ABC transporters, cytochromes P450 and their main transcription factors: expression at the human blood–brain barrier

Authors


Address correspondence and reprint requests to Xavier Declèves, PhD, INSERM U705 CNRS UMR 7157, Faculté de Pharmacie, 4 avenue de l’observatoire, Paris 75006, France. E-mail: xavier.decleves@univ-paris5.fr

Abstract

We have established the expression patterns of the genes encoding ATP-binding cassette (ABC) transporters and cytochromes P450 (CYPs) at the adult human blood–brain barrier (BBB) using isolated brain microvessels and cortex biopsies from patients with epilepsia or glioma. Microves synaptophysin (neurons) and neuron-glial antigen 2 (NG2) (pericytes). ABCG2 [breast cancer resistance protein (BCRP)] and ABCB1 (MDR1) were the main ABC transporter genes expressed in microvessels, with 20 times more ABCG2 and 25 times more ABCB1 in microvessels than in the cortex. The CYP1B1 isoform represented over 80% of all the CYPs genes detected in microvessels. There were 14 times more CYP1B1 in microvessels than in the cortex, showing that CYP1B1 is mainly expressed at the BBB. p-glycoprotein (ABCB1), BCRP (ABCG2) and CYP1B1 proteins were found in microvessels by western blotting. The expression of genes encoding three transcription factors [pregnane xenobiotic receptor (PXR), constitutive androstane receptor (CAR), aryl hydrocarbon receptor (AhR)] was also investigated. The AhR gene, involved in the regulation of CYP1B1 expression, was highly expressed in brain microvessels, whereas PXR and CAR genes were almost undetected. This detailed pattern of ABC and CYPs gene expression at the human BBB provides useful information for understanding how their substrates enter the brain.

Abbreviations used
ABC

ATP-binding cassette

AhR

aryl hydrocarbon receptor

BBB

blood-brain barrier

BCRP

breast cancer resistance protein

CAR

constitutive androstane receptor

Ct

crossing-threshold

CYP

cytochromes P450

GFAP

glial fibrillary acidic protein

MDR

multidrug resistance

MRP

multidrug resistance-associated protein

NG2

neuron-glial antigen 2

PECAM-1

platelet endothelial cell adhesion molecule 1

P-gp

P-glycoprotein

PXR

pregnane xenobiotic receptor

qPCR

quantitative PCR

SYP

synaptophysin

TBP

TATA box-binding protein

The blood–brain barrier (BBB) is composed of microvessel endothelial cells sealed by tight junctions and surrounded by pericytes, neuron endings and astrocyte foot processes. These form a dynamic neurovascular unit which is the first line of defence for the brain against unwanted compounds. The entry of many compounds into the brain, including numerous commercial drugs, is also restricted by ATP-binding cassette (ABC) efflux transporters, including P-glycoprotein [P-gp, ABCB1/multidrug resistance (MDR1)], several multidrug resistance-associated proteins (MRPs) (ABCCs) and breast cancer resistance protein (BCRP) (ABCG2), at the plasma membrane of brain microvessel endothelial cells (Scherrmann 2005). It is now agreed that the P-gp at the BBB lies mainly on the luminal membrane of brain endothelial cells in humans (Bendayan et al. 2006). BCRP, another ABC transporter, is also found in primary cultures of human brain endothelial cells (Lee et al. 2007) and has been immunolocalized at the luminal membrane of microvessel endothelial cells (Cooray et al. 2002). In contrast, the presence and function of MRPs at the BBB is still uncertain because of the use of poor specificity antibodies and putative species-specific differences in drug transporter profiles. MRP1, MRP4 and MRP5 are the only MRPs that have been detected in the endothelial cells of human brain microvessels (Nies et al. 2004; Bronger et al. 2005). But all these studies have provided only a qualitative pattern of the ABC transporters at the human BBB, without any quantitative data on the relative amounts of each ABC transporter that could help provide a clear picture of the impact of each transporter on the efflux of overlapping substrates.

The BBB can also be considered to be a metabolic barrier because of the drug-metabolizing enzymes in endothelial cells, especially cytochromes P450 (CYPs) (El-Bacha and Minn 1999). CYP mRNAs have been detected in several areas of the human brain (reviewed in Dutheil et al. 2008) and have been found to be functional in isolated rodent brain microvessels (Ghersi-Egea et al. 1994; Granberg et al. 2003). However, their relative expression at the human BBB, as reflected in isolated brain microvessels, remains unknown.

ATP-binding cassette transporters and CYPs also have some common substrates (P-gp and CYP3A4) and are co-expressed and/or co-regulated by transcription factors such as the pregnane xenobiotic receptor (PXR), the constitutive androstane receptor (CAR) and the aryl hydrocarbon receptor (AhR) in the intestine and liver (Xu et al. 2005). Although the expression of PXR, CAR and AhR at the rodent BBB has been previously investigated, results are still controversial (Bauer et al. 2004; Akanuma et al. 2008; Filbrandt et al. 2004) and no data for the human BBB have yet been published.

