The purpose of this study was to examine whether in vivo drug distribution to the brain can be reconstructed by integrating P-glycoprotein (P-gp)/mdr1a expression levels, P-gp in vitro activity, and drug unbound fractions in mouse plasma and brain. For 11 P-gp substrates, in vitro P-gp transport activities were determined by measuring transcellular transport across monolayers of mouse P-gp-transfected LLC-PK1 (L-mdr1a) and parental cells. P-gp expression amounts were determined by quantitative targeted absolute proteomics. Unbound drug fractions in plasma and brain were obtained from the literature and by measuring brain slice uptake, respectively. Brain-to-plasma concentration ratios (Kp brain) and its ratios between wild-type and mdr1a/1b(−/−) mice (Kp brain ratio) were obtained from the literature or determined by intravenous constant infusion. Unbound brain-to-plasma concentration ratios (Kp,uu,brain) were estimated from Kp brain and unbound fractions. Based on pharmacokinetic theory, Kp brain ratios were reconstructed from in vitro P-gp transport activities and P-gp expression amounts in L-mdr1a cells and mouse brain capillaries. All reconstructed Kp brain ratios were within a 1.6-fold range of observed values. Kp brain then was reconstructed from the reconstructed Kp brain ratios and unbound fractions. Kp,uu,brain was reconstructed as the reciprocal of the reconstructed Kp brain ratios. For quinidine, loperamide, risperidone, indinavir, dexamethasone, paclitaxel, verapamil, loratadine, and diazepam, the reconstructed Kp brain and Kp,uu,brain agreed with observed and estimated in vivo values within a 3-fold range, respectively. Thus, brain distributions of P-gp substrates can be reconstructed from P-gp expression levels, in vitro activity, and drug unbound fractions.
Quantitative evaluation of drug distribution in human brain is a key issue in drug development because the distribution of a drug is directly related to its pharmacological actions and toxic effects in the central nervous system (CNS). The brain-to-plasma concentration ratio (Kp brain) and unbound brain-to-plasma concentration ratios (Kp,uu,brain) are the most commonly used parameters and are accepted as good indexes for identifying drugs that would be effective in treating CNS disorders, because CNS drugs have been shown to have higher Kp brain and Kp,uu,brain values than non-CNS ones (Fridén et al., 2009).
Several strategies have been used to estimate Kp brain and Kp,uu,brain in humans. Clinical studies using imaging technologies, such as single-photon emission-computed tomography, positron emission tomography, and magnetic resonance imaging, are direct approaches in determining Kp brain in humans. However, the number of available imaging probes is limited (Kusuhara and Sugiyama, 2009). In vivo animal experiments are widely performed to estimate Kp brain and Kp,uu,brain, but the results are not very useful because of pronounced species differences in Kp brain; e.g., Kp brain of [11C](S)-(2-methoxy-5-(5-trifluoromethyltetrazol-1-yl)-phenylmethylamino)-2(S)-phenylpiperidine (GR205171) and [18F]altanserin were 8.6- and 4.5-fold higher in humans than in rats, respectively (Syvänen et al., 2009).
Another strategy is in vitro-to-in vivo prediction using cultured cells expressing transporters. P-glycoprotein (P-gp/MDR1/mdr1a) has a major role in drug distribution into the brain as an efflux transporter, which limits entry of drugs into the brain at the blood-brain barrier (BBB). It was reported that 72% of the 32 most prescribed CNS drugs have an in vivo ratio of Kp brain between mdr1a/1b(−/−) and wild-type mice (Kp brain ratio) that ranges from 1 to 3 (Liu et al., 2008). This indicates that many CNS drugs undergo weak mdr1a-mediated efflux at the BBB, suggesting that candidate drugs for the treatment of CNS diseases can to be chosen not only from nonsubstrates of P-gp but also from P-gp substrates with a low Kp brain ratio. Furthermore, the Kp,uu,brain can be calculated from the reciprocal of the Kp brain ratio when P-gp-mediated efflux is the only active process affecting brain distribution (Kalvass et al., 2007). Therefore, quantitative evaluation of P-gp transport activity (Kp brain ratio) using P-gp-expressing cells will provide important information for CNS drug development.
Several previous studies have evaluated the usefulness of P-gp-transfected cells for the prediction of the Kp brain ratio, which reflects P-gp activity at the BBB in vivo, and it has been found that the in vivo Kp brain ratio correlates well with the in vitro P-gp efflux ratio (also known as the corrected flux ratio), which is the basal-to-apical/apical-to-basal transport ratio in the P-gp-transfected cell monolayer divided by that in the parental cell monolayer (Adachi et al., 2001; Feng et al., 2008). However, the absolute values of the in vitro P-gp efflux ratio are not always the same as those of the Kp brain ratio. Adachi et al. (2001) showed that the in vitro P-gp efflux ratios of 10 compounds were approximately 4-fold smaller than the Kp brain ratio on average. In addition, Feng et al. (2008) showed that the in vitro P-gp efflux ratios of 39 compounds were approximately 2.4-fold smaller than the Kp brain ratio on average. One of the causes of the differences between in vitro and in vivo is considered to be the difference in P-gp expression levels between P-gp-transfected cell lines and brain capillary endothelial cells. This limitation of in vitro to in vivo prediction suggests that integration of the P-gp protein level is necessary to predict the P-gp activity at the human BBB.
