Evaluation of the role of P-glycoprotein in the uptake of
- 格式：pdf
- 大小：234.18 KB
- 文档页数：8
Evaluation of the role of P-glycoprotein in the uptake of paroxetine,clozapine,phenytoin and carbamazapineby bovine retinal endothelial cellsLynn W.Maines a ,David A.Antonetti b ,Ellen B.Wolpert b ,Charles D.Smith a ,*aDepartment of Pharmacology,The Pennsylvania State University College of Medicine,500University Drive,Hershey,PA 17033,USAbDepartment of Cellular and Molecular Physiology,The Pennsylvania State University College of Medicine,Hershey,PA 17033,USAReceived 1February 2005;received in revised form 6April 2005;accepted 15April 2005AbstractExpression of the drug transport proteins,including P-glycoprotein (Pgp),in the brain vascular endothelium represents a challenge for the effective delivery of drugs for the treatment of several central nervous system (CNS)disorders including depression,schizophrenia and epilepsy.It has been hypothesized that Pgp plays a major role in drug efflux at the blood e brain barrier,and may be an underlying factor in the variable responses of patients to CNS drugs.However,the role of Pgp in the transport of many CNS drugs has not been directly demonstrated.To explore the role of Pgp in drug transport across an endothelial cell barrier derived from the central nervous system,the expression and activity of Pgp in bovine retinal endothelial cells (BRECs)and the effects of representative CNS drugs on Pgp activity were examined.Significant Pgp expression in BRECs was demonstrated by western analyses,and expression was increased by treatment of the cells with hydrocortisone.Intracellular accumulation of the well-characterized Pgp-substrate Taxol was markedly increased by the non-selective transporter inhibitor verapamil and the Pgp-selective antagonist PGP-4008,demonstrating that Pgp is active in these endothelial cells.In contrast,neither verapamil nor PGP-4008affected the intracellular accumulation of [3H]paroxetine,[14C]phenytoin,[3H]clozapine or [14C]carbamazapine,indicating that these drugs are not substrates for Pgp.Paroxetine,clozapine and phenytoin were shown to be Pgp inhibitors,while carbamazapine did not inhibit Pgp at any concentration tested.These results indicate that Pgp is not likely to modulate patient responses to these drugs.Ó2005Elsevier Ltd.All rights reserved.Keywords:P-glycoprotein;Paroxetine;Clozapine;Phenytoin;Carbamazapine;Blood e brain barrier1.IntroductionThe blood e brain barrier (BBB)presents a challenge for the treatment of several CNS disorders,since many drugs are not able to effectively transverse the vascular endothelium to reach their targets in the parenchyma.It is therefore thought that treatment of these diseases could be improved if drug delivery across the BBB couldAbbreviations:BBB,blood e brain barrier;BRECs,bovine retinal endothelial cells;CNS,central nervous system;MEM,Dulbecco’s minimum essential medium;PBS,phosphate-buffered saline;Pgp,P-glycoprotein;SSRIs,selective serotonin reuptake inhibitors;TBS,Tris-buffered saline.*Corresponding author.Tel.:C 17175311672;fax:C 17175315013.E-mail address:cdsmith@ (C.D.Smith).0028-3908/$-see front matter Ó2005Elsevier Ltd.All rights reserved.doi:10.1016/j.neuropharm.2005.04.028Neuropharmacology 49(2005)610e617/locate/neuropharmbe increased(Abbott and Romero,1996;El Ela et al., 2004;Schinkel et al.,1996).Additionally,there is marked and unpredictable variability in patient response to chemotherapy for many CNS illnesses,including a large percentage of complete‘‘nonresponders’’for several diseases.For example,approximately half of all patients prescribed selective serotonin reuptake inhib-itors(SSRIs)for depression either do not respond to treatment or do not experience a sustained response (Kaplan,2002).Additionally,30%of patients with panic disorder do not respond to SSRIs(Zamorski and Albucher,2002).Despite recent advances in schizophre-nia,there also remains a significant proportion of patients who