening and inhibition constant (Ki) determination experiments, and a NADPH regenerating system were purchased from Sekisui XenoTech (Kansas City, KS, USA). HLMs (UltraPool 150 pooled, mixed-gender) used in IC50 determinations were from Corning (Woburn, MA, USA). Recombinant CYP Bactosomes (CYP1A2, CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP2E1, CYP2J2, CYP3A4, CYP3A5, and control Bactosomes) were from Cypex (Dundee, UK). Bovine serum albumin (BSA) was obtained from Biowest (Nuaillé, France). All solvents and commercially available reagents were of analytical grade and used without further purification.
Incubation Conditions and Sample Handling in Metabolism Experiments
All metabolism incubations were carried out at 37°C in sodium phosphate buffer (0.1 M, pH 7.4) in triplicate (CYP screening incubations in duplicate). HLM or recombinant CYP isoform and buffer were premixed and kept on ice until the start of the experiment. With the exception of studies including time-dependent CYP-selective inhibitors, experiments were started by premixing HCQ for 10 minutes with HLM or recombinant CYP buffer mixes on a heated shaker (37°C, 350 rpm), and followed by the addition of 1 mM NADPH to initiate the reactions. Reactions were stopped by moving a sample of the incubation mixture to acetonitrile containing internal standard (1:3). Samples were kept on ice for at least 10 minutes before centrifugation at 21,000 g for 10 minutes at 8°C and further processing (Supplemental Materials and Methods).
All stock solutions of HCQ, its metabolites, and inhibitors were prepared in methanol or acetonitrile. All incubations (including controls) contained the same concentration of organic solvent (1%). When the metabolite formation rate was measured in enzyme kinetic experiments, the incubation time was optimized within the linear range for metabolite formation, depending on the substrate turnover rate in each specific experiment (<20% turnover of substrate was required).
Metabolism by Recombinant CYPs
The metabolism of HCQ was first investigated in a recombinant CYP screening. HCQ (30 µM) was incubated with CYP enzyme (CYP1A2, CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP2E1, CYP2J2, CYP3A4, CYP3A5) or control Bactosomes at a protein concentration of 0.3 mg/ml with NADPH or without NADPH (negative controls) for 90 minutes. Based on the obtained data, seven CYP isoforms (CYP1A2, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP3A4, CYP3A5) were selected for a linearity experiment. Herein, the depletion of HCQ (1 and 10 µM) was measured at two protein concentrations (0.1 and 0.2 mg/ml) for up to 45 minutes. Samples were collected at 0, 7.5, 15, 30, and 45 minutes. Furthermore, the depletion of HCQ at a low initial concentration of 0.3 µM was studied in CYP2C8, CYP2D6, and CYP3A4 incubations (0.1 mg/ml). Samples were collected at the same time points as above. Finally, the enzyme kinetics of HCQ metabolite formation was tested in CYP2C8 (0.2 mg/ml), CYP2D6 (0.05 mg/ml), and CYP3A4 (0.2 mg/ml) incubations. The incubation times corresponded to 20, 10, and 20 minutes, respectively.
Metabolism in Human Liver Microsomes
The depletion of HCQ, DHCQ, and DCQ at initial concentrations of 0.3 and 3 µM were also studied in HLMs (0.5 mg/ml) in the presence of NADPH. Samples were collected at 0, 15, 30, 60, 90, and 120 minutes. To study the effects of CYP inhibition on HCQ metabolism, the time-dependent inhibitors gemfibrozil 1-O-β-glucuronide (75 µM; CYP2C8), paroxetine (15 µM; CYP2D6), and troleandomycin (100 µM; CYP3A) were first premixed with HLM (0.5 mg/ml) for 10 minutes before the addition of NADPH. After preincubation for 15 minutes, HCQ (3 µM) was included in the mix, and the reactions were allowed to incubate for 40 minutes. In addition, the effects of the reversible inhibitors montelukast (1 µM; CYP2C8), quinidine (10 µM; CYP2D6), and ketoconazole (1 µM, CYP3A) were studied by premixing the inhibitor with HCQ (3 µM) and HLM for 10 minutes before the addition of NADPH. The reactions were allowed to incubate for 30 minutes. In an additional experiment with a low initial HCQ concentration of 0.3 µM, the effects of the time-dependent inhibitors listed above were tested (identical incubation conditions) on HCQ depletion. Samples were collected at 0, 15, 30, 60, and 90 minutes.
Incubation Conditions and Sample Handling in Inhibition Experiments
The potential of HCQ, DCQ, DHCQ, and DDCQ to inhibit nine major CYP enzymes (CYP1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2J2, and 3A) by direct inhibition, slow-binding inhibition or time-dependent inhibition was investigated in HLMs, using a previously described automated probe substrate cocktail approach (Kahma et al., 2021).
