The epidermal growth factor receptor family has become a widely studied pharmacological target in the therapeutic area of cancer due to its central role in the transmission of mitogenic signals, and the frequent deregulation in cancers (Fabbro et al., 2002). PKI166, a pyrrolo-pyrimidine derivative (Fig. 1) (Traxler et al., 1999), is a selective inhibitor of the tyrosine kinase activity of two epidermal growth factor receptors (HER1/HER2). It has a potent in vitro antiproliferative activity and in vivo antitumor activity in a mouse model (Bruns et al., 2000; Solorzano et al., 2001; Baker et al., 2002; Motoyama et al., 2002). Its clinical development was hampered by liver toxicity findings (Traxler, 2003). Absorption, distribution, metabolism, elimination studies performed in rats and dogs demonstrated that PKI166 was almost exclusively excreted in feces (∼90% after peroral or intravenous dosing, >50% thereof being unchanged compound), with only about 3 and 5% of the dose excreted in urine of dog and rat, respectively. Bile analysis of rat indicated extensive direct O-glucuronidation. Glucuronidates apparently were then subject to colonic bacterial deconjugation before excretion or reabsorption of PKI166, the latter leading to enterohepatic recirculation (Novartis, nonpublished data). In clinical studies, a high plasma exposure to the main metabolite ACU154, an O-glucuronide (Fig. 1), was observed; the exposure to this metabolite was often higher than the plasma exposure to the parent compound. Against this background, hepatocyte trafficking of PKI166 and its metabolites must be regarded as critical for the overall disposition of this anticancer compound. These transport processes must be considered in the evaluation of the liver toxicity of PKI166.

Here we analyze the transport of PKI166 and its main metabolite ACU154 (Fig. 1) using Madin-Darby canine kidney II (MDCKII) cell monolayers and liver canalicular membrane vesicles (CMVs). The MDCKII cells used expressed one or two transporter proteins heterologously: the human multidrug resistance-associated protein 2 (MRP2) in their apical membrane and/or the human organic anion-transporting peptide 2 (OATP2) in their basal membrane (Evers et al., 1998; Sasaki et al., 2002), as it is found in human hepatocytes. CMVs were prepared from normal Wistar and TR^- rats, a mutant strain with a hereditary defect in mrp2 (Jansen et al., 1985). [³H]EG and [³H]TCA were used as model substrates for mrp2/MRP2/OATP2 and bsep, respectively. The aim was to understand the transport processes contributing to the hepatic elimination of PKI166 and its metabolite, ACU154, on a molecular level and to characterize the interaction of these two compounds with some pivotal transport proteins in the liver.

Materials and Methods

Materials. [³H]Taurocholic acid (74 GBq/mmol) and [³H]estradiol-17β-d-glucuronide (1499 GBq/mmol) were obtained from PerkinElmer Life and Analytical Sciences (Boston, MA). [¹⁴C]PKI166 (6.6 MBq/mg) and its glucuronide [¹⁴C]ACU154 (3.2 MBq/mg) were synthesized in the Novartis Isotope Laboratories, and unlabeled PKI166 and ACU154 were synthesized in Novartis Laboratories (Basel, Switzerland).

Cell Culture and Transport Studies. Mock-transfected MDCKII cells, or MDCKII cells expressing MRP2 (Evers et al., 1998), OATP2, or both transporters (Sasaki et al., 2002) were cultured in Dulbecco's modified Eagle's medium with 10% fetal bovine serum at 37°C under 5% CO₂ (Sasaki et al., 2002). Cells expressing OATP2 were cultured in the presence of 200 μg/ml Zeocin; cells expressing MRP2 were cultured in the presence of 200 μg/ml G418. Transport studies were performed as described (Sasaki et al., 2002): the cells were seeded in 24-well plates (Falcon; BD Biosciences Discovery Labware, Bedford, MA) at a density of 1.4 · 10⁵ cells per well; 10 mM sodium butyrate was included in the medium 24 h before the experiment. For initiation of transport, the medium at either the apical or basal side of the cell layer was exchanged with complete medium containing [¹⁴C]PKI166 or [¹⁴C]ACU154 (2 μM). The cells were incubated at 37°C, and aliquots of the medium were taken from each compartment at several points in time. Radioactivity in 100 μl of medium was measured in a liquid scintillation counter (TriCarb 2500 TR and 2700 TR Liquid Scintillation Systems; PerkinElmer Life and Analytical Sciences) after addition of a scintillation fluid (Irga-Safe Plus; PerkinElmer Life and Analytical Sciences). For the uptake studies, complete medium containing [¹⁴C]ACU154 (2 μM) was added at the basal side. After incubation at 37°C for 10 or 30 min, the cells were washed three times with 1.5 ml of ice-cold Krebs-Henseleit buffer and solubilized in 450 μl of 1 N NaOH. Then, 500 μl of distilled water were added before taking samples for liquid scintillation counting. The protein concentrations in cell experiments were determined by the method of Lowry et al. (1951) using bovine serum albumin as a standard.

