Abstract
A major functional component of the blood-brain barrier is P-glycoprotein. In principle, inhibition of this efflux transporter would permit greater distribution of its substrates into the brain and increased central effects. Tariquidar and elacridar, potent and selective P-glycoprotein inhibitors, were investigated in this regard using the opioid loperamide as an in vivo probe in mice. Pretreatment with both inhibitors converted intravenous loperamide from a drug without central effects to one producing antinociception. Radiolabeled loperamide tissue distribution studies indicated that inhibition was associated with increased uptake into brain and testes in the absence of changes in plasma levels, along with enhanced efflux of rhodamine 123 from CD3e+ T-lymphocytes. However, with tariquidar, the loperamide dose-response curves for testes/plasma and brain/plasma concentration ratios were shifted 6- (p = 0.07) and 25-fold (p < 0.01) to the right, respectively (ED50 = 1.48 and 5.65 mg/kg), compared with the rhodamine 123 efflux curve (ED50 0.25 mg/kg). Less pronounced shifts were noted with elacridar where the brain/plasma ratio was shifted only 2-fold relative to the rhodamine 123 efflux data (ED50 = 2.36 versus 1.34 mg/kg, respectively; p 0.01). These results indicate that the P-glycoprotein localized in the blood-brain barrier and, to a lesser extent, the testes-blood barrier is more resistant to inhibition than at other tissue sites such as the lymphocyte; moreover, the extent of this effect depends on the inhibitor. Such resistance can be overcome by a sufficiently high dose of an inhibitor; however, whether this is safely attainable in the clinical situation remains to be determined.
The importance of P-glycoprotein (MDR1, ABCB1) in the disposition of its substrates is now firmly established. For example, its presence in the intestinal tract, liver, and kidney tubule enhances drug elimination of both unchanged drug and metabolites. In addition, P-glycoprotein also has a barrier function and restricts the oral absorption of drugs and their distribution into the brain, testes, and fetus. Although such a protective function is teleologically highly desirable, it presents a problem from a drug delivery standpoint, because it may, for example, limit the attainment of sufficiently high drug levels in the brain to effectively treat central aspects of diseases, such as human immunodeficiency virus-1 infection, or tumors with intact brain-type vasculature.
The effectiveness of P-glycoprotein present in brain capillary endothelial cells in restricting the distribution of substrate drugs into the brain has been well documented by studies in “knockout” animals (Schinkel et al., 1994, 1995, 1996; Kim et al., 1998). For well transported drugs, such as ivermectin (Schinkel et al., 1994) and digoxin (Schinkel et al., 1995), the increase in brain concentrations may be 30 to 90-fold in the absence of the functional transporter relative to when it is present, and usually benign dosages elicit profound neurotoxicity in knockout mice (Schinkel et al., 1994, 1996).
Given the critical role for P-glycoprotein in the blood-brain barrier, the possibility of modulating the function of the transporter to increase drug distribution into the brain is an attractive concept. In fact, a number of in vivo proof-of-principle studies in animals have been undertaken using several different inhibitors of P-glycoprotein and various probe drugs. For example, drug brain levels relative to those in blood plasma of digoxin (Mayer et al., 1997, Song et al., 1999) and nelfinavir (Choo et al., 2000) were increased in mice pretreated with valspodar (PSC-833), and the neurotoxicities of ivermectin and cyclosporin A were greater following administration of the modulator (Didier and Loor, 1995). Similar studies have been reported with more recently developed inhibitors with increased potency, such as elacridar (GF-120918) and various drugs (Letrent et al., 1998, 1999; Kemper et al., 2003, 2004b). In addition, zosuquidar (LY-335979) markedly enhances nelfinavir concentrations in the brain to an extent comparable with those in mdr1 knockout animals (Choo et al., 2000) and, to a lesser degree, with those of paclitaxel (Kemper et al., 2004a). However, the logical extension of P-glycoprotein inhibition of the blood-brain barrier to humans is not straightforward.
