This study investigated the impact of the active efflux mediated by P-glycoprotein (P-gp) and breast cancer resistance protein (Bcrp) at the blood-brain barrier (BBB) on the predictability of the unbound brain concentration (Cu,brain) by the concentration in the cerebrospinal fluid (CSF) (Cu,CSF) in rats. Cu,brain is obtained as the product of the total brain concentration and unbound fraction in the brain (fu,brain) determined in vitro in brain slices. Twenty-five compounds, including P-gp and/or Bcrp substrates, were given a constant intravenous infusion, and their plasma, brain, and CSF concentrations were determined. P-gp and/or Bcrp substrates, such as verapamil, loperamide, flavopiridol, genistein, quinidine, dantrolene, daidzein, cimetidine, and pefloxacin, showed a higher CSF-to-brain unbound concentration ratio (Kp,uu,CSF/brain) compared with non-P-gp and non-Bcrp substrates. Kp,uu,CSF/brain values of P-gp-specific (quinidine and verapamil) and Bcrp-specific (daidzein and genistein) substrates were significantly decreased in Mdr1a/1b(−/−) and Bcrp(−/−) mice, respectively. Furthermore, consistent with the contribution of P-gp and Bcrp to the net efflux at the BBB, Kp,uu,CSF/brain values of the common substrates (flavopiridol and erlotinib) were markedly decreased in Mdr1a/1b(−/−)/Bcrp(−/−) mice, but only moderately or weakly in Mdr1a/1b(−/−) mice and negligibly in Bcrp(−/−) mice. In conclusion, predictability of Cu,brain by Cu,CSF decreases along with the net transport activities by P-gp and Bcrp at the BBB. Cu,CSF of non-P-gp and non-Bcrp substrates can be a reliable surrogate of Cu,brain for lipophilic compounds.
For drugs that act in the central nervous system (CNS), it is assumed that an unbound drug in the interstitial fluid in the brain (Cu,brain) is available to interact with the target site in the CNS. The Cu,brain is therefore a surrogate for the bound concentration at the target site, and thus, evaluation of the Cu,brain is important for the quantitative interpretation of a drug's effect on the CNS. Because of the difficulty in measuring the Cu,brain of a drug in humans, a surrogate of the Cu,brain is indispensable in drug development. Drugs given systemically have to cross the blood-brain barrier (BBB) to reach the CNS. The BBB is formed by the brain capillary endothelial cells whose tight junctions between adjacent cells are highly developed. Furthermore, ATP-binding cassette (ABC) transporters, such as P-glycoprotein (P-gp/MDR1/ABCB1), breast cancer resistance protein (BCRP/ABCG2), and multidrug resistance-associated protein-4/ABCC4, limit the penetration of drugs into the CNS by extruding their substrate drugs into the blood in the BBB (Schinkel, 1999; Leggas et al., 2004; Belinsky et al., 2007; Enokizono et al., 2007, 2008; Ose et al., 2009). Because the BBB limits the free exchange of a drug between the CNS and blood, the unbound concentration of a drug in the plasma is hardly used as a surrogate of the Cu,brain (Liu et al., 2009; Watson et al., 2009).
The unbound concentration of a drug in the cerebrospinal fluid (CSF) (Cu,CSF) is commonly used as the surrogate of the Cu,brain in clinical and preclinical studies (Lin, 2008). CSF is produced by the choroid plexus in the ventricles and bathes the brain. It is a heterogeneous compartment that slowly turns over by bulk flow and has poor mixing. Ependyma, the interface between the CSF and CNS, is thought to allow the free exchange of drugs, whereas the barrier between the blood and CSF is formed by choroid epithelial cells (blood-cerebrospinal fluid barrier), which express xenobiotic efflux transport systems (Kusuhara and Sugiyama, 2004; Abbott, 2005). It is noteworthy that expression of P-gp and Bcrp is detected in the choroid epithelial cells; however, their expression in the choroid epithelial cells is suggested to be in the cytoplasm or subapical membrane for P-gp and in the brush border membrane for Bcrp (Rao et al., 1999; Zhuang et al., 2006). Unlike the BBB, a lack of P-gp and Bcrp in the plasma membrane of the choroid epithelial cells facing the blood suggests that their substrates can easily penetrate into the CSF. This can cause overestimation of the Cu,brain for P-gp and Bcrp substrates and their effects in the CNS when the Cu,CSF is used as a surrogate. Actually, the Cu,CSF of P-gp substrates such as 9-OH risperidone and Pfizer's proprietary compounds overestimated their Cu,brain, although there were some exceptional P-gp substrates' Cu,CSF that were similar to that of Cu,brain (Kalvass et al., 2002; Maurer et al., 2005; Liu et al., 2006). Maurer et al. (2005) reported that buspirone, caffeine, carbamazepine, midazolam, phenytoin, and zolpidem exhibited a large concentration difference (>3-fold) between the Cu,CSF and Cu,brain in mice. The impact of the active efflux at the BBB on the concentration difference between the Cu,CSF and Cu,brain has not been investigated systematically.