Briefly, all incubations were performed in sodium phosphate buffer (0.1 M, pH 7.4) in triplicate in 96-well plates using an automated liquid handler (Tecan Freedom EVO 150 with Freedom EVOware software, Tecan Group, Männedorf, Switzerland) and a heated shaker (550 rpm, 37°C). Probe substrates with incubation concentrations approximating to their Michaelis-Menten constant (Km) values were mixed into two cocktails (Supplemental Table 1) to assess several CYP activities in one experiment. The HLM protein concentrations corresponded to 0.05 mg/ml and 0.1 mg/ml for cocktail 1 and 2, respectively. BSA (0.5% (w/v), final concentration) was included in cocktail 2 incubations to enhance CYP2C19 activity. The final solvent (methanol) concentration in all incubations was ≤1%.
In direct inhibition incubations, the inhibitors or solvent controls, probe substrates, BSA (for cocktail 2), and HLMs were diluted in buffer, and the mixture was prewarmed on the heated shaker for 3 minutes. Incubations were initiated by the addition of NADPH (1 mM)/a NADPH regenerating system (100 mM NADP, 500 mM glucose-6-phosphate, 100 units/ml glucose-6-phosphate dehydrogenase) and terminated after 5 minutes by mixing 30 µl of the incubation mixture with 90 µl of ice-cold methanol containing internal standards. All samples were kept at 4°C for 30 minutes before further processing and determination of metabolite concentrations (Supplemental Materials and Methods, Supplemental Table 2).
In slow-binding inhibition incubations, the inhibitors or solvent controls were preincubated with HLMs in buffer for 30 minutes on the heated shaker. Toward the end of the preincubation, the probe substrates and BSA (for cocktail 2) were included before CYP-mediated reactions were initiated by the addition of NADPH (1 mM)/a NADPH regenerating system (100 mM NADP, 500 mM glucose-6-phosphate, 100 units/ml glucose-6-phosphate dehydrogenase). Reactions were terminated after 5 minutes, and samples were handled as described above.
In time-dependent inhibition incubations, the inhibitors or solvent controls were preincubated with HLMs and NADPH (1 mM)/a NADPH regenerating system (100 mM NADP, 500 mM glucose-6-phosphate, 100 units/ml glucose-6-phosphate dehydrogenase) in buffer for 30 minutes on the heated shaker. Toward the end of the preincubation, BSA (for cocktail 2) was included before final incubations were initiated by adding the probe substrates. Reactions were terminated after 5 minutes, and samples were processed as described above.
Inhibition Screening and IC50 Experiments
In an initial screening, the direct, slow-binding, and time-dependent inhibitory potential of HCQ and its metabolites were tested at two inhibitor concentrations (10 and 50 µM). Based on the findings, IC50 experiments (direct and time-dependent inhibition) were carried out by incubating seven inhibitor concentrations (0.5–1,000 µM) with HLMs and the substrate cocktails. To determine the potential effect of BSA on the inhibition of the CYPs of cocktail 2, direct inhibition experiments were also carried out in the absence of BSA for HCQ and the primary metabolites.
Ki Experiments
Based on the findings of IC50 experiments, several incubations were carried out to determine the Ki values for the direct inhibition of CYP2D6 and 2J2 by HCQ and its metabolites, and to characterize the type of inhibition (competitive, noncompetitive, uncompetitive, or mixed inhibition). A series of inhibitor concentrations (1/4 to 5 times the direct IC50 values) were simultaneously incubated with four concentrations of dextromethorphan or astemizole (Km/3, Km, 3×Km, and 9×Km). All incubations were performed by hand without BSA.
Data Analysis of Metabolism Findings
The kinetics of substrate depletion in HLM and recombinant CYP incubations was analyzed using GraphPad Prism (version 7.03; GraphPad Software, Inc., San Diego, CA, USA). Depletion rate constants (kdep) were determined using nonlinear regression, and the intrinsic clearance (CLint) of HCQ and its metabolites was expressed as CLint = kdep/[M], where [M] is the HLM or recombinant CYP concentration used in the incubations. The kinetics of DCQ and DHCQ formation by CYP2C8, CYP2D6, and CYP3A4 were analyzed with the Michaelis-Menten, substrate inhibition, allosteric sigmoidal (Hill), and two enzymes models using GraphPad Prism. Selection of the best model for each reaction was based on the Akaike information criterion, R2 values, and a visual examination of Michaelis-Menten and Eadie-Hofstee plots. CLint values of each reaction were calculated according to CLint = Vmax/Km, where Vmax is the maximal velocity and Km is the Michaelis-Menten constant.
