Liver plays a crucial role in the disposition and elimination of a variety of physiological substrates and xenobiotics. The uptake of compounds into hepatocytes is primarily mediated by members of the solute carrier superfamily (SLC) located at the basolateral side of the plasma membrane of hepatocytes (Meier et al., 1997). Sodium and ATP-independent transporters involved in the hepatic uptake of anionic compounds can be categorized into two major families: the organic anion transporting polypeptide (OATP/Oatp; human/rodent) and the organic anion transporter (OAT/Oat; human/rodent). Members of the OATP/Oatp (solute carrier family 21: SLC21/slc21) family are predominantly expressed in the liver and are responsible for mediating the hepatocellular uptake of a wide spectrum of substrates, including bile salts, conjugates of steroids, hormones, and other large amphiphilic organic anions (Kanai et al., 1996a,b; Meier et al., 1997; Ambudkar et al., 1999; Eckhardt et al., 1999; Suzuki and Sugiyama, 1999). The substrate specificities of various OATPs/Oatps overlap considerably, although unique features of individual transporters have been demonstrated. For example, it has been shown that the cardiovascular drug digoxin is specifically transported with high affinity by Oatp2 (slc21a5) (Noe et al., 1997; Kodawara et al., 2002; Shitara et al., 2002). Because digoxin is also known to be a substrate for the ATP-dependent drug efflux pump, P-glycoprotein (P-gp), which is located in the canalicular membrane of hepatocytes (de Lannoy and Silverman, 1992; Cavet et al., 1996; Keppler and Arias, 1997; Stieger and Meier, 1998), it is reasonable to postulate that digoxin is actively taken up across the sinusoidal membrane into hepatocytes via Oatp2 and secreted into biliary canaliculi via P-gp. To estimate the contribution of Oatp2 to the hepatic disposition of digoxin, an inhibitor for Oatp2, rifampicin, was used to block the uptake function of Oatp2. Rifampicin has been shown to effectively block Oatp2-mediated transport of digoxin in Oatp2-transfected LLC-PK₁ cells with a K_i value of 1.46 μM (Shitara et al., 2002). On the basis of recent in vivo brain uptake studies, Zong and Pollack (2003) suggest that rifampicin at concentrations of 500 μM or greater could inhibit P-gp-mediated efflux. Earlier studies have also shown that rifampicin can inhibit P-gp activity in vitro (Fardel et al., 1995; Furusawa et al., 1997). However, the degree to which rifampicin inhibits P-gp-mediated efflux is dependent on the substrate molecule, and rifampicin inhibition on P-gp-mediated efflux of digoxin has not been specifically investigated. The antiarrhythmic agent quinidine, a well known P-gp inhibitor (de Lannoy et al., 1992; Su and Huang, 1996; Fromm et al., 1999), was used to examine P-gp-mediated transport of digoxin.

In human, digoxin is mostly excreted unchanged by the kidney; however, it is extensively metabolized (more than 70% of an i.p. dose) by cytochrome P450 3A (CYP3A) in rat (Harrison and Gibaldi, 1976; Schmoldt and Ahsendorf, 1980; Rodin and Johnson, 1988). Thus, we hypothesized that the regulation of the uptake/metabolism/efflux pathway for digoxin by Oatp2, CYP3A, and P-gp might be affected in the presence of uptake (Oatp2) and efflux transporter (P-gp) inhibitors.

The aim of the present study was to test the hypothesis that ex situ inhibition of hepatic uptake and efflux transporters would modify the disposition and metabolism of digoxin in the isolated perfused rat liver (IPRL) system. This system closely mimics the hepatic physiological condition independent of potentially confounding influence from other organs such as intestine and kidney. It has advantage over in vitro assays because cellular studies do not necessarily establish transport directionality and transporter-enzyme interactions at the organ level. Alterations in concentrations and amounts of digoxin and its metabolite digoxigenin bisdigitoxoside (Dg2) were monitored via a specific assay in perfusing medium and liver tissues.