Because the ABC transporters and CYPs form a first line of defence for the brain, we have carried out for the first time, a detailed examination of the expression of their gene at the human BBB. We also investigated the expression of genes encoding PXR, CAR and AhR. We isolated microvessels from human brain samples and checked their purity by quantitative real-time PCR (qRT-PCR) of the genes encoding various cell-specific markers. We then assessed the expression of 10 ABC transporter genes (ABCB1, ABCG2, ABCC1–6, ABCC11 and ABCC12) and those encoding 23 CYPs and their main transcription factors (PXR, CAR, AhR) in samples of cerebral cortex and the corresponding freshly isolated cortical microvessels. The samples and microvessels were from adult patients with epilepsia or glioma. Lastly, we investigated the expression of the main ABC and CYP proteins in isolated microvessels.

Materials and methods

Human brain tissue samples

Human tissues were collected after informed consent had been obtained from each patient, in accordance with the regulations of the Ethics Committee. Cerebral cortex samples were obtained from patients undergoing surgery for epilepsy or glioma at the Department of Neurosurgery at Sainte-Anne Hospital (Paris, France). The neurosurgeons provided samples of brain cortex (grey matter) located as far as possible from the tumour or the epileptogenic lesion. They were considered to be healthy tissue on the basis of neuroimaging. Information on the patients is given in supporting information. Samples were stored in RPMI buffer at 4°C for not more than 1 h.

Reagents

Commercial human brain cortex RNA was obtained from BD Biosciences (Woburn, MA, USA). The material was from pooled brain samples from 10 males/females who died suddenly. The antibodies used and their suppliers were: monoclonal mouse anti P-gp C219 antibodies and monoclonal mouse anti-BCRP BXP-21 (Abcam, Cambridge, UK), rabbit antihuman CYP1B1 peptide serum WB-1B1 (BD Gentest, Woburn, MA, USA), monoclonal mouse anti-β-actin antibody (Sigma-Aldrich, Saint Quentin Fallavier, France), horseradish peroxidase-conjugated anti-mouse secondary antibodies (Amersham, Buckinghamshire, UK) and alkaline phosphatase-conjugated anti-rabbit secondary antibody (Applied Biosystems, Foster City, CA, USA). Proteins were detected using the enhanced chemiluminescence system (Amersham Biosciences Europe GmbH, Orsay, France) or the CDPStar® chemiluminescent substrate (Sigma-Aldrich). Other chemicals and reagents were purchased from Sigma-Aldrich or from Invitrogen (Cergy-Pontoise, France).

Isolation of human microvessels

All procedures were carried out at 4°C on brain cortex samples weighing 200 mg to 1 g. Each sample was divided into two parts, one was used for brain microvessel isolation and the second for RNA extraction. Thus, the second part was considered to be a microvessel-containing fraction of whole cortex. Brain microvessels were isolated essentially as described previously (Yousif et al. 2007). The cortex samples were homogenized at 4°C with a Potter–Thomas homogenizer (Kontes Glass, Vineland, NJ, USA) using 20 up and down strokes at 400 rpm. The resulting homogenate was centrifuged at 1000 g for 10 min and the microvessel-enriched pellet was suspended in 17.5% dextran and centrifuged for 15 min at 4400 g in a swinging bucket rotor. The resulting pellet was suspended in Hanks’ balanced salt solution containing 1% bovine serum albumin, while the supernatant containing a layer of myelin was centrifuged once more and removed. The two suspended pellets were pooled and passed through a 20-μm nylon mesh. The microvessels retained by the nylon mesh were immediately collected and frozen at −80°C.

RNA extraction

Total RNA was extracted from each sample using the RNeasy micro kit (Qiagen GmbH, Hilden, Germany) according to the manufacturer’s instructions. Proteinase K was used to lyse the basal membrane surrounding the brain microvessels and samples were treated with DNAse I (Rnase-Free Dnase Set; Qiagen SA) to remove genomic DNA. The concentration and purity of the RNA samples were assessed spectrophotometrically at 260 nm using the Nanodrop® ND-1000 instrument (NanoDrop Technologies, Wilmington, DE, USA). About 100–400 ng of total RNA was extracted from isolated microvessels, depending on the size of the sample, whereas at least 1 μg was obtained from the corresponding brain cortex sample.

Reverse transcription

Reverse transcription was performed on the whole amount of total RNA extracted from the microvessels of each patient, and on the same amount of the total RNA from the corresponding microvessel-containing brain cortex fraction. RT was performed in a final reaction mixture of 20 μL containing 500 μM of each dNTP, 10 mM dithiothreitol, 1.5 μM random hexanucleotides primers (Amersham), 20 U Rnasin ribonuclease inhibitor (Promega, Charbonnières, France) and 100 U superscript II Rnase reverse transcriptase (Invitrogen). RT was performed on the commercial brain cortex total RNA under the same conditions for patient samples. All samples were incubated at 25°C for 10 min, then at 42°C for 30 min and at 99°C for 5 min in a thermal cycler (PTC-100 programmable thermal controller; MJ Research Inc, Waltham, MA, USA). cDNAs were stored at −80°C.

