Abstract
Intestinal drug efflux mediated by P-glycoprotein and other ABC transporters is widely accepted as a reason for low or variable oral absorption. However, little is known about species and regional differences in P-glycoprotein so the functional and predictive relevance of observations made in cell models such as Caco-2 is uncertain. The aim of this study was to define the kinetics of drug efflux in rat and human intestinal tissues in vitro using the “reference” substrates digoxin and vinblastine. The expression and functional role of other ABC transporters in the transport of these compounds was also investigated. Saturable, verapamil-sensitive efflux of digoxin was observed in all intestinal regions. Apparent affinity of the efflux process varied within a relatively narrow range (50–92 μM), increasing in rat from small to large intestine. In contrast, maximal transporter activity varied over a 4- to 5-fold range with ileum > jejunum > colon. Similar regional differences in efflux were also observed with vinblastine. Maximal efflux levels were similar in Caco-2 and ileum for both substrates, suggesting that Caco-2 may quantitatively predict small intestinal drug efflux. Digoxin efflux kinetics was virtually identical in rat and human colon. Inhibitor studies showed that digoxin and vinblastine efflux in intestinal tissues was mediated by P-glycoprotein, although a minor component could be attributed to multidrug resistance-related protein (MRP)-like transporters in Caco-2. This study has analyzed the differential functional expression of drug efflux along the gastrointestinal tract. Such data will be critical in developing predictive models of P-glycoprotein-mediated efflux using information gathered from in vitro systems.
Drug efflux mediated by broad-specificity xenobiotic transporters present in the intestinal epithelium may be an important factor in the poor or variable absorption of orally administered drugs. Although P-glycoprotein, a member of the ATP-binding cassette family, has been the most studied (for review, see Hunter and Hirst, 1997), there is increasing evidence that the intestine expresses several other members of this family, including multidrug resistance-related protein (MRP1) and the canalicular organic anion transporter (cMOAT/MRP2). The contribution that non-P-glycoprotein transporters make to limiting oral drug absorption has still to be determined but there is little doubt that they have the potential to transport a wide range of drug molecules (Evers et al., 1996, 1998; Paul et al., 1996; Twentyman and Versantvoort, 1996). Quite apart from their role in drug transport, efflux transporters, acting in concert with detoxifying enzyme systems (e.g., cytochrome P450), undoubtedly form a major component of the natural cellular defense against entry of potentially harmful xenobiotics from the gut lumen (Watkins, 1997; Kim et al., 1999).
Much of the information on drug interaction with efflux transporters has come from the use of cultured epithelial models such as Caco-2 (Hunter et al., 1993a,b; Cavet et al., 1996; Collett et al., 1996). Confluent monolayers of Caco-2 cells closely resemble the intestinal mucosa, are a well established model of efflux processes, and are often used to screen drug candidates for anomalous uptake that might be linked to such processes (Hunter and Hirst, 1993b; Gan and Thakker, 1997; Doppenschmitt et al., 1999). However, relatively little is known about how observations made in cell models relate to efflux in intestinal tissues. Tumor-derived cells such as Caco-2 often overexpress efflux transporters and may also show an altered pattern of expression of different ABC proteins (Van Hille et al., 1996;Doppenschmitt et al., 1999). Studies in intestinal tissues suggest regional differences in drug efflux (Saitoh and Aungst, 1995; Makhey et al., 1998) but there is a paucity of information on the kinetic properties of efflux transporters at different levels of the intestine. The contribution of non-P-glycoprotein transporters to drug efflux in gut tissues is also unclear.
To address these issues the kinetic indices of drug efflux in Caco-2, rat small and large intestine, and human large intestine in vitro have been compared using two “reference” compounds, the cardiac glycoside digoxin and the anticancer agent vinblastine. Both compounds have been shown to be effluxed in a range of models (Horio et al., 1990; Tanigawara et al., 1992; Ito et al., 1993a; Hunter et al., 1993a,b; Wils et al., 1994; Cavet et al., 1996; Fromm et al., 1999) and have recently been used as the basis for a physiologically based model of P-glycoprotein-mediated drug transport (Ito et al., 1999). Our data suggest that substrate affinities of the efflux pathways in the different models studied are similar but that marked differences in maximal efflux activity exist along the gastrointestinal tract. We also provide evidence for differential functional expression of non-P-glycoprotein transporters between Caco-2 and intestinal tissues.
