Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

MDR1 P-glycoprotein and the multidrug resistance proteins (MRPs) have been characterized as plasma membrane proteins that extrude amphipathic, toxic compounds from cells and thereby confer resistance against these compounds. Of the multidrug resistance transporters, MRP2 and MDR1 are localized in the apical membrane of epithelial cells including the intestine (Thiebaut et al., 1987; Keppler and Konig, 1997; Mottino et al., 2000). Thus, these transporters may play a role in the defense against orally ingested drugs, toxins, and carcinogens. Although the role of the MDR1 Pgp in oral bioavailability of drugs is currently being established (Lown et al., 1997; Sparreboom et al., 1997; van Asperen et al., 1997), little attention has been paid to MRP2 in this respect and to the role of both transporters in the defense against naturally occurring carcinogens, as has been done with MRP1 (Loe et al., 1997).

MRP2 was first identified as a hepatocellular canalicular organic anion transporter (Jansen et al., 1993). The range of molecules transported by this protein is broad (Oude Elferink et al., 1995; Konig et al., 1999), mainly including amphipathic anions and glucuronide-, glutathione-, and sulfate-conjugates. The gene was first cloned in the rat (Buchler et al., 1996; Paulusma et al., 1996); mutations in the human isoform were found to cause Dubin-Johnson syndrome (Paulusma et al., 1997; Toh et al., 1999).

MRP2 confers multidrug resistance against such compounds as methotrexate (Hooijberg et al., 1999), vincristine, or cisplatin (Kawabe et al., 1999) and is induced by cisplatin, 2-acetylaminofluorene (Kauffmann et al., 1997), rifampicin, and tamoxifen (Kauffmann et al., 1998). The transport of uncharged compounds such as vincristine by MRP1 can be stimulated by GSH; it was suggested that this involves cotransport (Loe et al., 1998). Because MRP2 is also expressed in the rat small intestine (Gotoh et al., 2000; Mottino et al., 2000), we hypothesized that it actively extrudes different amphipathic substances from the intestinal mucosa and thereby protects against food-derived xenobiotics, including carcinogens and drugs.

In the present study we tested this hypothesis by studying the role of rat MRP2 in the bioavailability of a food-derived carcinogen, 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP). PhIP is a heterocyclic amine belonging to the family of the aminoimidazoazaarenes. It was first described by Felton et al. (1986). PhIP is the most abundant heterocyclic amine formed during cooking, frying, and grilling (Skog et al., 1998) and was shown to induce colon and breast carcinomas in rats (Ito et al., 1991). The metabolism of the compound in different species is well defined (Alexander et al., 1995; Kadlubar et al., 1995; Lang et al., 1999).

Much less is known about the transport of PhIP. Transport of the parent compound is necessary for absorption, disposition, and for uptake in the cells of organs. In a very recent publication, it was shown that there might be active secretion of PhIP from Caco-2 cells (Walle and Walle, 1999); from the data presented in this work, it was suggested that MDR1 and/or MRP2 can play a role in the defense against this type of compounds.

Polymorphisms of phase I and phase II metabolizing enzymes play a role in type and extent of metabolism (Kadlubar et al., 1995; King et al., 1997; Lang et al., 1999). It already has been shown that polymorphisms in the MDR1 gene influence the bioavailability of digoxin (Hoffmeyer et al., 2000). It is obvious that polymorphisms in other transporter genes might also influence the absorption and distribution of xenobiotics in the body. In this study, we directly demonstrate MRP2-mediated transport of PhIP in MRP2-transduced epithelial cells, in small intestinal explants and in vivo.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals. [¹⁴C]PhIP (specific activity, 10 mCi/mM) was obtained from Toronto Research Chemicals, Inc (North York, Ontario, Canada). Unlabeled PhIP was bought from ICN (Costa Mesa, CA). All batches were found to be pure >98% by HPLC. L-Buthionine-[S,R]-sulfoximine (BSO) was bought from Sigma (St. Louis, MO). All solvents used for HPLC separation (HPLC- or analysis-grade) were from Merck (Darmstadt, Germany).

Animals. Female Wistar rats were obtained from Harlan-CPB (Zeist, The Netherlands). Age- and weight-matched female TR rats were from our own breeding stock, which has been characterized previously (Jansen et al., 1985). Rats were housed in cages with a 12-h light/dark cycle and given access to food and water ad libitum. Rats were used for experiments at age 12 to 16 weeks (200-250 g). All animal experiments in this work have been carried out in accordance with the Declaration of Helsinki.