Real-time quantitative PCR

The expression of genes encoding ABC transporters (ABCB1, ABCG2, ABCC1-6, ABCC11-12), transcription factors (PXR, CAR and AhR) and cell-specific markers like the platelet endothelial cell adhesion molecule 1 (PECAM-1), the glial fibrillary acidic protein (GFAP), the synaptophysin (SYP) and NG2 proteoglycan (NG2) was analysed by quantitative PCR (qPCR) on a Light-Cycler® instrument (Roche Diagnostics, Meylan, France) using SYBR Green fluorescence detection. The final reaction mixtures contained 5 μL diluted cDNA, 1 μL LC-FastStart DNA Master SYBR Green kit (Roche Diagnostics), 0.5 μL of each primer 10 μM, 1.2 μL of 10 mM MgCl2 and 1.8 μL nuclease-free water. Specific primers for each gene were designed using oligo 6.42 software (MedProbe, Oslo, Norway) and were on separate exons or on exon–exon junctions as far as possible to avoid amplification of genomic DNA. We performed no-template control assays for each primer pair; they always produced negligible signals [usually >40 in crossing-threshold (Ct) value]. The size of the specific amplicons of interest and the absence of other PCR products was checked by gel electrophoresis. The specificity of each reaction was also assessed by melting curve analysis to ensure the presence of only one product. Primer sequences are shown in Table S1. All the primers used for qPCR analysis of ABC transporter genes were checked with positive controls (liver and/or brain cortex). The qPCR of 23 CYPs genes was performed using SYBR Green fluorescence detection on an ABI Prism 7900 HT Sequence Detection System (Applied Biosystems, Foster City, CA, USA) as previously described (Girault et al. 2005). The final reaction mix contained 3 μL diluted cDNA, 5 μL Absolute QPCR SYBR Green ROX Mix (Abgene, Courtaboeuf, France), 0.3 μL of each primer 10 μM and 1.4 μL nuclease-free water. Primers were kindly provided by Dr I. de Waziers (Université Paris Descartes, France). The sequences of the primer for the 23 CYPs genes are available from Biopredic International (Rennes, France) (http://www.biopredic.com).

Genes of interest

The purity of isolated brain microvessels was investigated using specific markers for each cell type in the BBB: PECAM-1 for brain microvessel endothelial cells, GFAP for astrocytes, SYP for neurons and NG2 for pericytes. Genes of interest were (i) ABC transporters: ABCB1 (MDR1), ABCG2 (BCRP), ABCC1 (MRP1), ABCC2 (MRP2), ABCC3 (MRP3), ABCC4 (MRP4), ABCC5 (MRP5), ABCC6 (MRP6), ABCC11 (MRP8) and ABCC12 (MRP9), (ii) CYPs: 1A1, 1A2, 1B1, 2A6, 2A7, 2A13, 2B6, 2C8, 2C9, 2C18, 2C19, 2D6, 2E1, 2F1, 2J2, 2R1, 2S1, 2U1, 2W1, 3A4, 3A5, 3A7, 3A43 and (iii) transcription factors: PXR, CAR, and one member of the PAS (Per-ARNT-Sim) family, AhR.

Quantification of gene expression by qPCR

After proportional background adjustment, the second derivative method was used to determine the cycle at which the log-linear signal was distinguishable from the background, and this cycle number was used as the Ct value. Gene expression was evaluated using the Ct value from each sample. A target gene was considered to be quantifiable when the Ct obtained for the less diluted cDNA sample was lower than 30. A Ct value of 32 was taken as the detection limit.

The ΔΔCt method was first used to compare the expression of target genes in isolated microvessels and the corresponding brain cortex from each patient. The expression of the target gene in each sample was normalized to that of the housekeeping gene TATA box-binding protein (TBP). This comparison was possible because the expression of the TBP gene in microvessels and the corresponding cortex was not statistically different in any of the patients tested. The expression of each target gene in the microvessels and cortex was calculated from the comparative ΔΔCt parameter using the formula:

image

The 2ΔΔCt values are the difference (-fold) between the amounts of mRNA in microvessels and the brain cortex, arbitrarily defined as 1.

