Morphine, especially its subchronic and chronic use, is associated with the development of phenomena such as addiction, tolerance, and the appearance of withdrawal symptoms when it is stopped. The mechanisms underlying these events involve complex molecular pathways, most of which have been attributed to regulation of the main morphine target, the opioid receptors. However, there is increasing evidence that the blood–brain barrier (BBB) is also involved (Tournier et al. 2011). Morphine crosses the BBB to reach the brain parenchyma, but it does so less well than other opioid drugs like fentanyl and methadone (Dagenais et al. 2004). This relatively poor brain penetration of morphine has been linked, in part, to its active efflux from the brain to the blood by the P-glycoprotein (P-gp, MDR1, ABCB1) at the BBB (Zong and Pollack 2000; Hamabe et al. 2006). P-gp is the best known of the many drug transporters at the BBB (Scherrmann 2005). The breast cancer-resistance protein [breast cancer-resistance protein (BCRP), ABCG2], another drug efflux transporter, is also abundant at the BBBs of monkeys and humans (Dauchy et al. 2008; Ito et al. 2011; Shawahna et al. 2011) and has been shown to efflux some substrates that it has in common with P-gp (Decleves et al. 2011). P-gp and BCRP belong to the ATP-binding cassette (ABC) efflux transporter superfamily and they limit the access of many drugs to the CNS (Urquhart and Kim 2009). The effects of opioids on BCRP activity have been poorly evaluated. Recently, we identified buprenorphine and its metabolite norbuprenorphine as inhibitors of BCRP (Tournier et al. 2010). However, the effect of exposure to subchronic opioid treatment on the synthesis of BCRP at the BBB has never been studied. In contrast, there have been several studies on the P-gp-mediated transport of morphine in rodents and humans (Callaghan and Riordan 1993; Letrent et al. 1999; King et al. 2001). Inhibiting P-gp increased morphine-induced analgesia in rats (Letrent et al. 1999), and morphine has an increased anti-nociceptive effect in P-gp-deficient knockout mice (Thompson et al. 2000; Zong and Pollack 2000). Moreover, variation in the pain relief in patients with cancer has been associated with variants of the ABCB1 gene which alter the expression and/or activity of P-gp (Campa et al. 2008; Lotsch et al. 2009). The pharmacokinetic–pharmacodynamic relationships between the plasma concentration of morphine and its CNS effect EEG were clearly demonstrated by Groenendaal et al. 2007 who showed that P-gp modulates the distribution of morphine within the biophase and so helps delay its CNS effect by (Groenendaal et al. 2007). Therefore, P-gp is a key element in the regulation of the effect of morphine on the CNS. An increase in P-gp expression during morphine treatment may thus lead to a timely decrease in the brain uptake of morphine, which may explain the tolerance frequently observed in the CNS effects of morphine. Two studies have shown that P-gp is induced in the brains of mice (Zong and Pollack 2003) and rats (Aquilante et al. 2000) by chronic exposure to morphine. Recently, we showed that subchronic morphine treatment induced P-gp expression in rat brain cortex vessels 12 h after the last morphine dose (Yousif et al. 2008), but rate and amplitude of this stimulation and the mechanism underlying it remain unknown. It may be because of at least two mechanisms: (i) the direct induction of P-gp by morphine via increased transcription of the gene encoding P-gp, increased translation, and/or post-translational processing, (ii) an indirect effect of morphine during the withdrawal syndrome occurring after subchronic treatment. In this study, we have firstly investigated the kinetics of P-gp and Bcrp mRNA and protein levels in rat brain vessels at different times after the last dose of subchronic morphine treatment and then evaluated the roles of glutamate and cyclooxygenase-2 (COX-2) in this induction. During morphine withdrawal, there is an increase in the excitatory amino acid neurotransmission in the CNS area, particularly the glutamate and aspartate concentrations (Sepúlveda et al. 2004). Glutamate acts at four receptor subtypes: NMDA, AMPA, kainate, and metabotropic glutamate receptors, and this neurotransmitter increases P-gp expression in rat brain endothelial cells (Zhu and Liu 2004). P-gp was found to be regulated by a pathway that involves glutamate signaling through NMDA receptors and COX-2 activity, which have both been found in the brain capillary endothelial cells (Bauer et al. 2008; Zibell et al. 2009). In this way, a glutamate–NMDA receptor–COX-2 pathway could be involved in this morphine-mediated induction.