Recently, we developed a protein quantification method using LC-MS/MS, called quantitative targeted absolute proteomics (QTAP), which provides the absolute expression levels of transporter proteins in human, monkey, and mouse brain capillaries (Kamiie et al., 2008; Ito et al., 2011; Uchida et al., 2011). Thus, we hypothesized that LC-MS/MS-based quantification of P-gp protein in P-gp-transfected cells and brain capillaries would make it possible to reconstruct the Kp brain ratio from in vitro data in combination with transport assay using P-gp-transfected cells.
The purpose of the present study was to experimentally demonstrate in a mouse model that the Kp brain ratio could be reconstructed by integrating in vitro P-gp transport activity with P-gp protein levels in the brain capillaries and P-gp-transfected cells. The Kp brain of P-gp substrates was reconstructed by integrating the reconstructed Kp brain ratio with the plasma and brain unbound fractions, and the Kp,uu,brain was also reconstructed as the reciprocal of the Kp brain ratio from in vitro data. These three reconstructed parameters of 11 drugs were compared with those obtained from in vivo studies to validate the reconstruction from in vitro data. Furthermore, involvement of breast cancer resistance protein (bcrp) in the brain distribution was examined using bcrp-expressing cells for two compounds, which were suggested to be transported by efflux transporters other than P-gp on the basis that the predicted Kp brain and Kp,uu,brain were larger than the in vivo values.
Materials and Methods
Dexamethasone (Decadron), digitoxin (Crystodigin), digoxin (Lanoxin), fexofenadine hydrochloride (Allegra), flumethasone (Locacorten), loperamide hydrochloride (Imodium), (±)-metoprolol (+)-tartrate salt (Lopressor), and quinidine (Quinidex) were purchased from Sigma-Aldrich (St. Louis, MO). Cinchonine, diazepam (Valium), d-(−)-mannitol (Osmitol), haloperidol (Haldol), loratadine (Claritin), lorazepam (Ativan), paclitaxel (Taxol), risperidone (Risperdal), saquinavir (Invirase), verapamil hydrochloride (Calan), vinblastine sulfate (Velban), vincristine sulfate (Oncovin), vindesine sulfate (Eldisine), and xylitol were purchased from Wako Pure Chemicals (Osaka, Japan). Indinavir sulfate (Crixivan) was purchased from Toronto Research Chemicals Inc. (North York, ON, Canada). Mdr1a peptides were synthesized by Thermo Fisher Scientific, Waltham, MA with >95% peptide purity. All other chemicals were of reagent grade and were available commercially.
Male FVB wild-type and male FVB mdr1a/1b(−/−) mice were purchased from Taconic Farms Inc. (Germantown, NY), and male ddY mice were purchased from Charles River (Yokohama, Japan). Mice were maintained on a 12-h light/dark cycle in a temperature-controlled environment with free access to food and water. For the intravenous constant infusion study, FVB wild-type and mdr1a/1b(−/−) mice were used at 10 weeks of age with a body weight of 23 to 33 g. For the uptake experiment using brain slices, ddY mice were used at 10 weeks of age. The protocol was approved by the Institutional Animal Care and Use Committee at Tohoku University (Permission No. 20-Pharm-Animal-5 and -6).
Determination of Brain Distribution of Compounds in Wild-Type and mdr1a/1b(−/−) Mice.
Under anesthesia with a mixture of ketamine hydrochloride (125 mg/kg b.wt.) and xylazine hydrochloride (1.22 mg/kg b.wt.), mice were cannulated at the right jugular vein with polyethylene tubing (PE-10; Natsume, Tokyo, Japan) for infusion of test compounds (quinidine, loperamide, digoxin, dexamethasone, and vinblastine), which were dissolved in the infusion buffer (128 mM NaCl, 24 mM NaHCO3, 4.2 mM KCl, 2.4 mM NaH2PO4, 1.5 mM CaCl2, 0.9 mM MgCl2, 9.0 mM d-glucose, pH7.4). Mice received a 1-min rapid infusion of test compounds at a rate of 100 μl/min followed by a constant infusion at a rate of 200 μl/h for 100 min (Harvard pump 11; Harvard Apparatus Inc., Holliston, MA). Throughout the experiment, mice were kept warm using a hot plate. Blood were collected from the left jugular vein at 40, 60, 80, and 100 min and immediately centrifuged at 7720g for 5 min to obtain plasma. Immediately after the blood sampling at 100 min, the mice were sacrificed, and the cerebrum of each animal was collected, weighed, and homogenized with 10 mM ammonium acetate (4 ml:1 g brain). Acetonitrile (180 and 160 μl) containing 1% formic acid and internal standard (Supplemental Table 1) was added to 20 μl of plasma and 40 μl of brain homogenate, respectively. After having been shaken for 20 min, the samples were centrifuged at 4°C and 17,360g for 5 min. The supernatant (180 μl) then was evaporated by centrifugation under vacuum, and the residue was reconstituted in 100 μl of 0.1% aqueous formic acid and centrifuged at 4°C and 17,360g for 5 min. The supernatants were subjected to LC-MS/MS analysis.