Experimental Procedures
Materials
(G-[3H])Vinblastine sulfate (470 GBq · mmol−1, 12.7 Ci · mmol−1, radiochemical purity 97.8%) and (G-[3H])digoxin (703 GBq · mmol−1, 19 Ci · mmol−1, radiochemical purity >97%) were purchased from Amersham-Pharmacia Biotech (Amersham, Bucks, UK) and NEN Life Science Products (Hounslow, Middx, UK), respectively. Cell culture reagents were purchased from Gibco BRL Life Technologies Ltd. (Paisley, UK). All other compounds were obtained from Sigma-Aldrich Chemical Co. Ltd. (Poole, Dorset, UK).
Tissues
Rat Tissues.
Intestinal tissues were removed from nonfasting male Sprague-Dawley rats (250–350 g) that had been killed by a blow to the head followed by cervical dislocation. Distal ileum (15–20 cm beginning 3 cm proximal to the ileo-cecal junction) and distal colon (10–12 cm beginning 2 cm proximal to the rectum) were immediately removed, washed in Ringer's solution, and stripped of muscle layers by blunt dissection before mounting in modified Ussing chambers (0.64-cm2 surface area). Tissues were mounted within 20 min of removal from the animal. All procedures involving animals conformed to current UK Home Office legislation.
Human Tissues.
Human distal colon was obtained from patients undergoing surgery for benign or malignant tumors. Immediately after resection, macroscopically normal tissues, at least 5 cm from the tumor margin, were used for drug permeability studies. Tissues were placed immediately in cold, bicarbonate-buffered Ringer's solution containing 146 mM · l−1 Na+, 4.2 mM · l−1 K+, 1.2 mM · l−1 Ca2+, 1.2 mM · l−1 Mg2+, 126 mM · l−1 Cl−, 27 mM · l−1HCO3 −, 1.4 mM · l−1HPO4 −, and 10 mM d-glucose, which had been equilibrated to pH 7.4 by bubbling with 5% CO2, 95% O2. Tissues were transferred to the laboratory where the underlying muscle layers were removed by blunt dissection and the mucosa mounted in Ussing chambers for permeability studies. The time from resection to mounting was no more than 40 min.
Cell Culture
The human colonic adenocarcinoma cell line Caco-2 (passage 100–119) was cultured as previously described (Collett et al., 1996). For studies of transepithelial drug transport cells were seeded at a density of 1 × 105cells/cm2 onto 12-mm polycarbonate culture inserts (1.0-cm2 surface area) suitable for mounting in Ussing chambers (Snapwell, Costar Ltd., Bucks, UK). Culture medium was changed every 3 days and cells were used for transport studies 19 to 27 days post seeding. Monolayer formation was monitored by measurement of transepithelial electrical resistance (RT) using an EVOMeter fitted with chopstick electrodes (World Precision Instruments, Sarasota, FL).
Permeability Studies
Drug transport across intestinal tissues and Caco-2 monolayers was measured by methods similar to those described previously (Collett et al., 1996, 1999). Intestinal mucosa and Caco-2 monolayers were mounted in modified Ussing chambers and bathed on the mucosal and serosal surfaces with 5 ml of either bicarbonate-buffered Ringer's solution, pH 7.4 (mucosa) or Hanks' balanced salt solution containing 20 mM HEPES, pH 7.4 and 1 g · l−1 (5.6 mM) d-glucose (Caco-2). In a limited number of experiments with Caco-2, Hanks' balanced salt solution was substituted with serum-free Dulbecco's modified Eagle's medium to assess the effects of maintaining cells under “cell culture” conditions on substrate permeability. Buffer substitution had no significant effect on vinblastine or digoxin fluxes (data not shown). Spontaneous tissue potential difference, short-circuit current (ISC), and RT were monitored periodically throughout the experiment, otherwise tissues were maintained under open circuit conditions. A 30-min equilibration period was allowed before beginning permeability measurements to allow stabilization of electrical parameters. Mean electrical parameters in stripped mucosa and cell monolayers following equilibration are shown in Table 1. Tissues were excluded in cases where RT values fell by more than 15% from the value measured at the end of the equilibration period. Asymmetric permeability of vinblastine and digoxin was measured following addition of labeled (7.4 kBq · ml−1, 0.2 μCi · ml−1) and unlabeled drug (1–150 μM) to mucosal (apical) or serosal (basolateral) chambers. Drugs were added as stock solutions in dimethyl sulfoxide, giving a final solvent concentration of 0.02 to 0.3%. In the case of digoxin, unlabeled drug was added to an aliquot of transport buffer from the “donor” chamber that was then vortexed for 1 min and returned to the chamber. For both compounds, 1-ml samples were removed from the “receiver” chamber at time zero and after at least four successive 20-min periods and replaced with fresh transport buffer. Samples (100 μl) were also taken from the “donor” chamber for scintillation counting at the beginning of the first period and at the end of the experiment to monitor any changes in “donor” drug concentrations during the experiment. As a further check on the viability of human and rat tissues, the cAMP agonist forskolin (10 μM) was added to the basolateral chamber at the conclusion of the experiment. This causes a sharp and sustained rise in ISC in viable tissues caused by the stimulation of electrogenic Cl−secretion (Warhurst et al., 1996).