All CLint values were corrected for non-specific binding to protein by CLint,u = CLint/fu,mic, where CLint,u is the unbound intrinsic clearance and fu,mic is the unbound fraction of drug at various protein concentrations. fu,mic values were predicted as described in Supplemental Table 3. To estimate the relative contributions of CYP2C8, CYP2D6, and CYP3A4 to the metabolism of HCQ, Cypex LR intersystem extrapolation factors (ISEFs) and CYP expression values were obtained from the Simcyp Population-Based Simulator (V20; Simcyp Ltd, Certara, UK). The ISEFs corresponded to 0.982, 1.11, and 1.09, and the CYP expression levels to 24, 9.4, and 137 pmol/mg for CYP2C8, CYP2D6, and CYP3A4, respectively. In addition, for CYP2D6 and CYP3A4, CLint-based ISEFs (CLISEFs) were calculated based on the reported marker activities for the used lots of recombinant enzymes and HLM (Proctor et al., 2004). A CLISEF value was not calculated for CYP2C8, as different marker substrates had been used for recombinant enzyme and HLM. For CYP2D6 and CYP3A4, the CLISEFs corresponded to 0.57 and 0.42, respectively. The recombinant CYP CLint,u values were then multiplied with the respective ISEF or CLISEF and CYP expression value to obtain HLM CLint,u values. Measured and scaled HLM CLint,u values were further scaled to CLint,in vivo using 39.79 mg microsomal protein/g liver, and liver volume and density values of 1.65 l and 1,080 g/l liver (Simcyp Population-Based Simulator V20). In the final step, hepatic blood clearance (CLH) values were calculated using the well-stirred model (Yang et al., 2007),
where QH is the hepatic blood flow (1.610 l/min) (Pelkonen and Turpeinen, 2007) and fu,B is the unbound fraction of HCQ in blood. fu,B was calculated according to
, where fu,p and BP are the unbound fraction in plasma (0.48) and blood-to-plasma concentration ratio (7.2) of HCQ, respectively (Tett et al., 1988; McLachlan et al., 1993). The calculated fu,B equaled to 0.067.
Data Analysis of Inhibition Findings
IC50 and Ki values were determined by nonlinear regression using GraphPad Prism (version 8.4.3). Inhibitor concentration-response data were fitted to the following four-parameter log-logistic equation (variable slope sigmoidal model):
where Y is the percentage of remaining CYP activity compared to the solvent controls, X is the inhibitor concentration, and n is the Hill slope. The bottom plateau was set to zero when no bottom plateau could be reliably inferred; otherwise, parameters were not constrained. In cases where a stronger CYP inhibition was observed following preincubation of the inhibitor, IC50 shift (IC50 value obtained without preincubation/IC50 value obtained with preincubation) values were determined. An IC50 shift value ≥1.5 was used to denote time-dependent inhibition.
Rate versus probe substrate concentration data were fitted to the following equations for competitive inhibition (eq. 1), noncompetitive inhibition (eq. 2), uncompetitive inhibition (eq. 3), or mixed-type inhibition (eq. 4) (Copeland, 2000):
where v is the velocity of the reaction, Vmax is the maximum velocity, S is the substrate concentration, Km is the Michaelis constant (substrate concentration at Vmax/2), [I] is the inhibitor concentration, Ki is the inhibition constant describing the affinity of the inhibitor for the enzyme, and αKi describes the affinity of the inhibitor for the enzyme-substrate complex. The type of inhibition was determined based on the Akaike information criterion and further confirmed by visual examination of Michaelis-Menten and Eadie-Hofstee plots. The latter were created by plotting the transformed data and the following lines:
Prediction of Clinical Drug-Drug Interactions Due to CYP2D6 Inhibition
The combined effects of HCQ and its metabolites on the plasma exposure of the CYP2D6 substrate metoprolol were predicted using a static mechanistic model equation (Templeton et al., 2016),
where AUCR is the fold change in metoprolol area under the concentration-time curve (AUC) in the presence (AUCwith HCQ) and absence (AUCwithout HCQ) of the perpetrators, DCQ is desethylchloroquine, DDCQ is didesethylchloroquine, DHCQ is desethylhydroxychloroquine, HCQ is hydroxychloroquine, [I] is the inhibitor concentration, and fm is the fraction of the metoprolol dose cleared by CYP2D6. A fm value of 0.8 was used, estimated based on clinical pharmacogenetic and interaction data available in the UW Drug Interaction Database (DIDB, Copyright University of Washington, accessed on February 11, 2021). Experimental reversible inhibition constants (Ki) were used and adjusted for non-specific binding to HLM at 0.1 mg/ml (Supplemental Table 3). Because HCQ and its three metabolites accumulate largely in blood and tissues, six sets of interaction predictions were carried out, based on their blood (1), corresponding plasma (2), and estimated liver (3–6) concentrations. In (1), the total blood concentrations of the metabolites were estimated from HCQ concentrations and HCQ/metabolite ratios in blood (Supplemental Table 4). In (2), the corresponding total plasma concentrations of HCQ and its metabolites were calculated from their blood concentrations and blood-to-plasma values (Supplemental Table 4). In (3), their total liver concentrations were estimated from the blood concentrations and mouse tissue-to-blood concentration (Kp) values (Chhonker et al., 2018) (Supplemental Table 4). In (4), their total liver concentrations were estimated from plasma concentrations and tissue-to-plasma partition coefficients predicted in Simcyp Population-Based Simulator V20 as described in Supplemental Table 4. In (5–6), their unbound liver concentrations were estimated by multiplying the total liver concentrations by the unbound fraction values in plasma (thus assuming an unbound fraction in the hepatocytes that equals that observed in plasma for each compound; Supplemental Table 4).