The primary purpose of the present study was to elucidate the impact of the active efflux by P-gp and Bcrp on the predictability of the Cu,brain using the Cu,CSF as a surrogate. The Cu,CSF and Cu,brain of 25 compounds, including P-gp and/Bcrp substrates, were determined in rats. To quantitatively evaluate the impact of the active efflux by P-gp and Bcrp at the BBB, the Cu,CSF and Cu,brain were compared in wild-type, Mdr1a/1b(−/−), Bcrp(−/−), and Mdr1a/1b(−/−)/Bcrp(−/−) mice.
Materials and Methods
Erlotinib and sertraline were purchased from Toronto Research Chemicals Inc. (North York, ON, Canada), and flavopiridol was kindly supplied by sanofi-aventis (Bridgewater, NJ). Antipyrine, cephalexin hydrate, cimetidine, daidzein, genistein, midazolam, risperidone, and zolpidem were purchased from Sigma-Aldrich (St. Louis, MO). Buspirone, citalopram, dantrolene, fleroxacin, and pefloxacin were purchased from LKT Labs (St. Paul, MN). Quinidine was purchased from Tokyo Kasei (Tokyo, Japan). Benzylpenicillin, caffeine, carbamazepine, diazepam, phenytoin, (±)-sulpiride, thiopental, and verapamil were purchased from Wako Pure Chemicals (Osaka, Japan). Loperamide hydrochloride was purchased from MP Biomedicals (Solon, OH). All other chemicals were commercially available and of reagent grade.
Male Sprague-Dawley rats were obtained from Charles River Laboratories Japan, Inc. (Kanagawa, Japan). Male wild-type FVB, Mdr1a/b(−/−), Bcrp(−/−), and Mdr1a/b(−/−)/Bcrp(−/−) mice were obtained from Taconic Farms (Germantown, NY) or provided by Kyoto Pharmaceutical University (Kyoto, Japan) or the Netherlands Cancer Institute (Amsterdam, The Netherlands). The rats used in the present study were 7 to 9 weeks old and weighed 220 to 320 g, and the mice were 10 to 18 weeks old and weighed 23 to 35 g. All animals were maintained at a controlled temperature under a 12-h light/dark cycle. Food and water were available ad libitum.
Determination of the Plasma, Brain, and CSF Concentrations in Rats and Mice.
Test compounds were given to rats and mice by a constant infusion after a bolus injection. Priming dose of the test drugs was determined based on pharmacokinetic analyses of the systemic elimination after intravenous bolus injection in rats and wild-type mice. Under urethane anesthesia (1.5 g/kg i.p.), the right and left jugular veins of the rats were cannulated with a polyethylene tube (SP-31; Natsume Seisakusyo, Tokyo, Japan). Compounds were administered via the cannula by a continuous infusion for 120 min. The infusion rates and priming doses in rats are summarized in Supplemental Table 1. Under urethane anesthesia (1.25 g/kg i.p.), the right jugular vein of the mice was cannulated with a polyethylene tube (PE-10; BD Biosciences, San Jose, CA). Compounds were administered via the cannula by continuous infusion for 120 min. The priming dose (μmol/kg) and infusion rate (μmol/h/kg), respectively, in mice were as follows: 2 and 4 for erlotinib, 3 and 16 for daidzein, 8 and 32 for flavopiridol, 3 and 16 for genistein, 20 and 25 for quinidine, and 6 and 8 for verapamil.
