Article pubs.acs.org/molecularpharmaceutics
Quantification of Transporter and Receptor Proteins in Dog Brain Capillaries and Choroid Plexus: Relevance for the Distribution in Brain and CSF of Selected BCRP and P‑gp Substrates Clemens Braun,† Atsushi Sakamoto,‡ Holger Fuchs,† Naoki Ishiguro,‡ Shinobu Suzuki,‡ Yunhai Cui,† Klaus Klinder,† Michitoshi Watanabe,§,∥ Tetsuya Terasaki,∥ and Achim Sauer*,† †
Drug Discovery Sciences, Boehringer Ingelheim Pharma GmbH & Co. KG, 88397 Biberach an der Riss, Germany Kobe Pharma Research Institute, Nippon Boehringer Ingelheim Co., Ltd., Kobe 650-0046, Japan § Proteomedix Frontiers Co., Ltd, T-Biz, 6-6-40 Aoba, Aramaki, Aoba-ku, Sendai 980-8579, Japan ∥ Division of Membrane Transport and Drug Targeting, Graduate School of Pharmaceutical Sciences, Tohoku University, 6-3 Aoba, Aramaki, Aoba-ku, Sendai 980-8578, Japan ‡
S Supporting Information *
ABSTRACT: Transporters at the blood-brain barrier (BBB) and the blood-cerebrospinal fluid barrier (BCSFB) play a pivotal role as gatekeepers for efflux or uptake of endogenous and exogenous molecules. The protein expression of a number of them has already been determined in the brains of rodents, nonhuman primates, and humans using quantitative targeted absolute proteomics (QTAP). The dog is an important animal model for drug discovery and development, especially for safety evaluations. The purpose of the present study was to clarify the relevance of the transporter protein expression for drug distribution in the dog brain and CSF. We used QTAP to examine the protein expression of 17 selected transporters and receptors at the dog BBB and BCSFB. For the first time, we directly linked the expression of two efflux transporters, P-glycoprotein (P-gp) and breast cancer resistance protein (BCRP), to regional brain and CSF distribution using specific substrates. Two cocktails, each containing one P-gp substrate (quinidine or apafant) and one BCRP substrate (dantrolene or daidzein) were infused intravenously prior to collection of the brain. Transporter expression varied only slightly between the capillaries of different brain regions and did not result in region-specific distribution of the investigated substrates. There were, however, distinct differences between brain capillaries and choroid plexus. Largest differences were observed for BCRP and P-gp: both were highly expressed in brain capillaries, but no BCRP and only low amounts of P-gp were detected in the choroid plexus. Kp,uu,brain and Kp,uu,CSF of both P-gp substrates were indicative of drug efflux. Also, Kp,uu,brain for the BCRP substrates was low. In contrast, Kp,uu,CSF for both BCRP substrates was close to unity, resulting in Kp,uu,CSF/Kp,uu,brain ratios of 7 and 8, respectively. We conclude that the drug transporter expression profiles differ between the BBB and BCSFB in dogs, that there are species differences in the expression profiles, and that CSF is not a suitable surrogate for unbound brain concentrations of BCRP substrates in dogs. KEYWORDS: BCRP, P-gp, blood-brain barrier, blood-CSF barrier, choroid plexus, efflux transport, dog, canine, transporter expression, QTAP
■
INTRODUCTION
The relevance of drug transporters for drug distribution and elimination and thus for the pharmacokinetic profiles of compounds and their persistence at target sites is well accepted.1−3 In the brain, there are two major types of transporters involved in xenobiotic molecule disposition: ATPbinding cassette (ABC) and solute carrier (SLC) transporters. Among the ABC transporters, P-glycoprotein (P-gp) and breast
The blood-brain barrier (BBB) and blood-cerebrospinal fluid barrier (BCSFB) are major barrier functions in the central nervous system (CNS). The BBB is localized at the level of the cerebral capillary endothelial cells. Once an endogenous or xenobiotic molecule has crossed the BBB, diffusion distances to neuron and glial cells are short. The BBB is the physiological structure limiting access of such molecules to the brain. The BCSFB, comprising mainly choroid plexus (CP) epithelial cells, is the barrier separating the blood from the cerebrospinal fluid (CSF). © XXXX American Chemical Society
Received: Revised: Accepted: Published: A
May 30, 2017 August 8, 2017 September 7, 2017 September 7, 2017 DOI: 10.1021/acs.molpharmaceut.7b00449 Mol. Pharmaceutics XXXX, XXX, XXX−XXX
Article
Molecular Pharmaceutics
with human MDR1 P-gp (data not shown) indicating that they are no substrates for human and canine P-gp.