Results
Metabolism in Recombinant CYP incubations
In screening experiments, considerable HCQ depletion was observed in CYP2D6 and CYP2C8 incubations (Supplemental Fig. 1). The highest concentrations of the primary metabolites DCQ and DHCQ were formed in CYP2D6, CYP2C8, and CYP3A4 incubations (Supplemental Fig. 1). CYP3A5, CYP2C19, CYP2C9, and CYP1A2 also formed small amounts of these metabolites. The secondary metabolite DDCQ was formed abundantly in CYP2D6 incubations, while only very small amounts were produced by CYP2C8 and CYP3A4.
In follow-up experiments, when the metabolism of HCQ was studied, CYP2D6 and CYP2C8 formed the highest concentrations of HCQ primary metabolites (data of the 1 µM HCQ/0.1 mg/ml protein experiment is shown in Fig. 2, A–D). CYP3A4 also formed smaller amounts of the metabolites. DDCQ was only formed by CYP2D6. At a low HCQ concentration of 0.3 µM, the CLint of the CYP2D6-mediated depletion of HCQ was 4.3-fold and 12-fold higher than the depletion mediated by CYP2C8 and CYP3A4, respectively (Table 1, Supplemental Fig. 2A).
Depletion of HCQ (1 µM; A) and resulting metabolite formation (B–D) in recombinant CYP incubations (0.1 mg/ml, 45 minutes). An equal protein concentration was used in parallel experiments (resulting in variable CYP contents) to keep the possible nonspecific binding of HCQ equal across incubations. The data represent mean and standard deviation values of triplicate incubations of one experiment.
Measured intrinsic and scaled hepatic clearance values of HCQ depletion (0.3 µM) in HLM and recombinant CYP2C8, CYP2D6, and CYP3A4 incubations. Values shown are mean ± standard deviation of triplicates.
DHCQ and DCQ formation by CYP2C8, CYP2D6, and CYP3A4 followed Michaelis-Menten kinetics, with evidence of substrate inhibition type kinetics in some cases (Fig. 3, A–F). CYP2D6 displayed the lowest Km values of 20 and 7.5 µM for DCQ and DHCQ formation, respectively.
Enzyme kinetics of DCQ and DHCQ formation in recombinant CYP2C8, CYP2D6, and CYP3A4 incubations (A–F). Data represent mean and standard deviation values of triplicate incubations of one experiment.
Although the depletion rate of HCQ in recombinant CYP2D6 incubations was more than 10-fold higher than the one in CYP3A4 incubations, scaling of the recombinant data to HLM level resulted in almost equal CLint values for CYP2D6 and CYP3A4, due to the much greater abundancy of CYP3A4 compared with CYP2D6 (Table 1). The ISEF-based scaled HLM CLint value of CYP2C8 was approximately half of those of CYP2D6 and CYP3A4.
Metabolism in Human Liver Microsomes
An initial HLM experiment including negative controls (no NADPH) ruled out any HCQ metabolism that would take place without external cofactors (data not shown). In HLM incubations with HCQ (3 µM), both its primary metabolites were formed (Fig. 4A). In HLM incubations with DCQ and DHCQ (3 µM) as the substrate, only DDCQ was formed (Fig. 4, B–C). In experiments with both a reversible and a time-dependent inhibitor for each enzyme, CYP2D6, CYP3A, and CYP2C8 inhibitors inhibited the formation of DHCQ from HCQ by 34–49%, 46–47%, and 27–32%, respectively (Fig. 4D). For DCQ formation, the experiments with CYP-selective reversible and time-dependent inhibitors suggested a slightly higher importance for CYP3A, with 24–44% inhibition by the CYP2D6 inhibitors, 52–57% inhibition by the CYP3A inhibitors, and 23–29% inhibition by the CYP2C8 inhibitors (Fig. 4E). In these experiments, HCQ depletion was too slow to reliably measure partial inhibition of the depletion rate by time-dependent CYP2D6, CYP3A, and CYP2C8 inhibitors (data not shown).
Depletion and CYP-selective inhibition experiments in HLMs. The depletion of HCQ 3 µM (A), DHCQ 3 µM (B), and DCQ 3 µM (C), and resulting metabolite formation in HLM incubations (0.5 mg/ml, 120 minutes) are shown in the top panel. In the bottom panel, the effects of CYP-selective reversible and time-dependent inhibitors on DHCQ (D) and DCQ (E) formation from HCQ (3 µM) are illustrated. The data represent mean and standard deviation values of triplicate incubations of one experiment. GEM-G, gemfibrozil 1-O-β-glucuronide; KETO, ketoconazole; MON, montelukast; QUIN, quinidine; PAR, paroxetine; TAO, troleandomycin.