Subchronic morphine treatment induces P-glycoprotein (P-gp) up-regulation at the blood–brain barrier. This study investigates the rate and extent to which P-gp and breast cancer-resistance protein (Bcrp) increase at the rat blood–brain barrier following subchronic morphine treatment. Rats were given increasing doses of morphine (10–40 mg/kg) or saline i.p. twice daily for 5 days. The brain cortex large vessels and microvessels were then mechanical isolated 6, 9, 12, 24, and 36 h after the last injection. The gene and protein expression of P-gp and Bcrp in morphine-treated and control rats were compared by qRT-PCR and western blotting. The levels of Mdr1a and Bcrp mRNAs were not significantly modified 6 h post morphine, but the Mdr1a mRNA increased 1.4-fold and Bcrp mRNA 2.4-fold at 24 h. P-gp and Bcrp protein expression in brain microvessels was unchanged 6 h post morphine and increased 1.5-fold at 24 h. This effect was more pronounced in large vessels than in microvessels. However, extracellular morphine concentrations of 0.01–10 μM did not modify the expressions of the MDR1 and BCRP genes in hCMEC/D3 human endothelial brain cells in vitro. MK-801 (NMDA antagonist) and meloxicam (cyclo-oxygenase-2 inhibitor) given after morphine treatment completely blocked P-gp and Bcrp up-regulation. Interestingly, misoprostol and iloprost, two well-known agonists of prostaglandin E2 receptors induced both MDR1 and BCRP mRNA levels in hCMEC/D3. Thus, morphine does not directly stimulate P-gp and Bcrp expression by the brain endothelium, but glutamate released during morphine withdrawal may do so by activating the NMDA/cyclo-oxygenase-2 cascade.
breast cancer-resistance protein
bovine serum albumin
central nervous system
hank's buffered salt solution
TATA box-binding protein
tyrosine kinase inhibitors
Materials and methods
The adult male 8-week-old Sprague–Dawley rats weighing 200–240 g used in this study were purchased from Charles River laboratory (L'arbresle, France). They were housed in groups of four animals per cage in a temperature- and humidity-controlled room with 12 : 12 h light/dark conditions (light from 8:00 AM to 8 PM) and access to food and water ad libitum. They were acclimated for at least 3 days prior to experimentation. The care and treatment of animals conformed to the standards and guidelines promulgated by the European Union Council Directive (2010/63/EU).
Acute and subchronic morphine dosing
We used two treatment protocols. For acute treatment, rats were given a single dose (10 mg/kg or 40 mg/kg, ip) of morphine (or saline) and killed 3, 6, and 24 h later to study the levels of Mdr1a and Bcrp mRNAs in cerebral vessels. For subchronic treatment, animals were given several doses of morphine (or saline) over 5 days. The test rats were given morphine i.p. twice daily: 10 mg/kg on day 1, 20 mg/kg on day 2, 30 mg/kg on day 3, and 40 mg/kg on days 4 and 5. The increasing doses of morphine were used to overcome any tolerance developed and to better simulate addiction to morphine. Control rats were given physiological saline (1 mL/kg i.p.) twice daily. Rats in both groups were killed at 6, 9, 12, 24, and 36 h post-treatment (day 5).
Treatment with MK-801
We measured the levels of P-gp and Bcrp by western blotting in rats conditioned by subchronic morphine treatment and then given dizocilpine maleate (MK-801, Sigma-Aldrich, Lyon, France), an NMDA antagonist, to determine the influence of glutamate on the morphine-induced increases in P-gp and Bcrp. The rats were given two injections of 1 mg/kg MK-801 i.p. one 1 h and the other 4 h post morphine or saline treatment, because its half-life of elimination is only 2 h. They were killed 24 h after the last morphine or saline injection.
Treatment with Meloxicam
We studied the influence of COX-2 on the increases in P-gp and Bcrp induced by morphine treatment by giving them each a single i.p. injection of meloxicam (10 mg/kg), a COX-2 inhibitor, immediately after the last morphine or physiological saline injection. They were killed 24 h after the last morphine or saline injection.
Isolation of large vessels and microvessels from rat brain cortex
The brain cortex vessels were isolated from treated rats according to Yousif et al. 2008. Briefly, rats were anesthetized with isoflurane and decapitated. Their brains were immediately removed and placed in ice-cold Hank's buffered salt solution (HBSS). The cerebella, meninges, brainstems, and large superficial blood vessels were removed from the brains and the remaining cortices were minced in ice-cold HBSS (4 mL per gram of tissue). The minced samples were then homogenized in a Potter–Thomas homogenizer (Kontes Glass, Vineland, NJ, USA) (0.25 mm clearance), using 15–20 up-and-down strokes at 400 rpm. The resulting homogenates were centrifuged at 1000 g for 10 min. Each pellet was suspended in 17.5% dextran (64–76 kDa, Sigma-Aldrich, Lyon, France) and centrifuged for 15 min at 4400 g at 4°C in a swinging-bucket rotor. The resulting pellets were suspended in Hank's Buffered Salt Solution containing 1% bovine serum albumin (BSA), while the supernatants containing a layer of myelin were centrifuged once more for 15 min at 4400 g. The two pellets from each sample were pooled, suspended in HBSS with 1% BSA, and this suspension passed through a 100 μm nylon mesh. Large vessels (mainly 20–30 μm) were collected from the fraction retained on the 100 μm nylon mesh. The filtrate was passed through a 20 μm nylon mesh which retained the microvessels (mainly 4–6 μm).
