Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Multidrug resistance, a major problem in chemotherapy treatment of human cancers, is caused by the overexpression of a 140- to 170-kdalton phosphoglycoprotein, termed P-glycoprotein (Goldstein et al., 1989; Chan et al., 1991). This cell surface protein is capable of actively extruding a wide range of structurally dissimilar lipophilic compounds, such as Vinca-alkaloids, colchicine, adriamycin, actinomycin D, taxol, etoposide and doxorubicin from cells (Endicott and Ling, 1989; Gottesman and Pastan, 1993; Jette et al., 1995; Ueda et al., 1987). A variety of agents have been developed for reducing P-gP-mediated drug resistance of tumors during chemotherapy, e.g., calmodulin inhibitors, quinidine, verapamil, cyclosporin A (Ford and Hait, 1990) and the nonimmunosuppressive cyclosporin D analog, SDZ PSC 833, which exerts a higher MDR-modulating activity than cyclosporin A and verapamil in vitro and in vivo (Gaveriaux et al., 1991; Boesch et al., 1991; Friche et al., 1992; Watanabe et al., 1995). Biochemical evidence suggests that the mechanism of action of SDZ PSC 833 might be related to its capacity to bind to P-gP in a way similar to cyclosporin A (Archinal-Mattheis et al., 1995; Foxwell et al., 1989; Sakata et al., 1994; Shirai et al., 1994).

Recently, P-gP has been expressed in several normal tissues including brain, testes, kidney, pancreas, liver, colon and jejunum (Croop et al., 1989; Sugawara et al., 1988). More specifically, expression has been detected in the brush border of renal proximal tubules, at the biliary canalicular membrane of hepatocytes, at the apical surface of mucosal cells in small and large intestine (Thiebaut et al., 1987) and in the endothelial cells at the BBB and blood-testes barrier (Cordon-Cardo et al., 1989; Thiebaut et al., 1989). This tissue distribution suggests that the normal function of P-gP is to protect the organism against toxic compounds by actively excreting these compounds into bile, urine or intestinal lumen and by preventing accumulation in critical organs such as the brain and testes.

P-gP is a member of the ABC (ATP-binding cassette) superfamily of transporters. Within this family, mammalian P-gP consists of a group of closely related membrane proteins, containing two members in humans (MDR1 and MDR3) and three members in mice (mdr1a, mdr1b and mdr2). Only MDR1, mdr1a and mdr1b can confer MDR. The mouse mdr1a gene is expressed predominantly in the intestine, liver and blood capillaries of brain and testes, whereas the mdr1b gene is expressed predominantly in the adrenal, placenta, ovarian and pregnant uterus. Similar levels of mdr1a and mdr1b expression are found in the kidney (Devault and Gros, 1990; Ueda et al., 1987). These data suggest that mdr1a and mdr1b in the mouse together fulfill the same function as MDR1 in humans. Mdr1a and mdr1b P-gP may have overlapping but distinct substrate specificities (Raymond and Gros, 1989; Devault and Gros, 1990).

To learn more about the physiological role of P-gP, the group of P. Borst (Schinkel et al., 1994) has generated a mouse strain with a homozygous disruption of the mdr1a P-gP gene [mdr1a (-/-) mice]. They also have shown that its absence results in elevated concentrations of many drugs, such as [¹⁴C]ivermectine, [¹⁴C]vinblastine, [¹⁴C]digoxin, [¹⁴C]dexamethasone and [¹⁴C]cyclosporin A, in several tissues (especially in the brain) and in decreased drug elimination (Schinkel et al., 1995, 1996; van Asperen et al., 1996).

The BBB located at the endothelial cells of the blood capillaries of the brain makes the brain a pharmacological sanctuary by limiting the accumulation of many drugs in the brain. The mdr1a P-gP appears to be an important element of this defense and could participate to the integrity of this continuous, lipophilic physical barrier (Valverde et al., 1992). In mdr1a knock-out mice, the suppression of the P-gP efflux pump present at the BBB should only affect the transcellular passage of drugs. However, to check the tightness of the BBB in the transgenic mice, the paracellular passage of inulin, a vascular marker, was studied first.

Our study tested whether the blood pharmacokinetics and the tissue distribution of SDZ PSC 833 are affected by the absence of mdr1a P-gP. For this purpose, SDZ PSC 833 was administered intravenously by a constant-rate infusion to wild-type [mdr1a (+/+)] and knock-out [mdr1a (-/-)] mice for 4 h. At various times during infusion and after infusion, blood and tissues were sampled for total radioactivity (parent drug and radioactive metabolites) and parent drug analysis.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Compound. [¹⁴C]SDZ PSC 833 (44.4 µCi/mg) was supplied by the Isotope Laboratories of Novartis Pharma (Basel, Switzerland). Inulin labeled with [¹⁴C]carboxy was purchased from Amersham Laboratories (Buckinghamshire, England); the manufactured specific radioactivity was 1.2 µCi/mg.