Blood samples in rats and mice were collected by using a heparinized syringe from the left jugular vein at the appropriate time points and centrifuged at 4°C and 10,000g for 5 min to obtain plasma. Immediately after the final blood sample was obtained, mice or rats were sacrificed by exsanguination. CSF samples were collected via cisterna magna puncture, and then brain samples were collected. Plasma, brain, and CSF samples were stored at −20°C until use.
Determination of Unbound Fraction in Plasma.
An equilibrium dialysis apparatus was used to determine the unbound fraction in plasma for each compound (Banker et al., 2003). Spectra/Por 2 membranes with molecular cutoff of 12 to 14 kDa, obtained from Spectrum Laboratories Inc. (Rancho Dominguez, CA), were used for the dialysis. Plasma was taken from male FVB mice and Sprague-Dawley rats. Plasma in mice was diluted with an equal volume of 0.067 M phosphate-buffered saline (PBS; pH 7.4). Plasma or diluted plasma was spiked with the test compound (1 μM), and 120-μl aliquots were loaded into the 96-well equilibrium dialysis apparatus and dialyzed versus 120 μl of PBS. The 96-well equilibrium dialysis apparatus was maintained on a rotator (set at 120 rpm) in an incubator at 37°C for 9 h for all test compounds except benzylpenicillin and cephalexin, which were maintained at 37°C for 4 h. Fifty microliters of either plasma or diluted plasma obtained from the apparatus was mixed with 50 μl of control buffer. Fifty microliters of buffer was mixed with 50 μl of control plasma or diluted plasma. The samples were then mixed with 100 μl of acetonitrile. The acetonitrile mixtures were vortexed, then centrifuged and stored at −20°C until analysis.
The unbound fractions (fu) were corrected using eq. 1 to yield an estimate of fu in the intact tissue. where D and fu′ represent the dilution factor of diluted plasma and unbound fraction determined in the diluted plasma, respectively.
Determination of Unbound Fraction Using Brain Slices.
Brain slices were prepared from rats and mice as reported previously with a minor modification (Kakee et al., 1997). A cortex slice, 300 μm thick, was cut using a brain microslicer (DTK-2000; Dosaka, Kyoto, Japan) and kept in oxygenated ECF buffer (122 mM NaCl, 25 mM NaHCO3, 10 mM d-glucose, 3 mM KCl, 1.4 mM CaCl2, 1.2 mM MgSO4, 0.4 mM K2HPO4, and 10 mM HEPES, pH 7.4) equilibrated with 95% O2/5% CO2. After preincubation for 5 min at 37°C, the brain slices were transferred to 3 ml of oxygenated incubation medium containing up to 10 compounds (0.1, 0.2, or 1 μM) and kept at 37°C. At designated times between 0.17 and 8 h, the ECF buffer and brain slices were collected. The brain slices were weighed and homogenized in 9 volumes (w/v) of deionized water with an ultrasonic probe. The ECF buffer and the brain slice homogenate were stored at −20°C until analysis. The adherent water volume was estimated by using 14C-carboxyl inulin in a separate experiment. Protein normalization was performed to correct for dilution of brain homogenate. Protein concentrations of the brain homogenate samples were determined by the method of Lowry et al. (1951). The weights of brain slices were calculated by comparing the protein concentration of the brain homogenate with a control brain homogenate containing a known weight of brain. The ATP content of the slices to confirm the viability was determined using a Tissue ATP Assay Kit (TA100) from Wako Pure Chemicals.
Determination of Transcellular Transport Across Monolayers of Cell Lines Expressing P-gp or Bcrp.
In vitro mouse Bcrp transport experiments were performed as reported previously (Enokizono et al., 2007, 2008). In brief, MDCK II cells were seeded into 24-well Transwell plates (Corning Life Sciences, Lowell, MA) at a density of 1.4 × 105 cells/well and grown for 2 days in Dulbecco's modified Eagle's medium (Invitrogen, Carlsbad, CA) with 10% fetal bovine serum (Sigma-Aldrich) and 1% antibiotic-antimycotic solution (Sigma-Aldrich). The cells were infected with recombinant adenovirus harboring green fluorescent protein (GFP) or mouse Bcrp expression vector at a 200× multiplicity of infection. Details of the construction of these recombinant adenoviruses have been described previously (Ando et al., 2007). After 2 days in culture, GFP-expressing cells (MDCK II/GFP) and Bcrp-expressing cells (MDCK II/Bcrp) were used for transport studies. In vitro transport experiments to determine the transport activity by mouse Mdr1a were conducted using Mdr1a-expressing LLC-PK1 cells (L-Mdr1a) that had been established previously (Schinkel et al., 1995). Transport rates were calculated from the slopes of the time profiles of the apical-to-basal and basal-to-apical transport.