Measurement of transcript levels
Extraction of RNA from brain tissues
Total RNA was obtained from the isolated large vessels and microvessels by lysing the surrounding basal lamina with proteinase K and then extracting the RNA using the RNeasy Micro-Fibrous Tissue micro-kit (Qiagen GmbH, Hilden, Germany) according to the manufacturer's instructions. The isolated RNA samples were purified from contaminating genomic DNA by treatment with DNase (RNase-Free DNase Set, Qiagen, SA). The concentration and purity of the RNA samples obtained were assessed spectrophotometrically using a Nanodrop spectrophotometer (Nanodrop ND-1000; NanoDrop Technologies, Wilmington, DE, USA), and the integrity of the RNA was checked by electrophoresis through 0.8% agarose gels.
Total RNA samples (1 μg) from brain vessels or cultured hCMEC/D3 cells were reverse transcribed into cDNA in a final volume of 20 μL. The mixture consisted of 1 μg total RNA, 500 μM of each deoxyribonucleotide triphosphate, 10 mM dithiothreitol, 1.5 μM random hexa-nucleotide primers, 20 U RNAsin ribonuclease inhibitor, and 100 U Superscript II Rnase reverse transcriptase. Reagents were purchased from Invitrogen (Cergy-Pontoise, France). Reverse transcription was performed on a programmable thermal cycler (PTC-100 programmable thermal controller; MJ Research Inc., Waltham, MA, USA). Hexamers were annealed at 25°C for 10 min, the products were extended at 42°C for 30 min, the reaction terminated by heating at 99°C for 5 min, and samples were quick-chilled to 4°C.
Real-time quantitative PCR
The effect of morphine on the transcripts levels was investigated by qRT-PCR as previously described (Yousif et al. 2007). Specific primer sequences were designed using OLIGO 6.42 software (Medprobe, Oslo, Norway) (Table 1) and these primers were synthesized by Invitrogen Life Technologies (Invitrogen). Fluorescent PCR reactions were performed on a Light Cycler thermal cycler (Light-Cycler® instrument; Roche Diagnostics) using the LC-FastStart DNA Master SYBR Green I kit (Roche Diagnostics, Meylan, France). The PCR reaction mixture contained 1 μL LC-FastStart DNA Master SYBR Green 1 mix, 1.2 μL of 10 mM MgCl2, 0.5 μL of each upper and lower primer (final concentration 0.5 μM), and 1.8 μL water. The cDNAs were diluted 50-fold and 5 μL aliquots were mixed with an equal volume of PCR mixture to give a final volume of 10 μL. The thermal cycling conditions were 8 min at 95°C, followed by 40 amplification cycles at 95°C for 5 s, 64°C for 5 s, and 72°C for 5 s. Gene expression in each sample was normalized on the basis of its β-actin gene expression (rat samples) or TATA box-binding protein (TBP) gene expression (human samples). The change was calculated from the ratio of the expression of the gene of interest to that of the housekeeping gene (β-actin or TBP) in morphine-treated and control samples.
|Gene||Forward primer (5′–3′)||Reverse primer (5′–3′)||Length (bp)|
Immunoblots were performed on protein extracted from both large vessels and microvessels isolated at different times after the last dose of morphine or saline to measure, semi-quantitatively, the amounts of P-gp and Bcrp. β-actin was the reference protein. After its extraction, the protein content from samples of large or microvessels was determined with the Bradford reagent (Sigma-Aldrich, Lyon, France) and a BSA calibration curve. Equal amounts of proteins were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis on 8% acrylamide/bisacrylamide gels and the separated proteins were transferred electrophoretically to a nitrocellulose membrane (2 h at 80 V). Free sites on the membrane were blocked by incubation overnight at 4°C with 5% non-fat milk in Tris-buffered saline containing 0.1% Tween 20. The membranes were immunoblotted by incubation overnight at 4°C with mouse monoclonal antibody C219 against rat P-gp (diluted 1 : 100) (Abcam, Cambridge, UK), or rat monoclonal anti-rat Bcrp antibody BXP-53 (diluted 1 : 150) (Abcam) The membranes were then washed and incubated for 1 h at 25°C with horseradish peroxidase-conjugated anti-mouse IgG (diluted 1 : 20 000) or anti-rat IgG (1 : 10 000) (GE healthcare, Buckinghamshire, UK). The membranes were washed again and exposed to Amersham ECL blotting detection reagent (ECL, Amersham Biosciences Europe GmbH, Orsay, France). Signals were revealed using the Bio-Rad ChemiDoc® XRS imaging device (Marnes La Coquette, France). The blots were then stripped and reprobed with monoclonal mouse anti-β-actin antibody AC-74 (Sigma-Aldrich, Lyon, France) (diluted 1 : 10 000). Proteins were quantified using the Quantity One data analyser software v. 4. 6. 1 (Bio-Rad) and normalized to β-actin.