The brain-to-plasma concentration ratio (Kp brain) was calculated by dividing the brain concentration by the plasma concentration at 100 min. For risperidone, indinavir, paclitaxel, verapamil, loratadine, and diazepam, the Kp brain values were obtained from the literature (Hendrikse et al., 1998; Kim et al., 1998; Chen et al., 2003; Kemper et al., 2004; Doran et al., 2005). As a parameter describing the in vivo P-gp function at the BBB, the Kp brain ratio was defined and calculated as shown in eq. 1:
Cell Culture of L-mdr1a, Parental LLC-PK1, bcrp-Transfected MDCKII, and Parental MDCKII Cells.
Mouse P-gp/mdr1a-transfected LLC-PK1 cells (L-mdr1a) and the parental LLC-PK1 cells were kindly provided by Dr. Alfred H. Schinkel (The Netherlands Cancer Institute, Amsterdam, The Netherlands) and used under license agreement. Cells were cultured in medium 199 (Nissui Pharmaceutical, Tokyo, Japan) supplemented with 2 mM l-glutamine, 50 units/ml penicillin, 50 μg/ml streptomycin, and 10% (v/v) fetal bovine serum (Moregate, Bulimba, Queensland Australia). In addition to this medium, L-mdr1a was maintained in the presence of 640 nM vincristine. Mouse breast cancer resistance protein (bcrp/abcg2)-transfected MDCKII cells were cultured in Dulbecco's modified Eagle's medium (Nissui Pharmaceutical, Tokyo, Japan) supplemented with 2 mM l-alanyl-l-glutamine (GlutaMAX; Invitrogen, Carlsbad, CA), 800 μg/ml G418 (Geneticin), and 10% (v/v) fetal bovine serum. The parental MDCKII cells were cultured in Dulbecco's modified Eagle's medium supplemented with 2 mM GlutaMAX, 50 units/ml penicillin, 50 μg/ml streptomycin, and 10% (v/v) fetal bovine serum. Confluent monolayers were subcultured every 2 to 3 days by treatment with trypsin-EDTA. All cultures were incubated at 37°C in a humidified atmosphere of 5% CO2, 95% air.
L-mdr1a and the parental LLC-PK1 cells were seeded on porous (3.0-μm) polycarbonate membrane filters (Transwell; Corning, Cambridge, MA) at a density of 5 × 105 cells/cm2. Mouse bcrp-transfected MDCKII cells and the parental MDCKII cells were seeded on porous (1.0-μm) polyethylene terephthalate membrane filters (cell culture inserts; BD Biosciences, Franklin Lakes, NJ) at a density of 5 × 105 cells/cm2. Cells were supplemented with fresh medium on the 2nd day and used for the experiments on the 4th day after seeding.
Transcellular Transport Study across L-mdr1a, Parental LLC-PK1, bcrp-Transfected MDCKII, and Parental MDCKII Cell Monolayers.
The transcellular transport study was carried out as described previously (Yamazaki et al., 2001) with minor modifications. Opti-MEM (Invitrogen) was used for L-mdr1a and the parental LLC-PK1 cells, and extracellular fluid (ECF) buffer (122 mM NaCl, 3 mM KCl, 0.4 mM K2HPO4, 25 mM NaHCO3, 1.4 mM CaCl2, 1.2 mM MgSO4, 10 mM d-glucose, 10 mM HEPES, pH7.4) was used for bcrp-transfected MDCKII and the parental MDCKII cells. Cell monolayers were preincubated in the solution without test compounds at 37°C for approximately 2 h, and then the transport experiment was initiated by replacing the solution in each compartment with fresh solution with (donor compartment) and without (acceptor compartment) test compounds. At each sampling time, aliquots were taken from both the donor and acceptor compartments and then immediately replaced with fresh solution with (donor compartment) and without (acceptor compartment) test compounds to maintain sink conditions. To help maintain sink conditions, the solutions in both compartments were stirred with a pipette every few minutes during the transport experiment. The sampling times for each transport experiment were selected based on the results of a pilot experiment to keep the concentration in the donor compartment sufficiently higher than that in the acceptor compartment to prevent underestimation of donor-to-acceptor transport. Internal standards (Supplemental Table 1) were added to the collected samples, and then they were acidified with formic acid and centrifuged at 4°C and 17,360g for 5 min. The supernatants were subjected to LC-MS/MS analysis to quantify the amount of test compounds transported to the acceptor side and the concentration on the donor side.
For L-mdr1a and the parental LLC-PK1 cells, the apparent permeability (Papp) coefficient (centimeters per second) was calculated by dividing the slope (microliters per minute) of the time profiles of apical-to-basal or basal-to-apical transport by 1 cm2 of the insert area. Flux ratios were obtained by dividing the Papp in the basal-to-apical direction by that in the apical-to-basal direction. Flux ratios in L-mdr1a were divided by those in parental LLC-PK1 cells to give an in vitro P-gp efflux ratio. This in vitro P-gp efflux ratio was used as a measure of the in vitro P-gp transport activity. The paracellular flux was monitored in terms of the appearance of mannitol in the opposite compartment and was <0.7% of the total mannitol per hour.