Materials and Methods

Chemicals. Digoxin, quinidine, rifampicin, and corticosterone were purchased from Sigma-Aldrich (St. Louis, MO). The powder form of Dg2 was extracted and purified by HPLC in our laboratory.

Surgery and Perfusion of Isolated Rat Livers. Male Sprague-Dawley rats (300-400 g; Bantin and Kingman, San Leandro, CA) were anesthetized with ketamine/xylazine (80 mg; 12 mg/ml) before surgery. The hepatic portal vein and superior vena cava were cannulated following approval of protocols by the Committee on Animal Research, University of California, San Francisco. The livers were isolated for perfusion ex situ as described previously from our laboratory (Prueksaritanont et al., 1992; Wu and Benet, 2003). Oxygenated Krebs-Henseleit buffer (pH 7.4), supplemented with sodium taurocholate (220 nmol/min), 1% bovine serum albumin, and glucose (10 mM), was pumped through the liver at a flow rate of 40 ml/min via a catheter inserted in the portal vein. We chose the commonly used red blood cell-free perfusion technique, which requires higher flow rates than the physiological flow rate to provide sufficient oxygen carrying capacity. Perfusion was performed at 37°C in a recirculatory manner, from a reservoir containing 110 ml of perfusate, through a 10-μm in-line filter, oxygenator, and bubble trap placed before the liver. The perfusate in the reservoir was oxygenated directly using carbogen, 95% O₂/5% CO₂, and stirred continuously. Liver viability was judged on the basis of its appearance (uniformly pink to brown), oxygen consumption, portal vein pressure (20-30 mm Hg), and pH (in the range of 7.35-7.45), as well as metabolic capability.

After allowing time for the liver to stabilize for 20 min, the inhibitors (quinidine or rifampicin) were added to the reservoir 10 min before digoxin addition. Perfusate samples (0.5 ml) were collected immediately (0 min) and at 2, 5, 10, 15, 20, 30, 45, and 60 min after the addition of digoxin. At the end of experiment, the liver was removed, blotted dry, and weighed. An aliquot of liver was homogenized with ice-cold Krebs-Henseleit buffer in a 1:2 ratio and maintained frozen at -80°C before analysis. No attempt was made to quantitate digoxin and Dg2 in the bile due to fluctuating bile flow rate from cannulated common bile duct after liver isolation.

Experimental Design. To examine the influence of rifampicin and quinidine on the hepatic disposition of digoxin, 18 rats were divided equally into three groups, and each group was perfused with 10 μg of digoxin solution diluted in normal saline, added directly into the reservoir to yield an initial concentration of ∼110 nM. Whereas one group served as the control, the other two groups served as treatments. For inhibition studies, rifampicin or quinidine with a final perfusate concentration of 100 or 10 μM, respectively, was administered 10 min before digoxin addition.

Sample Preparation. Liquid-liquid extraction was performed for sample preparation. As a first step, 200 μl of each perfusate sample and 100 μl of each liver homogenate sample supplemented with 100 μl of perfusate buffer were transferred into glass tubes. The internal standard corticosterone (50 ng/ml) and 2 ml of methyl tertiary butyl ether were added to each tube. The final sample/internal standard/organic solvent mixtures were then mixed briefly followed by centrifugation at 4000 rpm for 10 min. Afterwards, a methanol ice-bath was prepared to freeze the bottom aqueous layers of each centrifuged sample so that the upper organic layers could be easily separated out. The organic layer of each sample was then evaporated under nitrogen. The dried solutes were reconstituted with 200 μl of methanol and transferred into HPLC screw cap vials with 250-μl inserts (Hewlett Packard, Palo Alto, CA).

Calibration control samples for digoxin and Dg2 with final concentrations ranging from 1 nM to 1 μM were extracted and prepared the same way as experimental samples. A calibration control sample set was prepared for each type of sample using perfused blank perfusate buffer and blank liver homogenate solutions, respectively.