Electrical parameters of monolayers and tissues used in this study
Samples were analyzed by liquid scintillation counting and results are expressed as transepithelial flux (J) in nmol · cm−2 · h−1 for either unidirectional (J A-B orJ B-A) or net flux (J net). The mean electrical data for the Caco-2 cells and tissues used in these studies are summarized in Table 1.
Inhibitor Studies
The efflux processes responsible for net secretion of vinblastine and digoxin in Caco-2 was investigated using verapamil (100 μM, applied apically) as a known inhibitor of P-glycoprotein and probenecid and MK571, inhibitors of MRP-dependent processes (applied basolaterally and apically). Unidirectional and net fluxes of vinblastine (15 μM) and digoxin (10 μM) were measured in the presence and absence of verapamil (100 μM) or probenecid (2 mM) and MK571 (20 μM). Comparative inhibitor studies were also performed on digoxin transport in rat ileum.
RT-PCR Screen for Transporter mRNA Expression
Total RNA was prepared from Caco-2 cells, human, and rat tissue using ULTRASPEC reagent according to the manufacturers' instructions (AMS Biotechnologies Ltd., Oxfordshire, UK). Two micrograms of total RNA was reverse transcribed into cDNA using random priming and Moloney murine leukemia virus (MMLV) reverse transcriptase (Life Technologies Ltd., Paisley, Scotland, UK), as per manufacturer's instructions in a 10-μl reaction volume. One microliter of the cDNA was then PCR-amplified on a Hybaid PCRSPRINT PCR block for 35 cycles, except for cMOAT in the rat colon where an apparently rare transcript necessitated an additional 10 cycles. Reaction volumes were 50 μl, comprising: 0.2 mM each of dATP, dCTP, dTTP, and dGTP, 1× PCR buffer (10 mM Tris/HCl pH 8.3, 1.5 mM MgCl2, 50 mM KCl), and 0.5 μM of each primer. The oligonucleotide primers used for RT-PCR screening of intestinal transporters are detailed in Table 2. Multiple restriction digests, followed by confirmation of the correct size restriction fragments by agarose gel electrophoresis, confirmed product identity.
Forward and reverse oligonucleotide primer sequences used for RT-PCR screen of intestinal transporters
Statistical Methods
Except in the case of kinetic parameters, values are expressed as mean ± S.E.M (n). Kinetic and statistical analyses were carried out using PRISM 2.01 (GraphPad Software Inc., San Diego, CA). Kinetic values (half-maximal effective concentration (EC50) and maximal net flux rate (J
max) for substrate dose-effect relationships were calculated by nonlinear regression with the method of least squares, fitting for a logistic sigmoid using the Hill equation:
where J is net flux, C is donor concentration, P is a constant (the Hill slope), and EC50 is the concentration at which half-maximal flux was achieved.
The error values associated with the kinetic parameters are asymptotic standard errors returned by the regression routine and are a measure of the certainty of the best fit value. To take account of this, differences in kinetic parameters returned by the regression routine were compared using unpaired Student's t test but with a higher threshold for significance of 0.01. Statistical comparisons of all other data (i.e., effects of inhibitors on substrate fluxes) were determined using unpaired Student's t test with a significance level of 0.05.
Results
P-glycoprotein-Dependent Digoxin Efflux in Rat Intestinal Tissues.