Determination of mdr1a Protein Expression Amounts in L-mdr1a Cell Monolayers Based on QTAP.
Mdr1a protein expression amounts were determined according to the reported method (Kamiie et al., 2008). On the 4th day after seeding, L-mdr1a cells were harvested from the Transwell filters, and then cells equivalent to 50 μg of total protein in cell mass were solubilized in 500 mM Tris-HCl (pH 8.5), 7 M guanidine hydrochloride, 10 mM EDTA, and the proteins were S-carbamoylmethylated as described previously (Kamiie et al., 2008). The alkylated proteins were precipitated with a mixture of methanol and chloroform, and the precipitates were dissolved in 6 M urea in 100 mM Tris-HCl (pH 8.5), diluted 5-fold with 100 mM Tris-HCl (pH 8.5), and treated with N-tosyl-l-phenylalanine chloromethyl ketone-treated trypsin (Promega, Madison, WI) at an enzyme/substrate ratio of 1:100 at 37°C for 16 h. The tryptic digests were mixed with internal standard peptides and formic acid and then centrifuged at 4°C and 17,360g for 5 min, and the supernatants were subjected to LC-MS/MS analysis.
Brain Slice Uptake Experiments.
The brain slice uptake experiments were performed as described previously (Kakee et al., 1996), with minor modifications. Mice were decapitated under anesthesia using a mixture of ketamine hydrochloride (125 mg/kg b.wt.) and xylazine hydrochloride (1.22 mg/kg b.wt.), and the brains were immediately removed and immersed in ice-cold oxygenated ECF buffer equilibrated with 95% O2, 5% CO2. Cerebral slices (500 μm thick) were cut using a microslicer (DTK-2000; Dosaka, Kyoto, Japan), and kept in ice-cold oxygenated ECF buffer equilibrated with 95% O2, 5% CO2. After preincubation at 37°C for 10 min in oxygenated ECF buffer, the brain slices (20–40 mg) were transferred to oxygenated ECF buffer containing test compounds at 37°C. At designated times (240, 300, and 360 min), slices were removed from the solution, dried on filter paper, and weighed. The slices were then homogenized in 9 volumes (w/v) of 10 mM ammonium acetate with an ultrasonic probe (Branson, Sonifier 150; Branson Ultrasonics Corporation, Danbury, CT). The homogenates were treated according to the same procedure as in the in vivo study and then subjected to LC-MS/MS analysis.
The unbound fraction in the brain (fu,brain) was calculated using eq. 2, where Aslice, Cbuffer, and Vi are the amount of test compound at 360 min in the slice, the medium concentration of test compound at 360 min, and the adherent water volume, respectively.
After incubation of brain slices in ECF buffer containing mannitol for 1, 3, and 5 min, Vi was estimated to be 0.0880 ml/g brain from a plot of Aslice/Cbuffer using zero time back-extrapolation. The uptakes at 360 min were confirmed to reach steady state for all 11 compounds by comparing those at 240, 300, and 360 min. The brain distribution volume (Vbrain) of quinidine was calculated using eq. 3.
The sample analysis was automated by coupling a triple quadrupole mass spectrometer (4000 QTRAP or API5000; Applied Biosystems, Foster City, CA) to an Agilent 1100 high-performance liquid chromatography (HPLC) system (Agilent Technologies, Santa Clara, CA). Samples were injected onto an Agilent XDB-C18 column (2.1 × 150 mm, 5 μm), an Inertsil NH2 column (2.1 × 250 mm, 5 μm), or an Agilent 300SB-C18 column (0.5 × 150 mm, 5.0 μm). The Inertsil NH2 column was maintained at 65°C, whereas all of the other columns were maintained at room temperature. Mobile phases A and B consisted of 0.1% formic acid in water and 0.1% formic acid in acetonitrile, respectively. The compounds or peptides were separated and eluted from the column under linear gradient or isocratic conditions. The eluted compounds and peptides were detected by electrospray ionization using multiple reaction monitoring (MRM) and multiplexed MRM modes, respectively. The detailed LC-MS/MS conditions, internal standards, and amino acid sequences for mdr1a peptides are shown in Supplemental Table 1.
Chromatogram ion counts were determined by data acquisition procedures using Analyst software version 1.4.2 (Applied Biosystems). For the determination of mdr1a expression amounts, the signal peaks with a peak area count of over 5000 detected at the same retention time as an internal standard peptide were defined as positive and used for quantification. The mdr1a protein expression amounts were determined as an average of 32 quantitative values, which were obtained from four different sets of MRM transitions of quadruplicate experiments (n = 4), with two sets of probe peptides as described in Supplemental Table 1.
Theory for Reconstruction of the Kp brain Ratio.