do not respond well to pharmacological intervention and are commonly labeled‘‘treatment resistant’’(Pantelis and Lambert,2003).Similar diffi-culties are also present with epilepsy patients,with drugs that have been in the clinics for several years producing highly varied effectiveness(Weiss et al.,2003a;Sisodiya et al.,2002).It has been hypothesized that at least one cause of these variations in response to treatment is due to variable P-glycoprotein(Pgp)expression and function in patients(El Ela et al.,2004;Schinkel et al.,1996; Loscher and Potschka,2002;De Luca et al.,2003;Lee and Bendayan,2004).P-glycoprotein is an energy-dependent drug transport protein that effectively effluxes a variety of drugs from cells.Since Pgp is expressed in vascular endothelial cells within the brain,it is thought that this protein provides a protective mechanism for excluding toxic compounds from the brain(Cordon-Cardo et al.,1989;Terasaki and Hosoya,1999).There-fore,varied expression of endothelial Pgp activity among individuals may underlie the inconsistency of response often seen in large numbers of patients treated for neurological disorders.For example,Pgp expression has even been shown to have significant gender differences associated with varied responses between men and women to both monoamine oxidase and tricyclic amine antidepressant treatments(Bies et al.,2003).Previous studies have suggested that a number of antidepressant,antipsychotic and antiepileptic drugs may be Pgp substrates(Abbott and Romero,1996; Loscher and Potschka,2002;Lee and Bendayan,2004). However,these studies have generally used indirect assays that do not distinguish Pgp substrates from allosteric inhibitors of drug transport(Weiss et al., 2003a,b;Boulton et al.,2002).Several other studies have used transgenic mice in which the genes encoding Pgps have been deleted,i.e.mdr1a/b knockout mice,and have shown increased penetration of drugs into the brain (Schinkel et al.,1996;Uhr et al.,2003;Rizzi et al.,2002). As discussed below,the results from these in vivo studies are also somewhat ambiguous because of uncertainty about the expression of the many other drug trans-porters expressed in the brain(de Lange,2004;Choudhuri et al.,2003;Ghersi-Egea and Strazielle, 2002).Therefore,clarification of the role of Pgp in the transport of these drugs is needed.In the present study,we have used bovine retinal endothelial cells(BRECs)as a cellular model of Pgp-mediated drug transport.Rat retinal endothelial cells have been successfully used in transport studies as a model of the blood e retinal barrier(Greenwood,1992; Shen et al.,2003).Similarly,the model of the present report has been established to be an excellent model of the BBB as it is a blood e nerve barrier that forms tight junctions(Burdo et al.,2003;Haselton et al.,1996; Antonetti et al.,2002).This model has also been shown to be responsive to hydrocortisone with increased occludin expression and tight junction formation consistent with that observed in the blood e brain barrier(Antonetti et al., 2002).In the current studies,P-glycoprotein is shown to be expressed and active in these cells,and is sensitive to established non-selective and selective inhibitors.Clo-zapine,paroxetine,phenytoin and carbamazapine are found to not be substrates for Pgp;however,clozapine and paroxetine are effective inhibitors of Pgp-mediated drug transport.2.Materials and methods2.1.Materials[3H]Taxol was obtained from Moravek Biochemicals (Brea,California),[3H]paroxetine and[14C]phenytoin were purchased from PerkinElmer(Boston,MA), [14C]carbamazapine and verapamil was obtained from Sigma(St.Louis,MO)and[3H]clozapine was purchased from American Radiolabeled Chemicals(St.Louis, MO).The radiochemical purity of all of the labeled compounds exceeded95%.Hydrocortisone sodium phosphate solution was purchased from Merck&Co (Rathway,NJ).PGP-4008was synthesized as previously described(Lee et al.,2004).All of the drugs were stored atÿ20 C as5mM solutions in ethanol until use. Therefore ethanol was used as a solvent control in the expression and uptake experiments described below. 