Cell Line Experiments. Transduction of the polarized canine kidney cell line MDCK II with human MRP2 and characterization of the respective product has been described previously (Evers et al., 1998). MDCK II cells were also transduced with human MDR1, using a method described previously for LLC-PK1 cells (Schinkel et al., 1995). Expression of the protein was verified using Western analysis immediately before experiments. All cell lines were cultured in 75 cm² flasks with Dulbecco's modified Eagle's Medium (DMEM; Life Technologies, Breda, The Netherlands), supplemented with 9% fetal calf serum (BioWhittaker, Verviers, Belgium), 2 mM glutamine, and 100 U/ml penicillin/streptomycin (both from Life Technologies, Breda, The Netherlands) in humidified atmosphere (10% CO₂). After trypsinization, cells were seeded in a density of 40,000 cells per milliliter of medium in filter inserts (4.7 cm², 3.0 µm pore size; Transwell, Costar-Corning, NY) and cultured 7 to 10 days, with a change of medium every other day. Filters were used for experiments only when macroscopically tight. Filters also were checked for tightness after the experiment using [¹⁴C]inulin.

Ussing Chamber Experiments. Rats were anesthetized (57 mg/kg ketamine, 5.7 mg/kg xylazine, intramuscular injection). The abdomen was opened via midline incision and the jejunum was identified and ligated. Incisions next to the ligation allowed rinsing of the lumen with ice-cold saline solution. After ligating the blood supply, the jejunal segment was removed and immediately placed in ice-cold DMEM without phenol red. The muscular layer was rapidly removed by stripping, cut pieces of jejunum were put on polycarbonate filters (Schleicher & Schuell, Dassel, Germany) and mounted in Ussing chambers (0.5-cm² surface, 10-ml volume of both compartments). After 5-min equilibration with carbogenated DMEM without phenol red at 37°C, PhIP (10 µM) was added to the donor compartment (serosal or mucosal). Samples (500 µl) were taken at 30, 60, and 90 min and 500 µl of medium was replaced. Transepithelial potential difference was continuously monitored with AgCl electrodes and voltage deflections induced by 10 µA bipolar current pulses through platinum wires. Increasing loss of integrity of the intestine beyond 90 min made longer experiments impossible.

Bioavailability Studies (4 Rats Each Group). 2 µCi of [¹⁴C]PhIP was given via gastric gavage in olive oil (500 µl). Blood samples (500 µl) were obtained 0.5, 1, 2, 4, 24, and 48 h after dosing by heart puncture (under light anesthesia with isoflurane). Feces and urine were collected in 24-h fractions.

BSO Pretreatment. To inhibit glutathione (GSH) synthesis, we treated Wistar rats with 4 mM BSO per kilogram 3 h and 1.5 h before beginning of the experiments (i.p. injection). In the bioavailability studies over 48 h, an additional 4 mM BSO per kilogram was given each 12 h after the first dosing. Cells were incubated with 100 µM BSO in culture medium 24 h immediately before the experiments were carried out. Efficiency of BSO-pretreatment was controlled by determining GSH levels [according to the method of Tietze (1969)] from gut tissue pieces or denatured cells.

Organs. After sample collection in each in vivo experiment animals were sacrificed by bleeding from the aorta. Liver, whole intestine (with feces), kidney, pancreas, heart, and lungs were excised. Feces were separated from the intestine and all organs were washed before storage. All material was stored at 80°C.

Sample Treatment. Directly after sampling, aliquots of cell medium samples were injected on the HPLC for quantification of PhIP. Weighed portions of the organs from at least two different sites were homogenized with the same volume (w/v) of ethanol then centrifuged, and the pellet was washed once more with ethanol. After another centrifugation step, dry pellets (as well as samples of whole blood) were completely solubilized in Soluene 350 (Packard, Meriden, CT) (1 ml per 100 mg of tissue or 100 µl of blood) and decolorized with 0.1 ml of H₂O₂. Feces were homogenized with 50% methanol in water (1:5, w/v); the homogenate was counted after adding scintillation fluid.