Digoxin exhibits marked asymmetric permeability across Caco-2 cell monolayers indicative of efflux by P-glycoprotein (1). Initial studies examined the suitability of digoxin to probe efflux pathways in intestinal tissues by investigating its bidirectional permeability across rat small and large intestine (Fig.1). At a concentration of 40 μM, digoxin permeability was strongly asymmetric in the basolateral to apical direction in all intestinal regions indicative of net secretory flux. Efflux ratio (ratio of B-A permeability to A-B permeability) did not vary significantly along the intestine being 4.4 ± 1.0 (4) in jejunum, 4.2 ± 0.5 (3) in ileum, and 3.6 ± 0.9 (4) in colon. This compares with a ratio of 15.4 ± 3.3 (3) for this concentration of digoxin in Caco-2 monolayers.
Comparison of bidirectional digoxin flux across Caco-2 monolayers and stripped rat jejunum, ileum, and colon mucosa. Values are mean flux ± S.E.M. in pmol · h−1 · cm−2. Apical-to-basolateral (A-B) fluxes are denoted by open bars, basolateral-to-apical (B-A) fluxes by filled bars. All A-B fluxes are significantly lower than the corresponding B-A fluxes (p < 0.05). [3H]Digoxin was present at a final concentration of 40 μM (0.2 μCi · ml−1).
An RT-PCR screen of rat and human intestine (Fig.2) showed that, in addition to P-glycoprotein, these tissues express mRNA for MRP-type transporters, which also transport a range of drug molecules in model systems (Paul et al., 1996; Twentyman and Versantvoort, 1996; Evers et al., 1998;Hirohashi et al., 2000). Rat small and large bowel express transcripts for MRP1 and cMOAT/MRP2 in addition to mdr1a, whereas a similar pattern of expression was observed in human tissues with the exception that no signal for cMOAT could be detected in human colon. Transcripts for all three transporters were expressed in the colonic carcinoma line Caco-2.
RT-PCR screen for mRNA expression of P-glycoprotein (MDR1), multidrug resistance-associated protein 1 (MRP-1), and canalicular multispecific organic anion transporter (cMOAT/MRP2) in intestinal epithelia. Gel images represent Caco-2 (A), human ileum (B), human colon (C), rat jejunum (D), rat ileum (E), and rat colon (F). Lanes are as follows: L, molecular size marker ladder with 600-bp band marked (▸); lane 1, P-gp (MDR1 in human or Caco-2, mdr1a in rat); lane 2, MRP1; and lane 3, cMOAT (MRP2). Figure shows ethidium bromide-stained PCR products separated by agarose gel electrophoresis.
In view of these findings it was deemed important to define the role of P-glycoprotein and non-P-glycoprotein processes in digoxin transport in these systems before undertaking a detailed kinetic analysis of digoxin efflux. As shown in Fig. 3, verapamil, a widely used P-glycoprotein inhibitor markedly reduced digoxin efflux in Caco-2 monolayers by 81.5 ± 2.3% (n = 4,p < 0.001) and in rat ileum by a similar degree (n = 4, 73.5 ± 3.0%, p < 0.001). Probenecid and MK571, both inhibitors of MRP-like transporters had differential effects on digoxin transport in Caco-2 and rat tissues. In Caco-2, probenecid (2 mM) and MK571 (20 μM) caused a small (20–25%) but significant (p < 0.05) inhibition of digoxin efflux. Verapamil and probenecid in combination caused an essentially additive inhibitory effect (Fig. 3). In marked contrast to Caco-2, neither probenecid nor MK571 inhibited digoxin transport in rat tissues. These data indicate that digoxin efflux is mediated overwhelmingly by P-glycoprotein in both Caco-2 and rat tissues with MRP-like transporters also playing a small but significant role in digoxin transport in Caco-2 cells.
Effect of the transport inhibitors probenecid (2 mM), verapamil hydrochloride (100 μM), and MK 571 (20 μM) on net efflux of vinblastine (15 μM) and digoxin (10 μM) across Caco-2 monolayers and net efflux of digoxin across stripped rat ileum mucosa. Data is represented as percentage of flux in the absence of inhibitors (i.e., control = 100%). C, control; P, probenecid; V, verapamil; PV, probenecid followed by verapamil; and M, MK571. Verapamil was added to the apical surface of the monolayers, whereas probenecid and MK571 were added to both apical and basolateral surfaces. Data are mean ± S.E.M. from four experiments in each group. *p < 0.05, † p < 0.001.
Kinetic Analysis of Digoxin Efflux in Caco-2 and Rat Tissues.