The depletion of HCQ, DCQ, and DHCQ at a low initial concentration of 0.3 µM resulted in CLint values of 5.6–12 µl/min/mg (Supplemental Fig. 2B; Table 1). For HCQ, scaling of the CLint resulted in a scaled hepatic clearance value of 4.1 l/h, which approximates to 71% of the published blood clearance of HCQ (Table 1).
Inhibition Screening Using CYP Probe Substrate Cocktail Assays
In preliminary screening experiments, HCQ was shown to cause direct inhibition of CYP2D6 (Supplemental Fig. 3); DCQ and DHCQ were direct inhibitors of CYP2D6, and of CYP2J2 to a lesser extent, as well as time-dependent inhibitors of CYP3A (Supplemental Fig. 4–5); while DDCQ appeared to have a broader inhibitory effect on CYP enzymes, with marked direct inhibition of CYP2D6 and CYP2J2, and time-dependent inhibition of CYP3A (Supplemental Fig. 6). There was no evidence for slow-binding inhibition of any CYP enzyme by HCQ and its metabolites (Supplemental Fig. 3–6).
The inhibitory effects of HCQ, DCQ, DHCQ, and DDCQ on dextromethorphan O-demethylation (CYP2D6 probe reaction), astemizole O-demethylation (CYP2J2 probe reaction), and midazolam 1’-hydroxylation (CYP3A probe reaction) in HLM incubations. IC50 values were determined following no preincubation (direct inhibition) or after a 30-minute preincubation of inhibitor in the presence of NADPH (time-dependent inhibition), as described in Materials and Methods. The obtained IC50 values are given in Table 2. The data points show mean and standard deviations of measured CYP activity in inhibitor incubations as compared with that in solvent control incubations (triplicate incubations of one experiment). As the screening indicated no inhibition of CYP3A by HCQ, no CYP3A IC50 values were determined for HCQ.
Direct inhibition of dextromethorphan O-demethylation (CYP2D6 probe reaction) by HCQ, DCQ, DHCQ and DDCQ in HLM incubations. The rate of metabolite formation was assessed at four substrate concentrations over a range of inhibitor concentrations to determine Ki values as described in Materials and Methods. The data points show mean and standard deviations of measured rates of metabolite formation in inhibitor incubations as compared with that in solvent control incubations (triplicate incubations of one experiment). DEX, dextromethorphan.
IC50 Experiments Using CYP Probe Substrate Cocktail Assays
Based on the screening data, HCQ and its metabolites were further tested for direct and time-dependent inhibition of selected CYPs. All test compounds were direct inhibitors of CYP2D6, and all three metabolites were direct inhibitors of CYP2J2, with no evidence of time-dependent inhibition (Table 2; Fig. 5, A–H). HCQ and its metabolites were moderate CYP2D6 inhibitors (IC50 values ranging from 18 to 135 µM; Table 2), while the observed inhibition of CYP2J2 by the metabolites was in general weaker (IC50 values between 63 and 504 µM; Table 2). Of the four compounds tested, DCQ was the most potent (direct) inhibitor of CYP2D6 (IC50 = 18 µM), while DDCQ was the most potent (direct) inhibitor of CYP2J2 (IC50 = 63 µM). BSA had little (≤1.3-fold difference) or no effect on the direct IC50 values of CYP2D6 and CYP2J2 (Supplemental Table 5).
Inhibitory effects of HCQ, DCQ, DHCQ, and DDCQ on CYP activities in HLM incubations. IC50 values were determined following no preincubation (direct inhibition) or after a 30-min preincubation of inhibitor in the presence of NADPH (time-dependent inhibition), as described in Materials and Methods. A full table of the obtained results of the IC50 experiments (also showing lack of inhibition and the effects of BSA exclusion in cocktail 2) can be found in the supplement (Supplemental Table 5). IC50 values are reported as means of three determinations with their 95% confidence intervals.
Compared with no preincubation, CYP3A inhibition increased after preincubation for 30 minutes with NADPH for all three metabolites, indicating that they are time-dependent CYP3A inhibitors (Table 2; Fig. 5, I–K; Supplemental Table 5). Preincubation of DCQ with NADPH resulted in a 3.2-fold decrease in its IC50 value for CYP3A inhibition from 149 µM to 46 µM. Preincubation of DHCQ with NADPH had a slightly weaker effect on CYP3A inhibition, with a 2.2-fold IC50 shift (from 260 µM to 117 µM). DDCQ inhibited CYP3A with the lowest IC50 values, with a 3.4-fold shift in IC50 from 40 µM to 12 µM after preincubation with NADPH.
Ki Determination Experiments for CYP2D6 and CYP2J2 Direct Inhibition
Statistical and visual examination of Michaelis-Menten and Eadie-Hofstee plots suggested that HCQ, DCQ, DHCQ, and DDCQ competitively inhibited CYP2D6 in pooled HLMs with Ki values of 32.5, 10.4, 48.1, and 24.1 μM, respectively (Fig. 6). DCQ, DHCQ, and DDCQ were mixed-type inhibitors of CYP2J2 with Ki values of 382, 508, and 41.8 μM, respectively (Fig. 7). All Ki values were in good agreement with the IC50 values determined during the IC50 shift experiments.