According to the pharmacokinetic model illustrated in Scheme 1, reported previously by Adachi et al. (2001), the Kp brain ratio and in vitro P-gp efflux ratio (also known as corrected flux ratio) are described in eqs. 4 and 5: where PSmdr1a,vivo and PSl,eff represent the PS products for the mdr1a-mediated efflux in the brain capillary endothelium and the luminal efflux, with the exception of mdr1a-mediated efflux, respectively; PSmdr1a,vitro and PSa,eff represent the PS products for the mdr1a-mediated efflux in the mdr1a-transfected cell monolayer and the apical efflux, with the exception of mdr1a-mediated efflux, respectively. If test compounds are specifically transported by mdr1a at the luminal membrane of the endothelium and the apical membrane of the mdr1a-transfected cell monolayer, PSl,eff can be assumed to be equal to PSa,eff because both processes occur by passive diffusion. On the basis of this assumption, the Kp brain ratio is obtained using the in vitro P-gp efflux ratio as shown in eq. 6:
Hoffmeyer et al. (2000) have suggested that the P-gp transport activity depends on its protein expression level in vivo in humans. Shirasaka et al. (2008) and Tachibana et al. (2010) reported that P-gp transport activity was approximately proportional to its protein expression levels in vitro. Therefore, it was assumed that the mdr1a activity is directly related to the mdr1a protein expression level and the in vitro intrinsic transport activity of mdr1a (transport rate per mdr1a protein) is identical to that in vivo. Hence, eq. 6 can be converted to eq. 7:
Accordingly, the in vivo Kp brain ratio can be reconstructed from in vitro experiments, i.e., transcellular transport study to determine the in vitro P-gp efflux ratio and QTAP to determine the mdr1a protein expression amounts.
Theory for Reconstruction of Kp brain.
The Kp brain ratio is defined as eq. 8: where Kp brain WT (= Kp brain) and Kp brain KO represent the brain-to-plasma concentration ratios in wild-type and mdr1a/1b(−/−) mice, respectively. In addition, the Kp brain KO is defined in eq. 9 based on unbound fractions in plasma (fu,plasma) and brain (fu,brain) and the ratio of the brain unbound concentration (Cu,brain KO) to the plasma unbound concentration (Cu,plasma KO).
Cu,brain KO/Cu,plasma KO can be replaced with the ratio of the blood-to-brain clearance to the brain-to-blood clearance across the BBB (PSblood-to-brain KO/PSbrain-to-blood KO). For compounds transported by only mdr1a at the BBB, PSblood-to-brain KO can be equal to PSbrain-to-blood KO because both processes occur by passive diffusion in mdr1a/1b(−/−) mice. Therefore, Cu,brain KO/Cu,plasma KO = 1 and then Kp brain KO is obtained as indicated by eq. 10:
Theory for Reconstruction of Kp,uu,brain.
The Kp,uu,brain is defined as eq. 12:
All data represent the mean ± S.E.M. unless otherwise indicated. The S.E.M. was calculated according to the following law of propagation of error; given the following functional relationship between several measured variables such as y, x1, x2, ……, xn (for example, in eq. 8, y can be the Kp brain ratio, and x1 and x2 can be the Kp brain in mdr1a/1b(−/−) mice or the Kp brain in wild-type mice) (eq. 14):
If the variables x1, x2, ……, xn are uncorrelated, standard deviations of y (S.D.y) and x (S.D.x) can be related as indicated in eq. 15: where ∂f/∂x is the partial derivative of function y with regard to x. The S.E.M. of y is given by eq. 16. where n(y) is assumed to be n(y) = n(x1) + n(x2) + …… + n(xn).
Determination of Kp brain in Wild-Type and mdr1a/1b(−/−) Mice and the Kp brain Ratio for 11 Compounds.
Quinidine, loperamide, digoxin, dexamethasone, and vinblastine were administered to wild-type and mdr1a/1b(−/−) mice by continuous intravenous infusion, and the Kp brain values were determined at a steady-state (100 min) plasma concentration (Supplemental Fig. 1). The plasma concentrations of five compounds at 100 min were 92 to 486 nM, which were lower than the reported Km values for P-gp determined by ATPase assay (Adachi et al., 2001). The values of Kp brain in wild-type mice showed a 49-fold variation among 11 compounds, including those taken from the literature (Hendrikse et al., 1998; Kim et al., 1998; Chen et al., 2003; Kemper et al., 2004; Doran et al., 2005) (Table 1). The Kp brain values of 10 compounds, other than diazepam, were increased over 2-fold in mdr1a/1b(−/−) mice compared with those in wild-type mice. Among the 11 compounds, the ratios of Kp brain in normal and mdr1a/1b(−/−) mice varied from 1.18 (diazepam) to 39.4 (quinidine).
Determination of the mdr1a Protein Expression Amount in L-mdr1a Cell Monolayers.
The mdr1a protein expression amount in L-mdr1a cell monolayers was determined by LC-MS/MS-based quantification with two sets of probe peptides listed in Supplemental Table 1. The average of the quantitative values obtained with two sets in quadruplicate experiments was 15.2 ± 2.7 fmol/μg protein (mean ± S.E.M.).
Determination of In Vitro mdr1a Transport Activity in L-mdr1a Cell Monolayers.
