1521-009X/44/5/624–633$25.00 http://dx.doi.org/10.1124/dmd.115.068932 DRUG METABOLISM AND DISPOSITION Drug Metab Dispos 44:624–633, May 2016 Copyright ª 2016 by The American Society for Pharmacology and Experimental Therapeutics Development of a Rat Plasma and Brain Extracellular Fluid Pharmacokinetic Model for Bupropion and Hydroxybupropion Based on Microdialysis Sampling, and Application to Predict Human Brain Concentrations

Thomas I.F.H. Cremers, Gunnar Flik, Joost H.A. Folgering, Hans Rollema, and Robert E. Stratford, Jr.

Brains On-Line BV, Groningen, The Netherlands (T.I.F.H.C., G.F. J.H.A.F.); Rollema Biomedical Consulting, Mystic, Connecticut (H.R.); and Division of Pharmaceutical Sciences, Mylan School of Pharmacy, Duquesne University, Pittsburgh, Pennsylvania (R.E.S.)

Received December 11, 2015; accepted February 24, 2016 Downloaded from

ABSTRACT Administration of bupropion [(6)-2-(tert-butylamino)-1-(3-chlorophenyl) concentration ratio at early time points was observed with propan-1-one] and its preformed active metabolite, hydroxybupropion bupropion; this was modeled as a time-dependent uptake clear- [(6)-1-(3-chlorophenyl)-2-[(1-hydroxy-2-methyl-2-propanyl)amino]- ance of the drug across the blood–brain barrier. Translation of the

1-propanone], to rats with measurement of unbound concentrations model was used to predict bupropion and hydroxybupropion dmd.aspetjournals.org by quantitative microdialysis sampling of plasma and brain extra- exposure in human brain extracellular fluid after twice-daily cellular fluid was used to develop a compartmental pharmacoki- administration of 150 mg bupropion. Predicted concentrations netics model to describe the blood–brain barrier transport of both indicate that preferential inhibition of the dopamine and norepi- substances. The population model revealed rapid equilibration of nephrine transporters by the metabolite, with little to no contribu- both entities across the blood–brain barrier, with resultant steady- tion by bupropion, would be expected at this therapeutic dose. state brain extracellular fluid/plasma unbound concentration ratio Therefore, these results extend nuclear imaging studies on dopa- estimates of 1.9 and 1.7 for bupropion and hydroxybupropion, mine transporter occupancy and suggest that inhibition of both at ASPET Journals on September 25, 2021 respectively, which is thus indicative of a net uptake asymmetry. transporters contributes significantly to bupropion’s therapeutic An overshoot of the brain extracellular fluid/plasma unbound efficacy.

Introduction bupropion-related plasma exposure, with one of its metabolites, 6 Bupropion [(6)-2-(tert-butylamino)-1-(3-chlorophenyl)propan-1- hydroxybupropion [( )-1-(3-chlorophenyl)-2-[(1-hydroxy-2-methyl- one] was originally introduced as an effective medication for the 2-propanyl)amino]-1-propanone], representing the predominant circu- treatment of depression (Wellbutrin). Among antidepressants, its lating entity (Ascher et al., 1995; Jefferson et al., 2005). The relative mechanism is considered atypical in that it has no effects on the systemic exposure of this metabolite to bupropion in repeated dosing serotonergic system; rather, its efficacy in depression has been ranges from 5- to 15-fold (Johnston et al., 2001; Learned-Coughlin attributed to its ability to increase dopaminergic and noradrenergic et al., 2003; Benowitz et al., 2013). Given this significant difference, tone, presumably via blockade of the respective synaptic reuptake it has been suggested that hydroxybupropion, which has similar transporters: the dopamine reuptake transporter (DAT) and the inhibitory activity at the DAT, but is a more potent NET inhibitor than norepinephrine reuptake transporter (NET) (Fava et al., 2005). After bupropion, contributes significantly to bupropion’s efficacy (Ascher its approval for depression, bupropion was also marketed as a smoking et al., 1995; Bondarev et al., 2003; Damaj et al., 2004; Lukas et al., cessation aid (Zyban). In addition to its effects on the DAT and NET, it 2010). Moreover, the higher exposure to hydroxybupropion may also has been suggested that bupropion’s efficacy in nicotine dependence be contributing to the side effects and adverse events associated with may be associated with alteration of central nicotinic acetylcholine bupropion therapy, particularly seizures observed at higher doses receptor function (Slemmer et al., 2000; Damaj et al., 2004). (Wooltorton, 2002). Single and repeat administration of therapeutic doses of bupropion in Use of animal models, particularly mice and rats, to understand the humans indicate that bupropion represents a minor fraction of the total complex pharmacology of bupropion is hampered by the very different disposition of bupropion in animals versus humans. In these rodent This research was sponsored by Brains On-Line (South San Francisco, CA). species, hydroxybupropion contributes approximately 10% to the total dx.doi.org/10.1124/dmd.115.068932. systemic and whole brain exposure observed (Suckow et al., 1986),

ABBREVIATIONS: 22LL, 22 Â log likelihood; AIC, Akaike information criterion; AUC, area under the curve; BAV, between-animal variability; BBB, blood–brain barrier; CL, clearance; DAT, dopamine reuptake transporter; ECF, extracellular fluid; HPLC, high-performance liquid chromatography;