Direct inhibition of astemizole O-demethylation (CYP2J2 probe reaction) by DCQ, DHCQ and DDCQ in HLM incubations. The rate of metabolite formation was assessed at four substrate concentrations over a range of inhibitor concentrations to determine Ki values as described in Materials and Methods. The data points show mean and standard deviations of measured rates of metabolite formation in inhibitor incubations as compared with that in solvent control incubations (triplicate incubations of one experiment). AST, astemizole.
Prediction of Clinical Drug–Drug Interactions Due to Direct CYP2D6 Inhibition
The predicted fold increase in metoprolol AUC (AUCR) following direct inhibition of CYP2D6 by HCQ and its three metabolites is shown in Fig. 8. AUCR were calculated based on HCQ, DCQ, DHCQ, and DDCQ total blood (1), total plasma (2), and total and unbound liver (3-6) concentrations. In a clinical study, administration of HCQ with metoprolol caused a 1.65-fold increase in metoprolol AUC in six patients, whose total HCQ blood concentrations ranged from 1.5 to 2.3 µM.
Prediction of the total inhibitory effect of HCQ and its metabolites on the pharmacokinetics of the CYP2D6 substrate metoprolol in vivo. In a clinical interaction study between HCQ and metoprolol in six healthy volunteers, an eight-day treatment with HCQ increased the area under the plasma concentration-time curve (AUC) of a single dose of metoprolol (on day 9) by 1.65-fold (average fold increase indicated by the bold dashed line, with individual values ranging from 1.1 to 2.4 indicated by the dashed lines) (Somer et al., 2000). In the clinical study, HCQ blood concentrations were 1.5–2.3 µM at single time points on days 8 and 9 (pink area). In predictions based on the present CYP2D6 direct Ki values and total blood (A) or plasma (B) concentrations of HCQ and its metabolites, HCQ concentrations of the measured magnitude did not explain the observed interaction (AUCR <1.1). When the predictions were based on estimated total liver concentrations (C–D), AUCR values of >2.9-fold were obtained. However, predictions based on unbound liver concentrations (E–F; assuming an unbound fraction in the hepatocytes that equals that observed in plasma for each compound), AUCR values of the clinically observed magnitude were obtained. A fraction metabolized by CYP2D6 (fm,CYP2D6) of 0.8 for metoprolol was used in the predictions, and several assumptions were made (Materials and Methods). Metoprolol AUCR, metoprolol AUCwith HCQ/metoprolol AUCwithout HCQ ratio; EMs, extensive CYP2D6 metabolizers.
Based on these blood (1) or corresponding plasma (2) concentrations (0.21–0.32 µM) combined with the corresponding (predicted) metabolite concentrations in blood or plasma, the predicted metoprolol AUCR was ≤1.1 (Fig. 8, A–B). However, when the corresponding (estimated) total liver concentrations (3–4) were used, the predicted metoprolol AUCR values exceeded 2.9 at the clinically relevant HCQ concentrations (Fig. 8, C–D). On the other hand, with the corresponding unbound liver concentrations (5–6), the predicted AUCR values were slightly lower (2.2–2.8) (Fig. 8, E–F). In all predictions, the contribution of metabolites was significant.
Discussion
Until now, there has been little direct evidence of the interactions of HCQ with CYP enzymes. In the present study, we thoroughly screened the CYP-mediated metabolism of HCQ, and tested the reversible and time-dependent inhibitory effects of HCQ and its three main metabolites on nine drug-metabolizing CYPs. Our collective findings from HLMs and recombinant enzymes suggest that HCQ is mainly metabolized by CYP3A4, CYP2D6, and CYP2C8. In addition, HCQ and its metabolites are reversible, competitive inhibitors of CYP2D6, and all three HCQ metabolites are mixed-type inhibitors of CYP2J2 and time-dependent inhibitors of CYP3A.
In the literature, CYP2D6, CYP3A4, and CYP2C8 are often claimed to be the enzymes responsible for HCQ metabolism. However, these studies generally refer to chloroquine data reported by Kim et al. (2003) and Projean et al. (2003). The present findings show for the first time that these same enzymes are, indeed, also responsible for HCQ metabolism. Although the HCQ depletion rate in recombinant CYP2D6 incubations was more than 10-fold higher than that in CYP3A4 incubations, scaling of the data to the HLM level resulted in almost equal CLint values for CYP2D6 and CYP3A4 (Table 1). Depending on the scaling method used (ISEF or CLISEF), different hepatic clearance values were obtained for CYP2D6 and CYP3A4. Of these, the CLISEF-based hepatic clearance values for CYP2D6 and CYP3A4 (1.6 and 1.5 l/h, respectively) were best in line with the microsomal-derived CLH value (4.1 l/h). Unfortunately, we were unable to carry out CLISEF-based scaling of the CYP2C8 CLint value. Nevertheless, according to depletion and inhibition data, CYP2D6 and CYP3A4 seem to be equally important in the formation of DHCQ, whereas CYP3A4 plays a slightly larger role in DCQ formation. CYP2C8 seems to contribute slightly less, approximately 20–25% to both pathways.