Quantification of Compounds in the Biological Samples.
Rat brain specimen was homogenized with 2 volumes of PBS, pH 7.4, to obtain a 33% homogenate. The proteins in rat plasma and brain homogenate were precipitated with 2 volumes of acetonitrile, and that in rat CSF samples was precipitated with one-half volume of acetonitrile. Then the specimens were centrifuged twice at 4°C and 10,000g for 5 min. The supernatants were evaporated, and the remaining residues were reconstituted in mobile phase and subjected to liquid chromatography (LC)-mass spectrometry (MS) or LC-tandem MS (MS/MS) analysis.
The plasma samples from mice were diluted with 2 volumes of PBS, pH 7.4. Mouse brain specimen was homogenized with 3 volumes of PBS, pH 7.4, to obtain a 25% homogenate. The proteins in these samples were precipitated with 2 volumes of acetonitrile, and the suspensions were centrifuged twice at 4°C and 10,000g for 5 min. The supernatants were evaporated, and the remaining residues were reconstituted in mobile phase and subjected to LC-MS(/MS) analysis. Mouse CSF samples were mixed with 3 volumes of PBS and an equal volume of acetonitrile and were directly injected on LC-MS(/MS).
The media (100 μl) mixed with 50 μl of acetonitrile or PBS in the in vitro transport studies were centrifuged at 4°C and 10,000g for 5 min. The protein in brain slice and ECF buffers were precipitated with 2 volumes of acetonitrile, and the suspensions were centrifuged twice at 4°C and 10,000g for 5 min. The supernatant solvents were evaporated, and the remaining residues were reconstituted in mobile phase and subjected to LC-MS(/MS) analysis. The acetonitrile mixtures in equilibrium dialysis studies were directly injected on LC-MS(/MS).
All compounds were analyzed in a multiple reaction monitoring mode using an API2000 instrument (Applied Biosystems, Foster City, CA) equipped with an Agilent 1100 series LC system (Agilent Technologies, Santa Clara, CA) or a selected ion monitoring mode using a LCMS-2010 EV equipped with a Prominence LC system (Shimadzu, Kyoto, Japan). A CAPCELL PAK C18 MGII column (3 μm, 3-mm i.d. × 35 mm; Shiseido, Kanagawa, Japan) or CAPCELL PAK C18 MG column (3 μm, 2-mm i.d. × 50 mm) was used at room temperature in the analysis of all compounds. The details of LC conditions and mass-to-charge ratios are shown in Supplemental Table 2.
The volume of distribution in brain slices (Vu,brain) was obtained by dividing the amount of the compound associated with the brain slice specimens by the concentration of drug in ECF buffer and corrected by the adherent water volume. The fu,brain value was defined as the reciprocal of Vu,brain. Because brain distribution in the slices for loperamide and sertraline did not achieve equilibrium under the experimental condition, their Vu,brain in equilibrium condition were obtained by fitting two- and three-compartment models to experimental data on Vu,brain using a nonlinear least-squares method (WinNonlin ver. 5.2.1; Pharsight, Mountain View, CA), respectively.
Brain concentrations (Cbrain) of the test compounds were calculated according to the method of Fridén et al. (2009) for correction of drug in the residual blood of brain vascular spaces. Brain-to-plasma unbound concentration ratio (Kp,uu,brain) was obtained by dividing the unbound concentration in the brain by that in the plasma. Unbound concentrations were defined as the product of the total concentrations and unbound fraction. Kp,uu,brain was calculated as follows: where Cp, fu,brain, and fp represent plasma concentration and unbound fractions in the brain and plasma, respectively.