Animals. The experiments were performed in the Leiden/Amsterdam Center for Drug Research (Leiden, the Netherlands). Thirty male FVB mdr1a (+/+) and thirty male FVB mdr1a (-/-) mice weighing 25 ± 3 g (10 ± 2 weeks of age) were used. The day before drug administration, the mice were anesthetized with a mixture of Hypnorm (fentanyl, 0.2 mg/ml; fluanisone, 10 mg/ml), Dormicum (midazolam, 5 mg/ml) and water (1:1:2, v/v/v). The volume of the mixture injected intraperitoneally was 7 ml/kg body weight. The right jugular vein was cannulated with a segment of silicone tubing containing heparinized saline solution. The tubes were passed subcutaneously to emerge at the back of the neck. The mice thereafter were placed on a paraffin pad (39°C) to recover from surgery until the first signs of movements (within 2 h) and then placed in their own cage. The presence of the catheter caused no obvious discomfort to the animals. Administration of the drug was carried out 1 day after surgery. Before and during the experiments, all animals had free access to food and water but were restricted in their movement during the experiment with their fixed tail.

Inulin infusion. [¹⁴C]Inulin dissolved in saline solution was infused into the jugular vein for 1 h at the rate of 100 µg/min (120 nCi/min) by means of a microinjection pump with a flow rate of 1 µl/min. At the end of the infusion, the mice [three mdr1a (-/-) mice and three mdr1a (+/+) mice] were sacrificed; blood and total brain were collected and analyzed for [¹⁴C]inulin concentrations.

SDZ PSC 833 infusion. [¹⁴C]SDZ PSC 833 was dissolved in a mixture of polyethylene glycol 200/ethanol/glucose 5% (55:5:10, v/v/v) with a concentration of 2 mg/ml (88.8 µCi/ml). This solvent has only a minor effect on the transport of SDZ PSC 833 into the brain (Lemaire et al., 1996). This radioactive solution was infused into the jugular vein for 4 h at the rate of 2 µg/min (80 µg/min/kg) by a microinjection pump (1 µl/min). Three mdr1a (+/+) and three mdr1a (-/-) mice were sacrificed at 0, 0.5, 1, 2 and 4 h during infusion and at 0.5, 1, 3, 8 and 24 h postinfusion. Blood, cerebrum (i.e., whole brain minus cerebellum), cerebellum, liver, kidney, heart, testes, muscle, small intestine, GIT and, when possible, the gall bladder were sampled, weighed and immediately frozen until total radioactivity and parent drug measurements.

Sample preparation. Tissues were homogenized in demineralized water (2 ml for heart, testes, muscle, cerebrum, cerebellum, gall bladder; 4 ml for kidney and GIT; and 20 ml for liver and small intestine).

Determination of radioactivity. Determination of radioactivity in eluate fractions was conducted by direct liquid scintillation counting. Before radiometric determinations, blood and homogenized tissues were solubilized in Solutron (Kontron Instruments, Zürich, Switzerland). After adding 10 ml of scintillation cocktail (Lumasafe, Lumac, Landgraaf, the Netherlands), all samples were counted in Tricarb liquid scintillation analyzer (Packard, Meriden, CT). Automatic external standard techniques (quench compensation) were used to determine the efficiency of the respective radiometric analyses; observed data (cpm) were converted to disintegrations per minute.