The critical issues are the potency, selectivity, and toxicity of the modulator; i.e., can effective inhibition of transporter function in specific tissues actually be achieved using available inhibitors at dosages/concentrations without untoward effects? Experience with the most clinically developed inhibitor, valspodar, indicates that this is not a readily achievable goal in patients receiving cancer chemotherapy (Leonard et al., 2002; Thomas and Coley, 2003). In addition, most first- and second-generation inhibitors affect other disposition processes by, for example, inhibiting drug metabolism (Leonard et al., 2002; Thomas and Coley, 2003), complicating interpretation of study findings since increased brain concentrations may simply be a reflection of a higher blood level. Finally, and in contrast to animal studies, it is difficult in humans to measure the brain levels of a drug to determine whether pretreatment with a P-glycoprotein inhibitor affects drug distribution into this tissue. On the other hand, centrally mediated drug effects indicate passage of drug across the blood-brain barrier. Accordingly, the requirements for in vivo studies of modulating P-glycoprotein function in human brain capillaries requires a sufficiently potent inhibitor that does not affect the metabolism of the probe drug, i.e., selectivity, and ideally the probe drug must in turn elicit a measurable and quantitative central effect, secondary to increased brain levels.
Therefore, the purpose of this study was to obtain proof-of-principle based on the above requirements in a mouse model that would provide the necessary underpinning for extension to human subjects. The primary experimental approach involved the use of the potent and selective P-glycoprotein inhibitor tariquidar (XR-9576) and loperamide as a probe drug. Selection of the latter was based on the fact that loperamide is a potent opioid with an affinity for the m-receptor similar to morphine (Mackerer et al., 1976; Stahl et al., 1977); however, it is normally devoid of central effects because the blood-brain barrier prevents its distribution into the brain. By contrast, in mdr1 knockout mice stereotypical opiate effects are present (Schinkel et al., 1996). Additional studies were also undertaken with elacridar, another third-generation inhibitor.
Materials and Methods
Drug Solutions. Tariquidar (Xenova Ltd., Slough, Bucks, UK) for injection was dissolved in propylene glycol, 5% dextrose solution, and ethanol (4:5:1) at an initial concentration of 10 mg/ml; subsequently, this was diluted with 5% dextrose solution to provide a concentration of 1.56 mg/ml. For the flow-cytometry studies, tariquidar (3 mM) was dissolved in dimethyl sulfoxide and diluted to 60 μM with Hanks' buffered salt solution (HBSS; Gibco/Invitrogen Corp., Carlsbad, CA). In the transport studies, the stock solution was diluted with Opti-MEM I (GibcoBRL, Gaithersburg, MD) to provide a solution of 40 nM. Elacridar (GlaxoSmithKline, Research Triangle Park, NC) solution for injection was prepared as a propylene glycol/ethanol/water solution (3:2:5) at a concentration of 5 mg/ml; subsequent dilutions were made with water. Loperamide hydrochloride (Sigma, St. Louis, MO) was initially dissolved in 10% ethanol and saline solution (2 mg/ml) and dilutions made with saline to provide appropriate strength solutions for injection. When tritiated loperamide (specific activity 596 GBq/nmol, 95% purity; Janssen Pharmaceutica, Beerse, Belgium) was used, it was added to the unlabeled solution at a concentration of 8.33 mCi/ml. In the in vivo studies, all drug solutions were injected in a total volume of 4 μl/g body weight.
In Vitro Inhibition of Loperamide Transport by Tariquidar. Cultured LMDR1 cells were grown under previously described conditions (Kim et al., 1998) and plated at a density of 2 × 105 cells/12-mm well on porous (3.0 μm) polycarbonate membrane filters (Transwell; Costar Corp., Cambridge, MA). Cells were supplemented with fresh medium every 2 days and used in transport studies on the 4th or 5th day after plating. Transepithelial resistance was measured in each well using a Millicell ohmmeter (model ERS; Millipore Corp., Bedford, MA); wells registering a resistance of 200 Ω or greater after correcting for that measured in control blank wells were used in the transport studies. Approximately 1 to 2 h before the start of the transport experiment, the medium in each compartment was replaced with OptiMEM I, a serum-free medium. The transport of 3H-labeled loperamide then was measured after replacing the medium in each compartment with 700 μl of OptiMEM I with or without the radiolabeled drug. Radioactivity appearing in the opposite compartment after 1, 2, 3, and 4 h was measured in 25-μl aliquots taken from each compartment after the addition of 5 ml of ScintiVerse scintillation fluid (Fisher Scientific Co., Fairlawn, NJ). Inhibition of transport was determined in a similar fashion after the addition of tariquidar (0.001–1 μM) to both the apical and basal compartments before the addition of radiolabeled loperamide. The same experiments were performed with parental LLC-PK1 cells, which do not overexpress human P-glycoprotein. The apparent permeability of loperamide across LLC-PK1 cells was estimated as described previously (Mahar Doan et al., 2002).