cancer resistance protein (BCRP) are expressed on the luminal membrane of brain capillary endothelial cells and maintain an efflux transport of substrate molecules. Thereby they can limit the distribution of drugs to the brain.4 Among SLC transporters, OATP1A2 is also expressed on the luminal membrane of BBB and plays a role in the transfer of drugs from the blood to the brain.5,6 Uchida et al.7 reported that the unbound brain-to-plasma concentration ratios (Kp,uu,brain) of P-gp substrates in the mouse can be reconstructed from P-gp protein expression, intrinsic transporter activity (transport function per P-gp molecule), and the unbound fraction in plasma and brain. Indeed, knowledge of the expression levels of drug transporters at the BBB is useful for predicting the pharmacokinetics of administered drugs and can be an essential factor for reconstructing the in vivo penetration of drugs in CNS when combined with intrinsic transporter activity data. To date, quantitative drug transporter protein expression data in the BBB have been published for the mouse,8,9 rat,10 marmoset,10 cynomolgus monkey,11 and humans.12 However, assessments of transporter and receptor expression in dog brain are lacking, even though the dog is widely used as a nonrodent species in nonclinical drug development and in particular for nonclinical drug safety studies. This information would be important for the interpretation of toxicological data and for the extrapolation of new drug pharmacokinetics to humans. Second, since it is not possible to collect brain samples in clinical studies, CSF is still used for estimation of human central drug exposure despite the discussed limitations of this matrix.13 Further, preclinical animal CSF data are used to extrapolate to humans. However, little is known about drug transporter expression and functioning at the BCSFB, its relevance for CSF drug exposure, and about possible species differences. The purpose of this study was therefore to determine the absolute protein levels of selected transporters and receptors both in dog BBB and BCSFB by using mass spectrometrybased protein quantification as described.9 Our selection of transporters and receptors were based on the relevance for drug brain disposition, the brain capillary expression data available from humans and other animal species, as well as on the feasibility of getting respective probes for dogs. Beside efflux transporters we were also interested in specific uptake transporters, carrier mediated transporters like GLUT14,15 and targets for receptor mediated transport to brain,15,16 which led to the selection of TfR1, LRP1, and INSR. Expression levels were compared with those reported for other species, including humans, to further clarify species similarities and differences. Additionally, the relevance of the expression of the two most important efflux transporters P-gp and BCRP on central distribution in brain and CSF was investigated by quantification of specific transporter substrates for P-gp (quinidine17−19 and apafant20,21) and BCRP (daidzein18,22 and dantrolene18,19,22,23). Selection of P-gp and BCRP substrates for this study was based on the following aspects: Apafant and quinidine are wellcharacterized P-gp substrates,17−21 both compounds showed also efflux in parenteral MDCK II cells which express endogenous canine P-gp (data now shown), indicating that both compounds are also substrates of canine P-gp. Little is known about the substrate specificity of canine BCRP, however, in vitro and in vivo data indicate that dantrolene and daidzein are both human and murine BCRP substrates 24 both compounds showed no efflux in MDCK II cells transfected
■