The kinetic parameters for efflux in Caco-2 and rat tissues were determined by measuring digoxin transport over a range of concentrations. Figure 4 shows concentration curves for Caco-2 (Fig. 4A) and rat ileum (Fig. 4B) for unidirectional and net fluxes of digoxin. In both cases, the basolateral to apical (J B-A) and net (J net) flux of digoxin was saturable over the range 0 to 150 μM indicative of P-glycoprotein-mediated efflux. Figure 5 compares the concentration-dependence of J net in different regions of the rat intestine. Kinetic analysis ofJ net showed that EC50 values for digoxin efflux tend to decrease down the bowel (jejunum, 81 μM; ileum, 74 μM; colon, 51 μM), suggesting a higher affinity for digoxin efflux in large intestine compared with small intestine, although the differences did not reach statistical significance (Table3). The EC50 value for the colon-derived cell line Caco-2 was virtually identical to rat colon (Table 3).
Concentration dependence of unidirectional and net fluxes of digoxin in Caco-2 monolayers (A) and stripped rat ileum mucosa (B). Basolateral-to-apical flux (●), apical-to-basolateral flux (▾), and net flux (○) are mean ± S.E.M. from three to five experiments at each concentration.
Comparison of concentration dependence of net flux (J net) of digoxin across Caco-2 monolayers (●), stripped rat jejunum (♦), ileum (▪), and colon (▾) mucosa. Values are mean ± S.E.M. from three to six experiments at each concentration.
Summary of apparent kinetic parameters for net basolateral to apical flux of vinblastine sulfate and digoxin in Caco-2 cells and rat and human intestine, mounted in Ussing chambers
Marked differences in J max values (an indicator of maximum transporter activity in each tissue) were also observed along the rat intestine (Table 3). Values in the ileum were approximately double those observed in the jejunum (9.3 ± 0.9 versus 4.7 ± 0.5 nmol · h−1· cm−2) and >4-fold higher than colon (2.2 ± 0.3 nmol · h−1 · cm−2). The J maxfor digoxin efflux in Caco-2 cells (13.0 ± 2.9 nmol · h−1 · cm−2) was not significantly different to rat ileum. To determine whether the differences in efflux kinetics observed with digoxin hold true for other substrates, comparative studies were performed with another well characterized P-glycoprotein substrate, vinblastine.
Kinetic Analysis of Vinblastine Efflux.
The overall pattern of vinblastine efflux in these systems was broadly similar to that observed for digoxin (Fig. 6). EC50 values for vinblastine were lower in Caco-2 than rat ileum (26.5 ± 4.1 versus 48.2 ± 4.3 μM).J max values for vinblastine were ∼35% higher in Caco-2 than rat ileum (5.7 ± 0.4 versus 3.6 ± 0.4 nmol · h−1 · cm−2). It was not possible to derive meaningful kinetic parameters for vinblastine in rat colon due to an inability to saturate at concentrations up to 100 μM and higher concentrations presented significant solubility problems. Nevertheless, efflux values at 100 μM vinblastine in colon (0.9 ± 0.2 nmol · h−1 · cm−2) were approximately 3-fold lower than those observed in ileum (3.4 ± 0.4 nmol · h−1 · cm−2), suggesting that differences in efflux activity between large and small bowel observed with digoxin also hold for vinblastine.
Comparison of concentration dependence of net flux (J net) of vinblastine across Caco-2 monolayers (●), stripped rat ileum (▪), and colon (▾) mucosa. Values are mean ± S.E.M. from three to six experiments at each concentration.
Comparison of Digoxin Efflux in Rat and Human Colon.
Interspecies differences in transporter characteristics are an important concern when assessing the relevance of observations made in animal models to human transport. To address this aspect preliminary studies were undertaken comparing digoxin transport in mucosal preparations from rat and human colon. As shown in Fig.7, the concentration dependence for digoxin efflux in human tissues was virtually identical to that of the rat colon with no significant difference in the calculated kinetic parameters between the two species (EC50, 59 ± 15 μM and J max, 3 ± 0.8 nmol · h−1 · cm−2 compared with EC50, 51 ± 8 μM and J max, 2.2 ± 0.3 nmol · h−1 · cm−2 in rat). Table 3 summarizes the kinetic data for digoxin and vinblastine in the intestinal systems investigated in this study.