Kp,uu, concentration ratios of brain ECF to plasma; MS/MS, tandem mass spectrometry; NET, norepinephrine reuptake transporter; OFV, objective function value; PD, pharmacodynamics; PET, positron emission tomography; PK, pharmacokinetics; SPECT, single photon emission computed tomography; SR, extended-release; Vb, brain distribution volume. 624 Bupropion and Hydroxybupropion Brain Pharmacokinetics 625 which is in complete reversal to the already-noted relative exposures 2181; ABDiets, Woerden, The Netherlands) and domestic-quality mains water observed in humans. Thus, superimposed upon remaining uncertainty were available ad libitum. regarding bupropion pharmacodynamics, either directly via inhibition Surgery for implantation of the microdialysis guide cannula was conducted of transporter-mediated reuptake of dopamine and norepinephrine or by under isoflurane anesthesia (2% with 400 ml/min N2O and 400 ml/min O2), indirect effects on these two neurotransmitter systems (Cooper et al., using bupivacaine/epinephrine for local analgesia and finadyne for perioperative/postoperative analgesia. One guide cannula was inserted into the 1994; Dong and Blier, 2001), or via effects on nicotinic acetylcholine medial prefrontal cortex to achieve the following probe tip coordinates: receptors (Slemmer et al., 2000; Damaj et al., 2004), the different anteroposterior, +3.4 mm from bregma; lateral, –0.8 mm from midline; and metabolic disposition in rodents complicates the ability to extrapolate to ventral, –5.0 mm from dura. A second 4.2-cm indwelling cannula was inserted humans. Although application of brain imaging modalities, such as into the jugular vein. Guide cannulae were exteriorized through an incision at the positron emission tomography (PET) and single photon emission top of the head. Animals were allowed at least 2 days to recover from surgery. computed tomography (SPECT), has received considerable attention MetaQuant Ultra-Slow Flow microdialysis probes (regenerated cellulose (Meyer et al., 2002; Learned-Coughlin et al., 2003; Árgyelán et al., membrane, 4-mm open membrane surface; BrainLink, Groningen, The Nether- 2005), these approaches cannot distinguish the contributions of lands) were inserted 1 day before the experiments. bupropion and hydroxybupropion to measured receptor occupancy. Drug Administration and Sample Collection. On the day of an experiment, Previous studies have shown that a translational pharmacokinetics probes were connected with flexible PEEK tubing to a CMA 102 microdialysis pump (CMA Microdialysis, Solna, Sweden) and perfused with a filtered (PK)/pharmacodynamics (PD) approach can be of great value to Ringer’s buffer containing 147 mM NaCl, 3.0 mM KCl, 1.2 mM CaCl , and increase understanding of the pharmacology of central nervous system 2 1.2 mM MgCl2 at a flow rate of 0.12 ml/min (CMA 142 pump). The slow flow Downloaded from drugs, particularly those with multiple targets and/or with active design of this probe maximizes sample recovery (Cremers et al., 2009), which metabolites. This approach describes plasma-to-brain transfer [blood– was 91% for bupropion and 100% for hydroxybupropion, as determined from in brain barrier (BBB) transport], using compartmental or physiologically vitro recovery experiments. Ultrapurified water was perfused through the based pharmacokinetic modeling in animals, with subsequent coupling dilution inlet of the probe at a flow rate of 0.8 ml/min; thus, the total dialysate of measured human systemic PK with the human brain disposition flow rate was 0.92 ml/min. After initiating flow, probes were allowed to stabilize derived from weight-based allometric scaling of the corresponding for 2 hours prior to administration of a single subcutaneous dose of bupropion animal parameters (de Lange, 2013). The method has been used to (10 mg/kg, n = 7 with partial overlap for each matrix) or hydroxybupropion dmd.aspetjournals.org predict atomoxetine and duloxetine exposure in human brain extracel- (2 mg/kg, n = 4), both dissolved in 0.9% NaCl at 2 mg/ml. Dialysates from the brain and jugular vein were collected every 30 minutes starting 1 hour prior to lular fluid (ECF) (Kielbasa and Stratford, 2012), risperidone and its administration and continuing for 360 minutes after drug administration. metabolite (9-OH-resperidone) receptor occupancy (Kozielska et al., Dialysate samples were stored at 280C until time of analysis. 2012), olanzapine receptor occupancy (Johnson et al., 2011), and Sample Analysis. Concentrations of compounds were measured in dialysate receptor occupancy of clozapine and its active metabolite (N-desme- collected from plasma and brain ECF probes using HPLC with MS/MS thylclozapine) at multiple receptors, based on brain ECF measures detection. An aliquot of 12 ml was taken from each MetaQuant dialysate sample using microdialysis (Li et al., 2014). In these examples, predicted (27.6 ml total volume) and mixed with 4 ml internal standard solution at ASPET Journals on September 25, 2021 exposure and/or occupancy in the human brain after subtherapeutic or (fenfluramine). Of this mixture, 10 ml was injected into the HPLC system by therapeutic doses was accordingly congruent with in vitro receptor an automated sample injector (SIL-20AD; Shimadzu, Kyoto, Japan). Chromato- potency (Kielbasa and Stratford, 2012) or with receptor occupancy graphic separation of the compounds was performed on a reversed phase column  m measured by PET imaging (Johnson et al., 2011; Kozielska et al., 2012; (100 3.0 mm, 2.5 m particle size; Phenomenex, Torrance, CA) held at a Li et al., 2014). Thus, the objective of our study was to apply this temperature of 35 C in a gradient elution run, using eluent B (acetonitrile) in eluent A (10 mM ammonium formate in ultrapurified water, pH 4) at a flow rate translational approach to predict bupropion and hydroxybupropion of 0.3 ml/min. The gradient profile was as follows: 0% B from 0 to 4 minutes, exposure in human brain ECF, the pharmacologically relevant biophase 40%–100% B from 4 to 5.5 minutes, remaining at 100% B until 6 minutes, and for the purported parenchymal membrane targets. To this end, single returning to 0% B from 6 to 6.5 minutes. The mass spectrometry analyses were doses of bupropion and its preformed metabolite were administered performed using an API 4000 MS/MS system consisting of an API 4000 MS/MS to two groups of rats, with subsequent application of population detector and a Turbo Ion Spray interface (both from Applied Biosystems, Foster compartmental modeling of measured unbound concentrations from City, CA). The acquisitions were performed in positive ionization mode, with quantitative microdialysis sampling of both plasma and brain ECF. ionization spray voltage set at 5.5 kV and a probe temperature of 500C. The Structural parameters describing brain ECF disposition were then instrument was operated in multiple reaction monitoring mode. Multiple reaction scaled to predict the time course of both entities in human brain ECF. monitoring transitions were 239.9 to 183.9 for bupropion and 256.0 to 166.0 for hydroxybupropion. Suitable in-run calibration curves were fitted using weighted (1/x) regression, and the sample concentrations were determined using these Materials and Methods calibration curves. Standard concentrations ranged from 0.05 to 50.0 nM. Accuracy Drugs and Chemicals. Bupropion (racemic) and (2S,3S)-hydroxybupropion was verified by quality-control samples after each sample series. Concentrations hydrochloride were purchased from Sigma-Aldrich (St. Louis, MO) and were were calculated with the Analyst data system (Applied Biosystems). used as received. Formulations for administration were prepared on the day of an Pharmacokinetic Analysis. All pharmacokinetic analyses were conducted experiment. Chemicals used in the preparation of microdialysis perfusion buffer with Phoenix NLME 1.3 (Pharsight Corporation, Certara, L.P., Princeton, NJ). A and solvents used for high-performance liquid chromatography (HPLC)–tandem population modeling approach was used for both bupropion and hydroxybupropion mass spectrometry (MS/MS) analysis were of reagent grade. using first-order conditional estimation. Initial models in plasma were built for Animal Preparation. Adult male Sprague-Dawley rats (280–350 g; Harlan, each compound based on their corresponding dose group (bupropion and Horst, The Netherlands) were used for the experiments. The experiments were preformed hydroxybupropion). Different model structures were evaluated for conducted in strict accordance with the National Institutes of Health Guide for each dose group. These included one- and two-compartment disposition and the Care and Use of Laboratory Animals, were in accordance with Dutch law, zero-order versus first-order absorption, with the latter including with versus and were approved by the Animal Care and Use Committee of the University of without a lag time. Model evaluation was based on the likelihood-ratio test, Groningen (Groningen, The Netherlands). After arrival, animals were housed in which uses the objective function value (OFV). The minimum OFV returned for groups of five in polypropylene cages (40  50  20 cm) with a wire mesh top in a model is approximately equal to 22  log likelihood (22LL). Reduction of a temperature-controlled (22 6 2C) and humidity-controlled (55% 6 15%) 22LL . 