2.2.Cell culturePrimary bovine retinal endothelial cell culture was carried out as described previously(Antonetti and Wolpert,2004)and rat brain endothelial cells were isolated by the same procedures as follows.For the isolation of BRECs,whole eyes from10to20 bovines were obtained from a local slaughterhouse and immediately packed in ice.In less than2h from extraction,eyes were bathed in a10%povidone-iodine solution for a minimum of5min,the vitreous and lens were extracted,and the retina was gently separated and611L.W.Maines et al./Neuropharmacology49(2005)610e617cut from the anterior portion of the eyeball using sterile tweezers in a laminar-flow hood.The retinas were rinsed three times in ice-cold MEM with D-valine containing 30mM HEPES buffer(Sigma,St.Louis,MO)and pooled in the same solution.The retinas were then passed through185m m nylon mesh stretched over a sterile porcelain funnel inserted into a sidearmflask connected to a vacuum to remove retinal pigment epi-thelial cells.The retinal tissue was suspended in ice cold MEM D-val/30mM HEPES and homogenized on ice six times in a Teflon/glass Potter-Elvehjem type tissue grinder with0.25mm clearance at250rpm.The homog-enate was centrifuged at400!g for10min at4 C. After resuspending the pelleted retinal tissue in10mL of 4 C Ca2C-Mg2C free PBS,the suspension was gently shaken and kept on ice.The isolated microvessel fragments were trapped on an88m m nylon mesh and resuspended in10mL of enzyme cocktail(made fresh at each isolation)consisting of10mL of Ca2C-Mg2C free Hank’s balanced salt solution with500m g/mL Type I collagenase(Sigma,St.Louis,MO)and200m g/mL DNase.The microvessels were incubated at37 C on a rocker for45e60min to separate the pericytes.The vessel preparation was passed over a53m m nylon mesh without suction,and the vessel fragments were washed twice with ice-cold MEM D-valine.After thefinal centrifugation,the resulting pellet was resuspended in 5mL of the standard growth medium consisting of MEM with D-valine supplemented with20%fetal calf serum(Gibco,Rockville,MD),50m g/mL of endothelial cell growth supplement(Vec Technologies,Rensslar, NY),16U/mL heparin(Fisher Scientific,Pittsburg, PA),0.01mL/mL MEM vitamins and glutamine(Sig-ma,St.Louis,MO),and0.02mL/mL antibiotic/ antimycotic(Gibco,Rockville,MD).The cells were plated on a25cm2tissue cultureflask precoated with fibronectin(Sigma,St.Louis,MO)at2m g/cm2and were grown in a humidified incubator at37 C with95% CO2,5%O2.The medium was removed and fresh medium was added24h following the plating.For the isolation of rat brain endothelial cells:Brains from rats were harvested and dissected into cortex and striatum.Microvessels and isolated endothelial cells were prepared as described above.For either source,after5e7days,colonies of cells were removed with0.05%trypsin and reseeded onto a75cm2tissue cultureflask precoated with1m g/cm2fibronectin.The cells were repeatedly subcultured with 0.05%trypsin when approximately80%confluent and expanded for experimental use at1:3ratio.The cells were routinely frozen at passage3for storage purposes. The endothelial cells were used experimentally at6e10 passages after primary culture.The medium used to culture the cells for experimentation was MCDB-131 media supplemented with10%fetal calf serum, 10ng/mL epidermal growth factor(R&D Systems,Minneapolis,MN),0.2mg/mL EndoGro,0.09mg/mL Heparin,and0.01mL/mL antibiotic/antimycotic.2.3.Western blotting for PgpProtein concentrations were determined using the fluorescamine assay(Bohlen et al.,1973)with bovine serum albumin as the standard.Samples were normal-ized for equal amounts of protein(50e100m g per lane), boiled for5min,separated using10%SDS-PAGE gels run at100mV for 2.5h,and electrotransferred to nitrocellulose membranes at40mV overnight.Mem-branes were blocked