The ΔΔCt method was also used to determine the expression of each target gene relative to those of others of the same family (CYPs and ABC transporters families) in the microvessels and cortex. The efficacy of each PCR was determined using a calibration curve with cDNA from tissues known to express genes of interest (liver for genes not expressed in our cortex or microvessel samples and brain cortex for the others). The PCR was performed using seven serial dilutions of cDNA (1 : 20 to 1 : 2000). Efficacy (E) was then calculated using the slope of the calibration curve: E = 10(−1/slope). The PCR was 100% efficient when E = 2, and better than 95% when 1.90 < E < 2, indeed when the slope was −3.45 to −3.32. The efficacy of each PCR for all CYPs genes was better than 95% (Girault et al. 2005) as was the efficacy of all the other genes tested, making it possible to determine the relative expression of each target gene. Results are expressed as percentages, with the total expression of all CYPs or ABC transporters being set at 100%.

Western blotting

Isolated microvessels were sonicated at 4°C in a buffer containing 50 mM Tris–HCl pH 7.4, 150 mM NaCl, 0.5% sodium deoxycholate, 0.5% Triton X-100 and complete™ protease inhibitor complex (Roche Diagnostics). Homogenates were centrifuged at 4°C for 15 min at 10 000 g, and the resulting supernatants were stored at −80°C. The protein content of samples was determined with the Bradford reagent (Sigma, Lyon, France) and a bovine serum albumin calibration curve; the mean amount of total protein in isolated microvessel fractions was 300 μg/g of brain sample. Samples were normalized for equal amounts of protein, separated on 8% (P-gp) or 10% (BCRP, CYP1B1) sodium dodecyl sulfate–polyacrylamide gel electrophoresis gels and electrotransferred to nitrocellulose membranes (HYBOND; Amersham). Non-specific binding sites were blocked by incubation overnight at 4°C with Tris-base buffer containing 0.1% Tween 20 and 5% non-fat milk for P-gp and BCRP or 0.2% I-Block™ (Applied Biosystem) for CYP1B1. The membranes were immunoblotted with the C219 (diluted 1 : 200), BXP-21 (1 : 200) or WB-1B1 (1 : 10 000) antibodies for 2 h at 25°C, washed several times in Tris-base buffer containing 0.1% Tween 20 and incubated with horseradish peroxidase-conjugated anti-mouse secondary antibody (1 : 10 000), or alkaline phosphatase-conjugated anti-rabbit secondary antibody (1 : 20 000) for 45 min at 25°C. The membranes were then exposed to the Amersham enhanced chemiluminescence system (BCRP, P-gp) or the CDPStar system (CYP1B1). Signals were revealed with the Bio-Rad ChemiDoc® XRS imaging device (Bio-Rad, Marnes-la-Coquette, France). The blot was then stripped by immersion in 0.1 M acetic acid for 2 h and reprobed with monoclonal mouse anti-β-actin antibody (1 : 10000). The human cerebral endothelial cell line hCMEC/D3, kindly provided by Dr P. O. Couraud (Institut Cochin, Paris, France), was used as a positive control for P-gp expression (Weksler et al. 2005). The human embronic kidney-BCRP cells supplied by Dr R. W. Robey (National Cancer Institute, Berkshire, UK) were used as positive controls for BCRP and human CYP1B1 supersomes™ (BD Gentest) were used for CYP1B1.

Statistical analysis

Statistics were carried out with GraphPad Prism® 4.0 software (GraphPad Software Inc., San Diego, CA, USA). The results are expressed as means ± SD. Student’s paired t-test was performed on ΔCt values obtained before normalization to unity to identify significant differences between brain cortex and isolated microvessels. All the tests were two-tailed and statistical significance was set at p < 0.05.

Results

Purity of microvessel preparation

The microvessel-enriched fraction obtained after dextran centrifugation and filtration through a 20-μm mesh provided enough microvessels to prepare total RNA. Microscopic analysis revealed mainly microvessels (4–7 μm), some cell debris and a few larger vessels (until 25 μm), but not very large vessels (>25 μm) (Fig. 1a). The purity of the microvessel preparation was evaluated by measuring the expression of cell-specific marker genes, like PECAM-1 (endothelial cells), NG2 (pericytes), GFAP (astrocytes) and SYP (neurons), and the relative abundance of the mRNAs in each sample was compared with that of the whole cortex sample from the same patient (Fig. 1b). PECAM-1 mRNA was 15-fold more abundant in microvessels than in the cortex and NG2 mRNA was four-fold more abundant. In contrast, there was 31 times less SYP mRNA and seven times less GFAP mRNA in microvessels than in the brain cortex, showing the acceptable purity of the brain microvessel preparations.

Figure 1.

 (a) Microscopic examination of samples of microvessel isolated from human brains (20× magnification). Scale bar = 25 μm. (b) Relative expression of marker genes for endothelial cells (PECAM-1), pericytes (NG2), astrocytes (GFAP) and neurons (SYP) in isolated microvessels and brain cortex samples (set at 1). Results are means (±SD) of seven patients (No. 1–7). *p < 0.05; **p < 0.01; ***p < 0.001. PECAM-1, platelet endothelial cell adhesion molecule 1; NG2, neuron-glial antigen 2; GFAP, glial fibrillary acidic protein; SYP, synaptophysin.