Effect of morphine, glutamate, iloprost, and misoprostol on the expression of P-gp and BCRP in the hCMEC/D3 cells
The hCMEC/D3 cells were kindly donated by Dr. P.O. Couraud of the Institut Cochin, University Paris Descartes, Paris, France. The immortalized hCMEC/D3 cells were cultured as previously described (Dauchy et al. 2009). Cells were grown on plates coated with rat-tail collagen type I in EBM-2 medium supplemented with 1 ng/mL bFGF, 2.5% FCS, ascorbate acid, 10 mM HEPES, penicillin–streptomycin at 37°C in a saturated atmosphere of 5% CO2, 95% air. The culture medium was changed every 3–4 days.
Cells were cultured for 24 h in medium alone (controls) or in medium containing morphine (0.01, 0.1, 1, or 10 μM), glutamate (10 or 100 μM), iloprost (1 or 10 μM), or misoprostol (1 or 10 μM). Total RNA was extracted from the hCMEC/D3 cells using the RNeasy mini kit (Qiagen GmbH, Hilden, Germany) according to the manufacturer's instructions, including DNAse I treatment to remove genomic DNA. The concentrations and purity of the RNA samples were assessed. The expressions of the MDR1 and BCRP genes were evaluated by qRT-PCR using the procedure outlined above. The reference genes were TBP and β-actin.
Data were analyzed using GraphPad Prism® 4.0 software (San Diego, CA, USA). The results are expressed as means ± SD. Student's unpaired t-test was used to identify significant differences between in vivo morphine and saline groups. The effect of morphine on hCMEC/D3 cells was analyzed by one-way anova and a post hoc Tukey's test. All the tests were two-tailed and statistical significance was set at p < 0.05.
Expression of Mdr1a and Bcrp gene in the cortex vessels of morphine-treated rats
The levels of Mdr1a and Bcrp mRNAs were determined in cortex large vessels and microvessels of rats 3, 6, and 24 h after a single dose of saline or morphine (n = 4 per group). The levels of neither the Mdr1a nor the Bcrp gene were significantly altered by morphine at any of the times tested (data not shown).
We also determined the effect of a 5-day morphine treatment on the levels and kinetics of Mdr1a and Bcrp mRNAs in cortex microvessels (Fig. 1a and 2a) and large vessels (Fig. 1b and 2b). The amounts of Mdr1a and Bcrp transcripts in both microvessels and large vessels were unchanged 6 h after the last dose of morphine. But, the levels of both Mdr1a and Bcrp genes increased starting at 9 h after the last morphine dose. The amounts of Mdr1a transcripts in microvessels were significantly increased by 22% at 12 h and by 36% at 24 h after the last dose of morphine (Fig. 1a). Mdr1a mRNA levels increased more in large vessels than in microvessels. The levels of Mdr1a transcripts 12 h after the last morphine dose were increased by 60% in large vessels and by 22% in microvessels (Fig. 1a and b). However, levels of Mdr1a transcripts in both types of vessels returned to baseline 36 h after the last dose of morphine (Fig. 1a and b).
The levels of Bcrp transcripts in both the microvessels and large vessels of morphine-treated rats were roughly double those of controls, with the increase occurring earlier in large vessels (Fig. 2a and b). Thus, the amount of Bcrp transcript was 120% greater than that of controls 12 h after the last morphine injection in large vessels, whereas it was only 40% greater in microvessels. Bcrp gene expression was still significantly greater than in controls 36 h after the last dose of morphine, in both the large vessels and microvessels, unlike Mdr1a expression (Fig. 2b). These data indicate that the levels of both Mdr1a and Bcrp transcripts were increased in rat brain vessels from 12 h after the last dose of a subchronic morphine treatment.
P-gp and Bcrp in the cortex vessels of subchronic morphine-treated rats
Western blot experiments with the anti-P-gp C-219 antibody revealed an immunoreactive protein of about 170 kDa, corresponding to P-gp, in isolated large vessels and microvessels, as previously reported (Yousif et al. 2007). The levels of P-gp were no higher than those of controls 6 h after the last dose of morphine. But, its level in both microvessels and large vessels was 1.5-times that of controls 24 h after the last dose of morphine. Like mRNA levels, morphine increased the expression of P-gp more in large vessels than in microvessels. The P-gp expression in microvessels decreased back to baseline at 36 h (Fig. 3).
We also determined the kinetics of Bcrp expression in the samples that were used for P-gp analysis. Just like its mRNA, the amount of Bcrp protein was unaltered 6 h after the last dose of morphine, but had increased 1.45-fold in large vessels and 1.6-fold in microvessels by 24 h post treatment (Fig. 4) and the amount of Bcrp in morphine-treated rats was still higher than in controls 36 h post treatment.