There seems to be no published interaction studies investigating the effects of inhibitors of CYP2D6, CYP2C8, and CYP3A on the pharmacokinetics of HCQ. However, the DHCQ/HCQ ratio was associated with the CYP2D6 genotype in Korean lupus patients receiving HCQ (Lee et al., 2016). Another study did not find any significant association between CYP genotypes and HCQ response in British patients, although there was a trend for CYP2C8*3 and CYP2C8*4 to be associated with greater odds of response (Wahie et al., 2011).
Except for a study showing no reversible inhibition of CYP3A4 by HCQ (Li et al., 2020), little is known about the inhibitory effects of HCQ on CYP enzymes in vitro. In our study, all three metabolites of HCQ inhibited CYP3A in a NADPH- and time-dependent fashion, as evidenced by IC50 shift values >1.5 (Table 2). Their inactivation constants will be determined in a follow-up study, since time-dependent inhibition may cause a longer-lasting inhibitory effect, as compared with reversible inhibition. In addition, it may result in hapten formation and in some cases trigger an idiosyncratic adverse reaction (Kalgutkar et al., 2007). There seems to be little data on the effects of HCQ on CYP3A substrates in vivo, but HCQ has increased the plasma exposure of the CYP3A substrate MK-2206 by 16–92% in cancer patients (Mehnert et al., 2019).
In healthy subjects, HCQ has increased the plasma exposure of metoprolol by 65%, suggesting that it acts as an inhibitor of CYP2D6 (Somer et al., 2000). According to our data, HCQ and all of its three metabolites are reversible, competitive CYP2D6 inhibitors. DCQ was the most potent inhibitor with an IC50 value (18 µM) more than 3-fold lower than those of the other compounds and a Ki value of 10.4 µM (Fig. 6B). Static predictions based on the in vitro inhibitory data, however, suggested only a minimal (<1.1-fold) increase in metoprolol AUC when making the predictions using the total blood concentrations of HCQ measured at the end of the HCQ treatment in the clinical study (1.5–2.3 µM) or corresponding total plasma concentrations. However, HCQ and its metabolites accumulate extensively into tissues, indicating that their intracellular concentrations are higher than those in the blood stream. Accordingly, in predictions based on intracellular hepatocyte concentrations, estimated using mouse Kp or predicted human Kp values, we obtained much higher predicted AUC increases that were close to the observed interaction with metoprolol. Hence, our predictions suggest that the extensive accumulation of HCQ and its metabolites into tissues must be taken into account when predicting CYP-mediated interactions with HCQ as the perpetrator drug. Of note, the contribution of the metabolites to the total inhibitory effect was significant in all predictions. Nevertheless, the inhibitory effect of HCQ and its metabolites on CYP2D6 is of particular concern when used concomitantly with CYP2D6 substrates that, similarly to HCQ, prolong QT interval, such as ondansetron and haloperidol. Moreover, autoinhibition of CYP2D6 may reduce its contribution to the overall metabolism of HCQ, thereby also increasing the relative importance of CYP2C8 and CYP3A4. To the best of our knowledge, there are no clinical reports to date suggesting time-dependent nonlinear pharmacokinetics for HCQ.
HCQ and its metabolites have very complex and unusual pharmacokinetic profiles. In addition to their extensive accumulation in blood and tissues, as a complicating factor, they display stereoselective pharmacokinetics and pharmacodynamics, as well as toxic properties (McChesney, 1983; McLachlan et al., 1993; Brocks et al., 1994; Ducharme et al., 1995; Lim et al., 2009). Unfortunately, the individual HCQ and metabolite enantiomers were not commercially available at the time of our study. Hence, the present in vitro CLint and scaled hepatic clearance values reflect those of the racemic compounds. The present hepatic clearance (4.1 l/h) calculated based on the depletion of HCQ in HLMs approximates to 71% of the measured total intravenous clearance (5.8 l/h). Our finding is in good agreement with clinical data showing that renal clearance accounts for approximately one third of the total plasma clearance of HCQ, whereas metabolism and biliary excretion is the predominant route of elimination (Tett et al., 1988; Tett et al., 1989).