The transcellular transport of 11 compounds was determined in the basal-to-apical and apical-to-basal directions across L-mdr1a and parental LLC-PK1 cell monolayers (Fig. 1). In L-mdr1a cells, the basal-to-apical transport of 10 compounds, other than diazepam, was more than 2-fold greater than the apical-to-basal transport (Table 2). In contrast, the transcellular transport of all of the compounds was within a 2-fold difference between the directions in parental LLC-PK1 cells. The in vitro P-gp efflux ratio, which is the basal-to-apical/apical-to-basal transport ratio in the L-mdr1a cell monolayer divided by that in the LLC-PK1 cell monolayer, reflects mdr1a-specific transport activities for each test compound and varied from 0.859 (diazepam) to 33.0 (quinidine) among the 11 compounds investigated.
Reconstruction of the Kp brain Ratio from the In Vitro mdr1a Transport Activity and mdr1a Protein Expression Amount.
The Kp brain ratio for the 11 compounds were reconstructed in Table 2 by means of eq. 7 from the in vitro P-gp efflux ratios and the mdr1a protein expression amounts in the L-mdr1a cell monolayers and in mouse brain capillaries (14.1 ± 0.9 fmol/μg protein, mean ± S.E.M.; reported by Kamiie et al., 2008). For the 11 substrates, the reconstructed Kp brain ratios agreed with the observed values in an in vivo study (Fig. 2). The largest difference between the reconstructed and observed Kp brain ratios was 1.61-fold in the case of verapamil.
Reconstruction of the Kp brain in Wild-Type Mice from the Reconstructed Kp brain Ratios and Unbound Fractions in Plasma and Brain.
The values of Kp brain for the 11 compounds in wild-type mice were reconstructed by means of eq. 11 from the reconstructed Kp brain ratios, the unbound fractions in mouse brain, and the reported unbound fractions in mouse plasma (Maurer et al., 2005; Kalvass et al., 2007) (Table 3). The reconstructed Kp brain values agreed with the observed ones within a 3-fold range for 9 of the 11 compounds (Fig. 3). By contrast, the reconstructed values for digoxin and vinblastine were more than three times the observed values, suggesting that other efflux mechanism(s) could limit the brain penetration of these compounds.
Reconstruction of Kp,uu,brain in Wild-Type Mice from the Reconstructed Kp brain Ratios.
According to eq. 13, the values of Kp,uu,brain for the 11 compounds in wild-type mice were reconstructed as the reciprocal of the reconstructed Kp brain ratios from in vitro data with the assumption that these compounds were transported only by mdr1a at the BBB (Table 4). The in vivo Kp,uu,brain values were estimated using the observed Kp brain in vivo and the unbound fractions by means of eq. 12. The reconstructed Kp,uu,brain values agreed with estimated in vivo values within a range of 3-fold for 9 of the 11 compounds (Fig. 4). The reconstructed values for digoxin and vinblastine were more than 3-fold greater than the estimated in vivo values, suggesting that some efflux mechanism(s) other than P-gp is involved in limiting the brain penetration of these drugs.
Transcellular Transport of Digoxin and Vinblastine across Mouse bcrp-Expressing MDCKII and Parental Cell Monolayers.
Bcrp is involved in the brain-to-blood efflux of a number of xenobiotics at the BBB (Breedveld et al., 2005; Enokizono et al., 2008). To evaluate the involvement of bcrp in the brain distribution of digoxin and/or vinblastine, transcellular transport was determined in the basal-to-apical and apical-to-basal directions across mouse bcrp-transfected and parental MDCKII cell monolayers (Fig. 5). No significant difference was observed in the basal-to-apical/apical-to-basal transport ratios between mouse bcrp-transfected and parental MDCKII cell monolayers.
The present study is the first to experimentally demonstrate in a mouse model that the Kp brain ratio values for 11 P-gp substrates can be reconstructed by integrating in vitro mdr1a transport activity and mdr1a protein expression levels in the brain capillaries and mdr1a-transfected cell monolayers. Furthermore, Kp brain and Kp,uu,brain were subsequently reconstructed for 9 of the 11 P-gp substrates from the reconstructed Kp brain ratios.
The Kp brain ratio values in the range of 1.18 (diazepam) to 39.4 (quinidine) were reconstructed within a 1.61-fold difference from the observed values (Table 1). The Kp brain ratio of P-gp substrates has been reported to range from approximately 1 to 50 in an animal experiment using wild-type and mdr1a/1b(−/−) mice (Kalvass et al., 2007). Therefore, the present study covered a wide variety of P-gp substrates, from weakly transported substrates to strongly transported substrates.
The reconstructed Kp brain and Kp,uu,brain agreed with the observed and estimated in vivo values within a 3-fold difference for 9 of the 11 P-gp substrates, respectively (Figs. 3 and 4). According to Fridén et al. (2009), there is a 4.5-fold difference in Kp brain between CNS (4.31) and non-CNS (0.962) drugs, and there is also a 5.3-fold difference in Kp,uu,brain between CNS (0.767) and non-CNS (0.145) drugs. Hence, in terms of brain distribution, the precision of pharmacoproteomics (PPx)-based reconstructions of Kp brain and Kp,uu,brain could be high enough to evaluate whether compounds are likely to be suitable for use as CNS or non-CNS drugs. However, the range of these parameters, especially Kp brain, among CNS or non-CNS drugs is large, so it would be difficult to identify CNS or non-CNS drugs based on Kp brain alone or Kp,uu,brain alone. A combination of Kp brain and Kp,uu,brain reconstructions might be more reliable.