6.63 for the addition of one parameter was taken as a significant model environment on a 12-hour light cycle (07.00–19.00). After surgery, animals were improvement (P , 0.01) for nested models. The Akaike information criterion housed individually (cages 30  30  30 cm). Standard diet (Diets, RMH-B (AIC), which is calculated as AIC = 22LL + 2  NP, where NP is the number of 626 Cremers et al. parameters, was used to compare models differing by more than one parameter. Results After optimization of the plasma model unique to each dose group, brain ECF Subcutaneous administration of a 10-mg/kg dose of bupropion concentrations were incorporated as a peripheral compartment. Transfer characteristics between the plasma and brain were modeled using either a single resulted in both parent drug and formed metabolite exposure in plasma intercompartmental clearance (Q) or separate parameters for the uptake and and brain ECF (Table 1). The hydroxybupropion area under the –‘ 6 efflux apparent distributional clearances (CLin and CLout, respectively). The curve (AUC0 ) as a percentage of bupropion exposure was 6% 6 unbound volume of distribution of bupropion in the brain (Vb,B) was tested with 0.6% (n = 6) in plasma and 4% 0.3% in brain ECF (n = 6). The versus without fixing this parameter to a literature-reported value of 16 ml/g relative exposure in plasma agrees reasonably well with the 13% brain (Fridén et al., 2011). For hydroxybupropion, a computational approach was reported after a 40-mg/kg intraperitoneal dose to rats (Suckow et al., used to estimate its volume of distribution in the brain (Vb,HB) (Spreafico and 1986). In addition, a recent microdialysis study (Yeniceli et al., 2011) Jacobson, 2013). Based on this approach, a value of 8 ml/g brain was used when measured comparable AUCs for bupropion (within 2-fold) and this parameter was fixed. In addition to minimization of OFV, the combined hydroxybupropion (within 3-fold) in brain ECF after a 10-mg/kg plasma-brain ECF model selection also included stability of the plasma model– specific estimates upon incorporation of the brain ECF compartment. intraperitoneal dose to rats. In our study, for both plasma and brain In parallel with development of the plasma-brain models for each compound, matrices, a decline of the metabolite proceeded in parallel with that of a combined plasma model that incorporated hydroxybupropion concentrations bupropion; in addition, the terminal half-life in brain ECF was similar to observed after both bupropion administration (formed metabolite) and hydroxy- the plasma for each compound (Table 1). With respect to preformed bupropion administration (preformed metabolite) was developed. Linkage of hydroxybupropion administration (2 mg/kg), its terminal half-life in the two separate plasma models was made through the estimated parameter, both plasma and brain ECF was approximately 50% of its half-life Downloaded from bupropion-to-hydroxybupropion formation clearance (CLf). In addition to this observed after bupropion administration. postabsorptive conversion, presystemic metabolism of bupropion was evaluated. The ratio of brain ECF concentration to unbound plasma concentra- The final structural model combined this unified plasma model with the separate tion (K ) was on average above 1 at all times for bupropion and plasma-to-brain transfer models for the two compounds. In total, there were 11 p,uu preformed hydroxybupropion (Fig. 1). AUC ratios, shown in Table 1, rats and 367 observations in this final model. The possibility of within-brain metabolism was also investigated in this final model. At each model building indicate a 2.1-fold asymmetry for bupropion and slightly lower values stage, between-animal variability (BAV) was estimated for each pharmacoki- for both preformed and formed hydroxybupropion. dmd.aspetjournals.org netic parameter by assuming a log-normal distribution based on the exponential Compartmental model structure definition for bupropion in plasma th relationship, Pi =Ptv  exp(hi), where Pi is the parameter estimate for the i and brain ECF indicated that a one-compartment model provided the h animal, Ptv is the population typical value, and i is the deviation from the best description of concentration time course in both matrices, in th population value for the i subject. Various residual error models were also agreement with visual inspection of monoexponential decline in evaluated. These were additive, proportional, and mixed additive-proportional individual animals (Fig. 2). The same was true for hydroxybupropion models. Residual error estimates of intra-animal variability between observed in both matrices, regardless of its disposition as formed or preformed and predicted concentrations were assumed to be normally distributed with a 2 metabolite. Estimation of bupropion and of preformed metabolite s at ASPET Journals on September 25, 2021 variance of . A proportional error model was eventually selected in the two . matrices for both compounds. bioavailability resulted in population estimates 95%. However, since Final plasma-specific and plasma-brain ECF models were evaluated based on precision of these estimates, as well as of the corresponding clearance visual inspection of conditional weighted residual versus either population and volume of distribution (VD) estimates was poor, bioavailability was predicted or time after dose plots as well as the observed versus individual fixed to 0.95 for bupropion to account for a small first-pass effect and to predicted and population predicted concentration plots. A visual predictive check 1 for preformed hydroxybupropion. With respect to the description of was also conducted, and parameter uncertainty was evaluated using a non- plasma-to-brain ECF transfer, for bupropion and for its metabolite, parametric bootstrap based on resampling 200 times from the original data set. estimation of CLin and CLout provided clearly superior results Human Exposure Simulations. The final plasma-brain ECF model was (minimization of OFV) relative to modeling this transfer as a single scaled to predict steady-state human brain ECF exposure to bupropion and intercompartmental clearance. Importantly, an attempt was made to hydroxybupropion. A 150-mg dose of the extended-release (SR) bupropion product administered twice daily was selected based on studies of bupropion PK estimate the unbound brain volume of distribution for both bupropion using this formulation and dose frequency (Johnston et al., 2001; Learned- (Vb,B) and hydroxybupropion (Vb,HB); however, it was necessary to fix Coughlin et al., 2003; Benowitz et al., 2013). Bupropion pharmacokinetic this parameter for both compounds to get reliable estimates of the parameters used for the simulations were based on these studies and corrected for distributional clearances. plasma protein binding of 85% (Jefferson et al., 2005). These were 17.5 l/min for The initial approach to combine the plasma disposition of bupropion CL/F and 12,400 liters for VD/F. An absorption rate constant (ka) of 0.01/min, and hydroxybupropion after their subcutaneous administration was to which yielded a Tmax of 3 hours, the reported value for the SR formulation fix the parameter estimates obtained from modeling the two adminis- (Jefferson et al., 2005), was used. Pharmacokinetic parameters for formed tration groups separately to define the formation clearance for the hydroxybupropion were adjusted to provide an average unbound steady-state formed hydroxybupropion. This approach led to a formation clearance concentration of 500 nM. This concentration was between the reported average concentrations from the two multiple dose studies after correcting for of approximately 3.5 ml/min. Subsequent population modeling of hydroxybupropion plasma protein binding of 77% (Johnston et al., 2002). Both plasma data simultaneously from both groups of animals, using this bupropion and hydroxybupropion pharmacokinetics are linear after long-term estimate and without fixing the parameters for the two compounds, bupropion administration of 300–400 mg/d (Jefferson et al., 2005). Weight- resulted in model nonconvergence. Therefore, various plasma model based allometric scaling of the plasma-brain ECF distributional clearance (CL) structures were investigated. These included 1) the possibility that and brain volumes was used to derive the corresponding human brain bupropion and hydroxybupropion had different bioavailability, 2) that parameters, using a rat brain weight of 2.45 g (1.8 g/250 g body weight; Davies bupropion was altering the clearance or VD of the formed metabolite, 3) and Morris, 1993) and a human brain weight of 1.35 kg. The exponential factors a model that included a presystemic component to bupropion metab- were 0.75 for CL and 1.0 for brain distribution volume (V ) (Sharma and b olism, and 4) a model that incorporated clearance of formed hydroxy- McNeill, 2009). A total of 1000 simulations were conducted. Data Presentation. Rat plasma and brain ECF matrix concentration data are bupropion prior to it reaching the systemic circulation (the so-called presented as unbound concentrations, and the corresponding time points are “sequential first-pass elimination of the formed metabolite” model; the midpoint of a given 30-minute collection interval. All pharmacokinetic Pang and Gillette, 1979). Neither different bioavailability nor the parameters are presented as the unbound estimates. various clearance models could describe the observed alteration in Bupropion and Hydroxybupropion Brain Pharmacokinetics 627