Because of the complex pharmacokinetics of HCQ and the wide variability in its concentration profile, it has been difficult to relate measured HCQ concentrations to its therapeutic and adverse effects (Rainsford et al., 2015). Nevertheless, several pharmacokinetic and physiologically-based pharmacokinetic models have been developed for HCQ, in particular after the outbreak of COVID-19 (Collins et al., 2018; Themans et al., 2020; Idkaidek et al., 2021). Our data can be used to update these models, and to simulate the effects of CYP-mediated drug-drug interactions and variants in CYP genes. In addition to CYPs, HCQ interacts with drug transporters. HCQ is a substrate of P-glycoprotein in vitro (Weiss et al., 2020). There seem to be no studies investigating HCQ and transporter pharmacogenetics, but ATP-binding cassette transporter A4, organic anion transporting polypeptide (OATP) 1A2, and OATP1B1 have been associated with chloroquine pharmacokinetics and response (Grassmann et al., 2015; Sortica et al., 2017). With respect to transporter inhibition, HCQ does not affect the activities of breast cancer resistance protein, multidrug resistance-associated protein 1, organic anion transporter (OAT) 1, OAT3, OATP1B1, or OATP1B3 in vitro (Weiss et al., 2020; Telbisz et al., 2021; Yee et al., 2021). However, it is a potent inhibitor of multidrug and toxin extrusion proteins 1 and 2 (IC50 2–4 and 1–7 µM, respectively), and a moderate or weak inhibitor of P-glycoprotein (IC50 52 µM), organic cation transporter (OCT) 1 (IC50 20–47 µM), OCT2 (IC50 ≥ 5 µM), OATP1A2 (IC50 9–19 µM), and OATP2B1 (IC50 ≥ 84 µM) (Xu et al., 2016; Weiss et al., 2020; Martinez-Guerrero et al., 2021; Telbisz et al., 2021; Yee et al., 2021). Inhibition of OATP1A2 by HCQ in the retinal pigment epithelium has been suggested to contribute to the retinal degradation observed in patients using HCQ (Xu et al., 2016). Together, these novel transport data and our metabolism data can be combined in physiologically-based pharmacokinetic models to simulate the role of enzyme-transport interplay in HCQ pharmacokinetics.
In conclusion, the present study shows for the first time that CYP2D6, CYP3A4, and CYP2C8 are responsible for the in vitro metabolism of HCQ. Furthermore, our data indicate that HCQ and its metabolites are reversible, competitive inhibitors of CYP2D6, and that HCQ metabolites are mixed-type inhibitors of CYP2J2 and time-dependent inhibitors of CYP3A. The current data can thus be applied to improve physiologically-based pharmacokinetic models and update drug–drug interaction risk estimations for HCQ. Collectively, our findings contribute to an improved understanding of the clinical pharmacology of HCQ.
Acknowledgments
The authors thank Orion Corporation (Espoo, Finland) for providing hydroxychloroquine sulfate for in vitro experiments.
Authorship Contributions
Participated in research design: Backman, Filppula, Kahma, Kurkela, Niemi, Paludetto.
Conducted experiments or drug concentration analysis: Filppula, Kurkela, Paludetto.
Performed data analysis: Filppula, Kurkela, Paludetto.
Wrote or contributed to the writing of the manuscript: Backman, Filppula, Kahma, Kurkela, Niemi, Paludetto.
Footnotes
- Received July 1, 2022.
- Accepted October 31, 2022.
This work was supported by the Academy of Finland (Grant decision 325667, 2019; Helsinki, Finland), Sigrid Jusélius Foundation (Grant number 8037 for J.T.B. and 1101 for M.N.), and State funding for university-level health research to HUS Helsinki University Hospital (TYH2019300 and TYH2021304 for J.T.B., TYH2019240, TYH2020323, and TYH2021324 for M.N.).
No author has an actual or perceived conflict of interest with the contents of this article.
Parts of this work were previously presented at the Pertti Neuvonen Symposium – Clinical Pharmacology and Therapeutics: from Research to Clinical Practice (November 29-30, 2021, Helsinki Finland) and at the 15th Congress of the European Association for Clinical Pharmacology and Therapeutics (June 25–28, 2022, Athens, Greece).
↵1Marie-Noëlle Paludetto and Mika Kurkela contributed equally to this work.
↵
This article has supplemental material available at dmd.aspetjournals.org.
ABBREVIATIONS
- AUC
- area under the concentration-time curve
- BSA
- bovine serum albumin
- CL
- clearance
- CLISEF
- intrinsic clearance-based intersystem extrapolation factor
- COVID-19
- Coronavirus Disease of 2019
- CYP
- cytochrome P450
- DCQ
- desethylchloroquine
- DDCQ
- didesethylchloroquine
- DHCQ
- desethylhydroxychloroquine
- fu,mic
- unbound fraction in microsomes
- HCQ
- hydroxychloroquine
- HLM
- human liver microsomes
- ISEF
- intersystem extrapolation factor
- Ki
- inhibition constant
- Km
- Michaelis-Menten constant
- Kp
- tissue-to-blood concentration coefficient or tissue-to-plasma concentration coefficient
- OAT
- organic anion transporter
- OATP
- organic anion transporting polypeptide
- OCT
- organic cation transporter
- Vmax
- maximal velocity
- Copyright © 2023 by The Author(s)
This is an open access article distributed under the CC BY-NC Attribution 4.0 International license.