Inhibition by Tariquidar of Rhodamine 123 Efflux from CD3e+T-lymphocytes. Studies of inhibition of efflux were performed in male FVB/NTTafcBr mice weighing 20 to 30 g (Taconic, Germantown, NY) after intravenous administration of tariquidar through a tail vein 30 min before sacrifice with isoflurane (Forane; Baxter Healthcare Corp, Deerfield, IL). Approval for these and other animal studies was obtained from the Vanderbilt University, Institutional Animal Care and Use Committee, and the mice were cared for in accordance with United States Public Health Service guidelines. The tariquidar dose ranged from 0.05 to 6.25 mg/kg along with a vehicle control injection, and three to four animals were used at each dose level. Similar studies were also performed after pretreatment of mice with elacridar (0.1–10 mg/kg; n = 3 at each dose level). The time course of inhibition was determined after administration of vehicle and tariquidar (1 and 6.25 mg/kg) with splenocytes obtained at 0, 0.5, 1, 2, and 4 h. After removing the spleen, it was washed 3 times with sterile ice-cold HBSS containing 15 mM HEPES (EM Science, Gibbstown, NJ) and 0.1% bovine serum albumin (Sigma) and then cut and mashed. The tissue was then centrifuged at 4°C and 250g for 7 min, and the isolated cells were washed 3 times with supplemented HBSS and then resuspended in the same buffer to a concentration of approximately 106 cells/ml.
Inhibition of P-glycoprotein-mediated efflux from the isolated splenocytes was determined using a modified procedure of a previously described method (Robey et al., 1999). Specifically, spleen cells were loaded with rhodamine 123 (Sigma) and counterstained with pan T-cell marker, anti-mouse CD3e conjugated to APC. This conjugate was chosen to maximize sensitivity and simultaneously minimize the requisite compensation, which in all of the experiments was less than 2%. Cell suspensions from each mouse were divided into 4 × 1-ml aliquots; 3 μM tariquidar in 50 μl of HBSS was added to two of these aliquots as controls, and the same volume of buffer was added to the remaining two aliquots. All preparations were kept in the dark and on ice, with the exception of when they were being incubated. Rhodamine 123 was added to each preparation (0.5 μg/ml), which was then incubated with shaking for 10 min at 37°C in the dark. Net cellular accumulation of rhodamine 123 was stopped by immediate cooling to 4°C, and the suspension was subsequently centrifuged at 250g for 5 min at 4°C. Cells were washed one time and resuspended in 1 ml of ice-cold HBSS and tariquidar (3 μM) to produce >99% inhibition of efflux, or HBSS was added to the appropriate cells. Each one of the HBSS- and tariquidar-treated preparations was kept on ice with shaking to serve as the zero time point, with respect to rhodamine efflux, whereas the other cell suspensions were incubated with shaking at 37°C for 30 min. Rhodamine 123 efflux was stopped by immediate cooling of the cells to 4°C, and APC-conjugated hamster, anti-mouse CD3e monoclonal antibody (BD Biosciences/PharMingen, San Diego, CA), was added. After a 30-min incubation with shaking in the dark at 4°C, the cells were spun down at 250g for 5 min at 4°C and washed twice before resuspension in 1 ml of ice-cold HBSS. Propidium iodide (0.2 μg; Molecular Probes, Eugene, OR) was added to allow assessment of viability.