TABLE 1 Bupropion and hydroxybupropion observational pharmacokinetic parameters in plasma and brain ECF after a subcutaneous 10-mg/kg dose of bupropion or a 2-mg/kg dose of hydroxybupropion by the same route

Results for AUC, Cmax, and AUCECF/AUCplasma,u ratio are presented as means (S.E.M.). Results for half-lives are presented as the geometric mean. There were a total of six rats in the bupropion administration group (five for plasma and five for brain), and four rats in the hydroxybupropion administration group.

Plasma Brain ECF AUC / Drug ECF AUCplasma,u AUC0–‘ Cmax Half-Life AUC0–‘ Cmax Half-Life

nM·min nM min nM·min nM min Bupropion 46,205 (5768.6) 198 (27.1) 132 98,616 (8552.2) 607 (74.3) 109 2.1 (0.18) Hydroxybupropion (formed) 2637 (309.8) 15 (1.8) 101 4249 (225.7) 23 (1.3) 94 1.8 (0.10) Hydroxybupropion (preformed) 46,026 (10,737.6) 496 (145.6) 50 56,683 (5779.6) 492 (70.8) 45 1.5 (0.21) formed metabolite disposition relative to the preformed metabolite. A (Jefferson et al., 2005), and the minimum concentration was 25 nM. model that invoked altered hydroxybupropion VD in the presence of These concentrations agree well with the reported maximum and bupropion was also unable to describe formed metabolite disposition. minimum concentrations of 59 nM and 16 nM, respectively (Johnston Ultimately, structure 3 above, a model that incorporated a presystemic et al., 2001), when corrected for bupropion’s plasma protein binding of Downloaded from (first-pass) component of bupropion conversion to hydroxybupropion, 85% (Jefferson et al., 2005). Median predicted hydroxybupropion was selected as the final plasma model. Table 2 provides a summary of plasma concentrations varied little over the 12-hour time course (from the final population model parameter estimates, as well as estimates of 729 nM at the end of the dose interval to a peak of 798 nM 6 hours BAV for several of the parameters, and residual error. Also shown are after administration) and agree well with reported steady-state the results of a nonparametric bootstrap to evaluate parameter stability hydroxybupropion plasma concentrations (Johnston et al., 2001; in the plasma model. Figure 3 summarizes individual and population Learned-Coughlin et al., 2003) when corrected for hydroxybupropion’s predicted concentrations versus the observed concentrations for both plasma protein binding of 77% (Johnston et al., 2002). A comparison dmd.aspetjournals.org bupropion and hydroxybupropion in plasma, as well as the conditional with reported IC50 values for DAT and NET (Damaj et al., 2004; weighted residuals distributions relative to both time and the population Lukas et al., 2010) shows that unbound brain concentrations of predicted concentrations. hydroxybupropion after twice-daily administration of 150 mg bupropion The final plasma-brain ECF model, which was based on fixed plasma SR exceed its IC50 values for DAT (630 nM) and NET (241 nM) over parameter estimates with simultaneous analysis of brain ECF concen- the entire time course, whereas bupropion concentrations were sub- trations from both administration arms, is shown in Fig. 2. The stantially lower than reported IC50s for the two transporters (660 nM assumption made in fixing the plasma estimates to derive the brain and 1850 nM, respectively) at all times. at ASPET Journals on September 25, 2021 ECF estimates is that brain kinetics do not influence plasma kinetics. This assumption is supported by the much larger systemic volumes of distribution relative to those in the brain, as well as the two administra- Discussion tion arm plasma-brain ECF–specific models that showed stability of Incorporation of a preformed hydroxybupropion administration the plasma estimates upon addition of the brain ECF component. The group into the study design revealed that disposition of the formed final model also incorporated a time-dependent component to CLin for metabolite was rate-limited by bupropion kinetics in both plasma and bupropion (CLinb), expressed as CLinb =CLinb0 2 slp  time, where brain ECF. Coupled with the observation that preformed metabolite CLinb decreases with time (slope) up to 195 minutes and is constant half-life in brain ECF was similar to that observed in plasma, these thereafter, and CLinb =CLinb0 when time = 0. A similar approach was findings indicate that metabolite uptake and egress across the BBB and recently used to describe time dependence in citalopram absorption in its intrabrain distribution were faster than metabolite systemic dispo- rats (Velez de Mendizabal et al., 2015) and was conceived in our case sition, both with respect to its formation and subsequent elimination. based on the observation in four of five rats of an overshoot in Kp,uu in Parenthetically, in humans, given that hydroxybupropion and bupropion the first half of the time course relative to the eventual Kp,uu,ss (Fig. 1). have a similar half-life after bupropion administration (Sweet Table 3 provides a summary of population parameter estimates as well et al., 1995; Hsyu et al., 1997), it is possible that hydroxybupropion as estimates of BAV for the various plasma-to-brain ECF distributional disposition, at least in plasma, is also rate-limited by bupropion. In our clearances and proportional residual error. rat study, comparing systemic AUCs of formed metabolite to bupropion Figure 4 summarizes individual and population predicted concen- and correcting for differences in clearance between bupropion and that trations versus the observed concentrations for both bupropion and of the preformed metabolite (Rowland and Tozer, 2011), the fraction of hydroxybupropion in brain ECF, as well as the conditional weighted bupropion metabolized to hydroxybupropion was 1.1%. In the compart- residual distributions relative to both time and the brain ECF population mental model, the ratio of formation clearance to total clearance of predicted concentrations. Figure 5 summarizes the results of the visual bupropion was 0.4%. The difference is attributed to a small presystemic predictive check for both compounds in plasma and brain ECF. Results clearance component prior to bupropion systemic absorption. Albeit from an evaluation of parameter precision in the final combined plasma small, this first-pass effect is surprising given that bupropion was and brain ECF model based on nonparametric bootstrap analysis are administered subcutaneously. In response to nonconvergence of the summarized in Table 3. combined formed and preformed metabolite plasma model that did not Figure 6 presents the predicted steady-state exposures in human include this first-pass effect, alternative explanations were investigated: plasma and brain ECF from 0 to 12 hours for bupropion and 1) bupropion-mediated alteration of formed metabolite disposition hydroxybupropion after twice-daily administration of the 150-mg (altered clearance or VD) or 2) different bioavailability for the two dose extended-release bupropion SR tablet. The median maximum bupropion groups. These alternative structures were found unable to generate a plasma concentration was 59 nM, occurring at 3 hours, which is model that included the two dose groups. In addition, the possibilities that the commonly observed peak time for the bupropion SR product hydroxybupropion probe recovery was enhanced in the presence of 628 Cremers et al. Downloaded from dmd.aspetjournals.org at ASPET Journals on September 25, 2021