Comparison of concentration dependence of net flux (J net) of digoxin across stripped human (●, solid line) and rat (○, broken line) colon mucosa. Values are mean ± S.E.M. from three to four experiments at each concentration.
Discussion
There is currently intense interest in intestinal drug efflux as a possible reason for the poor or variable absorption of drugs administered via the oral route (Artursson and Karlsson, 1991; Hunter and Hirst, 1997). In this regard, P-glycoprotein-expressing cell lines such as Caco-2 and Madin-Darby canine kidney are being used extensively as models of the intestinal epithelium, to screen drug candidates for likely efflux problems (Horio et al., 1989, 1990; Hunter et al., 1993;Collett et al., 1996; Gan and Thakker, 1997; Yee, 1997; Doppenschmitt et al., 1999). The relevance of data obtained from these models to tissue and in vivo systems is difficult to assess because of a lack of quantitative functional data on the characteristics of drug efflux in intestinal tissues. In this study we have provided in vitro kinetic data for the efflux of digoxin and vinblastine in different regions of rat intestine and compared these with the efflux characteristics of both drugs in Caco-2. Comparative data on the kinetics of digoxin efflux in rat and human colon are also provided.
Digoxin was chosen as the primary substrate for these studies because it is poorly metabolized in intestinal tissues (Fromm et al., 1999) and there is also a wealth of data demonstrating its interaction with P-glycoprotein in vitro and in vivo (De Lannoy and Silverman, 1992;Cavet et al., 1996; Fromm et al., 1999). The data presented here represent the first attempt to provide a kinetic profile of digoxin across different intestinal systems.
In all systems studied here digoxin showed concentration-dependent permeability in the basolateral-to-apical direction, consistent with transport by a saturable efflux mechanism. The EC50 for digoxin efflux was 40% lower in colon (51 μM) than jejunum (81 μM), suggestive of a modest increase in transporter affinity for digoxin in large intestine. This is borne out by data from human colon and the colon-derived Caco-2 line, both of which gave EC50 values similar to rat colon (58 μM for both human colon and Caco-2). However, given that these are relatively modest differences, it is important to consider that factors other than transporter affinity, including membrane binding and intracellular sequestration, may affect apparent EC50 in these systems. A recent study in rat intestine showed 3- to 7-fold variation in theK m for basolateral-to-apical permeability of both vinblastine and verapamil, suggesting major regional differences in transporter affinity (Makhey et al., 1998). One possible explanation for this discrepancy is that, in contrast to digoxin, both vinblastine and verapamil are subject to extensive metabolism by CYP3A4, the activity of which varies along the gut (Watkins, 1997). Saitoh and Aungst (1995) also found evidence that the ability of P-gp to transport verapamil may vary in different regions of the intestine, whereas Artursson and Karlsson (1991) described regional differences in passive permeability based largely on physicochemical properties of the compounds. However, in our hands the EC50 for vinblastine in rat ileum was ∼40% higher than in Caco-2, which mirrors the difference seen with digoxin in these two systems. These observations highlight the usefulness of digoxin as a stable and reliable probe for efflux that is applicable across different systems.
By far the biggest differences in efflux kinetics in rat intestine were in J max (indicative of the maximal level of efflux). Ileal tissues showed a 4- to 5-fold greater efflux than the colon with the jejunum having an intermediate activity. These findings are consistent with earlier molecular data from this laboratory showing much higher levels of P-gp protein expression in villus epithelial cells isolated from rat ileum compared with rat colonocytes (Collett et al., 1999). Maximal digoxin efflux in Caco-2 was similar to rat ileum (13.0 ± 3.0 versus 9.3 ± 1.0), suggesting that this model may be a reasonable predictor of efflux kinetics in the small intestine. However, Caco-2 would clearly overestimate efflux in both rat and human colon by 5- to 6-fold. This is interesting given that these cells are derived from colonic epithelium. One explanation of this is that in common with many tumor-derived cell lines, Caco-2 may overexpress P-gp in relation to its parent tissue (Van Hille et al., 1996). The virtually identical kinetics of digoxin efflux in rat and human colon would suggest that the rat might well be a quantitatively predictive model of human intestinal efflux although further studies are needed in human small intestine. A preliminary attempt to characterize digoxin efflux in human ileum was unsuccessful due to difficulties in maintaining acceptable viability in these tissues (data not shown).