Flow cytometry (LSRII; Beckman Dickinson) with lasers at 488, 407, and 635 nm was performed within 1 h of the completion of the final cell incubation. Cells were initially gated into a P1-gate for the exclusion of doublets and viability using a plot of side-scatter versus forward scatter. “Bright” (dye efflux negative) propidium iodide-stained cells were excluded by a P2 gate. Viable lymphocytes were gated into P3 based on their reactivity to anti-CD3e antibody, and rhodamine 123-containing cells were plotted as a single parameter histogram gated to the zero time point cut-off. “Dim” cells to the left of this point were considered to be effluxing cells. Photomultiplier tube voltages were as follows: forward scatter, 420; side-scatter, 450; rhodamine 123, 550 (LP 505, 530/30 DF); PI/7-aminoactinomycin D, 560 (LP 570, 595/40 DF); and APC, 580 (LP 660/20 DF).
Efflux was expressed as the reduction in “dim” cells from baseline (average of “dim” cells in vehicle-treated) in each animal according to eq. 1:
In Vivo Effect of Tariquidar on Antinociceptive Response of Loperamide. Male mdr1a/1b(–/–) FVB/TacfBR-[KO] Pgy2N7 and syngeneic FVB/NTTacfBR wild-type [mdr1a/1b(+/+)] mice weighing 20 to 30 g were obtained from Taconic. Antinociception was measured by a hot plate test (Analgesia Hotplate; Columbus Instruments, Columbus, OH). Mice were individually placed onto the hot plate (surface temperature, 52 ± 0.1°C), and the time taken before they licked their hind paw was recorded. If the animal did not perform this act within 30 s, it was removed from the hot plate to prevent tissue injury. Each mouse was used only once, and a predrug response time after 3 days of training was determined before the administration of any drug. In one series of studies, mdr1a/1b(+/+) and (–/–) mice [n = 3 at each dose level)] were pretreated with vehicle or tariquidar (6.25 mg/kg) intravenously administered into a tail vein 15 min before administration of loperamide (0.1–1 mg/kg) or vehicle by the same route. The antinociceptive response of individual mice was determined at 0, 5, 10, 15, 30, 45, 60, 90, 120, 180, and 240 min. In a second series of studies, the dose of tariquidar was varied (1–6.25 mg/kg) and that of loperamide was fixed (1 mg/kg), and studies were only undertaken in mdr1a/1b(+/+) mice (n = 3–4 at each dose level). Studies were also undertaken in a similar fashion to determine the antinociceptive response to loperamide (1 mg/kg) after pretreatment with elacridar (0.1–30 mg/kg; n = 3–4 at each dose level). Antinociception over the first 2 h of study was measured by eq. 2: where AUEC is the area-under the latency effect-curve estimated by the trapezoid rule under control conditions (AUECbaseline) after loperamide (AUEClop) and the 30-s maximal effect (AUECmax).
Determination of Plasma, Testes, and Brain Concentrations of Loperamide. Male mdr1a/b(+/+) mice weighing 20 to 30 g were administered either tariquidar (0.1–30 mg/kg; n = 3–4 for each dose) or elacridar (0.1–30 mg/kg; n = 1–2 for each dose) into a tail vein. Fifteen minutes later, radiolabeled loperamide (0.033 μCi/g) along with 1 mg/kg unlabeled drug was also injected intravenously. After a further 15 min, the animals were anesthetized using isoflurane, and blood was obtained from the inferior vena cava. After sacrifice, organs were harvested, blotted with filter paper to remove blood, weighed, and then homogenized with a 10-fold volume of 4% bovine serum albumin solution. Total radioactivity was determined by liquid scintillation after the addition of 100 μl of plasma or 500 ml of tissue homogenate to 6 ml of biodegradable counting scintillant reagent (Amersham Biosciences, Arlington Heights, IL).
Data Analysis. Data are expressed as mean ± S.D., and drug responses are shown as mean ± S.E.M. Log dose-response relationships were analyzed by nonlinear regression analysis based on a log-sigmoidal model with variable slope (Prism 4.0; GraphPad Software Inc., San Diego, CA). This statistical software provides estimates of means and confidence intervals for IC50/ED50 and Emax values. Univariate analysis of variance (ANOVA) with Bonferroni correction for multiple comparisons was used to compare log ED50 values for the different responses. All tests were two-tailed, and a p value of <0.05 was considered significant. For graphic representation, the response data are expressed as a percentage of the maximal effect.