Fig. 1. Bupropion and hydroxybupropion unbound concentration ratios of brain ECF to plasma (Kp,uu) in individual rats. The top panel presents bupropion ratios after a 10-mg/kg subcutaneous dose of bupropion, whereas the middle panel presents hydroxybupropion ratios observed in the same group of rats (formed metabolite case). The bottom panel presents hydroxybupropion ratios after a 2-mg/kg dose of the metabolite in a different group of rats by the same administration route (preformed metabolite case). The x-axis represents a given animal, with each point representing a sample time arranged in ascending order from left to right starting at 15 minutes and ending at 345 minutes in 30-minute increments. The solid horizontal line in each panel represents the plasma-to-brain equilibration model–predicted equilibrium Kp,uu value reported in Table 2. Bupropion and Hydroxybupropion Brain Pharmacokinetics 629

Fig. 2. Compartmental pharmacokinetic model used to describe bupropion, and its formed and preformed metabolite absorption and disposition in rats. Terms with a subscript containing B represent bupropion, whereas those with HB, hydroxybupropion. CbECF, unbound brain ECF concentration; Cu,p, unbound plasma concentration; CL/F, apparent elimination clearance; CLf, apparent formation clearance of bupropion to hydroxybupropion; CLin, uptake apparent distributional clearance; CLout, efflux

apparent distributional clearance; ka represents the Downloaded from first-order absorption rate constant; kf, first-order rate constant for presystemic conversion of bupropion to hydroxybupropion; Tlag, lag time for bupropion absorption; Vb, apparent brain volume of distribu- tion; Vp, apparent systemic volume of distribution. dmd.aspetjournals.org at ASPET Journals on September 25, 2021 bupropion or of bias in the analysis of the metabolite in the presence of brain equilibration half-life for each compound (based on the equation bupropion were considered and discounted. At this point, the cause of the t1/2,equil =Vb × ln2/CLout; Liu et al., 2005) was only 1.4 minutes and 4.6 presumed presystemic component is not understood. minutes for bupropion and hydroxybuprion, respectively. As shown in The observation that bupropion brain kinetics paralleled those in Fig. 1, in four of the five animals receiving bupropion, there was a plasma indicates that, like hydroxybupropion, bupropion disposition in distinct peak in Kp,uu for bupropion occurring by the second collection brain was faster than its overall (systemic) disposition in the rat. In fact, interval. A significant improvement in model performance (AIC and

TABLE 2 Plasma bupropion and hydroxybupropion population pharmacokinetic parameter estimates Data are given as percentages (percent relative standard error of the estimate) and medians (5th to 95th percentiles). BAV is calculated as the square root of the variance  100.

Bootstrap Analysis Parameter Estimate (RSE%) BAV (RSE%) Estimate BAV (RSE%)

%% Bupropion CLB/F (ml/min) 316 (10.6) 313 (270–366) Vp,B/F (ml) 56244 (19.8) 9.4 (3.9) 55,836 (43,053–79,639) 8.8 (6.4) kaB (1/min) 0.0355 (14.8) 6.1 (4.5) 0.0357 (0.0285–0.0485) 4.3 (5.5) Tlag (min) 11.8 (2.3) 11.8 (11.4–12.4) CLf/F (ml/min) 1.4 (29.9) 19.0 (9.9) 1.4 (0.7–2.5) 12.9 (10.3) kfp (1/min) 0.00027 (26.6) 18.4 (10.1) 0.00027 (0.00017–0.00045) 11.9 (12.6) FB 0.95 (fixed) 6.5 (1.7) 4.0 (2.9) Hydroxybupropion CLHB/F (ml/min) 67.3 (17.2) 67.4 (46.7–93.1) VP, HB/F (ml) 5329 (16.9) 5288 (3867–6989) kaHB (1/min) 0.1 (fixed) FHB 1 (fixed) 11.8 (3.4) 7.0 (5.3) Residual error (proportional) Bupropion-plasma 0.107 (13.7) 0.107 (0.083–0.132) Hydroxybupropion-plasma 0.159 (10.9) 0.160 (0.132–0.185)

B, bupropion; CL/F, apparent elimination clearance for bupropion or hydroxybupropion; CLf/F, hydroxybupropion formation clearance after bupropion administration; F, bioavailability fraction; HB, hydroxybupropion; ka, first-order rate constant for absorption; kfp, first order rate constant for presystemic metabolism of bupropion to hydroxybupropion; RSE, relative standard error of the estimate; Tlag, lag time for bupropion absorption; Vp, apparent systemic volume of distribution. 630 Cremers et al. Downloaded from

Fig. 3. Model diagnostic plots for plasma. (A and B) Individual predicted concentrations versus observed concentrations for bupropion and hydroxybupropion, respectively. The solid line is the line of unity. (C and D) Population predicted concentrations versus observed concentrations for bupropion and hydroxybupropion, respectively. (E and F) dmd.aspetjournals.org Conditional weighted residuals versus time for bupropion and hydroxybupropion, respectively. (G and H) Conditional weighted residuals versus population predicted concentrations for bupropion and hydroxybupropion, respectively. CWRES, conditional weighted residual; DV, concentration; IPRED, individual predicted; PRED, population predicted. goodness-of-fit plots) was achieved by incorporating a time depen- sensitive to intersubject variability caused by disease or genetic dency in CLin,B, with this decreasing over time. In the final population variability as well as drug–drug and drug–metabolite interactions. model, the ratio of CLin/CLout was 1.9 and 1.7 (relative standard error of The experimentally obtained pKa of bupropion has been estimated at the estimate = 5.6% and 6.1%) for bupropion and hydroxybupropion, 7.9 (Gondaliya and Pundarikakshudu, 2003) and 8.6 (Fridén et al., at ASPET Journals on September 25, 2021 respectively, both in good agreement with the AUC ratio estimates in 2011). Taking the average of these two values, and assuming that brain Table 1. The model did not include brain ECF bulk flow, since the ECF is 0.1 log unit more acidic than plasma (pH 7.3 versus pH 7.4; estimated value of 0.18 to 0.29 ml/min × g brain (Szentistványi et al., Fridén et al., 2011), results in a predicted diffusional equilibrium 1984) is well below 1% of the CLout estimates and thus is a negligible preference on the brain side of 1.01, which supports that the CLin/CLout contributing factor to brain clearance. The utility of modeling plasma ratios of approximately 2 are due to an uptake transporter for bupropion and brain distribution across the BBB is that this approach converts an and hydroxybupropion at the BBB. The net uptake asymmetry observation (Kp,uu) into parameters (CLin and CLout) that correlate with observed is similar to oxycodone BBB transport in rats, (Boström physiologic carrier-mediated processes that are potentially scalable et al., 2006), which showed a net 3-fold asymmetry favoring rat brain between species (Qiu et al., 2014), and whose function could be ECF relative to unbound plasma concentrations. Subsequent in vitro

TABLE 3 Brain ECF bupropion and hydroxybupropion population pharmacokinetic parameter estimates Data are given as percentages (percent relative standard error of the estimate) and medians (5th to 95th percentiles). BAV is calculated as the square root of the variance  100.