HPLC. Analysis and PhIP quantification was done on a system with Gynkotek-Pumps (Germering, Germany), connected to a Rheodyne 7125 injection valve (Rheodyne, Cotati, CA) and an Inertsil 5 ODS 3 column (Chrompack, Bergen op Zoom, The Netherlands). Injection volume was 20 µl. We used an UV-Detector Spectroflow 757 (Kratos, Ramsey, NJ) at 315 nm, a fluorescence detector (FP-920; Jasco, Tokyo, Japan) with excitation wavelength of 315 nm and emission wavelength of 370 nm. The mobile phase (0.4 ml/min) consisted of 10% methanol in 0.1% diethylamine, pH 6.0 (A), and 90% methanol in 0.1% diethylamine, pH 6.0 (B). The gradient was adjusted from 10% B to 75% B in 20 min. In this system, PhIP eluted as a single peak with a retention time of approximately 17 min.

Statistical Analysis. Data were compared using one-sided t test. P < 0.05 was considered to express significance.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Cell Line Experiments. Nontransduced MDCK II cells show equal rates of transport of PhIP in the apical as well as in the basolateral direction (11.56 ± 0.74 nM versus 12.74 ± 1.23 nM in 2 h, respectively). In the MDCK II cells transduced with human MRP2, transport from the apical to the basolateral compartment was lower than that in the parent cells (9.17 ± 2.1 nM in 2 h, p = 0.02). Higher transport was observed from basolateral to apical in these cells (26.05 ± 2.2 nM in 2 h, p < 0.001, Fig. 1). This effect of MRP2 transduction was completely reversed by pretreatment of the cells with BSO before the experiment (11.55 ± 0.83 nM in 2 h in basolateral direction, p = 0.52 compared with untreated transduced cells; 10.8 ± 1.2 nM in 2 h in apical direction, p < 0.001 compared with untreated transduced cells; no significance compared with parent cells) (Figs. 1 and 2).

View larger version (13K): [in this window] [in a new window]   Fig. 1.   Transport of PhIP in MRP2 transduced MDCK II cells from basolateral to apical () and apical to basolateral (). PhIP was added to either the basolateral or the apical compartment at a concentration of 50 nM. PhIP concentration in the acceptor compartment at indicated time points was determined by HPLC. Significance level compared with opposite direction: ***p < 0.001. Data represent mean ± S.D. of four experiments.

View larger version (22K): [in this window] [in a new window]   Fig. 2.   Transport rate of PhIP over 2 h in untransduced MDCK II cells (), MRP2 transduced cells (), BSO-pretreated MRP2 transduced cells (), and MDR1 transduced cells () in apical direction (left) and basolateral direction (right). Significance level, compared with the respective opposite direction of the same cell line, is ***p < 0.001. Significance levels, compared with the respective direction of the parent cell line (untransduced), are: p < 0.05; p < 0.001. Data are given as means ± S.D. (n = 4 for all groups).

Ussing Chamber Experiments. Explants from Wistar rat small intestine were mounted in Ussing chambers. Administration of PhIP (10 µM) to the serosal or mucosal compartment demonstrated a higher basal to apical transport than vice versa. In 90 min, the concentration on the mucosal side reached 51.2 ± 12.8 nM in the Wistar rat, but only 29 ± 5.9 nM in the TR rat (p = 0.003, Fig. 3A). Serosal transport in the TR rat (19.5 ± 6.2 nM) was still lower than mucosal transport (p = 0.01), but serosal and mucosal transport in the TR rat were not significantly different from serosal transport in the Wistar rat.

View larger version (14K): [in this window] [in a new window]   Fig. 3.   Transport rate of PhIP over 90 min in isolated jejunum from (A) Wistar () and TR rats () and (B) untreated Wistar () and BSO-pretreated Wistar rats () in both directions: serosal to mucosal (solid symbols) and mucosal to serosal (open symbols). PhIP was added at a concentration of 10 µM, samples were taken at indicated time points and analyzed by HPLC. Significance levels, compared with untreated Wistar rats, are **p < 0.01; ***p < 0.001. Data are given as means ± S.D. (n = 6 in all groups).

Oral Bioavailability and Disposition over 48 h. Because the previous experiments suggested that the parent compound could be transported by MRP2 from the gut mucosa into the lumen, we investigated whether MRP2 also influences the oral bioavailability of PhIP. We administered PhIP (dissolved in olive oil) by gastric gavage. Blood samples were collected on various time points over a subsequent period of 48 h (see under Materials and Methods). Thirty minutes after ingestion, Wistar rats had 0.79% ± 0.2 of the administered dose in their blood (assuming a blood volume of 4% of the body weight), whereas TR rats had a blood level that was 2-fold higher, 1.66% ± 0.38 (Fig. 4A, p = 0.006). This difference was sustained during the first 4 h, but 24 and 48 h after administration, the two strains had comparably low levels. In TR rats, there was a second peak at 2 h, which seemed to be absent in the Wistar rats. The area under the curve in the TR rats was more than doubled compared with the Wistar (4820 and 2240, respectively over 48 h, mean values).