Results
Tariquidar as an Inhibitor of P-glycoprotein. LMDR1 cells are LLC-PK1 cells stably transfected with human MDR1 cDNA and, therefore, overexpress the MDR1 gene product P-glycoprotein. Use of these two cell lines, accordingly, permits the assessment of P-glycoprotein-mediated efflux of substrate drugs in a side-by-side fashion. Consistent with the apical expression of the transporter, transfer of loperamide across cultured LMDR1 cells displayed polarization dependent on the direction of flux, basal-to-apical being more extensive than apical-to-basal with a ratio at 3 h of 35.7 ± 9.4 (n = 14). By contrast, uptake in parental LLC-PK1 cells was similar regardless of the direction of flux. The addition of tariquidar to LMDR1 cells reduced the extent of polarization, and inhibition of efflux transport exhibited a dose-response relationship (Fig. 1) with an IC50 value of 15 nM (95% CI, 10.1–21.5 nM) (∼10 ng/ml). The estimated B→ A/A→ B transport ratio for loperamide in the presence of 1 μM tariquidar was 1.23 ± 0.37, and its permeability in LLC-PK1 cells was 4.00 ± 0.95 (B→ A) and 4.37 ± 1.09 nm/s (A→ B). Similar values were obtained in LMDR1 cells [4.74 ± 2.36 (B→ A) and 5.04 ± 1.13 nm/s (A→ B) in the presence of 1 μM tariquidar].
Effect of Tariquidar on Antinociceptive Effect of Loperamide. In both mdr1a/1b(+/+) and (–/–) mice pretreated with 6.25 mg/kg tariquidar, increasing doses of intravenous loperamide resulted in enhanced antinociception (Fig. 2). At dosages of 0.5 mg or greater of loperamide in knockout animals, the maximal latency period was achieved within 5 min and maintained for almost3hor longer before, in the case of the 0.5 mg/kg dose, the effect began to decline after this time. In mdr1a/1b(+/+) mice, a more blunted response was seen at any particular dose level, but even so, a maximal response was maintained for nearly 1 and 2 h after 0.5 and 1 mg/kg loperamide, respectively. Pretreatment of mdr1a/1b(+/+) mice with varying doses of tariquidar followed by 1 mg/kg loperamide resulted in a dose-related antinociceptive response, especially at levels of 2 to 6.25 mg/kg (Fig. 3). The effect was maximal within 5 min, and subsequently, it declined toward baseline over the 4-h study period. After 4 and 6.25 mg/kg tariquidar, the maximal plateau effect was maintained for 1.5 h before it began to decline. Similar findings were also noted after pretreating the mice with 0.1 to 30 mg/kg elacridar (data not shown).
Dose-Response Relationships to P-glycoprotein Inhibition. Rhodamine 123 efflux from CD3e+ lymphocytes obtained from mdr1a/1b(+/+) mice pretreated with tariquidar 30 min previously was inhibited in a dose-response fashion (Fig. 4). The ED50 was estimated to be 0.25 mg/kg (95% CI, 0.06 –1.02 mg/kg), and the maximal inhibition was the same as that obtained in CD3e+ lymphocytes from mdr1a/1b(–/–) mice. Greater than 95% inhibition was maintained over the 4-h study period after intravenous administration of 1 and 6.25 mg/kg tariquidar to wild-type mice (data not shown). Elacridar pretreatment also inhibited rhodamine 123 efflux, but it was approximately 5-fold less potent than tariquidar with an ED50 value of 1.34 mg/kg (95% CI, 1.21–1.47 mg/kg). Furthermore, the dose-response curve was far steeper than that of tariquidar [Hill slope factor: 0.53 (95% CI, –0.18 –2.03) and 3.84 (95% CI, 2.66 –5.02) for tariquidar and elacridar, respectively (p < 0.05)].