EXPERIMENTAL SECTION In Vivo Part. Animals. Four male (5.0 to 10.7 years of age) and two female (6.7 and 7.6 years of age) Beagle dogs were obtained from Boehringer Ingelheim (Biberach, Germany) and BASF (Ludwigshafen, Germany) (Table S1). All animal care and experimental procedures were approved by the local German authorities and were in compliance with the German and European Animal Welfare Acts. Compounds and Formulation of Cocktails. Cocktail 1 containing the P-gp substrate apafant, the BCRP substrate dantrolene plus the nontransporter substrate antipyrine and cocktail 2 containing the P-gp substrate quinidine, the BCRP substrate daidzein plus antipyrine were administered by intravenous (i.v.) infusion to 3 dogs each. The transporter substrates quinidine and dantrolene were obtained from SigmaAldrich (Sigma-Aldrich Chemie GmbH, Germany), daidzein from TCI (TCI Deutschland GmbH, Germany), and antipyrine from Acros Organics (New Jersey, US via Fisher Scientific GmbH, Germany. WEB2086 (apafant) was synthesized by Boehringer Ingelheim Pharma GmbH & Co. KG, Germany. The formulation for cocktail 1 including apafant, dantrolene, and antipyrine for the i.v. loading dose and infusion was an aqueous solution containing 10% hydroxypropyl- beta-cyclodextrin and 0.03% glucose. The corresponding aqueous formulation for cocktail 2, including quinidine, daidzein, and antipyrine, was an aqueous solution containing 5% hydroxypropyl-beta-cyclodextrin and 0.04% glucose. For both formulations, the pH was adjusted to 8.9 with HCl or NaOH. The osmolality was 0.30 to 0.32 osmol/kg. All chemicals were commercial products of analytical grade. Anaesthesia and Euthanasia. Dogs were pretreated with 0.01 to 0.02 mg/kg intramuscular acepromazin (Vetranquil 1% solution, Ceva Tiergesundheit GmbH, Germany). Anaesthesia was induced with an i.v. bolus injection of 0.5 mg/kg midazolam (Dormicum, Roche AG, Switzerland), followed by a 4 mg/kg i.v. bolus of propofol (Propofol-Lipuro , B. Braun Melsungen AG, Germany). Animals were intubated and ventilated. Anaesthesia was maintained using an infusion of 24 mg/kg/h propofol. Subsequently, the 60 min infusions of the transporter substrate cocktails were started. Finally animals were euthanized with an i.v. bolus of 560 mg pentobarbital (3.5 mL Narcoren, Merial GmbH, Germany) followed by 20 to 35 mL of 3 molar potassium chloride solution. Death was confirmed by cardiac arrest. Infusion Scheme. For planning of the infusion scheme, pharmacokinetic profiles of the four transporter substrates and antipyrine were generated by i.v. bolus administration of the respective compound cocktails in the same dogs as used later for the infusion study. The key pharmacokinetic parameters determined by noncompartmental analysis are given in Table S2, and were used to estimate i.v. bolus loading doses and infusion rates (Table S3). The bolus loading doses were chosen to immediately reach the planned steady state plasma concentrations and the infusion rates were selected to maintain the planned plasma concentrations. Sample Collection. Brain and CSF: A skin incision was performed from the frontal bone to the neck, and dorsal neck muscles were carefully retracted in order to gain access to the cisterna magna via the atlanto-occipital ligament. Subsequently, approximately 150 μL CSF was collected by puncturing the
B
DOI: 10.1021/acs.molpharmaceut.7b00449 Mol. Pharmaceutics XXXX, XXX, XXX−XXX
Article
Molecular Pharmaceutics
solution B was then added to the suspension at a final concentration of 17.5% (v/v), and the mixture was centrifuged at 4400 × g for 15 min. The pellet was suspended in solution A (solution B containing 25 mM NaHCO3, 10 mM glucose, 1 mM pyruvate, and 0.5% (w/v) bovine serum albumin) and was passed through 210-, 85-, 45-, and 20-μm nylon mesh and the residue fractions of 85-, 45-, and 20-μm were suspended and collected in solution A as microvessel fractions. The resulting microvessel fractions were centrifuged at 1000 × g for 5 min. The pellet was suspended in solution B, and centrifuged at 1000 × g for 5 min. This process was repeated twice. Lysate Preparation from Brain Capillaries and CP and Protein Digestion. The pellets of brain capillaries and removed CP were suspended with hypotonic buffer (10 mM Tris-HCl, 10 mM NaCl, and 1.5 mM MgCl2, pH 7.4) and then sonicated. Protein concentration was measured using the Lowry method and the DC protein assay reagent (Bio-Rad Laboratories, Hercules, CA, USA) and lysates were stored at −80 °C. Lysates were dissolved in denaturing buffer [7 M guanidium hydrochloride, 500 mM Tris-HCl (pH 8.5), 10 mM EDTA], and proteins were S-carbamoylmethylated (Kamiie et al. 2008). Alkylated proteins were then precipitated