The present study assumed that the apical efflux (PSa,eff) in a LLC-PK1 cell monolayer is equal to the luminal efflux (PSl,eff) in the brain capillary endothelium, because both occur by passive diffusion if substrates are specific for mdr1a. Polli et al. (2000) observed a similar trend between in situ brain perfusion (Kin) and the Papp in culture cell monolayers derived from kidney epithelial cells. Furthermore, Summerfield et al. (2007) reported that the Kin of some non-P-gp substrates is of the same order as their passive membrane permeabilities in a cultured kidney epithelial cell monolayer. These data support the assumption that in vitro passive diffusion is identical to that in vivo. Good agreement was obtained between the reconstructed Kp brain ratio based on this assumption and the observed Kp brain ratios, thereby supporting the validity of the assumption.
In the present study, the mdr1a intrinsic transport activity (transport rate per mdr1a protein) in L-mdr1a cell monolayers was also assumed to be equal to that in brain capillaries. In rodents, the BBB influx clearance of quinidine, a good P-gp substrate, has been reported to be 25.5 μl · min−1 · g brain−1 by in situ brain perfusion (Kusuhara et al., 1997), corresponding to 4.2 × 10−6 cm/s using the surface area of the brain capillaries (100 cm2/g brain) (Hammarlund-Udenaes et al., 2008). This value is almost identical to the apparent permeability coefficient (Papp = 3.16 × 10−6 cm/s) of quinidine in the apical-to-basal direction in an L-mdr1a cell monolayer (Table 2). Likewise, the elimination rate constant (kel = 0.023 min−1) of quinidine determined by the brain efflux index method (Kusuhara et al., 1997) and the brain distribution volume (Vbrain = 25.2 ml/g brain) determined by incubation of mouse brain slices in ECF buffer containing quinidine at 37°C for 360 min in the present study give a BBB efflux clearance (kel × Vbrain) of 0.580 ml · min−1 · g brain−1. Using the surface area of brain capillaries, the clearance was calculated to be 96.6 × 10−6 cm/s. This value is close to the basal-to-apical Papp (146 × 10−6 cm/s) of quinidine in an L-mdr1a cell monolayer (Table 2). The mdr1a protein level in mouse brain capillary (14.1 fmol/μg protein; Kamiie et al., 2008) was almost the same as that in an L-mdr1a cell monolayer (15.2 fmol/μg protein). Therefore, these data support the assumption that the mdr1a intrinsic transport activity in L-mdr1a cell monolayers is identical to that in brain capillaries. However, the mdr1a intrinsic transport activity in brain capillaries has been suggested to vary in brain inflammation and pathological conditions associated with increased expression of vascular endothelial growth factor in brain, such as neurological disease and brain injury (Hawkins et al., 2010; Miller, 2010). For such conditions, a suitable in vitro model should be established and used for the prediction of P-gp function and brain distribution.
The reconstructed Kp brain and Kp,uu,brain values for digoxin and vinblastine were more than three times the observed and estimated in vivo values, respectively (Figs. 3 and 4). One possible explanation is that efflux mechanism(s) other than mdr1a could limit the brain penetration, because the Kp,uu,brain values of digoxin and vinblastine in mdr1a/1b(−/−) mice were much less than 1 (0.0942 and 0.198, respectively). In addition to mdr1a, bcrp also plays an important role in the brain-to-blood efflux of various xenobiotics at the BBB (Enokizono et al., 2008). Therefore, we investigated whether or not digoxin and vinblastine are transported via mouse bcrp by measuring their transcellular transport rates across bcrp-transfected and parental MDCKII cell monolayers. However, neither digoxin nor vinblastine was found to be transported by mouse bcrp (Fig. 5). Organic anion-transporting polypeptide 2 (oatp2) is expressed in mouse brain capillaries and contributes to brain-to-blood transport across the BBB (Ose et al., 2010). Digoxin is a substrate of oatp2, but oatp2 is involved in not only brain-to-blood but also blood-to-brain transport of substrates; thus, it is unclear whether oatp2 limits the brain distribution of digoxin in the steady state. Vinblastine is a substrate of multidrug resistance associated protein 1 (mrp1), which is localized on the luminal membrane of brain capillaries. However, the protein expression level is below the limit of quantification (Kamiie et al., 2008), and the Kp brain of etoposide, a good mrp1 substrate, is not altered in mdr1a/mdr1b double and mrp1/mdr1a/mdr1b triple KO mice (Wijnholds et al., 2000). Therefore, it remains questionable whether mrp1 limits the brain penetration of vinblastine. Although this study focused on mdr1a, further study is necessary in future to clarify the involvement of oatp2 and mrp1 or to identify the other non-mdr1a-mediated mechanism(s). In vitro to in vivo predictions based on PPx studies will be a useful strategy to quantitatively evaluate the contribution of each transporter, including P-gp, to drug distribution into the brain in vivo.