Testes/plasma and brain/plasma concentration ratios of radiolabeled loperamide measured 30 min after pretreatment with tariquidar and 15 min after intravenous loperamide administration to mdr1a/1b(+/+) mice also demonstrated dose relationships (Fig. 4). These appeared to reflect changes in tissue distribution of loperamide, because the plasma radioactivity concentration was similar regardless of the pretreatment tariquidar dose (data not shown). However, the curves were shifted markedly (6- and 25-fold) to the right relative to the rhodamine 123 efflux data (Fig. 4) with ED50 values of 1.48 mg/kg (95% CI, 0.79 –2.77 mg/kg) (testes/plasma) and 5.65 mg/kg (95% CI, 4.28 –7.66 mg/kg) (brain/plasma), respectively (p = 0.07 and p < 0.01 for comparison with rhodamine 123 efflux data, respectively). With elacridar, the ED50 for the brain/plasma concentration ratio was only 2-fold greater, 2.36 mg/kg (95% CI, 2.11–2.63), than that for rhodamine 123 efflux from CD3e+ lymphocytes (ED50 = 1.34 mg/kg, 95% CI, 1.21–1.47 mg/kg; p < 0.01). In addition, elacridar pretreatment (10 mg/kg) of mdr1a/1b(–/–) mice resulted in a similar brain/plasma ratio as seen in mdr1a/1b(+/+) animals administered with 5 to 30 mg/kg tariquidar. The ED50 for elacridar distribution into the testes of 1.09 mg/kg (95% CI, 0.75–1.58 mg/kg) was similar to that for rhodamine 123 efflux from lymphocytes (p > 0.05). The brain/plasma ratio after pretreatment with 10 mg/kg elacridar was similar in both mdr1a/1b(+/+) and (–/–) mice.
The antinociceptive effect of loperamide at 15 min after its intravenous administration (1 mg/kg) in mice pretreated with varying doses of tariquidar was essentially dichotomous—either no effect or maximal response (Fig. 3). Accordingly, the AUEC over 2 h was used as a measure of the central action of loperamide (Fig. 3). The resulting dose-response relationship was somewhat steeper than that observed for rhodamine 123 efflux from lymphocytes and also the brain/plasma concentration ratio; the estimated ED50 value, 2.18 mg/kg (95% CI, 1.76 –2.70 mg/kg), being 3-fold lower than the brain/plasma distribution ratio (p > 0.05) but approximately 10-fold higher than that for rhodamine 123 efflux (p < 0.05) (Fig. 4). With elacridar, the antinociceptive dose-response curve was also shifted to the left of the brain/plasma concentration ratio, and it was quite steep, with the no-effect to maximal effect dose ranging between 1 and 2 mg/kg. However, the ED50 [1.65 mg/kg (95% CI, 1.55–1.76 mg/kg)] for antinociception was not significantly different from that for rhodamine 123 efflux from lymphocytes (p > 0.05).
Discussion
The purpose of this study was to obtain proof-of-principle that pharmacological inhibition of P-glycoprotein localized at the blood-brain barrier would convert loperamide from a drug without opioid effects on the brain to one with such a characteristic. In turn, this would provide underpinning for studies in humans based on measuring changes in the central effects of the drug. The results with tariquidar, a potent and selective third-generation P-glycoprotein inhibitor (Martin et al., 1999; Mistry et al., 2001), provided such proof in that antinociception in mdr1a/1b(+/+) mice exhibited a dose-response relationship reflective of the extent of inhibition. In addition, the onset of the effect was rapid, indicating fast distribution into the brain followed by a decline dependent on the dose of loperamide. Similar observations were also noted with elacridar, thus confirming the generalizability of the phenomenon. In addition, elacridar, at concentrations similar to those associated with inhibition of P-glycoprotein, also inhibits breast cancer resistance protein 1 (Abcg2) (Allen and Schinkel, 2002); however, its administration to mdr1a/1b(–/–) mice did not increase the maximal brain/plasma concentration ratio relative to that seen in untreated animals. This would suggest that the observed effects at the blood-brain barrier are primarily accounted for by inhibition of P-glycoprotein, although other efflux transporters cannot be entirely excluded.
Neither tariquidar nor elacridar altered the plasma level of loperamide, consistent with their minimal inhibition of cytochrome P4503A (Xenova Ltd., personal communication) (Cummins et al., 2002), which is mainly responsible for the primary metabolism of the opioid, at least in humans (Kim et al., 2004). Thus, the observed increases in the brain/plasma concentration ratio of the loperamide with increasing dosages of the P-glycoprotein inhibitors reflected a change in its distribution into the brain. Such distribution of unbound drug generally is determined by the balance of its membrane permeability at the blood-brain barrier, influencing influx versus efflux mediated by P-glycoprotein. Thus, drug delivery to the brain is likely if the drug is not a good P-glycoprotein substrate as indicated by B→ A/A→ B transport ratio in LMDR1 cells of <2 at 3 h and has appropriate passive permeability (Lin and Yamazaki, 2003). Similar characteristics have been suggested using MDR1-Madin-Darby canine kidney II cells (Mahar Doan et al., 2002).