CSF-to-plasma unbound concentration ratios (Kp,uu,CSF) were calculated as follows: where CCSF and fu,CSF represent, respectively, CSF concentration and the unbound fraction in the CSF, which was calculated from fp according to the method of Fridén et al. (2009).
CSF-to-brain unbound concentration ratios (Kp,uu,CSF/brain) were calculated as follows:
The presented values are all mean ± S.E.M. For comparison between genotype groups in FVB, Mdr1a/b(−/−), Bcrp(−/−), and Mdr1a/b(−/−)/Bcrp(−/−) mice, log-transformed data were processed by using a one-way analysis of variance, followed by a Tukey post hoc test. For comparison between two groups, Student's two-tailed t test was used. Differences were considered significant at P < 0.05. All statistical calculations were performed using SAS software (version 9; SAS Institute, Cary, NC).
Determination of Unbound Fractions in Brain Slices.
The fu,brain values of the test compounds were determined in brain slices (Fig. 1). The ATP levels in the slice specimens were monitored during the incubation period to confirm the viability of the slices throughout the experiment. The ATP content in the brain slices did not decrease even after the 8-h incubation (Supplemental Fig. 1). The Vu,brain of all compounds except loperamide and sertraline reached a plateau during the 8-h incubation. The Vu,brain values of loperamide and sertraline under steady-state conditions were obtained by fitting a two-compartment and three-compartment model to the experimental data using a nonlinear least-squares method, respectively. The fu,brain values of the test compounds are summarized in Table 1.
Determination of CSF-to-Brain Unbound Concentration Ratios of 25 Compounds in Rats.
Twenty-five compounds were administered individually for 2 h by continuous infusion in rats. The concentrations in plasma samples were determined during the infusion and reached a plateau at 2 h (Fig. 2). The concentrations of the test compounds in the brain and CSF were determined at 2 h and used to calculate Kp,uu,brain, Kp,uu,CSF, and Kp,uu,CSF/brain (Table 1). The Kp,uu,CSF/brain values of the 25 compounds ranged from 0.48 to 4 (Fig. 3). The Cu,CSF values of verapamil, loperamide, flavopiridol, genistein, quinidine, dantrolene, daidzein, cimetidine, and sulpiride were 2-fold greater than the Cu,brain (Kp,uu,CSF/brain >2). Unlike the report by Maurer et al. (2005) in mice, buspirone, caffeine, carbamazepine, midazolam, phenytoin, and zolpidem did not show any concentration difference in this study.
Determination of Transcellular Transport of the Test Compounds Across the Monolayers of Control Cell Lines and Cells Expressing Either Mdr1a or Bcrp.
The transcellular transport in the basal-to-apical and apical-to-basal directions was determined in the cells expressing mouse Mdr1a or Bcrp (L-Mdr1a or MDCK II/Bcrp) and their corresponding mock cells. The time profiles of the directional transport in the basal-to-apical and apical-to-basal direction for the 25 compounds are shown in Supplemental Figs. 2 and 3. The transport rates in the basal-to-apical and apical-to-basal directions are summarized in Tables 2 and 3. The compounds, whose transport rates differed significantly between the control and P-gp- or Bcrp-expressing cells, were identified as substrates. Flavopiridol, loperamide, risperidone, quinidine, and verapamil showed higher transport rates in the basal-to-apical direction but lower transport rates in the apical-to-basal direction in L-Mdr1a cells compared with the mock LLC-PK1 cells. The transport rates of pefloxacin in L-Mdr1a cells showed a significant difference only in the basal-to-apical direction compared with mock LLC-PK1 cells. These compounds are substrates of P-gp. The transcellular transport of other compounds was nondirectional in L-Mdr1a cells compared with mock LLC-PK1 cells. In MDCK II/mBcrp cells, cimetidine, dantrolene, flavopiridol, and genistein showed higher transport rates in the basal-to-apical direction but lower transport rates in the apical-to-basal direction compared with mock MDCK II cells. The transport rates of daidzein, fleroxacin, and pefloxacin were significantly greater in MDCK II/mBcrp cells than in mock MDCK II cells only in the basal-to-apical direction. These compounds are substrates of Bcrp. Flavopiridol and pefloxacin are common substrates of P-gp and Bcrp.