Thus, the levels of P-gp and Bcrp proteins were increased in brain vessels of morphine-treated rats from 24 h post treatment, as were the levels of Mdr1a and Bcrp transcripts from 12 h after the last morphine dose.
Effect of morphine on the levels of MDR1 and BCRP transcripts in human hCMEC/D3 cells
We tested the effect of morphine on immortalized human brain capillary endothelial (hCMEC/D3) cells to determine whether the up-regulation of P-gp and Bcrp in brain vessels by morphine involved only the vascular endothelial cells or the complete neurovascular unit (Weksler et al. 2005). Cells were exposed to concentrations of morphine from 0.01 to 10 μM for 24 h and the amounts of MDR1 and BCRP gene transcripts were determined by qRT-PCR. Morphine did not alter the levels of either MDR1 or BCRP gene transcripts in hCMEC/D3 cells at any of the concentrations used (Fig. 5).
Effect of MK-801 and meloxicam on the expression of P-gp and Bcrp in cortex vessels
As endothelial cells may not be the direct target of morphine to increase P-gp and BCRP expression, we looked for the pathway by which it might act in rat cortex vessels. It was recently shown that exposure of isolated rat brain microvessels to glutamate ex vivo significantly increased P-gp expression by activation of the NMDA/COX-2 pathway (Bauer et al. 2008). We therefore treated rats with morphine with or without the NMDA antagonist MK-801 to determine whether activation of NMDA receptors by glutamate was involved in the up-regulation of P-gp and Bcrp. The amounts of P-gp and Bcrp measured by western blotting indicated that MK-801 totally blocked the morphine-dependent up-regulation of P-gp and Bcrp in large vessels and microvessels 24 h after the last morphine dose; their expressions in rats given morphine plus MK-801 were comparable to those in control rats (Fig. 6 and 7). MK-801 had a less pronounced effect on Bcrp expression in large vessels than in microvessels. Similarly, meloxicam, a COX-2 inhibitor, totally blocked the morphine-dependent up-regulation of P-gp and Bcrp in large vessels as well as in microvessels 24 h after the last morphine dose (Fig. 6 and 7). Neither MK-801 nor meloxicam given alone changed the expression of Mdr1a and Bcrp; their activities were the same as in saline-treated rats (data not shown). Taken together, these results suggest that NMDA/COX-2 pathway is implicated in morphine-dependent Bcrp and P-gp induction.
Effect of glutamate, iloprost, and misoprostol on the levels of MDR1 and BCRP transcripts in human hCMEC/D3 cells
We tested the effect of glutamate on MDR1 and BCRP transcripts in hCMEC/D3 cells to determine whether glutamate was directly involved in the regulation of P-gp and BCRP. Cells were exposed to 10 and 100 μM of glutamate for 40 min, washed, and then incubated in glutamate-free medium for 5.3 h before the amounts of MDR1 and BCRP gene transcripts were analyzed by qRT-PCR. The relative expressions of MDR1 and BCRP were not significantly different from those in non-treated cells (Fig. 8a and b). We also measured the transcript levels of the NR1 gene of the NMDA receptor in hCMEC/D3 cells. NR1 transcript levels were very low in hCMEC/D3 cells (cycle threshold > 32 for a dilution 1 : 20 of the cDNA reversed transcribed from 1 μg of total RNA), that may explain the absence of effect of glutamate on hCMEC/D3 cells.
As prostaglandin E2 (PGE2) is the major product of COX-2, we also tested the effect of two PGE2 analogs, iloprost and misoprostol, on hCMEC/D3 cells to determine whether COX-2 activity is involved in the regulation of P-gp and BCRP expression. Cells were exposed to 1 μM or 10 μM of iloprost or misoprostol for 6 h and the amounts of MDR1 and BCRP gene transcripts were also determined by qRT-PCR. Both concentrations of iloprost significantly increased the concentrations of MDR1 and BCRP gene transcripts. Thus, 10 μM iloprost increased MDR1 transcripts three-fold over controls and BCRP transcripts two-fold (Fig. 8c and d). Misoprostol had weaker effects on the levels of MDR1 and BCRP transcripts; 10 μM misoprostol produced just a significant increase in MDR1 expression over the control and a weaker increase in BCRP expression.
