Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Bioavailability of drugs after oral administration sometimes depends nonlinearly on concentration, which shows a dose-dependent increase and/or a decrease. An understanding of the mechanisms causing such nonlinear behavior would be helpful in the rational development of orally active drugs, as well as in the clinical context. Nonlinear phenomena in intestinal absorption could be caused by several factors. A decrease of bioavailability with increase of dose can be accounted for by limited solubility of the drug in the gut lumen or by capacity-limited permeation across the intestinal epithelial membranes, e.g., because of the involvement of a specialized carrier-mediated transport mechanism. Recent studies on drug transport in the intestine have demonstrated the presence of many carrier-mediated transport systems for various drugs as well as natural compounds (Tamai and Tsuji, 1996; Tsuji and Tamai, 1996). On the other hand, a dose-dependent increase of bioavailability may be mainly ascribed to saturable first-pass metabolism in the liver and/or gastrointestinal tract. The finding of drug-metabolizing enzymes such as cytochrome P-450 isozyme, CYP3A, in the small intestine has contributed to the mechanistic understanding of alterations of drug bioavailability arising from causes other than hepatic metabolism (Watkins et al., 1987). CYP3A is abundantly present in the intestinal tissues, so coadministration of drugs subject to metabolism by CYP3A or foods that contain enzyme inhibitor(s) may cause a significant increase in the plasma concentration of a drug similar to that which would be expected only from the saturation of a hepatic first-pass effect (Ducharme et al., 1995; Gomez et al., 1995). Another possible cause of dose-dependent alterations in bioavailability is the contribution of an intestinal luminal secretory system. P-Glycoprotein, which was originally found in multidrug-resistant tumor cells as an anticancer drug efflux pump, is also present in the luminal membrane of the intestinal epithelial cells (Thiebaut et al., 1987) and plays a role as a secretory transporter into intestinal lumen from the cells (Hsing et al., 1992; Hunter et al., 1993), in addition to its role as a component of the blood-brain barrier, transferring drugs out of the brain (Tsuji et al., 1992). P-Glycoprotein thus functions as an absorption barrier to various drugs, resulting in a lower absorption than would be expected from the lipophilicity of the drug molecule (Terao et al., 1996). If such secretory transporters, including P-glycoprotein, operate and become saturated at high doses of drugs, an apparent increase of bioavailability would be expected at high drug concentrations.

Azasetron is a selective 5-HT₃ receptor antagonist. It has a higher affinity for the 5-HT₃ receptor than pharmacologically and structurally analogous ondansetron and granisetron, and it is used to ameliorate the emetic effect of cytotoxic anticancer drugs (Fukuda et al., 1991). The lipophilicity of azasetron is relatively low, with a log P of 0.35 as assessed by n-octanol buffer (pH 7.4) partition. However, its bioavailability after oral administration in rats is high, ranging from 34 to 93%, depending on a dose in the range from 1 to 10 mg/kg, although these tested doses are much higher than the clinical dose (Nagamatsu, Y., Yamada, I. and Shibata, M., unpublished observation). This moderate to high absorbability of azasetron despite its hydrophilic nature and the dose-dependent increase of bioavailability suggest that azasetron could be absorbed and/or secreted by saturable transport mechanisms in the small intestine, although involvement of a saturable hepatic first-pass effect cannot be excluded (Nishimine et al., 1992). Recently, ondansetron, another 5-HT₃ receptor antagonist, was transported by the multidrug efflux transporter P-glycoprotein in mice and human (Schinkel et al., 1996). So, azasetron may also be transported into the intestinal lumen from epithelial cells by P-glycoprotein, which results in the nonlinear increase of bioavailability.

In the present study, to test the hypothesis that the nonlinear intestinal absorption of azasetron is caused by the involvement of concentration-dependent absorptive and secretory transporters, we examined the in situ absorption and in vitro intestinal tissue permeation of azasetron in rats, as well as the uptake and transport of the drug by cultured monolayers of human adenocarcinoma Caco-2 cells, which contain many intestinal transporters, including P-glycoprotein.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Materials. [¹⁴C]Azasetron (884 kBq/mg), unlabeled azasetron, granisetron and ondansetron were kindly supplied by Yoshitomi Pharmaceutical Industries, Ltd. (Fukuoka, Japan). [³H]Mannitol (1110 GBq/mmol) was purchased from New England Nuclear (Boston, MA). Cyclosporin A was kindly supplied by Sandoz (Basel, Switzerland). All other chemicals were commercial products of reagent grade.