with5%nonfat milk in Tris-buffered saline(TBS)with Tween20(TBST),probed with C219anti-Pgp antibody(Signet,mouse mono-clonal)at a1:50dilution,washed and incubated with anti-mouse antibodies conjugated to horseradish perox-idase(1h in3%nonfat milk).The blots were then washed four times for5min at room temperature, developed with SuperSignal development reagents (Pierce)and exposed to Kodak XARfilms.2.4.Drug accumulation studiesBRECs were plated on24-well plates pretreated with fibronectin and allowed to reach a confluent monolayer. The cells were then washed two times with PBS and switched to400m L of serum-free media.Cells in duplicate wells were pre-incubated for30min with the Pgp modulator verapamil or PGP-4008at afinal concentration of10m g/mL(established to be optimal for both compounds in previous dose e response experi-ments with concentrations ranging from1to100m g/mL; data not shown).Medium containing0.1m Ci of radio-labeled drug([3H]paroxetine,[14C]phenytoin,[3H]clo-zapine or[14C]carbamazapine)in100m L was then added,producingfinal drug concentrations of4.6nM, 18nM,100m M,and420m M,respectively.The cultures were incubated for1h at37 C with agitation every 15min,and the radioactive medium was then aspirated and cells were washed two times with ice-cold PBS.The cells were then permeabilized by incubation for15min with1%sodium dodecylsulfate,and radioactivity in the extracts was measured using a Beckman LS3801 Scintillation Counter.Metabolism of the compounds by the BRECs in the brief times of the assay is highly unlikely since drug metabolizing enzymes, e.g.cyto-chrome P450s,are not expressed to any significant extent in these cultured cells.Additionally,the in vivo half-lives of the drugs range from12to25h,indicating that the compounds are both chemically and metabol-ically stable.Differences between control and experi-mental groups were analyzed by two-tailed unpaired t-tests.612L.W.Maines et al./Neuropharmacology49(2005)610e6173.Results3.1.Expression and function of Pgp in retinaland brain endothelial cellsCapillary endothelial cells were isolated from either bovine retinas or rat brains and analyzed for Pgp expression by western blotting(Fig.1).For comparison, freshly isolated cells and cells that had been placed in culture for three passages were probed.As demonstrated in Fig.1A,immunoreactive Pgp migrating with an apparent molecular weight of approximately190kDa was clearly present in the BREC cells,and the level of expression was moderately reduced in the cultured cells. The endothelial cells from both the rat brain cortex and striatum also expressed high levels of immunoreactive Pgp at190kDa(Fig.1B).Isolated capillary fragments, which are used for seeding cell culture,from both cortex and striatum also contained a faster-migrating form of Pgp that is likely to be non-glycosylated Pgp.In-terestingly,this form of Pgp was not maintained in therat brain endothelial cells in culture.Since Pgp very efficiently effluxes Taxol from cells, the intracellular accumulation of[3H]Taxol can be used as a measure of Pgp activity in intact cells(Lee et al., 2004).In this assay,inhibition of Pgp activity with either a non-selective inhibitor of drug transporters such as verapamil or a Pgp-selective transporter inhibitor such as PGP-4008(Lee et al.,2004)results in increased accumulation of[3H]Taxol in the cells.As demonstrated in Fig.2,the basal accumulation of[3H]Taxol by BRECs was much lower than that of the brain endothelial cells.Furthermore,treatment of the BRECs with either verapamil or PGP-4008substantially in-creased the intracellular accumulation of[3H]Taxol. This indicates that the Pgp expressed in these cells is functionally active and can be attenuated with specific inhibitors.In contrast,neither verapamil nor PGP-4008 affected the accumulation of[3H]Taxol by capillary endothelial cells isolated from rat brain cortex or striatum.Thus,while the brain endothelial cells express Pgp,it is not