Measurement of Digoxin and Its Metabolites. Samples were analyzed on a liquid chromatography-mass selective detector system (Hewlett Packard) consisting of the 1100 HPLC components HPLCI and HPLCII as described previously (Christians et al., 2000). The two HPLC systems were connected via a 7240 Rheodyne six-port switching valve mounted on a step motor (Rheodyne, Cotati, CA). The system was controlled and data were processed using ChemStation Software revision A.06.01 (Hewlett Packard).

Samples (50 μl) were injected onto a 10 × 2-mm extraction column (Keystone Scientific, Bellefonte, PA) filled with Hypersil ODS-1 of 10-μm particle size (Shandon, Chadwick, UK). Samples were washed with a mobile phase of 20% methanol and 80% 0.1% formic acid supplemented with 1 mM sodium acetate. The flow was 2 ml/min, and the temperature for the extraction column was set to 65°C. After 0.75 min, the switching valve was activated, and the analytes were eluted in the backflush mode from the extraction column onto a 50 × 4.6-mm C₈, 3.5-μm analytical column (Zorbax XDB, C₈; Hewlett Packard)_. The mobile phase consisted of methanol and 0.1% formic acid supplemented with 1 mM sodium acetate. The following gradient was run: time 0 min, 55% methanol; 6 min, 100% methanol. The flow rate was 0.7 ml/min. The analytical column was also maintained at 65°C. Two minutes after sample injection, the mass-selective detector was activated.

Measurement of Rifampicin and Quinidine Levels in Perfusate Samples. The HPLC system used was described above. Rifampicin was resolved on a 250 × 4.6-μm C8, 4.0 100-Å analytical column (Microsorb-MV, C₈; Varian, Walnut Creek, CA). The mobile phase was composed of 0.05 M potassium dihydrogen phosphate-acetonitrile [55:45 (v/v)] with a flow rate of 1 ml/min. The UV absorbance was monitored at a wavelength of 340 nm. Quinidine was chromatographed on a 150 × 4.6-μm C8, 4.0 100-Å analytical column (Microsorb-MV, C₈; Varian) with a mobile phase of 0.5% acetic acid and 0.25% tetraethylamine-acetonitrile [75:25 (v/v)] with a flow rate of 1.5 ml/min. The UV absorbance was monitored at a wavelength of 245 nm.

Data Analysis. Values for area under the concentration-time curve (AUC) were calculated using the linear trapezoidal method. For the inhibition studies, concentrations in perfusate and homogenized liver at 60 min were used to calculate the ratios of the amount of digoxin to Dg2 recovered in liver, the ratio of the concentrations of rifampicin in liver to perfusate, and the ratio of the concentrations of quinidine in liver to perfusate. Student's two-tailed t test was used to assess statistical significance. Differences between groups were considered significant if p < 0.05.

Results

Using the perfusion system described above in the absence of a liver, perfusate concentrations of digoxin remained relatively constant (Fig. 1), indicating that little drug is lost in the system. In all experiments, Dg2 was the only digoxin metabolite detected.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Influence of rifampicin and quinidine on concentrations of digoxin in perfusate after addition of 10 μg of digoxin to the perfusate. Values are mean ± S.D., n = 6 per group; no liver, n = 3. ^*, p < 0.05; ^**, p < 0.01 for values compared with control.

Figure 1 illustrates the decline of digoxin concentrations in perfusate over time in the absence and presence of rifampicin or quinidine. In control livers perfused with digoxin only, digoxin concentrations in perfusate declined in a bioexponential manner in which they dropped rapidly within the first 20 min and decreased steadily throughout the 60-min perfusion period (Fig. 1). The concentration of Dg2 was detectable 2 min after the start of perfusion and increased steadily up to 60 min (Fig. 2), indicating that hepatic enzymatic activity of CYP3A was functional throughout the perfusion period.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Influence of rifampicin and quinidine on concentrations of Dg2 in perfusate after addition of 10 μg of digoxin to the perfusing medium. Values are mean ± S.D., n = 6 per group. ^*, p < 0.05; ^**, p < 0.01 for values compared with control.