View larger version (14K): [in this window] [in a new window]   Fig. 4.   Blood content of radioactivity after luminal administration of [14C]PhIP in the gut. A, levels in peripheral blood (heart puncture) between 0.5 and 48 h after administration of [14C]PhIP by gastric gavage in Wistar () and TR () rats. B, levels in portal vein blood between 5 and 120 min after intraduodenal injection of [14C]PhIP in Wistar () and TR () rats. C, levels in peripheral blood (heart puncture) between 0.5 and 48 h after administration of [14C]PhIP by gastric gavage in Wistar () and BSO-pretreated Wistar rats (). Significance levels (*p < 0.05; **p < 0.01; ***p < 0.001) of TR and BSO-pretreated rats are compared with untreated Wistar rat. Data are given as means ± S.D. (n = 4 in all groups).

View larger version (20K): [in this window] [in a new window]   Fig. 5.   Radioactivity content after ethanol extraction in several organs of Wistar (), TR (), and BSO-pretreated Wistar rats () 48 h after luminal administration of [14C]PhIP in the gut. Significance levels (*p < 0.05; **p < 0.01) of TR and BSO-pretreated rats are those compared with untreated Wistar rat. Data are given as means ± S.D. (n = 4 in all groups).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

This study is the first to show that both human and rat MRP2 are involved in oral bioavailability of a xenobiotic. We present clear evidence that the heterocyclic amine PhIP, an abundant food-derived carcinogen, is transported by MRP2. This confirms and extends recent data, showing that inhibitors of MRP2 impair PhIP transport in Caco-2 cells (Walle and Walle, 1999). PhIP transport from the basolateral to the apical compartment in polarized epithelial cells transduced with the human MRP2 is almost 3-fold higher than vice versa. Corresponding to that result, we found that transport from serosal to mucosal approximately doubled in the isolated intestine from normal Wistar rats compared with MRP2-deficient TR rats. The difference in concentrations needed in the donor compartments in the two in vitro models (50 nM in MDCK II cells and 10 µM in intestinal explants) reflects the technical and anatomical differences of the used models: whereas the barrier in the cell line experiments is only a monolayer, in the Ussing chamber, the compound has to cross the mucus, epithelial, and submucosal layers. In addition, experiments with the Ussing chamber could not last longer than 90 min because of subsequent deterioration of the tissue. Thus, in the Ussing chamber, a higher PhIP concentration had to be added to the system to reach significant transport within the experimental period. It is possible that other transporters play a role as well in intestinal PhIP secretion, given the fact that the serosal to mucosal transport in the TR rat is still significantly higher than vice versa. Quantitatively, however, the contribution of MRP2 seems to be most important. It is questionable whether MDR1 contributes significantly to transport of PhIP as suggested by Walle and Walle (1999). We found slightly but significantly higher transport of PhIP in the apical direction in MDR1 transduced cells compared with the untransduced cells, but compared with the basolateral direction of the transduced cells, the difference was not significant. This indicates that MDR1 plays a minor role, if any, in PhIP transport. It is an interesting finding that MRP2 contributes much more than MDR1, because PhIP is an uncharged amphipath.

To test the role of MRP 2 in genuine oral bioavailability of PhIP, we administered PhIP by gastric gavage to Wistar and TR rats and determined blood levels up to 48 h after administration. Blood levels of radioactivity were about twice as high in TR rats than in Wistar rats. There are two possible explanations for that effect: Firstly, higher absorption from the gut of the TR rat and secondly, reduced excretion by liver or kidney via bile or urine, respectively. To distinguish between these two possibilities we drew blood samples directly from the portal vein after intraduodenal administration of PhIP; the increased PhIP levels in portal blood supported the first hypothesis. The 3- to 4-fold higher portal blood levels in both animal groups (compared with the heart blood) demonstrated the importance of metabolism of PhIP in the liver; the difference in portal blood levels between Wistar and TR confirmed a higher initial PhIP absorption in the TR rats of ~2- to 3-fold and higher overall absorption of 1.5- to 2-fold, as can be calculated from the area under the curve.