The broadly similar efflux kinetics observed for digoxin and vinblastine in Caco-2 and rat ileum is interesting given their different physicochemical properties (log D: vinblastine, 3.70; digoxin, 1.70; Terao et al., 1996). Vinblastine is recognized as having a high affinity for P-gp-based on radioligand binding and calcein accumulation (Doppenschmitt et al., 1998, Holló et al., 1996). In contrast, preliminary calcein accumulation studies in Caco-2 indicate an EC50 for digoxin that is an order of magnitude greater than vinblastine, suggesting a significantly lower affinity for P-gp (data not shown). In addition uptake of [3H]vinblastine into Madin-Darby canine kidney cells has been reported to be 8-fold greater than [3H]digoxin (Ito et al., 1999). However, in common with the present study these same workers noted the similarity in kinetics of P-glycoprotein-mediated efflux of digoxin and vinblastine (Ito et al., 1993a,b, 1999). Several other groups (Cavet et al., 1996; Woodland et al., 1997) have observed a greater efflux of digoxin than vinblastine in both kidney and intestinal cell lines. Taken together, such observations suggest that in vitro measures of P-glycoprotein affinity are not direct indicators of how P-gp will modulate permeability of the drug and that other factors, including passive membrane permeability, morphology, and gut wall metabolism will need to be taken into account (Yamashita et al., 1997; Doppenschmitt et al., 1999).
RT-PCR analysis showing expression of MRP1 and MRP2/cMOAT in addition to P-glycoprotein suggests that drugs have access to a variety of polyspecific transporters capable of drug efflux in the intestine. MRP1 and cMOAT appear to be important in the excretion of glutathione (GSH) or glucuronide-conjugated compounds. MRP seems to absolutely require GSH for transport activity since it is stimulated by GSH (Deeley and Cole, 1997) and inhibited by the GSH-depleting agent buthionine sulfoximine (Versantvoort et al., 1995; Rappa et al., 1997). Sugiyama and coworkers have shown that GSH and glucuronide conjugate transport occurs in both Caco-2 cells and rat intestine (Gotoh et al., 2000; Hirohashi et al., 2000). Unconjugated compounds such as vinblastine, vincristine, and etoposide can also be transported, in the presence of GSH, probably by drug/GSH cotransport (Paul et al., 1996;Twentyman and Versantvoort, 1996; Deeley and Cole, 1997; Rappa et al., 1997; Evers et al., 1998). In the present study, ∼25% of vinblastine and digoxin efflux in Caco-2 could be inhibited by agents with selectivity for MRP-like transporters (probenecid and MK571). Although cross-reactivity of these inhibitors with P-gp cannot be ruled out such data imply that drug efflux in Caco-2 may be a multifactorial process involving both P-gp and non-P-gp-mediated efflux systems. There was no clear evidence of this in intestine with MRP inhibitors having no effect on digoxin efflux in rat ileum. However, given the evidence that MRP2/cMOAT expression decreases from proximal-to-distal small intestine (Gotoh et al., 2000) there is clearly a need for further studies.
In conclusion, the data presented here represent the first systematic and comparative analysis of drug efflux in intestinal tissues and the Caco-2 cell line. Using digoxin as a stable and reproducible efflux probe, quantitative data are provided showing the variation in functional affinity and maximal efflux activity along the gastrointestinal tract. A similar profile was also observed using a second efflux substrate, vinblastine. Such data will be useful in developing predictive models of P-glycoprotein-mediated efflux along the intestine using information gathered from cell systems such as Caco-2.
Footnotes
- Received July 26, 2000.
- Accepted October 27, 2000.
-
Send reprint requests to: Dr. R. H. Stephens, Section of Gastroenterology, Clinical Division I, University of Manchester, Hope Hospital, Eccles Old Rd., Salford M6 8HD, UK. E-mail:rstephen{at}fs1.ho.man.ac.uk
-
This work was supported by Pfizer Central Research, Sandwich, UK.
Abbreviations
- MRP
- multidrug resistance-related protein
- RT
- transepithelial electrical resistance
- cMOAT
- canalicular multispecific organic anion transporter
- ISC
- short-circuit current
- JA-B
- apical-to-basolateral flux
- JB-A
- basolateral-to-apical flux
- Jnet
- net flux
- RT-PCR
- reverse transcription-polymerase chain reaction
- Jmax
- maximal net flux
- P-gp
- P-glycoprotein
- GSH
- glutathione
- The American Society for Pharmacology and Experimental Therapeutics