Determination of CSF-to-Brain Unbound Concentration Ratios in Wild-Type, Mdr1a/1b(−/−), Bcrp(−/−), and Mdr1a/1b(−/−)/Bcrp(−/−) Mice.
The Kp,uu,CSF/brain values of P-gp substrates (quinidine and verapamil), Bcrp substrates (daidzein and genistein), and their common substrates (erlotinib and flavopiridol) were compared between wild-type and knockout mice. The fu,brain values of the test compounds were determined in brain slices of wild-type mouse (Fig. 4A). The fu,brain values of quinidine and verapamil did not differ between wild-type mice and Mdr1a/1b(−/−) mice, and the fu,brain values of daidzein and genistein did not differ between wild-type mice and Bcrp(−/−) mice (Supplemental Fig. 4). Test compounds were given by continuous infusion after a bolus injection. The concentrations in the plasma were determined during infusion and reached a plateau at 2 h (Fig. 4B). The brain and CSF concentrations were determined at 2 h after administration (Table 4). Cu,brain was calculated as the product of the total concentration in the brain and fu,brain. The Kp,uu,CSF/brain values of quinidine and verapamil were greater in wild-type mice than in Mdr1a/1b(−/−) mice, and the Kp,uu,CSF/brain values of daidzein and genistein were greater in wild-type mice than in Bcrp(−/−) mice (Fig. 4C). The Kp,uu,CSF/brain values of these compounds in the knockout strain was 1. The concentrations of erlotinib and flavopiridol in the plasma and brain were obtained from our previous report (Kodaira et al., 2010), and their CSF concentrations had been determined in the same animals (Table 4). The Kp,uu,CSF/brain values of erlotinib and flavopiridol were more than 1 in wild-type mice and decreased only weakly or moderately in Mdr1a/1b(−/−) mice and negligibly in Bcrp(−/−) mice. A simultaneous defect of P-gp and Bcrp significantly reduced the Kp,uu,CSF/brain of erlotinib and flavopiridol to almost 1 (Fig. 4D).
In the present study, the impact of the active efflux by P-gp and Bcrp at the BBB on the concentration difference between the brain and CSF was investigated quantitatively in rats and mice. Our findings will contribute to the understanding of the discrepancy in the effect of P-gp and Bcrp substrates in the CNS when the Cu,CSF is used as a surrogate of the Cu,brain in drug discovery and development studies.
The Cu,brain was obtained as a product of the total drug concentrations in the whole brain and fu,brain. Brain slices and brain homogenate have been used to determine the fu,brain in vitro (Kalvass and Maurer, 2002; Fridén et al., 2007). Acidic and basic compounds show relatively large differences in the fu,brain determined by the brain slice and homogenate methods because of the low membrane permeability of anionic compounds and the large distribution of cationic compounds into the intracellular acidic compartments (Fridén et al., 2011). A similar tendency was observed in this study (Supplemental Fig. 5). Because the brain slice method produces a better prediction of the unbound distribution volumes in the brain (Fridén et al., 2007), the fu,brain determined in brain slices was used for further calculations in this study.