Bootstrap Analysis Parameter Estimate (RSE%) BAV (RSE%) Estimate BAV (RSE%)

%% Bupropion CLin,B0/F (ml/min) 18.3 (11.2) 3.4 (1.9) 18.6 (16.9–43.8) 9.7 (14.4) CLout,B/F (ml/min) 5.4 (16.5) 5.6 (0.3) 5.5 (4.4–16.2) 14.2 (19.0) Vb,B/F (ml) 10.67 (fixed) Slp 0.04 (fixed) Hydroxybupropion CLin,HB/F (ml/min) 1.3 (30.1) 132.9 (24.5) 1.4 (0.9–1.9) 128.6 (40.2) CLout,HB/F (ml/min) 0.8 (26.9) 99.1 (15.9) 0.8 (0.5–1.2) 94.3 (31.0) Vb,HB/F (ml) 5.33 (fixed) Residual error (proportional) Bupropion-brain 0.165 (8.8) 0.166 (0.143–0.188) Hydroxybupropion-brain 0.163 (6.3) 0.163 (0.146–0.178)

B, bupropion; CLin,B0/F, apparent influx clearance for bupropion at initial time; CLin,HB, apparent influx clearance for hydroxybupropion; CLout, apparent efflux clearance; HB, hydroxybupropion; RSE, relative standard error of the estimate; slp, slope relating time-dependent change in CLin B;Vb, apparent brain volume of distribution. Bupropion and Hydroxybupropion Brain Pharmacokinetics 631 Downloaded from

Fig. 4. Model diagnostic plots for brain ECF. (A and B) Individual predicted concentrations versus observed concentrations for bupropion and hydroxybupropion, respectively. The solid line is the line of unity. (C and D) Population predicted concentrations versus observed concentrations for bupropion and hydroxybupropion, respectively. (E and F) Conditional weighted residuals versus time for bupropion and hydroxybupropion, respectively. (G and H) Conditional weighted residuals versus population predicted concentrations for bupropion and hydroxybupropion, respectively. DV, concentration; IPRED, individual predicted; PRED, population predicted; WRES, weighted residual. dmd.aspetjournals.org studies with oxycodone (Okura et al., 2008)—as well as more recent in replacement of protons by bupropion and/or hydroxybupropion on the vitro transport or in situ brain perfusion studies with several other low abluminal side. This may have resulted in a decrease in CLin over time molecular weight weakly basic drugs, such as apomorphine (Okura due to reduced function, or just to loss of the proton gradient over time, et al., 2014), clonidine (André et al., 2009), codeine (Fischer et al., much like occurs with proton-coupled dipeptide transport in the small 2010), diphenhydramine (Shimomura et al., 2013), and nicotine intestine (Ganapathy and Leibach, 1983). Interestingly, a similar time- (Cisternino et al., 2013)—implicate a pH-dependent proton-coupled dependent Kp,uu was observed with oxycodone (Boström et al., 2006). It antiporter at the BBB (Gharavi et al., 2015). It is possible that this is possible that the overshoot observed with bupropion is due to an at ASPET Journals on September 25, 2021 transporter may also apply to bupropion and hydroxybupropion. experimental artifact (namely, a decline in probe recovery of bupropion Moreover, the proton-coupled antiporter has been suggested to be after administration and lasting for 2 to 3 hours, at which point it capable of bidirectional transport (Cisternino et al., 2013). It is possible reached a constant value). As is standard practice in microdialysis that the overshoot in bupropion Kp,uu, which was modeled as a time- research, probes in this study were inserted 24 hours before an dependent distributional clearance, may have been due to eventual experiment and perfused for 2 hours prior to drug administration to

Fig. 5. Visual predictive checks for plasma (top) and brain ECF (bottom). The left plot in each panel represents bupropion after a 10-mg/kg dose bupropion, the middle plot represents formed hydroxybupropion after the 10-mg/kg bupropion dose, and the right plot in each panel represents preformed hydroxybupropion after a 2-mg/kg dose of the metabolite. The solid line in each plot represents the median of the observed concentrations, the dashed line represents the median predicted concentrations, and the two dotted lines represent the 5% and 95% limits of the predicted 90% confidence intervals of the median. Individual observed concentrations are shown as the open circles. 632 Cremers et al. Downloaded from dmd.aspetjournals.org

Fig. 6. Simulated bupropion (left) and hydroxybupropion (right) concentrations over 12 hours in humans after multiple every-12-hour daily dosing of 150 mg of the bupropion SR formulation. The top panel represents plasma unbound concentrations, and the bottom panel represents brain ECF concentrations. The solid line describes the median predicted concentration, whereas the gray shaded lines represent the 5% and 95% predicted limits of the 90% confidence interval of the median concentrations. The dashed line in the brain ECF plots refers to the human IC50 value reported for the NET, and the dotted line refers to the IC50 for the DAT. at ASPET Journals on September 25, 2021 reduce the possibility for such an artifact. At a minimum, our for NET. Importantly, our predictions of the much larger exposure to observations point to additional studies to develop a better mechanistic the metabolite in brain ECF agree with its reported 7-fold higher understanding of bupropion and hydroxybupropion transport across the concentrations than bupropion in human CSF (Cooper, et al., 1994). BBB. Receptor occupancy studies using PET (Meyer et al., 2002; Learned- Bupropion’s effects in rodent models of depression have been linked Coughlin et al., 2003) or SPECT (Argyelán et al., 2005) tracers specific to its ability to increase dopamine and norepinephrine concentrations in for DAT, and after attainment of steady-state kinetics based on several brain ECF, as determined using microdialysis sampling (Nomikos et al., days of dosing 150 mg of the SR formulation twice daily, indicate 1989, 1992). More recently, Li et al. (2002) evaluated the effect of a modest occupancy of around 20% in both patients with and without 10-mg/kg bupropion subcutaneous dose on dopamine and norepi- depression. In one of these studies (Learned-Coughlin et al., 2003), nephrine levels in the ECF of medial prefrontal cortex of rats, thus DAT occupancy was sustained for at least 24 hours. Our predicted employing the same dose, route, and microdialysis probe placement as sustained exposure to hydroxybupropion (Fig. 6) agrees with the used in our study. In the study by Li et al. (2002), the increase in reported sustained occupancy at this transporter and suggests that it was dopamine peaked at 262% over baseline 90 minutes after administra- due to the metabolite. However, the low occupancy reported in these tion. A dopamine increase at that time is in accord with the median peak studies is below that which our predictions would suggest (approxi- bupropion concentration of 504 nM observed during the 60- to mately 50%) based on the simulated concentrations being similar to the 90-minute collection interval (Fig. 5), and its proximity to its reported reported IC50 for hydroxybupropion at this transporter (630 nM; Lukas IC50 at DAT of 550–660 nM (Damaj et al., 2004; Lukas et al., 2010). et al., 2010). This difference identifies the need for studies to investigate Although hydroxybupropion also possesses inhibitory activity at DAT the mechanism of bupropion and hydroxybupropion transport across in the same range (790 nM) in rats, its contribution to increasing blood–brain surrogate models and, if a transporter is involved, to apply extracellular dopamine would be expected to be minimal, given the mass spectrometry–based proteomics to support interspecies scaling median concentration of 22 nM observed. By contrast, hydroxybupropion (Qiu et al., 2014). The predicted exposures for bupropion and exposure in plasma in humans is approximately an order of magnitude hydroxybupropion, together with their IC50 values for NET, indicate greater than bupropion (Johnston et al., 2001; Learned-Coughlin et al., that, if norepinephrine levels are increased in humans via NET 2003; Benowitz et al., 2013). Translation of the rat bupropion and inhibition, this would largely be attributed to hydroxybupropion, and hydroxybupropion plasma-brain distributional clearances to humans not bupropion. suggests that monoamine contributions to bupropion’s antidepressant In conclusion, a pharmacokinetic model that predicts bupropion and effects via monoamine reuptake inhibition are due to hydroxybupropion. hydroxybupropion concentrations in brain ECF of rats has been Median hydroxybupropion brain ECF concentrations are similar to its developed. The model asserts a carrier-mediated mechanism contrib- DAT IC50 and are approximately 3-fold above its NET IC50 (Fig. 6), utes to the plasma to brain transfer of both compounds. Similar to what whereas peak bupropion brain ECF concentrations are approximately has been done with atomoxetine and duloxetine (Kielbasa and Stratford, 7-fold below the DAT IC50 and are even more so relative to that 2012) and clozapine and its active metabolite, N-desmethylclozapine Bupropion and Hydroxybupropion Brain Pharmacokinetics 633