in a mixture of methanol and chloroform; precipitates were dissolved in 6 M urea, and diluted 5-fold with 100 mM Tris-HCl (pH 8.5). ProteaseMax surfactant (Promega, Madison, Wisconsin) and lysyl endopeptidase (Wako Pure Chemical Industries, Osaka, Japan) were added at 0.05% final concentration, and an enzyme/substrate ratio of 1:100, respectively. This mixture was incubated at 25 °C for 3 h, followed by digestion with TPCKtreated trypsin (Promega, Madison, Wisconsin) at an enzyme/ substrate ratio of 1:100 at 37 °C for 16 h. Quantitative Targeted Absolute Proteomics. All peptides used for LC-MS/MS analysis were selected according to the in silico selection criteria reported previously9 and were synthesized by the Thermo Electron GmbH, Ulm, Germany) at a purity of >95%. Concentrations of peptide solutions were determined by quantitative amino acid analysis (Lachrom Elite, Hitachi, Tokyo, Japan). LC/MS/MS-based Protein Quantification. All samples were analyzed using a nano-LC system after the addition of internal standards (Ultimate 3000; Dyonex, Amsterdam, The Netherlands). The system was connected to an ESI triple quadrupole mass spectrometer (QTRAP5500; AB Sciex, Foster City, CA, USA), and the nano-LC system comprised a trapping column (L-column Micro, 300 μm inner diameter, 5 mm length, 5 μm particles; Chemicals Evaluation and Research Institute, Tokyo, Japan), and a separation column (L-column Micro, 100 μm inner diameter, 100 mm length, 3 μm particles, Chemicals Evaluation and Research Institute, Tokyo, Japan). A linear gradient was used from 0% to 45% acetonitrile in 0.1% formic acid to elute peptides at a flow rate of 200 nL/min, and the spectrometer was set up to perform MRM analysis for peptide detection with a 10 ms dwell time per channel. Calculation of Protein Expression Levels. Four MRM transitions were set for each target peptide and its internal standard because a single transition may be hidden by nontarget components (Table S4). Ion counts in chromatograms were determined by data acquisition procedures using Analyst software 1.5.1 (AB Sciex, Foster City, CA, USA). When peaks were detected in three or four channels, target molecules were judged to be expressed, and the protein expression level (fmol/μg protein) of the corresponding target protein was calculated by dividing the amount (fmol) of target peptide by
cisterna. Then, the large temporal muscles were removed from the cranial bone, and the cranial cavity was opened using an electrical autopsy bone saw. Using a scalpel, the brainstem was cut at the medulla oblongata, just before the foramen magnum, and the whole brain was carefully removed and stored on ice until further dissection. A sample of temporal muscle was taken and approximately 100 mg was transferred to a Precellys vial, avoiding any visible fascia. Choroid plexus: the whole brain was weighed and a midsagittal section was performed. The lateral ventricles of both hemispheres were accessed after carefully opening the region between the corpus callosum and the fornix, and the CP was removed. Because the larger CP of the lateral ventricle (LV), and the smaller CP of the third ventricle (3 V) are connected, the sampled LV also contained a minor fraction of 3 V CP tissue. In this manuscript, the LV with minor contaminations of 3 V are referred to as LV only. Subsequently, the CP of the fourth ventricle (4 V) were removed from both hemispheres. The LV and the 4 V CP samples from both hemispheres were pooled separately in preweighed 1.5 mL microreaction tubes and snap frozen on dry ice. Choroid plexus of the 4 V and LV were investigated separately, because morphological, histological and ultrastructural differences exist.25 Pictures illustrating the location and separation of the CPs after midsagittal dissection of the brain and LV and 4 V CP samples are contained in the Figure S1. Both brain hemispheres were further dissected into brainstem, cerebellum, and cerebrum (telencephalon, containing also subcortical structures, such as the hippocampus, amygdala, nucleus caudatus, and putamen). Cerebrum was collected as reference brain section to be comparable to other species investigated in the past with the same