Opioids, particularly morphine, are the standard analgesics used in the clinical management of chronic and severe pain. However, patients require greater and greater doses to maintain the same level of analgesia as they become tolerant to their central effects, a phenomenon known as central tolerance. The animal treatment protocol, with frequent increasing doses reflects the typical pattern of opioid abuse. The subchronic morphine treatment, twice daily for 5 days, allows the serum concentration of morphine to fluctuate, with peaks and troughs of morphine. These fluctuations generate withdrawal after the last dose of morphine as already reported in our laboratory (Desjardins et al. 2008). This profile of drug administration is also responsible for the development of tolerance to morphine (Way et al. 1969; Kalivas and Duffy 1987). P-gp, a key-element in the control of the brain distribution of several opioids at the BBB, including morphine, may be responsible for this tolerance, as its expression is also increased by morphine. There have been two reports of P-gp induction in the whole brains of mice (Zong and Pollack 2003) and rats (Aquilante et al. 2000) after subchronic exposure to morphine. Aquilante et al. 2000 measured the amount of P-gp the day after the last dose of morphine, while Zong and Pollack 2003 did so 24 h after the last morphine dose. But, neither of these studies distinguished the effect of morphine during treatment from that occurring during withdrawal. We have previously used isolated rat brain microvessels to show that the amount of P-gp at the BBB increased as early as 12 h after a 5-day morphine treatment; this treatment was similar to that used in the studies cited above (Yousif et al. 2008). The subchronic morphine treatment used in this study is widely used for behavioral and neurobiological studies in our laboratory. We showed, recently, that it triggers withdrawal that started as early as 9 h after the last dose of morphine (Desjardins et al. 2008). The working hypothesis for this study was that the induction of P-gp and Bcrp in the rat BBB was because of the withdrawal syndrome following subchronic morphine treatment rather than from the direct effect of morphine at the BBB during the morphine treatment. A single dose of morphine (10 or 40 mg/kg, i.p.) did not increase the levels of P-gp mRNA in either type of rat brain vessels at 3, 6, and 24 h. Similarly, the levels of P-gp transcripts and protein were not modified 6 h after a 5-day morphine treatment, suggesting that morphine itself does not stimulate P-gp synthesis during this treatment. Conversely, the levels of Mdr1a transcripts and P-gp protein increased between 9 h and 24 h after this morphine treatment. Therefore, these in vivo experiments suggest that morphine does not directly induce P-gp up-regulation at the brain endothelial, but triggers other mechanisms during withdrawal that follows its chronic administration which then induce P-gp synthesis. This is supported by our in vitro experiments showing that morphine did not stimulate the synthesis of MDR1 transcripts in hCMEC/D3 human endothelial cells, although P-gp synthesis by these cells is increased by rifampin (Zastre et al. 2009). Our data are also supported by other in vitro experiments demonstrating the absence of P-gp induction in several cancer cell lines exposed to morphine (Pajic et al. 2004). In contrast, Pal et al. 2011 have shown that prolonged exposure (15 days) of caco-2 cells to 3 μM morphine increased their MDR1 mRNA contents, suggesting that effect of morphine on P-gp expression may depend on tissue of origin and length of exposure (Pal et al. 2011).
We therefore postulate that morphine-mediated stimulation of P-gp synthesis occurs during the spontaneous withdrawal syndrome that follows the end of a 5-day morphine treatment rather than to a direct effect of morphine on P-gp expression. Guo et al. (2005) found that the extracellular concentration of glutamate in the hippocampus of mice determined by microdialysis was decreased after acute or subchronic morphine treatment. But, they also showed that the extracellular glutamate concentration was significantly increased in an artificial morphine withdrawal situation induced by the opioid antagonist naloxone (Guo et al. 2005). There is also good evidence from studies on rodents with experimental epilepsy that the massive release of glutamate or intracerebroventricular injection of kainate results in the activation of glutamate receptors in the hippocampus and the increase in P-gp expression (Zhu and Liu 2004; Potschka 2010). Glutamate also increases P-gp expression in primary cultures of brain endothelial cells (Zhu and Liu 2004). It has also been shown recently that exposure of isolated rat brain microvessels to glutamate ex vivo significantly increases P-gp synthesis through the activation of NMDA receptors, which increases the activity of COX-2, which, in turn produces prostaglandin E2. It is believed that the effect of COX-2 on P-gp expression is mediated by PGE2 acting on E-type prostanoid receptor (EP) 1 receptors. PGE2 is a major product of COX-2 signaling in the brain. MRP4, a member of the multidrug-resistance proteins (MRPs) belonging to the ABC protein superfamily, takes part in the brain-to-blood efflux transport of PGE2 at the BBB. It is thus likely to be involved in the release of PGE2 from the endothelial cells and so to act as a signaling molecule (Reid et al. 2003; Akanuma et al. 2010). It acts on four different G protein-coupled receptors (EP1, EP2, EP3, and EP4), which mediate diverse effects (Hata and Breyer 2004). EP1 receptor expression has already been found in isolated rat brain capillaries by (Pekcec et al. 2009), and its blockade by SC-51089, a specific antagonist of EP1 receptor, abolished the glutamate-induced increase of P-gp expression in brain capillaries, suggesting that the EP1 receptor plays a key role in the signaling events that regulate P-gp expression via glutamate.