Determination of the parent drug. The concentrations of [¹⁴C]SDZ PSC 833 in whole blood (200 µl) and tissues (1 ml homogenate) were determined by liquid chromatography-reversed isotope dilution (Everett et al., 1989). The procedure involved the addition of 5 µg of nonlabeled SDZ PSC 833 to each sample as an internal standard. After adding 1 ml of Borat buffer, pH 9 and 4 ml of diethyl ether, the samples were shaken for 15 min and centrifuged (15 min, 4000 × g). The organic phase was transferred into a conical tube and evaporated in a Speedvac. The residue was reconstituted in 250 µl of acetonitrile/water phase (40:60, w/w) and after adding 75 µl of n-hexane, transferred into an autosampler vial. After a short centrifugation (1 min, 3000 × g), the hexane layer was pipetted off and discarded, and 200 µl of the remainder were injected onto the high-performance liquid chromatography system (MT2, Kontron Instruments, Zürich, Switzerland). SDZ PSC 833 was separated from potential metabolites and endogenous compounds on an analytical column (Phenomenex IBSil Phenyl, 5 µ, 150 × 3.2 mm). The mobile phase consisted of acetonitrile/t-butylmethyl ether/water (49:9:42, w/w). The flow rate was 0.8 ml/min; the effluent was monitored by an UV-detector set at 210 nm. The peak corresponding to the unchanged compound was collected in a polyethylene vial by a fraction collector and analyzed for radioactivity. The concentration of [¹⁴C]SDZ PSC 833 in each sample was calculated from the ratio of the amount of radioactivity in the eluate fraction and the area of the ultraviolet absorbance of the nonlabeled SDZ PSC 833 used as internal standards (Everett et al., 1989). Recoveries for blood averaged 72 ± 4%, and those for tissues were lower (30-65%). The limit of quantification was 5 ng/g or ml.

Data analysis. Tissue and blood AUC_0-28h (area under the blood and tissues mean concentration versus time curve up to 28 h) were obtained by the trapezoidal rule.

View this table: [in this window] [in a new window]   TABLE 1 Mdrla (/)/mdrla (+/+) ratios of SDZ PSC 833 tissue concentrations during and after a 4-h intravenous constant-rate infusion of [14C]SDZ PSC 833 (2 µg/min)

View this table: [in this window] [in a new window]   TABLE 2 Fraction unchanged (%) of SDZ PSC 833 in the tissues of wild-type and mdrla (/) mice during and after a 4-h intravenous constant-rate infusion of [14C]SDZ PSC 833 (2 µg/min) Fraction unchanged (%) represents the ratio of parent drug AUC0-28h (area under the parent drug blood and tissue concentration versus time curve up to 28 h) over radioactivity AUC0-28h (area under the radioactivity blood and tissue concentration versus time curve up to 28 h). Three mice were analyzed in each group.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Inulin brain penetration. Steady state blood levels were obtained after a 1-h constant-rate infusion of [¹⁴C]inulin. At this time, the brain/blood concentration ratio K_p characterized the brain penetration of inulin. The measured K_p values of 6.4 ± 1.2% for mdr1a (+/+) mice and 8.9 ± 1.5% for mdr1a (-/-) mice were not significantly different and corresponded roughly to the brain vascular volume.

SDZ PSC 833 tissue distribution. To investigate whether SDZ PSC 833 pharmacokinetics is affected by the P-gP transport, we infused this cyclosporin analog intravenously to mdr1a (+/+) and (-/-) mice, and determined the tissue distribution at various time points. Figure 1 shows the impact of the mdr1a gene disruption on the blood and tissue SDZ PSC 833 concentration time profiles. The (-/-)/(+/+) SDZ PSC 833 blood and tissue concentration ratios, listed in table 1, roughly amounted to 1, except for cerebrum, cerebellum and somewhat for testes. The mdr1a (-/-) mice brain (cerebrum and cerebellum) concentrations were 2- to 10-fold higher than those of mdr1a (+/+) mice. The SDZ PSC 833 blood pharmacokinetics could not be considered as responsible for this accumulation because the blood concentration/time profiles were roughly similar in both types of mice (fig. 1). In addition one can note that the difference in brain distribution was more pronounced at the beginning of the infusion and at the end of the postinfusion (table 1, fig. 1). More precisely, for cerebrum, this effect decreased over infusion time, from 5.5-fold after 0.5 h infusion to 2.4-fold after 4 h infusion, and increased during the postinfusion period, from 2.2-fold after stopping the infusion to 9.9-fold at the end of experiment.

View larger version (30K): [in this window] [in a new window]   Fig. 1.   Blood and tissue concentration (µg/ml or g) versus time curves of SDZ PSC 833 in wild-type (continuous line) and mdr1a knock-out (dotted line) mice during and after a 4-h constant-rate i.v. infusion (2 µg/min). SDZ PSC 833 concentrations were measured in blood (A), cerebrum (B), cerebellum (C), testes (D), muscle (E), kidney (F), small intestine (G), liver (H), heart (I) and GIT (J). Values represent means ± S.D. of three mice.

View larger version (36K): [in this window] [in a new window]   Fig. 2.   Tissue-to-blood coefficient of distribution Ktissue/blood of SDZ PSC 833 in wild-type and mdr1a knock-out mice during and after a 4-h constant-rate i.v. infusion (2 µg/min). Ktissue/blood values were calculated for cerebrum, cerebellum, testes, kidney, liver and small intestine (Small I.).