Measurement of intestinal absorption by the Doluisio method. Intestinal absorption was evaluated by the Doluisio method (Doluisio et al., 1969). The ileum of male Wistar rats weighing 200 to 280 g (Japan SLC, Hamamatsu, Japan) was exposed by midline abdominal incision, and two L-shaped glass cannulas were inserted through small slits at the proximal and distal ends (20 cm). Each cannula was secured by ligation with a silk suture, and the intestine was returned to the abdominal cavity to maintain its integrity. A 4-cm portion of Tygon tubing was attached to the exposed end of each cannula, and a 2.5-ml hypodermic syringe fitted with a three-way stopcock and containing perfusion solution warmed at 37°C was attached to the proximal cannula. To clear the gut, perfusion solution was passed slowly through it to the distal cannula and discarded until the effluent was clear. The remaining perfusion solution was carefully expelled from the intestine by means of air pumped through the syringe, and 2 ml of drug solution was immediately introduced into the intestine. The distal cannula was connected to a 30-ml syringe fitted with a three-way stopcock. At 15, 30, 45 and 60 min after administration of a drug solution, a 0.2-ml aliquot of luminal solution was removed through the attached syringe. The absorption rate constant was evaluated from the slope of decline of the concentration in the luminal fluid with time. Here, the change in volume of water in the intestinal lumen was corrected by the measurement of the change of concentration of unabsorbable marker, phenolsulfophthalein (100 µM), administered simultaneously with azasetron.

Transport experiments by the Ussing-type chamber method. Rat ileal tissue sheets were prepared as described previously (Tamai et al., 1997). The tissue preparation, consisting of the mucosa and most of the muscularis mucosa, was made by removing the submucosa and tunica muscularis with fine forceps. The tissue sheets were mounted vertically in an Ussing-type chamber that provided an exposed area of 0.5 cm². The volume of bathing solution on each side was 10 ml, and the solution temperature was maintained at 37°C in a water-jacketed reservoir. The test solution was composed of 128 mM NaCl, 5.1 mM KCl, 1.4 mM CaCl₂, 1.3 mM MgSO₄, 21 mM NaHCO₃, 1.3 mM KH₂PO₄ and 10 mM NaH₂PO₄ at pH 7.4, and the solution was gassed with 95% O₂/5% CO₂ before and during the transport experiment.

Cultivation of Caco-2, K562 and K562/ADM cells. Caco-2 cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and 1% non-essential amino acids, 2 mM L-glutamine, 100 units/ml penicillin G and 100 µg/ml streptomycin, as described previously (Tsuji et al., 1994). For the transport experiments, Caco-2 cells were grown on Transwell microporous polycarbonate membrane (Costar, Bedford, MA) and cultured for about 3 weeks before use for the transport experiments. For the uptake experiments, Caco-2 cells were seeded on collagen-coated culture dishes and grown for about 2 weeks. Human myelogenous leukemia K562 cells and their adriamycin-resistant variant K562/ADM cells were kindly supplied by Dr. Tsuruo (Tokyo University, Tokyo, Japan), and cultured in RPMI 1640 medium containing 5% fetal bovine serum, 100 µg/ml kanamycin and 2 mg/ml sodium bicarbonate (Tsuruo et al., 1986). The adriamycin-resistant cells were maintained in culture medium containing 30 ng/ml adriamycin and were grown in drug-free medium 1 week before the transport experiments.

Uptake and transport experiments with Caco-2 cells. Uptake of azasetron by cultured monolayers of Caco-2 cells was examined by use of the method reported previously (Tsuji et al., 1994). Cultured cells were first washed three times with 1 ml of HBSS (0.952 mM CaCl₂, 5.36 mM KCl, 0.441 mM KH₂PO₄, 0.812 mM MgSO₄, 136.7 mM NaCl, 0.385 mM Na₂HPO₄, 25 mM D-glucose and 10 mM HEPES, pH 7.4; the osmolarity was 315 mOsm/kg) at 37°C. Uptake was initiated by adding 200 µl of incubation solution containing [¹⁴C]azasetron (usually 2 µM) to the cells. At designated times, the cells were washed three times with 1 ml of ice-cold HBSS to terminate the uptake. To solubilize cells, a 500-µl aliquot of 5 N NaOH was added and the mixture was left at room temperature for 2 h. After neutralization with 500 µl of 5 N HCl, radioactivity in the cell precipitate was measured.