ABC gene expression profile in brain microvessels and cortex

The expression of 10 ABC transporter genes was investigated by RT-qPCR in cortex samples and in the corresponding isolated microvessel fraction from each patient (Table 1). Neither ABCC2 nor ABCC3 mRNAs were detected in the brain cortex or in isolated microvessels, whereas these mRNAs were readily detected in the liver (data not shown). In contrast, ABCB1, ABCG2, ABCC1, ABCC4, ABCC5, ABCC6, ABCC11 and ABCC12 transcripts were detected in both isolated microvessels and cortex samples. The expression profiles of the genes encoding the major ABC transporters that actively efflux xenobiotics (ABCB1, ABCG2, ABCC1, ABCC4 and ABCC5) were established. The ABCC1, ABCC4 and ABCC5 genes were much less expressed (relative expression below 2%) than the ABCB1 (relative expression 12%) and ABCG2 genes (relative expression 85%) (Table 1, Fig. 2a).

Table 1.   Expression of genes encoding ABC transporters, CYPs and three transcription factors in the human brain cortex and in human brain isolated microvessels
 GeneExpression
Brain cortexMicrovessels
  1. ABC, ATP-binding casette; CYP, cytochromes P450; NQ, not quantifiable (30 < Ct < 32); ND, not detected (Ct > 32).

  2. Results are from seven patients (No. 1–7).

ABC transportersABCB1 ++
ABCG2++
ABCC1 ++
ABCC2 NDND
ABCC3 NDND
ABCC4 ++
ABCC5 ++
ABCC6 ++
ABCC11++
ABCC12 ++
CYPCYP1A1++
CYP1A2NDND
CYP1B1++
CYP2A6NDND
CYP2A7NDND
CYP2A13NDND
CYP2B6++
CYP2C8++
CYP2C9NDND
CYP2C18NDND
CYP2C19NDND
CYP2D6++
CYP2E1++
CYP2F1NDND
CYP2J2++
CYP2R1++
CYP2S1++
CYP2U1++
CYP2W1NDND
CYP3A4NDND
CYP3A5NDND
CYP3A7NDND
CYP3A43NDND
Transcription factorsPXR (NR1I2)NDND
CAR (NR1I3)NQNQ
AhR++
Figure 2.

 Comparative expressions of five ATP-binding cassette (ABC) transporter genes in isolated microvessels (a) and in total cerebral cortex samples from donors and from normal brain (b). Results are means (±SD) from seven patients (No. 1–7) and from four independent experiments (commercial RNA). (c) Relative expressions of genes encoding eight ABC transporters in isolated microvessels and in brain cortex samples (set at 1). Results are means (±SD) of seven patients (No. 1–7). **p < 0.01; ***p < 0.001.

The transcripts encoding eight ABC transporters were also detected in the microvessel-containing cortex samples (Table 1), but the expression profiles of the genes in the cortex, mainly ABCG2 (47%) and ABCC5 (33%), were different from those of the corresponding isolated microvessels. The relative amounts of ABCB1, ABCC1 and ABCC4 transcripts were below 10% (Fig. 2b). The ABC gene expression profiles in cortex samples from all seven patients enroled in this study were similar to that of the commercial normal human cortex (Fig. 2b), suggesting that the overall expression of ABC transporters in the brains of these patients was essentially unaffected by pre-surgery antiepileptic treatment. Furthermore, the expression profile of ABC transporter genes varied very little between individuals.

The expression profile of the eight ABC transporter genes in isolated microvessels and the corresponding cortex was also compared for each patient (Fig. 2c). The expression of the ABCB1 (25-fold), ABCG2 (20-fold), ABCC6 (11-fold) and ABCC12 (4.5-fold) genes was greater in brain microvessels than in cortex samples. Thus, the expression profiles of ABCB1, ABCG2 and ABCC6 resembled that of the endothelial marker PECAM-1, whereas the ABCC12 profile resembled that of NG2. In contrast, there was no statistical difference between the expression of the ABCC1, ABCC4 and ABCC5 genes in the microvessels and the cortex, whereas ABCC11 was significantly higher in the cortex (five-fold) than in the microvessels.

CYP gene expression profile in brain microvessels and cortex

The expression profiles of the genes encoding the 23 CYPs from families 1–3 (Table 1) showed that only a few isoforms were significantly expressed in brain microvessels (Table 1, Fig. 3a); the main isoforms were CYP1B1 (77%) and CYP2U1 (14%). The CYP2D6, CYP2E1, CYP2J2 and CYP2R1 genes were also weakly expressed (below 4%), whereas the CYP1A1, CYP2B6, CYP2C8 and CYP2S1 genes were barely detectable (below 0.5%). None of the other CYPs isoforms tested (CYP1A2, CYP2A6, CYP2A7, CYP2A13, CYP2C9, CYP2C18, CYP2C19, CYP2F1, CYP2W1, CYP3A4, CYP3A5, CYP3A7 and CYP3A43) was detected in either the brain microvessels or the cortex samples.