method. The brainstem was removed from the telencephalon by blunt separation at the capsula interna, and each of these large parts was further dissected into a rostral (frontal), medial, and caudal fraction. A minimum of 3 g tissue was needed for the capillary isolation. This in combination with the difficulty to exactly separate substructures of the telencephalon in a native brain without prior staining prevented us from sampling with a higher spatial resolution. Brain tissue samples from the right hemisphere were snap frozen on dry ice and kept for brain micro capillary isolation. Tissue samples from the left hemisphere (approximately 2 g each) were transferred into Dispomix tubes, frozen on dry ice, and stored until subsequent homogenization and bioanalysis. The complete sampling procedure from cardiac arrest until sample freezing was performed within a maximum of 45 min. Sample Preparation. Prior to bioanalysis, plasma samples were diluted with acetonitrile to precipitate proteins. No special sample preparation was done for CSF. Brain was homogenized in an aqueous buffer prior to protein precipitation with acetonitrile. Compounds in plasma, CSF, and brain were quantified by means of HPLC-MS/MS. Brain Capillary Preparation. Brain capillary preparation was performed by Proteomedix Frontiers Co., Ltd., Sendai Miyagi, Japan. Brain microvessels were isolated using a combination of dextran density gradient and size filtration. Briefly, brain tissue samples stored at −80 °C were thawed under running water and homogenized in solution B (101 mM NaCl, 4.6 mM KCl, 2.5 mM CaCl2·2H2O, 1.2 mM KH2PO4, 1.2 mM MgSO4· 7H2O, 15 mM HEPES, pH 7.4) with a Potter-Elvehjem homogenizer using 14 up-and-down stokes with no rotation. The homogenate was centrifuged at 1000 × g for 10 min and the pellet was suspended in solution B. 35% (w/v) dextran in C
DOI: 10.1021/acs.molpharmaceut.7b00449 Mol. Pharmaceutics XXXX, XXX, XXX−XXX
Article
Molecular Pharmaceutics
membrane with a 5 kDa molecular weight cutoff. Stock solutions for each test compound were prepared in DMSO at 1 mM and diluted to the final test concentration. The subsequent dialysis solutions were prepared in plasma. Aliquots of 200 μL dialysis buffer (100 mM potassium phosphate, pH 7.4) were dispensed into the buffer chamber. Aliquots of 200 μL test compound dialysis solution were dispensed into the plasma chambers. Final compound concentration was 1 μM except for quinidine where, due to known nonlinear plasma protein binding,31 concentration was increased. Quinidine, in contrast to the other four compounds investigated, is a basic compound with a preference for binding to alpha-1-acidic glycoprotein (AGP). The binding to AGP is concentration dependent31 and the AGP concentration in dog plasma is approximately 320 μg/ mL.32 Thus, even at low quinidine plasma concentrations this binding runs into saturation and the fraction unbound in plasma increases. For human plasma a 3-fold increase of unbound fraction with increasing concentrations was reported.31 Also rat plasma protein binding increased with quinidine concentration.33 As the fraction unbound is important for the calculation of Kp,uu, we investigated the PPB for quinidine at 4 μM, the steady state target concentration in this infusion study. Dialysis was carried out for 2 h under rotation at 37 °C. At the end of the dialysis period, the dialysate was transferred into reaction tubes and mixed with an analytical internal standard. Plasma was added to buffer samples and buffer was added to plasma samples to ensure identical matrix composition for bioanalytical measurement. After protein precipitation using acetonitrile, the samples were measured by HPLC-MS/MS (CTC PAL autosampler attached to Sciex API5000 mass spectrometer). Percent bound was calculated with the formula:
the total protein amount of brain capillaries (cerebrum, cerebellum, and brain stem) and CP (LV and 4 V) in lysate (μg protein). Data demonstrating the reliability of the analytical assays is given in Supporting Information. Assuming that the yield of brain capillary isolation is same among different brain regions, i.e., cerebrum, cerebellum, and brain stem, the following equation can be used to determine the transporter or receptor protein concentrations of the capillaries for the averaged brain (Cavg brain): Wcerebrum (Wcerebrum + Wcerebellum + Wbrain stem) Wcerebellum + Ccerebellum × (Wcerebrum + Wcerebellum + Wbrain stem) Wbrain stem + C brain stem × (Wcerebrum + Wcerebellum + Wbrain stem)