functional,and consequently the BRECs appear to be a superior model for drug transport into capillary endothelial cells.Since hydrocortisone has been previously shown to increase the barrier properties of BRECs in culture associated with increased tight junction protein expres-sion(Antonetti et al.,2002),we examined the effect of this steroid on the expression of Pgp.As demonstrated in Fig.3,treatment of the BRECs with hydrocortisone increased the amount of Pgp expression relative to untreated control cells.This was accompanied by a small decrease in the basal accumulation of[3H]Taxol in the hydrocortisone-treated cells(data not shown).Thus,the BRECs express functional Pgp that appears to be related to their efflux properties.3.2.Evaluation of paroxetine,clozapine,phenytoinand carbamazapineThe ability of Pgp in BRECs to efflux representative CNS drugs was evaluated using the same methodsas Fig.1.Pgp expression in capillary endothelial cells.(A)Bovine retinal endothelial cells(BRECs)were isolated as described in Section2and either immediately lysed(lane3)or placed into culture for three passages(lanes1and2).In each case,whole cell lysates were prepared,subjected to SDS-PAGE and probed for Pgp expression as described in nes1and2are two samples from different wells of cultured BRECs.(B) Endothelial cells were isolated from rat brain cortex(lanes1and2)or rat brain striatum(lanes3and4)and either immediately lysed(lanes2and4) or placed into culture(lanes1and3).Pgp levels in whole cell lysates were then determined by western blotting.A blot representative of three experiments is shown.Rat Cortex Rat Striatum BRECs[3H]Taxolaccumulation(picomoles)Fig. 2.Evaluation of Pgp activity in endothelial cells.Rat brainendothelial cells or BRECs were cultured in24-well plates toconfluence and incubated for60min with[3H]Taxol in the presenceand absence of verapamil(V),PGP-4008(8)or ethanol(E,as thesolvent control for verapamil and PGP-4008).The intracellularaccumulation of[3H]Taxol was determined as described in Section2.Bars represent mean G SEM for duplicate samples in one of threesimilar experiments.613L.W.Maines et al./Neuropharmacology49(2005)610e617described above for [3H]Taxol.Specifically,[3H]pa-roxetine,[14C]phenytoin,[3H]clozapine,and [14C]carba-mazapine were incubated with BRECs in the absence or presence of verapamil or PGP-4008.As demonstrated in Fig.4,the extent of accumulation of these drugs varied over 1000-fold,with levels of carbamazapine>pheny-toin>paroxetine>clozapine.More importantly,neither verapamil nor PGP-4008increased the accumulation of [3H]paroxetine,[3H]clozapine,[14C]phenytoin or [14C]carbamazapine.These data demonstrate that these CNS drugs are not substrates for Pgp.Since several other compounds have been shown to act as allosteric modulators of Pgp rather than competitive substrates (Maki et al.,2003;Litman et al.,1997),the four drugs were also evaluated as Pgp inhibitors.In this case the abilities of the compounds to increase the intracellular accumulation of [3H]Taxol was used as an indicator of inhibition of Pgp activity.As demonstrated in Fig.5,clozapine,paroxetine and phenytoin each caused dose-dependent inhibition of Pgp transport of [3H]Taxol,while carbamazapine had no affect on this activity.For clozapine,25m g/mLproduced the maximal inhibitory of Pgp,while higher doses reduced [3H]Taxol accumulation readings.Similar effects were observed for both paroxetine and phenytoin with maximal inhibitory concentrations at 16and 137m g/mL,respectively,and pronounced cellular toxic-ity at higher doses.The toxicity of high doses of clozapine,paroxetine and phenytoin was directly due to these drugs since it was also present in parallel cultures that did not contain Taxol.This was easily observed microscopically as cell rounding and detachment.While high doses of verapamil and PGP-4008were also cytotoxic,their abilities to inhibit Pgp occur at much lower doses.The fact that they promoted high levels