For the other nine compounds, the reconstructed values were within a 3-fold range of the observed ones; however, both values were not identical completely. It is considered that active influx and/or efflux mechanism(s) other than mdr1a could slightly influence brain penetration. In addition, the present study determined fu,brain using ddY mice, a strain different from that (FVB mice) in which observed Kp brain was determined. The differences between the reported fu,brain determined from FVB brain homogenate and the present values obtained in ddY brain slices were within a 1.9-fold range for the nine compounds (Maurer et al., 2005; Kalvass et al., 2007). This difference may be partly due to both strain and method differences. The brain slice method can estimate in vivo unbound drug concentration better than the homogenate method (Fridén et al., 2007). However, the possibility cannot be ruled out that the strain difference in fu,brain influenced the reconstruction, although the effect is likely to be minor.
Using the present PPx-based reconstruction method, P-gp transport activities at the BBB and Kp,uu,brain in human can be predicted from in vitro data by integrating the protein level of P-gp in human brain capillaries (Uchida et al., 2011), in vitro P-gp transport activity, and P-gp protein level in human P-gp-transfected LLC-PK1 cell monolayers. Furthermore, the Kp brain in humans can also be predicted from in vitro data by integrating the Kp brain ratio with human fu,plasma and fu,brain. The prediction of Kp brain can be validated by comparison with in vivo data determined by positron emission tomography using available probes for P-gp substrates. However, human brain is usually obtained in a frozen state. Therefore, human fu,brain may not be accurately determined by the brain slice method, because the cells could be partially ruptured in the frozen brain. Alternatively, Summerfield et al. (2008) determined human fu,brain by the brain homogenate method with commercially available frozen human brain. Although fu,brain measured by the homogenate method alone is less relevant to the in vivo situation than that measured by the brain slice method (Fridén et al., 2007), the combination of the homogenate method with a pH partition model is likely to give a value of fu,brain that is more relevant to the in vivo situation (Fridén et al., 2011). Therefore, the homogenate method using a pH partition model might be useful for the reconstruction of human Kp brain.
The PPx-based reconstruction method can be applied to various animals, as well as humans, by using materials from the corresponding animals. P-gp protein levels in monkey brain capillaries have been determined (Ito et al., 2011). Therefore, the reconstruction method should be useful in evaluating interspecies differences in P-gp transport activities at the BBB and the brain distribution of P-gp substrates.
In conclusion, on the basis of BBB PPx, the present study experimentally demonstrated the in vitro to in vivo reconstruction of P-gp function at the mouse BBB (Kp brain ratio), and subsequently, the brain-to-plasma total drug concentration gradient (Kp brain). Furthermore, the brain-to-plasma unbound drug concentration gradient (Kp,uu,brain) could be reconstructed from the reciprocal of the reconstructed Kp brain ratio. This showed that the brain-to-plasma unbound drug concentration gradient is associated with the transport activity of P-gp at the BBB. Therefore, the present study illustrates well the value of quantitative targeted absolute proteomics for understanding in vitro/in vivo differences in the transport activities of membrane transporters and the mechanism governing the drug concentration gradient between plasma and brain. Although further study is necessary to clarify and reconstruct non-P-gp-mediated mechanisms, the present study is the first to offer a rational approach for the in vitro to in vivo prediction of drug penetration into the human brain.
Participated in research design: Uchida, Ohtsuki, Kamiie, and Terasaki.
Conducted experiments: Uchida.
Contributed new reagents or analytic tools: Kamiie.
Performed data analysis: Uchida.
Wrote or contributed to the writing of the manuscript: Uchida, Ohtsuki, and Terasaki.
We thank Dr. Alfred H. Schinkel for providing the L-mdr1a and the parental LLC-PK1 cells. We also thank A. Niitomi, Y. Yoshikawa, and K. Tsukiura for secretarial assistance.
This study was supported in part by Grants-in-Aid for Scientific Research (S) [KAKENHI: 18109002]; the Japan Society for the Promotion of Science (JSPS) Fellows [KAKENHI: 207291] from the JSPS; a Grant-in-Aid for Scientific Research on Priority Area [KAKENHI: 17081002] from The Ministry of Education, Culture, Sports, Science and Technology; and a Grant for Development of Creative Technology Seeds Supporting Program for Creating University Ventures from Japan Science and Technology Agency. This study was also supported in part by the Industrial Technology Research Grant Program from New Energy and the Industrial Technology Development Organization of Japan.
T.T. is a full professor and S.O. is an associate professor of Tohoku University, respectively, and are also directors of Proteomedix Frontiers. This research was not supported by Proteomedix Frontiers and their position at Proteomedix Frontiers does not present any financial conflicts.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
- central nervous system
- blood-brain barrier
- breast cancer resistance protein
- extracellular fluid
- unbound fraction in brain
- unbound fraction in plasma
- high-performance liquid chromatography
- Kp brain
- brain-to-plasma concentration ratio
- unbound brain-to-plasma concentration ratio
- liquid chromatography-tandem mass spectrometry
- multidrug resistance protein 1
- multidrug resistance protein 1a
- multiple reaction monitoring
- quantitative targeted absolute proteomics
- apparent permeability
- organic anion-transporting polypeptide 2
- wild type
- Madin-Darby canine kidney
- Received May 23, 2011.
- Accepted August 8, 2011.
- Copyright © 2011 by The American Society for Pharmacology and Experimental Therapeutics