Digoxin concentrations at all sampling times were elevated by the addition of rifampicin (Fig. 1), which is reflected by the significant 34% increase in AUC from 3880 ± 210 to 5200 ± 240 nM·min (p < 0.01) as shown in Table 1. In contrast, concurrent perfusion with quinidine decreased the concentration of digoxin markedly with a 14% decrease in the AUC to 3220 ± 340 nM·min (p < 0.05) (Table 1). In vitro rat hepatic microsome studies showed no effect for these concentrations of the two inhibitors on digoxin metabolism (data not shown). Figure 3, A and B, show the concentrations of rifampicin and quinidine in perfusate, respectively. Immediately after administration, rifampicin levels declined rapidly within the first 10 min (before digoxin was dosed) and decreased only slightly afterwards. A similar pattern was observed for quinidine, indicating rapid hepatic uptake during the initial distribution phase.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Perfusate time course of rifampicin (A) and quinidine (B). Values are mean ± S.D., n = 6.

The Dg2 concentration-time profiles are shown in Fig. 2. Dg2 was detectable as early as 2 min after the start of perfusion and increased steadily up to 60 min. Comparing the AUC of the control group and the treatment groups (Table 1), the AUC of Dg2 was decreased 24% by rifampicin (p < 0.05) and increased 14% by quinidine (p < 0.05). The ratio of the metabolite AUC to parent AUC (Dg2 AUC/digoxin AUC) was also calculated and was significantly increased by quinidine (0.530 ± 0.076 versus 0.382 ± 0.029 in control; p < 0.01) and significantly decreased by rifampicin (0.217 ± 0.037 versus 0.382 ± 0.029 in control; p < 0.05).

Figure 4 shows the amount of digoxin and Dg2 recovered in liver tissue (expressed as percentage of original dose) for control, rifampicin, and quinidine-treated livers. Approximately 17% of digoxin was retained in control livers after 60 min of perfusion, whereas 12 and 13% of digoxin dose, respectively, was detected in rifampicin and quinidine-treated groups. Dg2 in liver has a significant increase of 168% in the quinidine-treated group (p < 0.01) versus control, whereas it decreased by 31% in the rifampicin-treated group (p < 0.01). The ratios of Dg2 to digoxin retention in liver were comparable for control and rifampicin-treated livers, whereas that for quinidine-treated livers showed a highly significant increase by ∼4-fold (0.488 ± 0.192 versus 0.131 ± 0.023 in control; p < 0.01) (Table 1). Mass balance calculations (digoxin and Dg2 in perfusate and liver) show no significant differences between the three studies (90.5 ± 9.8% for control, 89.3 ± 6.6% with rifampicin, and 88.9 ± 5.1% with quinidine).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Amounts of digoxin retained and Dg2 formed in control, rifampicin-treated, and quinidine-treated livers. Values are mean ± S.D., n = 6 per group. ^*, p < 0.05; ^**, p < 0.01 for values compared with control.

Discussion

In the present study, we investigated the ex vivo disposition and excretion of digoxin and its metabolite Dg2 in the isolated rat perfused liver system in the presence of inhibitors for Oatp2 and P-gp. Dg2 was the only metabolite detected during the perfusion period. This is consistent with previous data from our laboratory where rat liver microsome studies showed that the formation of Dg2 from digoxin is rapid (CL_int = 2.92 μl/min/mg protein) and is much faster (∼20-fold higher) than the cleavage of Dg2 to digoxigenin mono-digitoxoside and the aglycone digoxigenin (Salphati and Benet, 1999).