Overall, the blood levels in the portal vein confirmed the in vitro results qualitatively and quantitatively. Thus, this physiological setting demonstrated an important role of Mrp2 in the direct elimination of PhIP from the gut mucosa, thereby reducing the oral bioavailability. Given the results from transduced cells and the fact that MRP2 is also expressed in human small intestine and is inducible there (Fromm et al., 2000), it is likely that this mechanism also plays a role in man, depending on the individual expression level.

HPLC analysis of samples from the cell line experiments and the Ussing chamber studies showed no metabolism in MDCK II cells or the rat intestine, indicating that the demonstrated effect on bioavailability is caused by MRP2-mediated transport of the parent compound. However, PhIP that reaches the liver is metabolized extensively (Alexander et al., 1995).

We speculate that the transport of PhIP is taking place in cotransport with GSH, because PhIP itself is uncharged. We have shown previously that incubation of MRP2 transduced cells with 100 µM BSO for 24 h before the experiment results in GSH depletion to 17% of controls and that this treatment did not affect ATP levels or cellular integrity (Paulusma et al., 1999). Transport in such pretreated cells, as well as in small intestine from BSO-pretreated animals in Ussing chambers, was reduced significantly. Blood levels of radioactivity were significantly higher in the BSO-pretreated animals than in normal Wistar rats. Although GSH depletion by BSO is an established method (Drew and Miners, 1984; Minchinton et al., 1984; Paulusma et al., 1999), we cannot totally exclude the possibility that other mechanisms are at play as well. In all of our experimental systems, however, BSO causes almost the same results as the absence of MRP2, which suggests that GSH plays a role in MRP2-mediated transport of PhIP.

If this is true, the role of GSH in defense against PhIP is complex. In the past it was shown that glutathione depletion caused a 5-fold increase of DNA adducts in the liver (Kaderlik et al., 1994a) or a 15-fold increase in hepatocytes (Kaderlik et al., 1994b), respectively, in the latter case without changing the metabolism profile. It was also shown that a concentration of 40 mM GSH in vitro inhibited DNA-binding by N-acetoxy-PhIP (Kaderlik et al., 1994b). Our data suggest that, in addition to reduction of DNA adduct formation, GSH may play a role in direct extrusion of the carcinogen. It is likely, therefore, that there is addition of several effects in GSH action, which may explain discordant results between BSO-pretreated Wistar and TR rats in the bioavailability studies.

We show here that the absence of MRP2 leads to altered GSH homeostasis in the small intestine of the TR rat (Table 1) as is the case in the liver (Paulusma et al., 1999). Thus, in this animal, one effect could counteract the other. On one hand, absence of MRP2 will lead to increased tissue levels of PhIP and reactive metabolites, as shown here. On the other hand, the increased tissue levels of GSH could reduce adduct formation. This may explain the relatively low tissue binding in the small intestine of the TR rats and the very high tissue levels of radioactivity in small intestine of the BSO-pretreated Wistar rats. However, BSO-pretreatment produced lower levels of radioactivity in the liver of Wistar rats than MRP2-deficiency in the liver of TR rats. Our data cannot resolve this difference, but it may be possible that this difference in susceptibility to tissue radioactivity binding between intestine and liver is a result of different degrees of GSH depletion (Table 1).

The importance of the described defense mechanism in MRP2 is highly emphasized by the differences in tissue levels 48 h after administration. In our study, MRP2 deficiency led to significantly higher, ethanol-insoluble radioactivity contents in several organs after 48 h. After 48 h, excretion and metabolism of the administered compound is almost completed (over 97% recovery of the dose in urine and feces). Therefore, ethanol-insoluble tissue contents of radioactivity at that time indicate covalent binding of reactive metabolites to the tissue (Watkins et al., 1991). This may suggest, but does not yet prove, that there will be higher carcinogenicity. Because DNA-adduct formation cannot simply be used to estimate carcinogenicity (Ochiai et al., 1996), a long-term feeding study is, in our opinion, the only way to discover the relevance of this difference for carcinogenicity of PhIP.

Our results shed new light on the role of MRP2. Obviously, MRP2 is, in two respects, not only a canalicular multispecific organic anion transporter, as was initially assumed. First, in addition to its role in biliary transport, it plays an important role in the first-line defense in the intestine. Second, it transports a wide variety of neutral or amphipathic compounds, not only organic anions. As such, it probably plays a role in the oral bioavailability of many drugs, toxins, and carcinogens.