Among the 25 compounds tested, P-gp and/or Bcrp substrates such as verapamil, loperamide, flavopiridol, genistein, quinidine, dantrolene, daidzein, cimetidine, and pefloxacin showed higher Kp,uu,CSF/brain values compared with non-P-gp and -Bcrp substrates (Fig. 3). P-gp- and/or Bcrp-mediated efflux at the BBB is a determinant of the concentration of the substrate compounds in the brain (Schinkel et al., 1994; Enokizono et al., 2008; Kodaira et al., 2010) except for cimetidine, whose Cbrain is unaffected by the defect of Bcrp (Zhao et al., 2009). Therefore, the active efflux by P-gp and/Bcrp at the BBB must be the mechanism underlying the higher Kp,uu,CSF/brain. To confirm this, Cu,brain and Cu,CSF were determined in the knockout mice. In wild-type mice, the Kp,uu,CSF/brain values of the test compounds ranged from 1.4 to 3.6 (Fig. 4, C and D). A defect of P-gp or Bcrp significantly increased both the Kp,brain and Kp,CSF (Table 4). Because the increase was greater for Kp,brain than for Kp,CSF, the Kp,uu,CSF/brain value of P-gp- and Bcrp-specific substrates was significantly lower in Mdr1a/1b(−/−) and Bcrp(−/−) mice, respectively, compared with wild-type mice (Fig. 4C). Furthermore, for the common substrates whose efflux at the BBB is mediated by both P-gp and Bcrp (Kodaira et al., 2010), the Kp,uu,CSF/brain decreased significantly to 1 in Mdr1a/1b(−/−)/Bcrp(−/−) mice (Fig. 4D). The Kp,uu,CSF/brain of flavopiridol also decreased moderately in Mdr1a/1b(−/−) mice, whereas that of erlotinib decreased slightly (Fig. 4D). The Kp,uu,CSF/brain of both drugs was unchanged in Bcrp(−/−) mice. The difference in the effect of impaired P-gp and Bcrp on the Kp,uu,CSF/brain of erlotinib and flavopiridol is attributable to the contribution of P-gp and Bcrp to the net efflux at the BBB (Kodaira et al., 2010). An increase in the Kp,CSF caused by the defect of P-gp and/or Bcrp is probably attributable to the increase in the influx rate from the brain parenchyma across the ependyma interface. That is, the exchange across the ependyma interface is a determinant of the CSF concentration of these lipophilic compounds. This provides a rationale for using Cu,CSF as a surrogate of Cu,brain, although the net efflux activity at the BBB affects the predictability of Cu,brain using Cu,CSF.
There are some discrepancies between the in vitro/in vivo transport activities by P-gp and Bcrp and Kp,uu,CSF/brain. Hydrophilic organic cations such as cimetidine and sulpiride showed a relatively high Kp,uu,CSF/brain despite the limited impact of P-gp and Bcrp on their Cbrain values (Doran et al., 2005; Zhao et al., 2009). Cimetidine is actively eliminated from the CSF by the choroid plexus via the organic anion transport system, and its inhibition by probenecid significantly increased the CSF-to-plasma ratio of cimetidine in rats but did not have an effect on the brain-to-plasma ratio (Nagata et al., 2004). Therefore, the contribution of the exchange across the ependyma interface on the Cu,CSF may be limited for such a hydrophilic compound. Therefore, prediction of the Cu,brain using the Cu,CSF as a surrogate requires quantitative evaluation of not only the BBB transport, but also the influx and efflux transport across the choroid plexus and bulk flow. There is no information on the transport systems for sulpiride at the BBB and blood-cerebrospinal fluid barrier. Because the physicochemical properties of sulpiride (cLogD and parallel artificial membrane permeability assay permeability) are similar to those of cimetidine (cationic charge at pH 7.4, cLogD −1.49 versus −0.45, and parallel artificial membrane permeability assay 0.0421 versus 0.0584 × 10−6 cm/s; H. Kodaira, unpublished data), it is also possible that transporters are involved in the CSF disposition of sulpiride.
Because risperidone undergoes extensive efflux at the BBB (Doran et al., 2005), its Kp,uu,CSF/brain value is more then 1. However, the observed value was lower than expected from its transport activity by P-gp. Despite the similar transport activity by P-gp (Table 2), the Kp,uu,CSF/brain value of risperidone was 2.6-fold lower than that of verapamil (Table 1). The Kp,uu,CSF value of risperidone was lower than that of verapamil, whereas the Kp,uu,brain values were similar. Another transporter may limit the penetration of risperidone into the CSF. Maurer et al. (2005) and Doran et al. (2005) found that the Cu,brain value of risperidone in the brain of Mdr1a/1b(−/−) mice was higher than that in the plasma, suggesting that the BBB transport of risperidone involves active uptake at the BBB. This may compete with the efflux at the BBB, thereby lowering the Kp,uu,CSF/brain.