Rifampicin has been shown to be a potent inhibitor of Oatp2 in LLC-PK₁ cells (Shitara et al., 2002) and rat hepatocytes (Fattinger et al., 2000). At 100 μM, rifampicin substantially inhibits Oatp2-mediated transport of digoxin by 60% (Shitara et al., 2002). In microsomal incubations, no effect of 100 μM rifampicin on digoxin was observed (data not shown). Our results indicate that a single bolus dose of rifampicin at the same concentration significantly increased the AUC of digoxin in perfusate (Table 1) and decreased the amount of recovered digoxin in liver (Fig. 4), suggesting that rifampicin markedly reduced digoxin's sinusoidal uptake. We also monitored the disposition of rifampicin itself in perfusate (Fig. 3A) and observed higher concentrations of rifampicin in liver versus perfusate (Table 1). Given that the hepatocellular uptake of digoxin was reduced in the presence of rifampicin, we would expect a reduction in its hepatic metabolism because the bulk of drug entering the hepatocytes is reduced and the availability to CYP3A is decreased subsequently. Consequently, in our results, significantly less Dg2 was detected in liver tissues (Fig. 4). The change of Dg2 in liver seems to be reflective of what is observed in the perfusate (Fig. 2). The ratio of Dg2 to digoxin retention in liver is consistent between the control and the rifampicin-treated group, suggesting linear kinetics in which reduction in digoxin entry into liver leads to a proportional reduction in Dg2 formed. These results support our hypothesis that an uptake transporter such as Oatp2 may control the access of dual Oatp2/CYP3A substrates to the enzyme in the liver. Because the Dg2 to parent ratio in the liver did not change in the presence of rifampicin versus control (Table 1), we believe that any potential rifampicin inhibition of P-gp as reported by Zong and Pollack (2003) is not significant here.

P-gp functions as an energy-dependent drug efflux pump and prevents the accumulation of drugs in various organs, including intestine and liver. Wu and Benet (2003) have performed an IPRL study using a known potent P-gp inhibitor, GG918 (1 μM), to inhibit the P-gp-mediated efflux of a dual CYP3A and P-gp substrate, tacrolimus. The AUC of tacrolimus was significantly decreased from control (2260 ± 430 ng-min/ml) by GG918 (1730 ± 270 ng-min/ml; p < 0.05), indicating the importance of P-gp in affecting CYP3A metabolism.

Quinidine is another potent inhibitor of this efflux pump. Fromm et al. (1999) have shown that 5 μM quinidine effectively inhibits P-gp-mediated transport of digoxin by approximately 57% in the Caco-2 MDR1-transfected cell line. Kakumoto et al. (2002) also demonstrated that quinidine inhibits the transport of digoxin in MDR1-overexpressing LLC-GA5-COL150 cells with an estimated IC₅₀ value of 9.52 μM. Here, we tested whether coperfusion of quinidine in isolated perfused rat livers could inhibit the P-gp-mediated transport of digoxin. In the liver, absorbed compounds enter the hepatocytes from the sinusoidal blood and then the drugs are either biotransformed, transported/diffused back into blood, or eliminated via biliary secretion. Because P-gp is located on the canalicular membrane of hepatocytes, drug molecules confront CYP3A before P-gp efflux, in an opposite manner to the Oatp2/CYP3A interaction. Because quinidine can inhibit P-gp from pumping drug molecules out to the bile canaliculi, the parent drug will have a prolonged intracellular residence time in liver tissue, thereby, increasing its availability to CYP3A for metabolism; as a result, an increase in metabolite formation is expected with a relative decrease in parent compound compared with control. Others have shown that Oatp2-mediated transport of digoxin is inhibited by quinidine with a K_i value of 120 ± 27 μM in Oatp2-stably expressing LLC-PK₁ cells (Shitara et al., 2002). This is not surprising because often inhibitors that alter P-gp transport activity also affect the function of uptake transporters. To overcome this complication, choosing an appropriate concentration of quinidine to serve as a specific inhibitor of P-gp becomes crucial. In our study, an initial perfusate concentration of 10 μM quinidine was used based on its inhibition constants (K_i) for Oatp2 and P-gp-mediated transport of digoxin, respectively. As depicted in Fig. 3B, quinidine perfusate concentrations affecting Oatp2 rapidly fell to 15% of this value, about one-hundredth of the K_i for Oatp2. However, intracellular concentrations affecting P-gp were probably maintained at or above the 10 μM concentration due to the preferential intracellular accumulation of quinidine (Table 1), where the liver to perfusate concentration ratio at 60 min is ∼8.