In the absence of tariquidar, the B→ A/A→ B transport ratio of loperamide in LMDR1 cells was greater than 30, consistent with it being a good MDR1 substrate (Schinkel et al., 1996). Furthermore, treatment of the cells with tariquidar resulted in concentration-dependent inhibition of transport. Thus, the observed results are entirely consistent with an increasing distribution of loperamide into the brain resulting from a decrease in the effectiveness of the blood-brain barrier, secondary to P-glycoprotein inhibition. Interestingly, the in vitro permeability of loperamide across cultured LLC-PK1 cells was 5-fold less than the suggested critical value for distribution into the brain of 20 nm/s (Mahar Doan et al., 2002; Lin and Yamazaki, 2003). However, in vivo the brain levels and antinociceptive effect of the drug were rapidly achieved, indicating ready passage into central tissue. Moreover, direct measure of the permeability of loperamide in vivo is also consistent with the lack of any difficulty in the drug entering the brain in the absence of P-glycoprotein (Dagenais et al., 2004). This suggests the possible involvement of transport-mediated uptake at the blood-brain barrier perhaps by an organic anion-transporting polypeptide transporter, which is not expressed in the cultured cells (Marzolini et al., 2004).
Although the major and original goal of the studies was focused on P-glycoprotein in the blood-brain barrier, inhibition was determined at other sites to obtain a more global perspective and, in the case of lymphocytes, to relate the results to those previously obtained in humans (Stewart et al., 2000). Tariquidar is a potent inhibitor of rhodamine 123 efflux from human CD56+ lymphocytes with an ex vivo IC50 of 33 nM (∼20 ng/ml). Furthermore, after intravenous administration in humans, this inhibition was maximal after a dose of 2 mg/kg; moreover, it was sustained for almost 24 h (Stewart et al., 2000). In addition, after an intravenous dose of 150 mg of tariquidar to cancer patients, the uptake of the P-glycoprotein substrate 99mTc-sestamibi into liver and various drug-resistant tumors was substantially increased, and excretion into the gastrointestinal tract and by the kidneys was markedly reduced (Agrawal et al., 2003). Such findings along with those from tissue distribution studies in animals (Xenova Ltd., 2005) all indicate that tariquidar at dosages/concentrations associated with complete inhibition of lymphocyte P-glycoprotein also inhibits the function of the transporter at other tissue sites, including tumors. Therefore, it was unexpected to find that measures reflecting P-glycoprotein-dependent drug distribution into the brain, in particular, exhibited greater resistance to inhibition than did lymphocytes. Such differential sensitivity is even more remarkable since, within the microvasculature, P-glycoprotein located in the lymphocyte and endothelial cell membrane is presumably exposed to the same luminal concentration of the inhibitor.
Therefore, it would seem that P-glycoprotein function localized in the blood-brain barrier and, to a lesser extent, in the blood-testes barrier is more resistant to inhibition by tariquidar than that in lymphocytes and at other sites. This makes teleological sense given the critical nature of these organs and the desirability of protecting them from potentially damaging xenobiotics. However, the involved mechanism is unknown. Generally, P-glycoprotein mediated transport is considered to be associated with a 170-kDa protein, (Ambudkar et al., 1999), but functional “mini P-glycoproteins” rather than the classic transporter have been reported in certain lymphocytes (Trambas et al., 2001). Such a difference or other structural distinctions in the transporter may account for the differential tissue sensitivity. An alternative explanation could be related to the fact that P-glycoprotein is localized in caveolae within brain microvascular endothelial cell membranes, whereas such microdomains are absent in lymphocytes (Fra et al., 1994; Demeule et al., 2000). Because caveolae are enriched in cholesterol (Smart et al., 1999) and P-glycoprotein activity is increased in an enhanced cholesterol microenvironment (Troost et al., 2004; Gayet et al., 2005), this could affect the interaction of the transporter with inhibitors dependent on cell type. A further possible factor may be the difference in the amount/number of P-glycoprotein pumps present in the two involved cell membranes. This could be important because tariquidar apparently acts on the cytoplasmic side of the membrane rather than at the extracellular face, and in such a situation, the apparent half-inhibition concentration (IC50) increases with the number of pumps (Litman et al., 2003). Thus, the more abundant expression of P-glycoprotein in the brain capillary endothelial cell compared with lymphocytes and other cell types (Ambudkar et al., 1999) would account for its greater resistance to inhibition. Additional studies will be required to investigate these possible causal differences and their relative importance.