Figure 3.

 Comparative expressions of cytochromes P450 (CYPs) genes in isolated microvessels (a) and in total cerebral cortex from patients and from normal brain (b). Results are means (±SD) from seven patients (No. 1–7) and from four independent experiments (commercial RNA). (c) Relative expressions of the genes encoding the two main CYPs expressed in isolated microvessels and in brain cortex samples (set at 1). Results are means (±SD) of seven patients (No. 1–7). *p < 0.05.

The most expressed CYP genes in the microvessel-containing cortex samples (Fig. 3b) were CYP2J2 (45%), CYP2U1 (22%) and CYP1B1 (12%). This profile was similar to that of commercial human normal cerebral cortex RNA (Fig. 3b).

Furthermore, there was 14 times more CYP1B1 mRNA in microvessels than in the cortex (Fig. 3c), and this preferential expression in microvessels resembled that of the endothelial marker gene, PECAM-1. In contrast, the relative expressions of the CYP2U1 gene in isolated microvessels and in the brain cortex were similar.

Transcription factor gene expression profile in brain microvessels and cortex

The expression of the genes encoding the PXR, the CAR and the AhR in isolated microvessels and in the corresponding microvessel-containing cortex samples were also assayed by qPCR, because the ABC transporters and CYPs are both regulated by these transcription factors in several human tissues (Table 1). No PXR transcripts were detected in the microvessels or in the cortex samples, while CAR mRNA was detected but not quantifiable in both isolated microvessels and the brain cortex. The AhR gene was expressed in isolated microvessels and in the cortex; its expression was 2.7 times higher in microvessels than in the cortex (data not shown).

Major ABC transporters and CYP proteins in isolated microvessels

Western blotting experiments were performed to confirm the presence of the main ABC transporter proteins and CYP (P-gp, BCRP, CYP1B1) in isolated microvessels. Because of the small amount of proteins extracted from isolated microvessels from one patient (a mean of 150 μg per patient), it was not possible to perform several western blots for P-gp and the two other proteins (BCRP, CYP1B1). Figure 4 shows the western blots of P-gp, BCRP, CYP1B1 and the reference β-actin in isolated microvessels from four patients. P-gp protein was detected in the isolated microvessels from all patients (Nos: 8, 9, 10 and 11) with a variable intensity for each patient. BXP-21 antibody was used to detect BCRP protein in isolated microvessels. A band at the expected molecular weight (70 kDa) was observed in the three samples studied. Finally, CYP1B1 protein was detected using the WB-1B1 polyclonal antibody (band at 61 kDa) in all three patients studied.

Figure 4.

 p-glycoprotein (P-gp), breast cancer resistance protein (BCRP), CYP1B1 and β-actin proteins detected by western blotting in isolated human microvessels. Positive controls were hCMEC/D3 cells for P-gp, human embryonic kidney BCRP cells for BCRP and CYP1B1 supersomesTM for CYP1B1. MW represents approximate molecular weights of these proteins. Each lane was loaded with 20 μg protein.

Discussion

We first validated our protocol of isolating human brain microvessels by assessing the mRNAs of BBB cell-specific markers. Fifteen times greater amount of PECAM-1 mRNA in isolated microvessels than in the cortex indicates a marked enrichment of endothelial cells in the preparation, in agreement with the 10-fold difference previously observed in the rat (Yousif et al. 2007). The main advantage of this mechanical method is the significant reduction in contamination by astrocytes and neurons. There was seven-fold less GFAP mRNA and 31-fold less SYP mRNA in microvessels than in the brain cortex, indicating that this protocol is a reliable procedure for quantitatively assessing gene expression at the human BBB.

There have been a few reports that P-gp (ABCB1), BCRP (ABCG2) and probably MRP1 (ABCC1), MRP4 (ABCC4) and MRP5 (ABCC5) are at the human BBB (Virgintino et al. 2002; Cooray et al. 2002; Nies et al. 2004). But there is no published information about their relative expression at the human BBB using the same samples. Our analyses of isolated microvessels from human brain cortex samples indicate that ABCB1 and ABCG2 are the most expressed ABC drug efflux transporters genes, in terms of mRNA (more than 90%). Although one group found P-gp in the astrocyte foot processes of the human BBB (Golden and Pardridge 1999), there is now no doubt that P-gp is abundant at the luminal membrane of brain endothelial cells (Bendayan et al. 2006). Much more ABCB1 mRNA was detected in microvessels than in the cortex, like PECAM-1, a specific endothelial cell marker. We have also confirmed the finding of P-gp protein in human brain microvessels as already described (Virgintino et al. 2002). However, information available on patients did not provide any explanation of the observed variations in the intensity of blots between individuals.