A significant increase in digoxin metabolism could also be observed in the liver tissues as reflected by the 3.7-fold increase in the ratio of Dg2 to digoxin retention in the quinidine-treated group (Table 1), compared with the 1.4-fold increase in perfusate AUCs. An efflux transport study using Madin-Darby canine kidney-expressing MDR1 cells demonstrated that Dg2, which is structurally similar to digoxin, is a substrate for P-gp (data not shown), explaining the increased Dg2 in liver relative to perfusate. Because mass balance was the same for all three groups, the significant decrease in liver digoxin concentrations in the quinidine group (Fig. 4) may be explained by a marked increase in metabolite formation in the liver, whereas the significant decrease in liver digoxin concentrations in the rifampicin (Fig. 4) group is explained by decreased hepatic uptake.

The working concentration of digoxin was carefully chosen based on its metabolic parameter to ensure that the metabolizing enzyme does not reach saturation. Here, 110 nM digoxin was selected, which is well below the K_m value (125 μM) for the CYP3A catalysis of Dg2 from digoxin reported by our laboratory (Salphati and Benet, 1999). Quinidine had no effect on digoxin metabolism up to 100 μM (Salphati and Benet, 1999) suggesting that it is unlikely that quindine inhibits CYP3A in the perfused rat liver. The present study suggests that 10 μM quinidine is an appropriate concentration with minimal inhibitory effects on Oatp2 and CYP3A.

The mechanism of the clinically important drug interaction between digoxin and quinidine is well documented in the literature (Leahey et al., 1978; Bussey, 1982; Mordel et al., 1993) When administered concomitantly, quinidine is known to increase the plasma concentrations of digoxin in both humans and rats (Bigger and Leahey, 1982; Sakai et al., 1988). In vitro studies addressing the digoxin-quinidine interaction suggest that P-gp is involved (Tanigawara et al., 1992; Su and Huang, 1996; Fromm et al., 1999). P-gp is highly expressed in the apical membrane of epithelial cells of small intestine (Su and Huang, 1996), the apical canalicular membrane of hepatocytes (Keppler and Arias, 1997), and the renal tubular cells of the kidney (Hori et al., 1993). The orientation of P-gp in these locations allows the rapid efflux of substrates from the basolateral surface to the lumen of small intestine, renal tubules, and into the bile. Hence, it is likely that the reported increase in digoxin plasma levels in rats is caused by the inhibitory effect of the coadministered quinidine on the basolateral-to-apical transport of digoxin across the enterocytes, hepatocytes, and renal tubules (Hori et al., 1993; Su and Huang, 1996). However, because digoxin is extensively metabolized by CYP3A in rat, our data demonstrate that quinidine increases the extent of digoxin metabolism by blocking the P-gp-mediated biliary secretion of digoxin, yielding lower in vivo digoxin concentrations in the hepatic compartment alone. Thus, our results point out a caution, as we have noted previously for P-gp knockout animals (Cummins et al., 2002b): although inhibition of P-gp may decrease overall systemic drug clearance, it would always be expected to increase hepatic metabolism of P-gp substrates.

Recent studies from our laboratory (Cummins et al., 2002a,b, 2003; Wu and Benet, 2003) have examined the interplay between transporters and metabolic enzymes, proposing that transporters may control the access of drug molecules to the enzymes and therefore that changes in transporter function can change intestinal and hepatic metabolism without apparently affecting enzyme activity (Benet et al., 2003). The present work supports this hypothesis with respect to hepatic metabolism. Here, we demonstrated that ex situ inhibition of Oatp2 and P-gp modifies the transport and metabolism of digoxin in the isolated perfused rat liver system. Rifampicin significantly inhibits the uptake function of Oatp2 in transporting digoxin into rat hepatocytes and decreases digoxin metabolism. Quinidine significantly increases the metabolism of digoxin by inhibiting P-gp-mediated efflux. These recent studies from our laboratory emphasize that considerations of the interplay of uptake and efflux transporters with metabolic enzymes must be considered in evaluating drug disposition.