(Li et al., 2014), to translate rat BBB kinetics to humans to predict Johnson M, Kozielska M, Pilla Reddy V, Vermeulen A, Li C, Grimwood S, de Greef R, Groothuis GMM, Danhof M, and Proost JH (2011) Mechanism-based pharmacokinetic- human brain exposure, our approach predicts that the dopamine pharmacodynamic modeling of the dopamine D2 receptor occupancy of olanzapine in rats. transporter occupancy observed in humans using PET and SPECT is Pharm Res 28:2490–2504. Johnston AJ, Ascher J, Leadbetter R, Schmith VD, Patel DK, Durcan M, and Bentley B (2002) due to hydroxybupropion and that the metabolite is responsible for a Pharmacokinetic optimisation of sustained-release bupropion for smoking cessation. Drugs 62 direct effect on increasing synaptic dopamine and norepinephrine via (Suppl 2):11–24. Johnston JA, Fiedler-Kelly J, Glover ED, Sachs DPL, Grasela TH, and DeVeaugh-Geiss J (2001) DAT and NET inhibition, respectively. Relationship between drug exposure and the efficacy and safety of bupropion sustained release for smoking cessation. Nicotine Tob Res 3:131–140. Kielbasa W and Stratford RE, Jr (2012) Exploratory translational modeling approach in drug Acknowledgments development to predict human brain pharmacokinetics and pharmacologically relevant clinical – The authors thank Robert Bies (University at Buffalo, State University of New doses. Drug Metab Dispos 40:877 883. Kozielska M, Johnson M, Pilla Reddy V, Vermeulen A, Li C, Grimwood S, de Greef R, York) and Serge Guzy (POP-PHARM) for collegial support during the modeling Groothuis GMM, Danhof M, and Proost JH (2012) Pharmacokinetic-pharmacodynamic analysis portion of this work. Erin Gorse is also thankfully acknowledged for modeling of the D₂ and 5-HT (2A) receptor occupancy of risperidone and paliperidone in rats. – support with table and figure preparation. Pharm Res 29:1932 1948. Learned-Coughlin SM, Bergström M, Savitcheva I, Ascher J, Schmith VD, and Långstrom B (2003) In vivo activity of bupropion at the human dopamine transporter as measured by Authorship Contributions positron emission tomography. Biol Psychiatry 54:800–805. Li CH, Stratford RE, Jr, Velez de Mendizabal N, Cremers TIFH, Pollock BG, Mulsant BH, Participated in research design: Cremers, Flik, Rollema, Stratford. Remington G, and Bies RR (2014) Prediction of brain clozapine and norclozapine concen- Conducted experiments: Flik, Folgering. trations in humans from a scaled pharmacokinetic model for rat brain and plasma pharmaco- Performed data analysis: Stratford. kinetics. J Transl Med 12:203. Li SXM, Perry KW, and Wong DT (2002) Influence of fluoxetine on the ability of bupropion to Downloaded from Wrote or contributed to the writing of the manuscript: Cremers, Flik, modulate extracellular dopamine and norepinephrine concentrations in three mesocorticolimbic Folgering, Rollema, Stratford. areas of rats. Neuropharmacology 42:181–190. Liu X, Smith BJ, Chen C, Callegari E, Becker SL, Chen X, Cianfrogna J, Doran AC, Doran SD, and Gibbs JP, et al. (2005) Use of a physiologically based pharmacokinetic model to study the References time to reach brain equilibrium: an experimental analysis of the role of blood-brain barrier permeability, plasma protein binding, and brain tissue binding. J Pharmacol Exp Ther 313: André P, Debray M, Scherrmann JM, and Cisternino S (2009) Clonidine transport at the mouse 1254–1262. + blood-brain barrier by a new H antiporter that interacts with addictive drugs. J Cereb Blood Lukas RJ, Muresan AZ, Damaj MI, Blough BE, Huang X, Navarro HA, Mascarella SW, Eaton Flow Metab 29:1293–1304. JB, Marxer-Miller SK, and Carroll FI (2010) Synthesis and characterization of in vitro and in dmd.aspetjournals.org Argyelán M, Szabó Z, Kanyó B, Tanács A, Kovács Z, Janka Z, and Pávics L (2005) Dopamine vivo profiles of hydroxybupropion analogues: aids to smoking cessation. J Med Chem 53: 99m transporter availability in medication free and in bupropion treated depression: a Tc- 4731–4748. TRODAT-1 SPECT study. J Affect Disord 89:115–123. Meyer JH, Goulding VS, Wilson AA, Hussey D, Christensen BK, and Houle S (2002) Bupropion Ascher JA, Cole JO, Colin JN, Feighner JP, Ferris RM, Fibiger HC, Golden RN, Martin P, Potter occupancy of the dopamine transporter is low during clinical treatment. Psychopharmacology WZ, and Richelson E, et al. (1995) Bupropion: a review of its mechanism of antidepressant (Berl) 163:102–105. activity. J Clin Psychiatry 56:395–401. Nomikos GG, Damsma G, Wenkstern D, and Fibiger HC (1989) Acute effects of bupropion on Bondarev ML, Bondareva TS, Young R, and Glennon RA (2003) Behavioral and biochemical extracellular dopamine concentrations in rat striatum and nucleus accumbens studied by in vivo investigations of bupropion metabolites. Eur J Pharmacol 474:85–93. microdialysis. Neuropsychopharmacology 2:273–279. Benowitz NL, Zhu AZ, Tyndale RF, Dempsey D, and Jacob P (2013) Influence of CYP2B6 Nomikos GG, Damsma G, Wenkstern D, and Fibiger HC (1992) Effects of chronic bupropion on genetic variants on plasma and urine concentrations of bupropion and metabolites at steady interstitial concentrations of dopamine in rat nucleus accumbens and striatum. Neuro- – – state. Pharmacogent Genomics 23:135 141. psychopharmacology 7:7 14. at ASPET Journals on September 25, 2021 Boström E, Simonsson USH, and Hammarlund-Udenaes M (2006) In vivo blood-brain barrier Okura T, Hattori A, Takano Y, Sato T, Hammarlund-Udenaes M, Terasaki T, and Deguchi Y transport of oxycodone in the rat: indications for active influx and implications for pharma- (2008) Involvement of the pyrilamine transporter, a putative organic cation transporter, in cokinetics/pharmacodynamics. Drug Metab Dispos 34:1624–1631. blood-brain barrier transport of oxycodone. Drug Metab Dispos 36:2005–2013. Cisternino S, Chapy H, André