We also found large amounts of ABCG2 mRNA and the corresponding protein in isolated microvessels. Moreover, there is more ABCG2 mRNA, like ABCB1 mRNA, in microvessels than in the microvessel-containing cortex, confirming that ABCG2 lies mainly at the BBB (Lee et al. 2007; Aronica et al. 2005). We also found that ABCG2 is the main ABC transporter, in terms of mRNA, in brain microvessels (85% of total ABC transporters). The ABCG2 was eight times more expressed than ABCB1, as also reported for cultured porcine brain endothelial cells (Eisenblatter et al. 2003). However, a very recent study using quantitative proteomic analysis of isolated mouse brain microvessels (Kamiie et al. 2008) reported that there was three times more Mdr1a P-gp than Bcrp. This may be related to species differences between mice and humans or to the post-transcriptional and/or post-translational regulation of Mdr1a (P-gp) and Abcg2 (Bcrp). Our study provides a quantitative analysis of ABCB1 and ABCG2 gene transcription, but a similar comparison was not possible from the western blotting experiments. However, the extent to which BCRP actually contributes to BBB function remains unclear. P-gp is abundant at the luminal membrane of the brain capillary endothelium and thus remains of utmost importance at the BBB. Indeed, BCRP is a ‘half transporter’ acting as a drug efflux pump when trafficked as a homodimer at the plasma membrane. Moreover, the substrate specificity of BCRP is lower than that of P-gp, although it partially overlaps that of P-gp. Several studies have shown that the accumulation of substances known to be BCRP substrates is moderate [imatinib (Breedveld et al. 2005)] or even lacking [dehydroepiandrosterone sulphate (DHEAS) and mitoxantrone (Lee et al. 2005)] in the brain of Abcg2−/− knock-out mice than it is in the brain of wild-type mice. Nevertheless, BCRP has been shown to work with P-gp to limit distribution of prazosine and mitoxantrone (Cisternino et al. 2004) and topotecan (de Vries et al. 2007) in the brain.

The expression of MRPs (ABCCs) at the human BBB remains controversial. We detected ABCC1, ABCC4, ABCC5, ABCC6, ABCC11 and ABCC12 mRNAs in isolated brain microvessels, but their expression was lower than those of ABCB1 and ABCG2. These results are in agreement with previous work showing that ABCC1, ABCC4 and ABCC5 and to a much lesser extent ABCC6, are the only ABCC genes expressed in isolated bovine brain microvessels (Zhang et al. 2000). Although MRP4 was the only MRP protein detected in mouse brain microvessels (Kamiie et al. 2008), our finding of ABCC1, ABCC4 and ABCC5 transcripts in human microvessels is in agreement with published immunohistochemical data on MRPs in human brain endothelial cells (Nies et al. 2004). We did not detect ABCC2 in either isolated microvessels or the brain cortex, in accordance with a previous data showing the lack of Mrp2 in the human brain (Nies et al. 2004). However, this is at odds with reports of functional Mrp2 at the BBB of rats and mice (Potschka et al. 2003; Soontornmalai et al. 2006; Bauer et al. 2008). This discrepancy may be attributed to interspecies differences. ABCC2 transcripts have also been detected in human cultured cerebral endothelial cells taken from brain samples of an epileptic region (Dombrowski et al. 2001). Thus, the lack of ABCC2 transcripts in our study may be because the samples were obtained far from an epileptic region. We also detected six ABCC isoforms in cortex samples. However, the ABCC6 and ABCC12 genes are more expressed in microvessels than in the cortex, suggesting that they are mainly expressed at the BBB, whereas the ABCC1, ABCC4, ABCC5 and ABCC11 genes are not preferentially expressed in brain microvessels as already described (Nies et al. 2004; Decleves et al. 2006). ABCC12 and ABCC11 mRNAs have been detected in the human brain (Kruh et al. 2007). We have now shown that ABCC12 is more expressed at the BBB while ABCC11 is more expressed in the cortex.

Our results on the expression of ABC transporter genes must be interpreted with caution, as they were obtained from patients suffering from refractory epilepsy or epileptogenic brain tumours. Although the cortex samples were collected from perilesional zones that were considered to be ‘healthy’ by the neurosurgeons, they may not have been normal tissue. Some ABC transporters are up-regulated in epileptic and tumour brain tissues. ABCB1 and the P-gp protein, ABCC2 and ABCC5 are over-expressed in epileptic human brain microvessel endothelial cells, while ABCC1 (Dombrowski et al. 2001) and the BCRP protein (Sisodiya et al. 2003) are not. Also, four patients in our study were treated for low grade (WHO grade II) oligodendrogliomas (three patients) and glioblastoma (one patient), and the expression of BCRP is known to increase with the grade of glioma (Aronica et al. 2005). Moreover, an up-regulation of MDR1 and MRP1 have been found in human glioma, and their expressions also increase with the grade of tumour (Calatozzolo et al. 2005; Nies 2007). However, we found no ABCC2 mRNA and low amounts of ABCC1 and ABCC5 mRNAs in all the microvessel-enriched fractions. Similarly, there was more ABCG2 mRNA than ABCB1 mRNA in the isolated microvessels from all patients, regardless of their pathology or treatment. We find that the mRNA profiles in our brain cortex samples and in a commercially normal human brain cortex are similar. This suggests that our results are unlikely to be affected by any pathology or treatment, but actually reflect the profile of ABC transporter gene expression in the brain cortex and brain microvessel endothelial cells.