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

This study was designed to assess the impact of the mdr1a P-gP on the SDZ PSC 833 tissue distribution. The most pronounced pharmacological effect of mdr1a P-gP inactivation was seen in the brain (cerebrum and cerebellum), which indicates that this protein plays an important role in the BBB.

The endothelial cells of the brain capillaries are responsible for the barrier properties. These capillary cells possess tight junctions, sparse pinocytic vesicular transport and an absence of pores or fenestrations which restrict nonspecific transport into the brain via para-cellular route (Lefauconnier and Hauw, 1983, 1984; Bradbury, 1985).

Increase in SDZ PSC 833 brain penetration in mdr1a (-/-) mice compared with mdr1a (+/+) mice could be caused by a modification of the physiological function of the BBB, such as opening of the tight junctions in the knock-out mice, thereby increasing paracellular transport to the brain. However between both types of mice, no difference in brain penetration of [¹⁴C]inulin, a hydrophilic compound considered as a vascular marker (Sisson and Oldendorf, 1971) was observed. Thus, this result excludes the possibility that the disruption of the mdr1a gene increases the paracellular transport through the BBB. Consequently, this knock-out mouse model could be considered as an appropriate tool for studying the role of mdr1a P-gP in the transport of drug into the brain.

The difference between the wild-type and the mdr1a (-/-) brain penetration varies during and after the i.v. infusion of SDZ PSC 833. The lowest difference between both types of mice is observed at the end of infusion, i.e., at the highest blood concentration of 5.3 ± 1.6 µg/ml [in mdr1a (+/+) mice], whereas the highest difference is observed 24 h after the end of infusion, i.e., at the lowest blood concentration of 0.39 ± 0.04 µg/ml. This result is consistent with the hypothesis reported previously (Lemaire et al., 1996) that the SDZ PSC 833 brain penetration is blood concentration-dependent. Indeed, the decrease in (-/-)/(+/+) brain concentration ratios observed at higher blood concentrations agrees with the hypothesis of an inhibition of the P-gP pump present in the endothelial cells of the brain capillaries of mdr1a (+/+) mice. Based on this hypothesis, the blood concentrations of SDZ PSC 833 observed after a 4-h infusion should be sufficient to inhibit the active efflux pump of mdr1a (+/+) mice, whereas after a short infusion time the blood concentrations of SDZ PSC 833 should be insufficient to do so. The inhibition of the P-gP efflux pump at this blood concentration of 5.3 µg/ml is not complete, however, because the brain concentrations observed in (-/-) mice are still twice those observed in wild-type mice. This result suggests that a complete inhibition of the P-gP efflux pump could be obtained only after a longer infusion time. However, because the SDZ PSC 833 blood concentration threshold for P-gP inhibition was roughly 1 µg/ml, the difference in brain penetration observed between the mdr1a (+/+) and (-/-) mice probably can be explained by other barrier processes being involved in the brain penetration of SDZ PSC 833. The hypothesis of SDZ PSC 833 governing its own brain penetration also could explain the increase in (-/-)/(+/+) brain concentration ratios observed during the postinfusion. Indeed, at the end of the experiment (at a blood concentration of 0.39 ± 0.04 µg/ml), the P-gP pump activity was restored in mdr1a (+/+) mice and SDZ PSC 833 could be transported from brain to blood whereas in mdr1a (-/-) mice the brain elimination of this lipophilic compound was slower in the absence of active efflux pump. All these results demonstrate the interaction of SDZ PSC 833 with the P-gP present at the BBB.

Clinically and experimentally, it is well established that SDZ PSC 833 treatment combined with cytotoxic P-gP substrates, such as doxorubicin, etoposide and taxol, affects the pharmacokinetics of these compounds, notably by an increase in blood AUC and by a decrease in clearance (Keller et al., 1992; Gonzales et al., 1995). Consequently, it is reasonable to assume that the pharmacokinetics of SDZ PSC 833, in turn, might be affected by the presence of these P-gP substrates, depending on their relative ability to compete for P-gP transport.

Others clinical observations have suggested that the ataxia phenomenon and locomotory impairment observed after SDZ PSC 833 treatment might have a cerebellum origin. Possible differences in anatomical distribution of P-gP in capillaries of cerebrum and cerebellum could result in a different distribution of SDZ PSC 833 between these brain compartments; further this different brain distribution could explain the specific adverse effect on the central nervous system. However, the similar concentration-time profiles observed in cerebrum and cerebellum suggest that there are no differences in the anatomical distribution of P-gP in endothelial capillary cells of the cerebellum and cerebrum.