P, Smirnova M, Debray M, and Scherrmann JM (2013) Co- Okura T, Higuchi K, Kitamura A, and Deguchi Y (2014) Proton-coupled organic cation existence of passive and proton antiporter-mediated processes in nicotine transport at the mouse antiporter-mediated uptake of apomorphine enantiomers in human brain capillary endothelial blood-brain barrier. AAPS J 15:299–307. cell line hCMEC/D3. Biol Pharm Bull 37:286–291. Cooper BR, Wang CM, Cox RF, Norton R, Shea V, and Ferris RM (1994) Evidence that the acute Pang KS and Gillette JR (1979) Sequential first-pass elimination of a metabolite derived from a behavioral and electrophysiological effects of bupropion (Wellbutrin) are mediated by a nor- precursor. J Pharmacokinet Biopharm 7:275–290. adrenergic mechanism. Neuropsychopharmacology 11:133–141. Qiu X, Zhang H, and Lai Y (2014) Quantitative targeted proteomics for membrane transporter Cremers TIFH, de Vries MG, Huinink KD, van Loon JP, v d Hart M, Ebert B, Westerink BH, proteins: method and application. AAPS J 16:714–726. and De Lange EC (2009) Quantitative microdialysis using modified ultraslow microdialysis: Rowland M and Tozer TN (2011) Clinical Pharmacokinetics and Pharmacodynamics Concepts direct rapid and reliable determination of free brain concentrations with the MetaQuant tech- and Applications, 4th ed, Walters Kluwer, Lippincott Williams and Wilkins, Philadelphia. nique. J Neurosci Methods 178:249–254. Sharma V and McNeill JH (2009) To scale or not to scale: the principles of dose extrapolation. Br Damaj MI, Carroll FI, Eaton JB, Navarro HA, Blough BE, Mirza S, Lukas RJ, and Martin BR J Pharmacol 157:907–921. (2004) Enantioselective effects of hydroxy metabolites of bupropion on behavior and on Shimomura K, Okura T, Kato S, Couraud P-O, Schermann JM, Terasaki T, and Deguchi Y (2013) function of monoamine transporters and nicotinic receptors. Mol Pharmacol 66:675–682. Functional expression of a proton-coupled organic cation (H+/OC) antiporter in human brain capillary Davies B and Morris T (1993) Physiological parameters in laboratory animals and humans. endothelial cell line hCMEC/D3, a human blood-brain barrier model. Fluids Barriers CNS 10:8. Pharm Res 10:1093–1095. Slemmer JE, Martin BR, and Damaj MI (2000) Bupropion is a nicotinic antagonist. J Pharmacol de Lange ECM (2013) The mastermind approach to CNS drug therapy: translational prediction of Exp Ther 295:321–327. human brain distribution, target site kinetics, and therapeutic effects. Fluids Barriers CNS 10: Spreafico M and Jacobson MP (2013) In silico prediction of brain exposure: drug free fraction, 12. unbound brain to plasma concentration ratio and equilibrium half-life. Curr Top Med Chem 13: Dong J and Blier P (2001) Modification of norepinephrine and serotonin, but not dopamine, 813–820. neuron firing by sustained bupropion treatment. Psychopharmacology (Berl) 155:52–57. Suckow RF, Smith TM, Perumal AS, and Cooper TB (1986) Pharmacokinetics of bupropion and Fava M, Rush AJ, Thase ME, Clayton A, Stahl SM, Pradko JF, and Johnston JA (2005) 15 years metabolites in plasma and brain of rats, mice, and guinea pigs. Drug Metab Dispos 14: of clinical experience with bupropion HCl: from bupropion to bupropion SR to bupropion XL. 692–697. Prim Care Companion J Clin Psychiatry 7:106–113. Sweet RA, Pollock BG, Kirshner M, Wright B, Altieri LP, and DeVane CL (1995) Pharmaco- Fischer W, Bernhagen J, Neubert RHH, and Brandsch M (2010) Uptake of codeine into intestinal kinetics of single- and multiple-dose bupropion in elderly patients with depression. J Clin epithelial (Caco-2) and brain endothelial (RBE4) cells. Eur J Pharm Sci 41:31–42. Pharmacol 35:876–884. Fridén M, Bergström F, Wan H, Rehngren M, Ahlin G, Hammarlund-Udenaes M, and Bredberg Szentistványi I, Patlak CS, Ellis RA, and Cserr HF (1984) Drainage of interstitial fluid from U (2011) Measurement of unbound drug exposure in brain: modeling of pH partitioning different regions of rat brain. Am J Physiol 246:F835–F844. explains diverging results between the brain slice and brain homogenate methods. Drug Metab Velez de Mendizabal N, Jackson K, Eastwood B, Swanson S, Bender DM, Lowe S, and Bies RR Dispos 39:353–362. (2015) A population PK model for citalopram and its major metabolite, N-desmethyl Ganapathy V and Leibach FH (1983) Role of pH gradient and membrane potential in dipeptide citalopram, in rats. J Pharmacokinet Pharmacodyn 42:721–733. transport in intestinal and renal brush-border membrane vesicles from the rabbit. Studies with Wooltorton E (2002) Bupropion (Zyban, Wellbutrin SR): reports of deaths, seizures, serum L-carnosine and glycyl-L-proline. J Biol Chem 258:14189–14192. sickness. CMAJ 166:68. Gharavi R, Hedrich W, Wang H, and Hassan HE (2015) Transporter-mediated disposition of Yeniceli D, S¸ener E, Korkmaz OT, Dogrukol-Ak D, and Tuncel N (2011) A simple and sensitive opioids: implications for clinical drug interactions. Pharm Res 32:2477–2502. LC-ESI-MS (ion trap) method for the determination of bupropion and its major metabolite, Gondaliya D and Pundarikakshudu K (2003) Studies in formulation and pharmacotechnical hydroxybupropion in rat plasma and brain microdialysates. Talanta 84:19–26. evaluation of controlled release transdermal delivery system of bupropion. AAPS PharmSci- Tech 4:E3. Hsyu PH, Singh A, Giargiari TD, Dunn JA, Ascher JA, and Johnston JA (1997) Pharmacokinetics Address correspondence to: Robert E. Stratford, Jr., Mylan School of of bupropion and its metabolites in cigarette smokers versus nonsmokers. J Clin Pharmacol 37: Pharmacy, Duquesne University, 600 Forbes Avenue, Pittsburgh, PA 15282. 737–743. Jefferson JW, Pradko JF, and Muir KT (2005) Bupropion for major depressive disorder: phar- E-mail: [email protected] macokinetic and formulation considerations. Clin Ther 27:1685–1695.