Analytical methods. To assay radioactivity, all samples were transferred into counting vials, mixed with scintillation fluid (Cleasol I; Nacalai Tesque, Kyoto, Japan) and counted in a liquid scintillation counter (Aloka, Tokyo, Japan). Nonradioactive compound was measured by HPLC. The HPLC system consisted of a constant-flow pump (880-PU; Japan Spectroscopic Co., Tokyo, Japan), fluorescence detector (RF-550; Shimadzu Co., Kyoto, Japan), integrator (Chromatopac CR3A; Shimadzu Co.) and automatic sample injector (AS-L350; Japan Spectroscopic Co.). The analytical column was reversed-phase TSK-gel ODS-50Ts (4.6 mm × 15 cm, Tosoh, Tokyo, Japan). The mobile phase was 0.1 M ammonium acetate-acetonitrile-tetrahydrofuran (85.8:8:6.2) adjusted to pH 5.0 with acetic acid. Azasetron was detected with a fluorescence detector at excitation and emission wavelengths of 318 and 382 nm, respectively. The retention time of azasetron was approximately 5.6 min with detection limit of about 0.01 µM, and reproducibility of the assay was 100 ± 9.3% (n = 20). Cellular protein was measured by the method of Lowry et al. (1951) with bovine serum albumin as the standard.

Data analysis. Uptake (µl/1.5 × 10⁶ cells or µl/mg protein) and permeation (µl/mg protein) were measured by dividing the amount transported (dpm/1.5 × 10⁶ cells or mg protein) by the initial concentration of test compound on the donor side (dpm/µl). The permeation coefficient (µl/min/mg protein or µl/min/cm²) was obtained from the slope of the initial linear portion of the plots of permeation against time (min). To estimate the kinetic parameters of saturable uptake by Caco-2 monolayers, the uptake rate (v) was fitted to the following equation, with use of a nonlinear least-squares regression analysis program, MULTI (Yamaoka et al., 1981): where v and s are the uptake rate and concentration of substrate, and K_m and V_max are the half-saturation concentration (Michaelis constant) and the maximum transport rate, respectively. All data are expressed as means ± S.E. and statistical analysis was performed by use of Student's one-tailed t test. The criterion of significance was taken to be P < .05.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Concentration dependence of intestinal absorption of azasetron in rats. The relationship between the first-order absorption rate constant and azasetron concentration was evaluated by the Doluisio method. When 0.01 to 60 mM azasetron was perfused into the intraileal loop the intestinal absorption rate changed nonlinearly as shown in figure 1. There was a slight increase of intestinal absorption rate from 0.01 to 1 mM, followed by a sharp increase up to 10 mM, then a decrease until 60 mM. The absorption rate constants at concentrations of 5, 10, 20 and 60 mM were significantly higher than the absorption rate constant at the lower concentration of 0.01 mM (P < .05). This result suggests that at least two nonlinear events are involved in the intestinal absorption of azasetron.

View larger version (14K): [in this window] [in a new window]   Fig. 1.   Concentration dependence of ileal absorption of azasetron in rats. The intestinal absorption rate constant of azasetron (10 µM-60 mM) was evaluated from the time course of the decrease in luminal azasetron concentration measured by the Doluisio method. Each point represents the mean ± S.E. of three to six experiments.

View larger version (16K): [in this window] [in a new window]   Fig. 2.   Concentration dependence of azasetron permeation across rat ileal tissue. Permeation of azasetron was determined by use of an Ussing-type chamber mounted with rat ileum. Open and closed circles represent the permeation coefficient in the mucosal-to-serosal and serosal-to-mucosal directions, respectively. Open and closed squares represent mucosal-to-serosal and serosal-to-mucosal transfer of azasetron in the presence of 10 µM cyclosporin A, respectively. Each point represents the mean ± S.E. of four or five experiments.

Transcellular transport of azasetron across Caco-2 cells. To determine whether the transport of [¹⁴C]azasetron across Caco-2 cells was unidirectional, transepithelial fluxes were measured by adding 2 µM [¹⁴C]azasetron to either the apical or basolateral side of the monolayer of Caco-2 cells and monitoring the appearance of radioactivity on the opposite side (fig. 3). The permeation of [¹⁴C]azasetron was linear over 60 min with an initial lag time of a few minutes. The flux from the basolateral-to-apical side (3.36 ± 0.085 µl/mg protein/min) was about 1.5 times larger than the reverse flux (2.30 ± 0.025 µl/mg protein/min). HPLC analysis showed that more than 99% of the permeated radioactivity was caused by unchanged [¹⁴C]azasetron (data not shown). The fluxes of [¹⁴C]azasetron across the Caco-2 monolayer were significantly higher than the flux of [³H]mannitol (0.092 ± 0.003 µl/mg protein/min), which is a measure of paracellular permeability. Therefore, the transport of [¹⁴C]azasetron can be ascribed to transcellular permeation.