Although the protective role of this drug-transporting pump was demonstrated for cerebral tissue, the P-gP function of excretion of toxic compounds into bile, urine and intestine lumen is not shown clearly. Indeed, in both mdr1a (-/-) and wild-type mice, the metabolism and the elimination, described by the blood, tissue and bile concentrations of parent compound and radioactive metabolites, were not significantly different. The interpretation of these results in terms of mdr1a gene disruption is difficult because the elimination of SDZ PSC 833 depends not only on the rate and extent of P-gP present in the liver, the kidney and the intestine but also on the rate and extent of the metabolism. Actually, the major route of elimination for SDZ PSC 833 is via biotransformation by the cytochrome P-450 3A isoenzyme (CYP3A). A strong overlap between P-gP and cytochrome P-450 substrates and modulators was reported recently (Wacher et al., 1995) that could make interpretation of the metabolism more difficult. Concerning the SDZ PSC 833 excretion, it was shown that radioactive metabolites of SDZ PSC 833 were essentially eliminated through the bile. The analysis of gall bladder samples showed that SDZ PSC 833 unchanged represented only 3.0 ± 1.8% and 6.9 ± 2.2% in mdr1a (+/+) and (-/-) mice, respectively. This result indicates that the extent of hepatic metabolism is not significantly different in both types of mice.

Furthermore, the absence of significant differences in SDZ PSC 833 kidney and liver elimination may be related to the similar distribution of SDZ PSC 833 in these tissues for both types of mice. In both tissues, the disruption of mdr1a P-gP gene leads to increased levels of mdr1b P-gP RNA (Schinkel et al., 1994). The Borst's group (Schinkel et al., 1994, 1995) also suggests that mdr1b P-gP contributes to the distribution and excretion of the drug in tissue, and the increase of expression of mdr1b P-gP probably compensates for the absence of mdr1a P-gP and therefore limits the effect of the mdr1a gene disruption. The use of mdr1a1b double knock-out mice could be of great significance for understanding the role of mdr1a and mdr1b P-gP in the distribution of SDZ PSC 833 in the peripheral tissue.

The difference in SDZ PSC 833 accumulation in the testes and small intestine, where the mdr1a gene is expressed predominantly, was not as clear as in the brain. Nevertheless, the concentrations in the testes and small intestine tended to be higher in mdr1a knock-out mice than in wild-type mice, which suggests a P-gP saturation in wild-type mice caused by a relatively high SDZ PSC 833 blood concentration, even at the beginning and the end of the experiment. This agrees with the results of Borst's group, which suggests that the mdr1a P-gP in these organs could provide protection only against low concentrations of xenobiotics, whereas in the brain, the mdr1a function protects it against high blood concentrations of drugs. For the structurally similar cyclosporin A (Schinkel et al., 1995), 4 h after an i.v. bolus of [³H]cyclosporin A, the accumulation of radioactivity in mdr1a (-/-) mice was increased in the heart (factor 1.6), liver (factor 1.2), small intestine (factor 1.9) and testes (factor 2.6), but to a lesser extent than those found in the brain (factor 17). The smaller differences observed between both types of mice with SDZ PSC 833 could be explained by: 1) the partial or complete saturation and inhibition of P-gP in mdr1a (+/+) mice by SDZ PSC 833, 2) the different pharmacokinetics of cyclosporin A and SDZ PSC 833 and 3) their different affinity for P-gP transport. The tissue distribution of SDZ PSC 833 could be driven by many factors, such as the drug concentration and free fraction in the central (blood) compartment, drug affinity for intracellular constituents, tissue vascularization and influx and efflux rates through cellular membranes. Thus the absence of mdr1a P-gP may cause physiological and pharmacological modifications which differ according to the drug (routes of distribution, metabolism and elimination) and the tissues (vascularization, rate and extent of transporters, sites of actions).

In conclusion, the impact of the absence of the mdr1a P-gP on SDZ PSC 833 brain penetration suggests that SDZ PSC 833 strongly interacts with the P-gP present at the BBB. This has great significance in the clinical development of SDZ PSC 833 for circumvention of MDR in cancer chemotherapy because these results confirm the potential importance of the mdr1a P-gP.

Nevertheless the difference in accumulation in the brain between both types of mice observed at high blood levels, i.e., when P-gP presumably was inhibited by SDZ PSC 833 in mdr1a (+/+), suggests that other processes are involved in the brain penetration. Likewise, the lack of difference in SDZ PSC 833 metabolism and elimination reveals that others factors could be involved in the pharmacokinetics of this MDR-reversing agent, e.g., the mdr1b P-gP and the cytochrome P450 3A.