of Taxol accumulation which did not kill the cells in the timeframe of the assay also indicates that clozapine,paroxetine and phenytoin are directly cytotoxic at high doses.Thus,the data indicate that clozapine,paroxetine and phenytoin inhibit,but are not transported by,Pgp.4.DiscussionThe roles of drug transport proteins expressed in brain capillary endothelial cells in preventing the delivery of drugs into the brain parenchyma have been explored using in vitro and in vivo models.In vitro studies have included assays of the stimulation of P-glycoprotein-associated ATPase activity and transport of drugs into or across cultured cells that model the BBB,typically capillary-derived endothelial cells.In vivo models have included studies on the penetration of test drugs into the brain of animals treated with putative Pgp inhibitors or into the brain of transgenic mice in which the mdr1aE80.000.250.500.751.00[3H ]C l o z a p i n e a c c u m u l a t i o n (p i c o m o l e s )V12[14C ]C a r b a m a z a p i n e a c c u m u l a t i o n (n a n o m o l e s )E8V012345[3H ]P a r o x e t i n e a c c u m u l a t i o n (p i c o m o l e s )E8V255075[14C ]P h e n y t o i n a c c u m u l a t i o n (p i c o m o l e s )E 8V Fig.4.Pgp-mediated drug transport in BRECs.BRECs were cultured in 24-well plates to confluence and incubated for 60min with [3H]clozapine,[3H]paroxetine,[14C]carbamazapine or [14C]phenytoin in the presence of verapamil (V;diagonal hatch),PGP-4008(8;horizontal hatch)or ethanol (E;unfilled)as the control.Cells were rinsed,lysates collected and intracellular radioactivity accumulation was determined as described in Section 2.Bars represent mean G SEM forduplicate samples in one of three to five similar experiments.Fig. 3.Effects of hydrocortisone on Pgp expression in BRECs.Confluent cultures of BRECs were left untreated (lanes 1and 3)or treated with 103nM hydrocortisone (lanes 2and 4)for either 2days (lanes 1and 2)or 5days (lanes 3and 4).Whole cell lysates were subjected to SDS-PAGE and probed for Pgp expression as described in Section 2.A blot representative of two experiments is shown.614L.W.Maines et al./Neuropharmacology 49(2005)610e 617and mdr1b genes (encoding the two isozymes of Pgp in mice)have been genetically ablated.Each of these systems has advantages and disadvantages.For exam-ple,the knockout of one member of the ABC family of transporters can alter the expression of other trans-porters (Cisternino et al.,2004),making interpretation of brain penetration data somewhat ambiguous.Addi-tionally,pharmacologic studies using antagonists of drug transporters have utilized non-selective inhibitors making it impossible to ascribe effects specifically to Pgp.Conversely,the cellular models may not maintain functional expression of Pgp and related drug transport proteins.We have observed high levels of immunoreac-tive Pgp in capillary endothelial cells isolated from rat brain microvessels;however,these cells did not have functional Pgp activity as measured by the intracellular drug accumulation assay and verified using a Pgp-specific inhibitor.In the present study,we have explored the utility of BRECs as a model for drug transport across a blood e nerve barrier.This model has been previously shown to form tight junctions,and has also been shown to respond to hydrocortisone with increased occludin protein expression and tight junction formation,consis-tent with that observed in the blood e brain barrier (Antonetti et al.,2002).Our data demonstrate that Pgp is expressed by these cells,but more importantly,that it is active in effluxing substrate drugs from the BRECs.Additionally,the enhanced expression of Pgp in BRECs treated with hydrocortisone is consistent with previous reports of induction of mdr1and several other trans-porters in hepatic,renal and adrenal cells exposed to the steroid (Turncliffet al.,2004;Romiti et al.,2002;Seree et al.,1998).Since large numbers of BRECs can be isolated and maintained