View larger version (15K): [in this window] [in a new window]   Fig. 3.   Transport of [14C]azasetron and [3H]mannitol across a Caco-2 cell monolayer. Open and closed circles represent transport in the apical-to-basolateral and basolateral-to-apical directions, respectively, at a concentration of 2 µM [14C]azasetron. Open squares represent [3H]mannitol (3 µM) transport from the apical to the basolateral side. Each point represents the mean ± S.E. of three experiments.

Uptake of azasetron by Caco-2 cells. Figure 4 shows the time course of [¹⁴C]azasetron uptake by Caco-2 cells. Uptake of [¹⁴C]azasetron is time-dependent and the steady-state was attained by 10 min. Because the uptake of [¹⁴C]azasetron increased linearly for 1 min, the influx characteristics were evaluated by measuring the uptake at 1 min, and the uptake at 60 min was studied to evaluate the efflux characteristics of azasetron in the following experiments. The effect of ATP depletion by metabolic inhibitors and that of cyclosporin A on the apparent uptake of [¹⁴C]azasetron were examined. ATP-depletion by NaF, NaN₃ and 3-O-methylglucose in the absence of D-glucose, significantly increased the steady-state uptake of [¹⁴C]azasetron, but not the initial influx. Similarly, an increase in the steady-state uptake of azasetron was observed in the presence of 10 mM cyclosporin A. These results demonstrate that efflux of azasetron in Caco-2 cells is metabolic energy-dependent and cyclosporin A-sensitive.

View larger version (17K): [in this window] [in a new window]   Fig. 4.   Time course of apical uptake of [14C]azasetron by Caco-2 cells. Uptake of [14C]azasetron (4 µM) was measured in the absence () and presence () of 10 µM cyclosporin A or in ATP-depleted Caco-2 cells (). Cellular ATP was depleted with a combination of 10 mM NaN3, 10 mM NaF and 25 mM 3-O-methylglucose for 20 min before the uptake experiment. Each point represents the mean ± S.E. of three or four experiments.

View larger version (19K): [in this window] [in a new window]   Fig. 5.   Concentration dependence of apical uptake of [14C]azasetron by Caco-2 cells. Initial uptake rate of azasetron (4 µM-50 mM) was measured at 37°C for 1 min.The solid line represents the uptake rate evaluated by use of the MULTI-fitted Km and Vmax. Each point represents the mean ± S.E. of four experiments. Insert: Eadie-Hofstee plot of azasetron uptake by Caco-2 cells. Each point represents the mean ± S.E. of four experiments.

View larger version (35K): [in this window] [in a new window]   Fig. 6.   Concentration-dependent effect of unlabeled azasetron and ondansetron on the apical uptake of [14C]azasetron by Caco-2 cells. Uptake of [14C]azasetron (4 µM) was measured in the presence and absence of unlabeled azasetron and ondansetron at 37°C for 1 min (open columns) and 60 min (dotted columns). Each column represents the mean ± S.E. of four experiments. (*) Significantly different from the control value by Student's t test (P < .05). N.D.: not determined.

Uptake of azasetron by K562 and multidrug-resistant K562/ADM cells. To determine whether azasetron is a substrate of P-glycoprotein, its accumulation by human leukemic drug-sensitive K562 cells and their multidrug-resistant variant K562/ADM cells was examined. The uptake of [¹⁴C]azasetron was measured at the steady state (50 min) to evaluate the efflux process, and the result is shown in table 3. The uptake of [¹⁴C]azasetron by K562/ADM cells was significantly less than that by sensitive K562 cells. The effect of a multidrug-resistance reversing agent, cyclosporin A, and metabolic inhibitors on the uptake by K562 and K562/ADM cells was also examined. When [¹⁴C]azasetron uptake was measured in the ATP-depleted condition, the uptake by K562/ADM cells was significantly increased, whereas that by K562 cells was unchanged. In the presence of cyclosporin A, an increase of [¹⁴C]azasetron uptake by K562/ADM cells was observed, with even greater accumulation of [¹⁴C]azasetron at a high concentration (10 µM) than at a lower concentration (1 µM) of cyclosporin A, whereas such an increase was not observed in K562 cells.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

The concept that carrier-mediated intestinal absorption and luminal secretion mechanisms, as well as intestinal tissue metabolic activity, regulate the bioavailability of various drugs has recently been established (Tamai and Tsuji, 1996; Tsuji and Tamai, 1996; Benet et al., 1996; Saitoh et al., 1996). Such saturable physiological mechanisms may sometimes produce nonlinear pharmacokinetic phenomena. The results obtained in the present study represent the first evidence to our knowledge that a complex pattern of nonlinear bioavailability is indeed generated by the interaction of intestinal absorptive and secretory transport systems, when the drug concentration in the intestinal luminal fluid is outside the range where these systems operate linearly.