Cytochromes P450s have been found in isolated rat brain microvessels (Ghersi-Egea et al. 1994) and in whole human brain (Dutheil et al. 2008), but no data are available for the human BBB. We have established the expression profiles of an extensive range of CYP genes in isolated human brain microvessels. CYP1B1 is clearly the most expressed CYP gene in microvessels, where the corresponding CYP1B1 protein was also detected. CYP1B1 is a novel extrahepatic CYP, already found in the brain (Rieder et al. 1998), that is involved in the metabolism of endogenous compounds and in the activation of many procarcinogens and promutagens (Shimada et al. 1996). CYP1B1 has been immunolocalized at the human blood-brain interface (Rieder et al. 2000), but was not found in isolated mouse brain microvessels (Granberg et al. 2003). We also find that CYP1B1 is more expressed in microvessels than in the cortex, as is PECAM-1, suggesting that CYP1B1 is predominantly expressed at the BBB. A recent study found increased CYP1B1 expression in oligodendroglioma and glioblastoma (Barnett et al. 2007), but we find that the CYP expression profiles, like those of the ABC transporter genes, differ little between individuals despite their different treatments and pathologies. Moreover, the CYP expression profile in the cortex resembles that of the commercial normal human brain cortex RNA. Thus, our results probably reflect the expression profile of the normal adult brain. We also find that genes encoding CYP3A4, CYP2C9 and CYP2D6 involved in the hepatic metabolism of about 50% of drugs, which are not present at the BBB, show the specificity of the BBB CYP profile.

We also investigated the expression of genes encoding transcription factors known to be involved in the regulation of ABC transporters and CYPs. PXR, whose activation increases the expressions of P-gp and CYP3A4 proteins (Synold et al. 2001), ABCC2 (Kast et al. 2002) and ABCC3 (Teng et al. 2003), was not detected here, whereas it has been detected in rat brain microvessels (Bauer et al. 2004) and shown to be functional at the BBB of transgenic mouse expressing human PXR (Bauer et al. 2006). Neither did we detect any PXR transcripts in human brain samples, in agreement with a study showing that PXR is expressed in the human thalamus but not in human brain temporal and frontal lobes (Lamba et al. 2004). CAR is reported to induce ABCB1 (Burk et al. 2005), ABCC2 (Kast et al. 2002) and mMrp4 (Assem et al. 2004), but only small amounts of its mRNA were detected. A most interesting finding is increased expression of AhR in isolated microvessels, in agreement with data for mouse brain endothelial cells (Filbrandt et al. 2004). AhR is activated by polycyclic aromatic hydrocarbons, including dioxin and benzo[a]pyrene, which is present in tobacco smoke. Among its target genes are CYP1B1 (MacDonald et al. 2001) and ABCG2 (Ebert et al. 2005), both of which are present in brain microvessels. Our study therefore raises the question about the AhR-mediated regulation of CYP1B1 expression at the human BBB, particularly in smokers. However, we cannot draw any conclusions about this because of the limited number of cigarette smokers included in the study. However, we did look at the mRNAs encoding three transcription factors (PXR, CAR, AhR) that may be involved in the regulation of some ABC transporters and CYPs at the BBB without any in vivo data indicating the presence of regulatory pathways at the human BBB.

In conclusion, we have established for the first time the pattern of the expression of genes encoding ABC transporters and CYPs at the adult human BBB. The ABCB1 and ABCG2 are the main ABC drug efflux transporter genes expressed in isolated microvessels. There is also more ABCB1 mRNA and ABCG2 mRNA in microvessels than in the microvessel-containing brain cortex. CYP1B1 is clearly the most expressed CYP gene in microvessels and it is more expressed in microvessels than in the cortex. Finally, AhR is expressed in isolated microvessels, whereas the genes encoding PXR and CAR, the main transcription factors involved in ABC transporters and CYPs gene regulation, are almost undetectable. Therefore, our findings will be useful for understanding the capacity of drugs to enter the brain and their pharmacological or toxic effects on the CNS.

Acknowledgements

This research was supported in part by a grant from Servier Laboratories. We would like to thank Dr Elisabeth Landré, Dr Bertrand Devaux, Dr Baris Turak and Dr Philippe Page from the Neurology and Neurosurgery Departments and Dr Pascale Varlet and Dr Chiara Villa from the Anatomopathology Department (Sainte-Anne Hospital, Paris, France) for providing tissue specimens obtained during neurosurgery. The English text was edited by Dr Owen Parkes.

Ancillary