Cavg brain = Ccerebrum ×
(1)
where Cavg brain, Ccerebrum, Ccerebellum, and Cbrain stem are the protein concentration of capillaries for averaged brain, cerebrum, cerebellum, and brain stem, respectively. Wcerebrum, Wcerebellum, and Wbrain stem are the wet organ weights of cerebrum, cerebellum, and brain stem, respectively. Assuming that the percentage of epithelial cell amount per CP is same among LV, 3, and 4 V, we used the following equation to determine the protein concentration of the averaged CP (CCP): CCP = C LV ×
W4V WLV + C4V × (WLV + W4V ) (WLV + W4V ) (2)
In this expression, CLV and C4 V are the protein expression levels and WLV and W4 V the weights of LV and 4 V, respectively. Calculations of the Partition Coefficients. Unbound tissue-to-plasma concentration coefficient (Kp,uu,tissue) of the transporter substrates and antipyrine were calculated by dividing unbound tissue concentrations by unbound plasma concentrations. Unbound tissue, plasma, and CSF concentrations were defined as the product of the total concentration and unbound fraction. The principle of calculating the Kp,uu of a specific matrix (matrix1) compared to a second matrix (matrix2 = plasma) is as follows: K p,uu,matrix1 =
%bound =
Frozen brains were homogenized in PBS buffer (dilution factor D = 4). The homogenate was spiked with test compounds (10 μM final concentration, final concentration of DMSO is 0.1%) and subsequently dialyzed against PBS buffer for 6 h at 37 °C in rapid equilibrium dialysis (REDdevice, Pierce) cells. At the end of the dialysis period, the dialysate was transferred into reaction tubes and mixed with an analytical internal standard. Tissue homogenate was added to buffer samples and buffer was added to tissue samples to ensure identical matrix composition for bioanalytical measurement. After protein precipitation using acetonitrile samples were measured by HPLC-MS/MS (CTC PAL autosampler attached to Sciex API5000 mass spectrometer). The free fraction was calculated as
(3) 26
f u,CSF was estimated as proposed by Friden et al. according to the equation: fu,CSF =
1 ⎛ 1 ⎞ − 1⎟ 1 + Q alb⎜ f ⎝ u,plasma ⎠
(5)
× 100
Cmatrix1 × fu,matrix1 Cmatrix2 × fu,matrix2
(plasma concentration − buffer concentration) plasma concentration
(4)
The albumin CSF to plasma ratio, Qalb, was assumed to be same in rat and dog (0.003) as data show that they are in the same range for rat,27,28 cat, and dog29 and taking into account that cisternal samples tend to have slightly lower levels compared with lumbosacral samples.30 Plasma Protein and Brain Homogenate Binding. Equilibrium dialysis technique was used to determine the in vitro fractional binding of the transporter substrates and antipyrine to plasma proteins. Dianorm Teflon dialysis cells (micro 0.2) were used. Each cell consisted of a donor and an acceptor chamber, separated by an ultrathin semipermeable
fu,app = (C PBS/C homogenate)
(6)
Since the brains were homogenized in the buffer, the observed free fraction values (f u,app) were extrapolated to the fractions unbound in the brain (f u*) based on the dilution (D) using the following formula: fu* = (1/D)/((1/fu,app − 1) + 1/D)
(7)
Bioanalytics. Quantitative bioanalysis was performed with a fit-for-purpose assay using protein precipitation followed by D
DOI: 10.1021/acs.molpharmaceut.7b00449 Mol. Pharmaceutics XXXX, XXX, XXX−XXX
Article
Molecular Pharmaceutics
Table 1. Protein Expression Levels of Transporters and Receptors in Brain Capillaries in Different Parts of Canine Brain absolute protein expression level (fmol/μg protein lysate)a alias/gene symbol ABC transporters MRP1/ABCC1 MRP4/ABCC4 BCRP/ABCG2 P-gp/ABCB1 SLC transporters OCT2/SLC22A2 OAT3/SLC22A8 PEPT2/SLC15A2 OATP1A2/SLC21A3 OATP2B1/SLC21A9 ENT1/SLC29A1 LAT1/SLC7A5 MCT1/SLC16A1 GLUT1/SLC2A1 receptors INSR LRP1 TfR1 other 4F2hc/SLC3A2 Na+K+ ATPase
cerebrum n 0 0 6 6
cerebellum
b
b
brain stem
meand
b