in culture,we believe that these cells provide a useful alternate model for drug penetra-tion across an endothelial monolayer.In view of the ambiguities with BBB models as discussed above,it is not surprising that studies of the abilities of CNS drugs to be transported by Pgp has produced disparate conclusions.Therefore,we have used the BREC model to evaluate the transport of representative antipsychotic and antiepilepsy drugs.The most direct assay for transport of a compound by Pgp (or other drug transporters)is measurement of its intracellular accumulation in cells that express the functional protein(s).To facilitate these studies,we have selected paroxetine,clozapine,carbamazapine and phenytoin because radiolabeled samples of these drugs are commercially available.The drug transport studies conclusively show that none of these drugs are significantly transported by Pgp,and paroxetine,clozapine and phenytoin are low-affinity inhibitors of Pgp.The lack of transport of carbamaza-pine by Pgp is consistent with a previous report (Weiss et al.,2003a),but conflicts with a report that accumu-lation of this drug in the brains of mdr1a/b knockout mice is greater than in control mice (Rizzi et al.,2002).For the first time,we show that phenytoin is not a substrate for Pgp;although it is a weak inhibitor of51015Clozapine ( g/ml)[3H ] T a x o l a c c u m u l a t i o n (p i c o m o l e s )[3H ] T a x o l a c c u m u l a t i o n (p i c o m o l e s)Paroxetine ( g/ml)[3H ] T a x o l a c c u m u l a t i o n (p i c o m o l e s)0.02.55.07.5Carbamazapine ( g/ml)[3H ] T a x o l a c c u m u l a t i o n (p i c o m o l e s )0.02.55.07.510.0Phenytoin ( g/ml)Fig.5.Effects of clozapine,paroxetine,carbamazapine and phenytoin on Pgp activity in BRECs.BRECs were cultured in 24-well plates to confluence and incubated for 60min with [3H]Taxol with the indicated concentrations of the test drugs.Parallel cultures were treated with ethanol (E)or verapamil (V)as negative and positive controls,respectively.Cells were rinsed,lysates collected and intracellular radioactivity accumulation was determined as described in Section 2.Bars represent mean G SEM for duplicate samples in one of two similar experiments.615L.W.Maines et al./Neuropharmacology 49(2005)610e 617Pgp activity,consistent with the report of Weiss et al. (2003a).These data also conflict with in vivo studies of phenytoin accumulation in the brains of mdr1a/b knock-out mice where a modest increase was seen relative to control mice(Schinkel et al.,1996).Another in vivo study indicated that phenytoin uptake into the brains of rats was enhanced by the administration of non-selective inhibitors of drug transport proteins,including cyanide, verapamil and PSC833(Potschka and Loscher,2001).It is likely that these in vivo effects are mediated by transport of phenytoin by proteins other than Pgp,and in fact this drug has been shown to be a substrate for transport by MRP2(Potschka et al.,2003).Although paroxetine was not a substrate for Pgp-mediated trans-port,it did inhibit Pgp function at doses as low as3m g/ mL,consistent with the data of Weiss et al.(2003b). Again,this contrasts with data indicating enhanced uptake into the brains of mdr1a/b knockout mice(Uhr et al.,2003).Finally,our data demonstrate for thefirst time that clozapine is not a substrate for transport by Pgp,even though it acts as an inhibitor of Pgp.This data is consistent with previous in vitro(Boulton et al., 2002)and cellular(El Ela et al.,2004)studies,as well as studies with mdr1a/b knockout mice(Schinkel et al, 1996).In addition to brain capillary endothelial cells,Pgp is expressed in a limited number of other normal tissues.It is well established that Pgp expression in intestinal epithelial cells limits the oral absorption of many drugs (Chan et al.,2004;Kunta and Sinko,2004),requiring them to be administered by injection.Similarly,Pgp expression in renal and hepatic tissues regulates the excretion of substrate