The regulation of P-gp gene transcription at the BBB is rather complex and is far from being completely understood (Miller 2010). Several factors that are known to be involved in the regulation of the MDR1 promoter. NF-kB is a ubiquitous transcription factor that has been shown to up-regulate P-gp expression and function in several cancer cells (Bentires-Alj et al. 2003). Upon activation of NF-kB, PI3K/Akt are upstream signals which lead to the phophorylation of IkB proteins by IkB kinase. The phosphorylated IkB allows the release of NF-kB, which is translocated to the nucleus, where it activates its target genes (Karin 1999). EP1 receptor signaling is coupled to the phospholipase C/inositol triphosphate pathway, leading to activation of PI3K/Akt and mobilization of intracellular calcium (Asbóth et al. 1996). PGE2 also activates NF-kB through EP1 receptors in colonocytes and strongly induces IkB phophorylation, (Kim et al. 2007). Hence, we postulate that these mechanisms might be present at the BBB and the downstream effect of COX-2 on the regulation of P-gp and BCRP expression may be through a PGE2-EP1-PI3K/Akt-NF-kB signaling pathway. The β-catenin signaling pathway might also be behind the effect of NMDAr/COX-2 on the regulation of P-gp and BCRP expression. Lim et al. 2008. showed that stabilizing and activating β-catenin signaling increases P-gp expression and that this was functionally significant (Lim et al. 2008). Moreover, manipulating β-catenin signaling also affects the expression of other ABC transporters, such as BCRP (Scotto 2003). The EP1 receptor contributes significantly to the PGE2-mediated stabilization of β-catenin by activating phospholipase C and PI3K, and thus strongly increases the amount of β-catenin in the nucleus (Lee et al. 2004). The stabilization and subsequent translocation of β-catenin leads to the activation of the Tcf-/Lef-1 transcription factor, which alters the expression of several genes, including the MDR1 gene (Fujino et al. 2002).
Thus, these data on the release of glutamate and its effect on P-gp support our hypothesis that the induction of P-gp synthesis in brain large vessels and microvessels of rats treated with morphine for 5 days involves the glutamatergic system and is because of the glutamate released during spontaneous morphine withdrawal. This induction driven by morphine withdrawal may occur through activation of the NMDA receptor by glutamate and COX-2 activity. We tested the direct effect of glutamate on MDR1 and BCRP transcript production by incubating hCMEC/D3 cells with two concentrations of glutamate (10 and 100 μM). Glutamate had no influence on MDR1 and BCRP expression, the transcript levels were the same as in the control (Fig. 8). We also checked the mRNA levels encoding for the NR1 subunit of the NMDA receptor by qRT-PCR on cDNA from hCMEC/D3 cells. There was very low expression of the NR1 subunit of the NMDA receptor in these cells, which agrees well with the finding that glutamate had no effect on MDR1 and BCRP mRNA levels in hCMEC/D3.
As we observed the greatest expression of P-gp in brain vessels 24 h after the last morphine dose, we dosed morphine-treated rats with two injections of 1 mg/kg MK-801 or a single injection of meloxicam after the last dose of morphine to see whether the NMDA/COX-2 pathway could be involved. Both MK-801 and meloxicam completely blocked the over-expression of P-gp in both microvessels and large vessels. Although meloxicam is not a very specific inhibitor of COX-2, it consistently proved to be a preferential inhibitor of COX-2 rather than of COX-1, in contrast to a group of standard NSAIDs (Engelhardt 1996; Degner et al. 1998). COX-1 has not been shown to be constitutively expressed in the endothelial cells of the human BBB (Tomimoto et al. 2002), whereas there is evidence that COX-2 is (Zibell et al. 2009). This indicates that the observed effect of meloxicam is more likely to be because of inhibition of COX-2 than of COX-1. As hCMEC/D3 cells express no NMDA receptor, exposing this human cell line to MK-801 or meloxicam should have no effect on the expression of MDR1 and BCRP genes. We incubated hCMEC/D3 cells with PGE2 analogs, iloprost and misoprostol, to determine the involvement of COX-2 in the regulation of P-gp and BCRP in these cells and to study the downstream cascade. Iloprost is a structural analog of prostacyclin whose affinity for EP1 receptors is similar to that of PGE2, whereas misoprostol is a PGE1 analog whose agonist activity is mediated by EP2, EP3, and EP4 receptors (Abramovitz et al. 2000). Iloprost increased the amounts of both MDR1 and BCRP gene transcripts, whereas the highest concentration tested of misoprostol increased MDR1 transcription but had little effect on BCRP expression (Fig. 8c to 8f). These findings also point to COX-2 activity mediated by PGE2 should act via EP1 receptor rather than via EP2–EP4 receptors. We believe this is clear evidence that the glutamate release and COX-2 activity after the last morphine dose are involved in the induction of P-gp up-regulation. We therefore suggest that repeated doses of morphine without the withdrawal syndrome may not affect P-gp synthesis during exposure to morphine. However, the glutamate released during the spontaneous morphine withdrawal syndrome does affect P-gp expression and may be part of the mechanism underlying morphine tolerance, even though the small increase in Mdr1a mRNA expression found here cannot be the major cause of tolerance as morphine is a poor substrate of P-gp. Schinkel et al. 1995 showed that the concentration of morphine in the brains of Mdr1a ‘knockout’ mice was only 1.7-fold greater than that in the brains of normal mice (Schinkel et al. 1995). These pre-clinical data may have clinical relevance as patients treated with opiates for pain relief (oxycodone, morphine, hydromorphone) or heroin maintenance (methadone) suffer from withdrawal syndrome. This could increase P-gp expression and thus decrease the pharmacological effect of opiates that are P-gp substrates, but most importantly decrease the efficacy of anti-cancer drugs that are frequently given with morphine, and of methadone for heroin addiction, as these drugs are better substrates of P-gp than morphine.