It was reported that the Cu,CSF overestimated the Cu,brain for 9-OH risperidone, quinidine, risperidone, and thiopental (Liu et al., 2009) and for loperamide, morphine, its glucuronide conjugates (M3G and M6G), nelfinavir, rifampicin, and verapamil (Fridén et al., 2009). Consistent with this study, those results included P-gp substrates such as 9-OH risperidone, quinidine, risperidone, loperamide, morphine, nelfinavir, rifampicin, and verapamil. The drugs tested by Fridén et al. (2009) included some Bcrp substrates, such as levofloxacin and norfloxacin (Merino et al., 2006), methotrexate (Suzuki et al., 2003), nitrofurantoin (Merino et al., 2005), and sulfasalazine (van der Heijden et al., 2004), whose Kp,uu,CSF/brain values were close to 1. For levofloxacin and norfloxacin, the Bcrp activity may not be sufficient to have an impact on the Kp,uu,CSF/brain, as in the case of fleroxacin in this study. For methotrexate, nitrofurantoin, and sulfasalazine, because of their small distribution volume in the brain (essentially the capillary space), the calculated Cu,brain is not reliable. The relationship between Bcrp transport activity and its impact on the brain and CSF concentrations should be investigated in the future.
Maurer et al. (2005) and Doran et al. (2005) investigated the effect of P-gp on the Cbrain and CCSF of drugs in mice. There are two discrepancies between these reports and this study. First, in the previous reports, the outliers included quinidine and verapamil. The Kp,uu,CSF/brain values of P-gp substrates such as metoclopramide, risperidone, loperamide, quinidine, and verapamil were less than 2 in wild-type mice, whereas those of morphine and 9-OH risperidone were 2.8 and 8.6, respectively. The Kp,uu,CSF values of quinidine and verapamil were 4- and 10-fold higher in this study than in the previous report. Second, in the previous report, the defect of P-gp had almost no effect on the Kp,uu,CSF/brain of P-gp substrates. The Kp,uu,CSF/brain values of morphine and 9-OH risperidone decreased to only 2.0 and 4.9 in Mdr1a/1b(−/−) mice, respectively, although their Kp, brain values were increased 1.7-and 17-fold in Mdr1a/1b(−/−) mice, respectively. The reason for this discrepancy is unclear. The previous reports determined the Kp values in the brain and CSF based on the area under the curve 5 h after subcutaneous drug administration, but the samples from many mice were pooled to create the time profile. By contrast, we measured these parameters under the steady-state condition. Theoretically, both methods should provide the same value, but in practice, interanimal differences may lead to overestimation.
In conclusion, the ability to predict the Cu,brain using the Cu,CSF decreases along with the net transport activities by P-gp and Bcrp at the BBB. The Cu,CSF of non-P-gp and non-Bcrp substrates may be a reliable surrogate for Cu,brain for lipophilic compounds.
Participated in research design: Kodaira, Kusuhara, and Sugiyama.
Conducted experiments: Kodaira.
Contributed new reagents or analytic tools: Fujita.
Performed data analysis: Kodaira, Kusuhara, and Sugiyama.
Wrote or contributed to the writing of the manuscript: Kodaira, Kusuhara, Ushiki, Fuse, and Sugiyama.
We thank Dr. Alfred H. Schinkel (The Netherlands Cancer Institute, Amsterdam, The Netherlands) for supplying L-Mdr1a cells and Bcrp(−/−) mice, and sanofi-aventis (Bridgewater, NJ) for supplying flavopiridol.
This study was supported in part by the Japan Society for the Promotion of Science [Grant-in-Aid for Scientific Research (A) 20249008 (to Y.S.) and Grant-in-Aid for Scientific Research (B) 23390034 (to H.K.)].
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
- central nervous system
- blood-brain barrier
- cerebrospinal fluid
- total drug concentration in cerebrospinal fluid
- total drug concentration in brain
- unbound drug concentration in cerebrospinal fluid
- unbound drug concentration in brain
- brain-to-plasma drug concentration ratio
- cerebrospinal fluid-to-plasma drug concentration ratio
- brain-to-plasma unbound drug concentration ratio
- cerebrospinal fluid-to-plasma unbound drug concentration ratio
- cerebrospinal fluid-to-brain unbound drug concentration ratio
- multidrug resistance protein
- breast cancer resistance protein
- Madin-Darby canine kidney
- green fluorescent protein
- LLC-PK1 cells expressing mouse Mdr1a
- liquid chromatography
- tandem mass spectrometry
- phosphate-buffered saline
- ATP-binding cassette.
- Received February 8, 2011.
- Accepted September 19, 2011.
- Copyright © 2011 by The American Society for Pharmacology and Experimental Therapeutics