Intestinal absorption of azasetron assessed by the Doluisio method in rats exhibited distinctive nonlinearity, which showed a gradual increase of the apparent first-order absorption rate constant at drug concentrations up to 1 mM, followed by a significant and sharp increase up to 10 mM and then a decrease at higher concentrations (fig. 1). Because the disappearance of intact azasetron from the intestinal luminal fluid was measured in this experiment, the observations could be accounted for by saturable secretory and absorptive transport mechanisms, but not by saturation of metabolism or by a solubility limitation. Absorptive-directed (mucosal-to-serosal) flux of azasetron across ileal tissue preparations mounted on an Ussing-type chamber increased with increasing concentration of azasetron up to 10 mM (fig. 2). This result is consistent with that obtained above by the Doluisio method, and both can apparently be explained by the participation of a saturable secretory mechanism. This hypothesis is further supported by the measurement of serosal-to-mucosal flux by the Ussing-type chamber method, which showed a marked decrease in permeation coefficient with increase of azasetron concentration (fig. 2). The break point observed in figures 1 and 2 was between 1 and 5 mM in both experiments, although the Ussing-type chamber method exhibited an increase of the flux at a lower concentration (less than 1 mM) than was observed in the Doluisio method (more than 1 mM). This small difference may be ascribed to the difference in the thickness of the unstirred water layer in these experimental systems, because the Doluisio method is expected to show a thicker unstirred water layer than that in the isolated ileal sheet chamber method.

The decrease of absorption rate constant at higher concentrations observed with the Doluisio method can be explained by the participation of a saturable absorptive transport mechanism. The observation of a decreased mucosal-to-serosal flux in the Ussing chamber method at higher concentration than 20 mM also supports the involvement of a saturable uptake mechanism. To confirm the results above, we used an in vitro method with cultured cells, because it was easier to interpret the underlying events. Caco-2 cells derived from human adenocarcinoma have been shown to form an intestinal epithelial-like monolayer (Hilgers et al., 1990) and to have many of the small-intestinal membrane transport activities for amino acids (Hu and Borchardt, 1992), peptides (Dantzig and Bergin 1990; Inui et al., 1992), hexoses (Riley et al., 1991), monocarboxylic acids (Tsuji et al., 1994; Takanaga et al., 1994; Ogihara et al., 1996) and P-glycoprotein (Hsing et al., 1992; Hunter et al., 1993; Terao et al., 1996). It should be relatively easy to identify specialized transporters in these cells, so they were used in the present study.

The vectorial transport of [¹⁴C]azasetron across Caco-2 cell monolayers was observed at a concentration of 2 mM, with a higher permeation in the basolateral-to-apical direction than in the reverse direction (fig. 3). Accordingly, mechanisms may operate in Caco-2 cells similar to those observed in rat intestinal tissue. Here, apical-to-basolateral flux (2.3 µl/min/mg protein) was significantly higher than that of the paracellular transport marker [³H]mannitol (0.09) and similar to that of compounds that have been shown to be transported across Caco-2 cells by a carrier-mediated transport mechanism (Tsuji et al., 1994, Takanaga et al., 1994; Ogihara et al., 1996). This relatively high permeation despite the hydrophilic character of the drug and the possible contribution of a secretory transport system may be ascribed to the presence of an efficient absorptive transporter, as discussed below. Furthermore, concentration-dependent transport of azasetron in both directions was observed (table 1). Apical-to-basolateral transport was increased when azasetron concentration was increased from trace (2 µM) to 0.5 and 5 mM, whereas at the higher concentration of 50 mM its permeation was decreased to 41% of that at a trace concentration. The basolateral-to-apical transport changed in a manner completely opposite to the reversely directed transport discussed above, with a decrease and increase at lower and higher concentrations of azasetron, respectively. These concentration-dependent changes are consistent with those observed in the rat intestinal tissues.