We have also measured the kinetics of P-gp induction in large vessels and microvessels. P-gp synthesis is not uniform throughout the rat brain vasculature, as there is less P-gp in small arteries than in microvessels and small veins of the brain cortex (Saubamea et al. 2012). Our results show that the increase in P-gp expression was more pronounced in large vessels than in microvessels, suggesting that small arteries and veins are more sensitive to the released glutamate. We also find that the expression of Bcrp, like that of P-gp, is increased following subchronic morphine treatment. There is a lack of information on the effects of opioids on BCRP activity despite increasing evidence that BCRP at the BBB modulates the brain distribution of drug substrates. We showed recently that buprenorphine and its main metabolite in human, norbuprenorphine, are inhibitors of BCRP, whereas opioids appeared to be poor substrates of BCRP (Tournier et al. 2010). However, we clearly showed that the concentration of Bcrp in rat brain vessels was doubled 36 h after the last dose of a 5-day morphine treatment. The increase in Bcrp synthesis was higher than that of P-gp and more prolonged. The concentration of Bcrp did not return to baseline 36 h after the last dose of morphine, unlike that of P-gp. This difference in Bcrp and P-gp induction might mean that there is a distinct transcriptional mechanism or post-transcriptional regulation during the activation of the Mdr1a and Bcrp genes, although they both involve NMDA receptor activation and COX-2 activity. These higher concentrations of Bcrp protein are positively correlated with greater Bcrp gene expression 36 h after the last dose of morphine, in contrast to the findings for Mdr1a and P-gp. These distinct results could also be related to the difference in the half-lives of these two transporters: P-gp has a half-life of 14–17 h (Muller et al. 1995), whereas the estimated half-life of BCRP is longer around 60 h (Peng et al. 2010). Thus, de novo synthesized P-gp might be eliminated faster than BCRP. This is, to our knowledge, the first time that Bcrp expression has shown to be regulated via glutamate-NMDA-COX-2 activation. The increase in BCRP expression that follows morphine treatment may thus decrease the penetration of drugs that are BCRP substrates into the brain. The substrates of BCRP that are used to treat humans are mainly anti-cancer drugs like cytotoxic drugs and tyrosine kinase inhibitors. Erlotinib and gefitinib, two tyrosine kinase inhibitors used to treat several solid cancers, are well-known substrates of BCRP and their brain distributions are restricted because of the high concentration of BCRP at the BBB (Chen et al. 2011; de Vries et al. 2012). This low brain penetration is a major issue in the treatment of brain metastasis that is often observed in patients suffering lung cancer and metastatic renal cancer. There is also evidence that the expression of BCRP in isolated human brain microvessels is higher than that of P-gp (Shawahna et al. 2011), making BCRP a key element in controlling the access of its substrates to the brain. As morphine is widely prescribed to manage severe pain in patients with cancer, the up-regulation of BCRP following intermittent morphine treatment and withdrawal syndromes may increase the expression of BCRP in the human BBB, which could restrict the distribution within the CNS of any anti-cancer drugs that are substrates of BCRP.
In conclusion, we have clearly shown that the expression of P-gp and Bcrp at the rat BBB is not increased at the end of a subchronic morphine treatment that triggers tolerance to the analgesic effects of morphine. Thus, our data do not support the concept that P-gp auto-induction is involved in the mechanism of tolerance often observed with morphine. Conversely, the expression of P-gp and Bcrp begins to increase 9 h after the last dose of a subchronic morphine treatment, which coincides with the appearance of the morphine withdrawal syndrome. We suggest that morphine withdrawal syndrome occurring after a subchronic morphine treatment leads to the activation of the NMDA/COX-2 pathway whose consequences are increases in P-gp and Bcrp expression at the rat BBB.
Catarina Chaves acknowledges Fundação para a Ciência e Tecnologia (FCT) for her PhD grant [SFRH/BD/79196/2011].
The authors have no conflict of interest to declare.