A further unexplainable observation is the greater difference between lymphocyte and brain capillary endothelial cell inhibition after pretreatment with tariquidar (20-fold) compared with that with elacridar (2-fold). In addition, despite the fact that it was not statistically significant, there was a trend for a greater differential sensitivity of the blood-testes barrier to inhibition relative to lymphocyte efflux for tariquidar (6-fold; p = 0.07) compared with elacridar (no difference). In contrast to earlier P-glycoprotein modulators, these third-generation drugs are not substrates for the transporter, rather they interact with multiple distinct binding sites on P-glycoprotein, which results in complex allosteric changes leading to altered transport function (Martin et al., 1999, 2000). Moreover, the binding sites for tariquidar and elacridar seem to be different (Martin et al., 2000). Perhaps this difference in the interaction with the transporter accounts for the greater similarity between P-glycoprotein inhibition in lymphocytes and that in the blood-brain and blood-testes barrier after elacridar administration compared with the more marked difference with tariquidar. It is also possible that the two drugs may possess different and unknown inhibitory properties to alter uptake and/or efflux transporters that are differentially expressed in different cell types.
The described studies clearly demonstrate and confirm that pharmacologic modulation of the blood-brain barrier through inhibition of P-glycoprotein can markedly increase brain levels of a substrate for the transporter that is normally excluded from this site. In the case of loperamide, measurable central effects of the opioid become manifest. However, this situation appears to be more complex than at first sight since, in addition to considerations related to in-hibitor potency, the nature of the molecular interaction between the inhibitor and P-glycoprotein also seem to be important. Thus, the greater resistance to inhibition of the transporter located in the brain compared with other sites requires doses/concentrations greater than those associated with inhibition at other tissue sites, including tumors. Moreover, the difference seems to depend on the individual inhibitor being greater for tariquidar and elacridar; whether other third-generation noncompetitive inhibitors, such as zosuquidar and laniquidar, or earlier used inhibitors that are substrates, e.g., quinidine, cyclosporin A, or valspodar, have similar characteristics is unknown. Certainly, the fact that P-glycoprotein inhibition is present in lymphocytes and other tissue does not necessarily indicate that this situation also applies to the brain. Accordingly, strategies to increase drug delivery to the brain through inhibiting P-glycoprotein will have to take these considerations into account.
Acknowledgments
We gratefully acknowledge Janssen Pharmaceutica, Xenova Ltd., and GSK Ltd., for the gifts of radiolabeled loperamide, tariquidar, and elacridar, respectively,
Footnotes
-
This work was supported by United States Public Heath Service Grants GM31304 and CA68485, and Merck Sharp and Dohme International Fellowships in Clinical Pharmacology (to D.K. and M.M.).
-
E.F.C., D.K., and M.M. contributed equally to this work.
-
doi:10.1124/jpet.105.099648.
-
ABBREVIATIONS: MDR1, multidrug resistance protein 1; GF-120918. elacridar; HBSS, Hanks' buffered salt solution; LMDR1, LLC-PK1 cell line stably expressing human MDR1; LLC-PK1, porcine kidney-derived renal epithelial cell line; APC, allophycocyanin; PSC-833, valspodar; LP, long-pass filter; DF, discrimination filter; AUEC, area-under the latency effect curve; ANOVA, univariate analysis of variance; A→ B, apical-to-basal; B→ A, basal-to-apical; LY-335979, zosuquidar; XR-9576, tariquidar; Tc, technetium.
- Received December 12, 2005.
- Accepted March 13, 2006.
- The American Society for Pharmacology and Experimental Therapeutics