The initial uptake of azasetron at 1 min by Caco-2 cells was saturable, as shown in fig. 5, and only a single saturable component was included as judged from Eadie-Hofstee plot analysis, with a K_m value of 15 mM (inset, fig. 5). The K_m value rationally explains the decrease of concentration-dependent apical-to-basolateral flux (table 1). The uptake of [¹⁴C]azasetron by Caco-2 cells at 1 min was reduced in the presence of several cationic compounds, including imipramine, desipramine, serotonin and ondansetron (table 2, fig. 6). Serotonin is taken up into several types of cells by a specific serotonin transporter (Balkovetz et al., 1989; Hoffman et al., 1991) and a specialized transport mechanism may also exist in the intestine (Takayanagi et al., 1995). Indeed, we observed sodium-dependent serotonin transport in Caco-2 cells which is similar to those observed in other tissues such as neuronal cells and placenta. Serotonin and relatively low-affinity inhibitors of the serotonin transporter, imipramine and desipramine (Hoffman et al., 1991), reduced the uptake of [¹⁴C]azasetron by Caco-2 cells, whereas clomipramine, a high-affinity inhibitor of the serotonin transporter, was not inhibitory. The concentration-dependent and selective effects by the several cationic compounds suggest the specificity of the inhibitory effects observed in the present study, although nonspecific effect of these compounds on the apparent uptake of azasetron cannot be denied. Ondansetron also reduced initial uptake of [¹⁴C]azasetron. Serotonin transport is characterized by an absolute sodium and chloride ion dependence (Balkovetz et al., 1989; Hoffman et al., 1991). However, no sodium ion dependence was observed in azasetron transport. Accordingly, azasetron is thought to be taken up in the intestine by a transporter which exhibits distinct properties from those of the known serotonin transporter, even though serotonin competes with the transport of azasetron.

P-Glycoprotein, an ATP-dependent drug efflux pump with a broad substrate specificity, has been demonstrated to have a significant role as an intestinal secretory transporter (Hsing et al., 1992; Hunter et al., 1993; Terao et al., 1996) and ondansetron, which has the same pharmacological effect with azasetron, is transported by P-glycoprotein encoded by mouse mdr1a and human MDR1 genes (Schinkel et al., 1996). Accordingly, the nonlinear increase of absorptive transport of azasetron observed in the present study can be interpreted in terms of a contribution of P-glycoprotein. Several results in the present study support this idea: (1) The serosal-to-mucosal flux of [¹⁴C]azasetron in rat intestinal sheets mounted on an Ussing-type chamber was suppressed, but the mucosal-to-serosal flux was increased by a high-affinity P-glycoprotein inhibitor, cyclosporin A. (2) Apical-to-basolateral transport of [¹⁴C]azasetron across Caco-2 cells was significantly increased in the presence of cyclosporin A, whereas basolateral-to-apical transport was reduced (table 1). (3) Initial uptake of [¹⁴C]azasetron by Caco-2 cells was not affected by cyclosporin A or by ATP depletion induced by metabolic inhibitors, whereas the steady-state uptake was enhanced by inhibiting P-glycoprotein-mediated efflux with cyclosporin A and by ATP depletion (fig. 4). (4) Although ondansetron inhibited the initial uptake of [¹⁴C]azasetron by Caco-2 cells, it increased the steady-state uptake of [¹⁴C]azasetron (fig. 6). Furthermore, (5) steady-state uptake of [¹⁴C]azasetron by multidrug-resistant K562/ADM cells that overexpress P-glycoprotein was significantly less than that by their parental drug-sensitive K562 cells and was also increased in the presence of cyclosporin A and by ATP depletion of the cells (table 3). The results obtained by use of cancer cells, K562 and K562/ADM, directly demonstrate that azasetron is transported by P-glycoprotein. Observations (1) to (4) also strongly suggest that P-glycoprotein makes a predominant contribution to the nonlinear increase of bioavailability of azasetron after oral administration.

In conclusion, 5-HT₃ receptor antagonist, azasetron, and presumably also ondansetron, showed nonlinear intestinal absorption with an increase and a subsequent decrease in absorption at the lower and higher concentration ranges, respectively. Such nonlinearity can be explained by the operation of both absorptive and secretory transporters in the intestine. P-Glycoprotein made a substantial contribution in the secretory transport, whereas the absorptive transporter has not been identified. Clarification of the mechanistic and kinetic features of nonlinear intestinal absorption is important for rational development of orally active drugs, as well as in clinical applications.