drugs(Ayrton and Morgan,2001; Fromm,2003).Consequently,inhibition of Pgp activity by other substrate drugs or inhibitors is considered to be a prominent mechanism for adverse drug e drug inter-actions.Although paroxetine,clozapine and phenytoin are low-affinity inhibitors of Pgp,it is unlikely that they cause adverse drug e drug interactions via this mecha-nism since the doses required to inhibit Pgp are greater than10-fold higher than their circulating concentrations in patients.The combined results of this and the cited studies indicate that markedly different conclusions about the transport of CNS drugs can be drawn when comparing in vitro and cellular assays of Pgp function with in vivo brain penetration studies using the mdr1a/b knockout model.Drug accumulation by Pgp-expressing cells is the most direct and conclusive assay for determining if a compound is a substrate for transport by this protein, and the BREC model appears to be a suitable endothe-lial system for this assay.Since the expression of Pgp and other transporters is regulated by the prior history of exposure of the cells to a variety of agents,it is not surprising that ablation of the expression of Pgp in the mdr1a/b knockout model may lead to alterations in the expression of other transporters.This may underlie the lack of concordance of the in vivo studies of CNS transport with the cellular and biochemical studies. While cell culture studies may be confounded by altered transport protein expression,the present results clearly demonstrate an inability of Pgp to transport the drugs tested while the control,Taxol,was a substrate.Since a large number of transporter proteins are expressed in the brain(de Lange,2004;Choudhuri et al., 2003;Ghersi-Egea and Strazielle,2002),it appears likely that transporters other than Pgp are responsible for the BBB for many CNS drugs.Discovery and characteriza-tion of these transporters could identify sources of inter-patient variability of response,as well as lead to the development of new drugs that would be beneficial to treatment of CNS diseases.Inhibitors of these trans-porters would not only reduce response variability,but could also reduce doses needed for efficacy,providing decreased drug-drug interactions and decreased periph-eral side effects and toxicity.AcknowledgementThis project is funded by grants1R01EY016448-01 to CDS,1R01EY016413-01to DAA,and a grant from the Pennsylvania Department of Health using Tobacco Settlement Funds.The Department specifically dis-claims responsibility for any analyses,interpretations or conclusions.ReferencesAbbott,N.J.,Romero,I.A.,1996.Transporting therapeutics across the blood-brain barrier.Mol.Med.Today2(3),106e113. Antonetti, D.A.,et al.,2002.Hydrocortisone decreases retinal endothelial cell water and soluteflux coincident with increased content and decreased phosphorylation of occludin.J.Neurochem.80(4),667e677.Antonetti,D.A.,Wolpert,E.B.,2004.Isolation and characterization of retinal endothelial cells.Methods Mol.Med.89,365e374. Ayrton,A.,Morgan,P.,2001.Role of transport proteins in drug absorption,distribution and excretion.Xenobiotica31(8-9), 469e497.Bies,R.R.,Bigos,K.L.,Pollock,B.G.,2003.Gender differences in the pharmacokinetics and pharmacodynamics of antidepressants.J.Gender Specific Med.6(3),12e20.Bohlen,P.,Stein,S.,Dairman,W.,Udenfriend,S.,1973.Fluormetric assay of proteins in the nanogram range.Arch.Biochem.Biophys.155,213e220.Boulton,D.W.,DeVane,C.L.,Liston,H.L.,Markowitz,J.S.,2002.In vitro P-glycoprotein affinity for atypical and conventional anti-psychotics.Life Sci.71(2),163e169.Burdo,J.R.,et al.,2003.Mechanisms and regulation of transferrin and iron transport in a model blood-brain barrier system.Neuroscience 121(4),883e890.Chan,L.M.,Lowes,S.,Hirst,B.H.,2004.The ABCs of drug transport in intestine and liver:efflux proteins limiting drug absorption and bioavailability.Eur.J.Pharm.Sci.21(1),25e51.616L.W.Maines et al./Neuropharmacology49(2005)610e617。