Preparation of CMVs and Uptake Studies. CMVs were prepared from Wistar and TR^- rats (Jansen et al., 1985) as described (Niinuma et al., 1997). Uptake of [¹⁴C]ACU154, [³H]EG, or [³H]TCA was measured in buffer (250 mM sucrose, 10 mM Tris-HCl, 10 mM MgCl₂, pH 7.4) containing 10 mM creatine phosphate, 100 μg/ml creatine phosphokinase, and a 5 mM concentration of either ATP or AMP (final concentrations after addition of CMVs). After preincubation at 37°C, the reactions were started by addition of CMVs (0.15-0.5 mg/ml protein) and stopped by addition of a 50-fold excess of ice-cold stop buffer (250 mM sucrose, 10 mM Tris-HCl, 100 mM NaCl, pH 7.4). Separation of vesicle-bound and -free labeled compound was achieved by rapid filtration and washing on a sampling manifold (model 1225 with 0.45-μm filter type HA; Millipore Corporation, Bedford, MA) as described (Suzuki et al., 2003). Protein concentrations were determined according to the method of Bradford (1976) with bovine serum albumin as a standard.

Data Analysis. The kinetic parameters for transcellular transport of [¹⁴C]ACU154 were estimated from the Michaelis-Menten equation: v₀ = (V_max × S)(K_m + S), where v₀ is the initial uptake rate of substrates (pmol/min/mg of protein), S is the substrate concentration in medium (μM), K_m is the Michaelis constant (μM), and V_max is the maximum uptake rate (pmol/min/mg of protein); the data points determined over 2 h of incubation were used. The kinetic parameters for the inhibition of transcellular [³H]EG transport by PKI166 were estimated from the following equation: v́/v = 1(1 + I/K_i), where v́ and v are the transport in the presence and absence of inhibitor, respectively; I is the inhibitor concentration (μM) and K_i the inhibition constant (μM); the data points determined over 0.5 h of incubation were used. The kinetic parameters for uptake of [¹⁴C]ACU154 into CMVs were estimated from the Michaelis-Menten equation including the nonsaturable transporter clearance P_diff (μl/min/mg of protein): v₀ = (V_max × S)/(K_m + S) + P_diff× S. Data were fitted to the mentioned equations by a nonlinear least-squares method with the MULTI program (Yamaoka et al., 1981) to obtain estimates of the kinetic parameters. The input data were weighted as the reciprocals of the squares of the observed values.

Results

Transcellular Transport across MDCKII Monolayers. The transport of [¹⁴C]PKI166 across MDCKII monolayers expressing OATP2, MRP2, or both transporters was compared with the transport across control transfected MDCKII cells. In control cells the basal-to-apical flux of [¹⁴C]PKI166 was about 2-fold higher than the flux into the basal direction; no major changes were evident upon expression of OATP2, MRP2 or both transporters (Fig. 2). The more hydrophilic glucuronide [¹⁴C]ACU154 did permeate the monolayer of control transfected MDCKII cells much more slowly than did PKI166 itself (Figs. 2A and 3A). In the presence of both OATP2 and MRP2, the vectorial basal-to-apical flux did increase strongly and was about 5-fold higher than the apical-to-basal flux (Fig. 3D).

PKI166 inhibited the transcellular transport of [³H]EG across OATP2/MRP2 double-transfected cells in a dose-dependent way (Fig. 4). A K_i of 17 ± 2 μM was determined by nonlinear least-squares fitting. The transcellular basal-to-apical transport of [¹⁴C]ACU154 by double-transfected cells was saturable (Fig. 5), with a K_m value of 42 ± 3 μM and a V_max of 84 ± 3 pmol/min/mg protein.

Uptake into MDCKII Cells Expressing OATP2. The uptake of [¹⁴C]ACU154 into OATP2 expressing MDCKII cells was compared with the uptake into control transfected MDCKII cells. In a 30-min incubation containing 2 μM [¹⁴C]ACU154, the clearance was 8.6 ± 0.8 μl/mg protein in control cells and 11.7 ± 1.0 μl/mg protein in OATP2-expressing cells. Assays with [¹⁴C]PKI166 had a very high background due to either uptake of [¹⁴C]PKI166 or binding to the cells. No difference between control cells and OATP2-expressing cells was apparent in incubations containing 2 μM [¹⁴C]PKI166. However, the OATP2-mediated uptake of [³H]EG into MDCKII cells was reduced in the presence of 10 and 100 μM PKI166 (data not shown).

Uptake into Rat Canalicular Membrane Vesicles. To quantify the ATP-dependent transport of [¹⁴C]ACU154 and [¹⁴C]PKI166, uptake into CMVs was analyzed in the presence of either ATP or AMP. CMVs prepared from Wistar and mrp2-deficient TR^- rats were used, allowing us to identify the contribution of mrp2 to the active transport. For [¹⁴C]ACU154, an increased uptake in the presence of ATP as compared with AMP was evident into CMVs prepared from Wistar but not from TR^- rats (Fig. 6). The concentration dependence of the [¹⁴C]ACU154 uptake into CMVs prepared from Wistar rats could be best explained assuming a saturable and a nonsaturable component of uptake. A K_m of approximately 1 μM and a V_max of 95 ± 24 pmol/min/mg of protein for the saturable component and a nonsaturable transport clearance of 1.5 ± 0.1 μl/min/mg of protein were determined by nonlinear least squares fitting (Fig. 7; quantification of [¹⁴C]ACU154 was difficult at concentrations <1 μM, so that the K_m could not be defined accurately). The uptake of [¹⁴C]PKI166 into CMVs could not be measured. Different experimental conditions were tested (data not shown); however, since the background binding of [¹⁴C]PKI166 to vesicles and/or filters was high, any possibly present minor ATP-dependent uptake could not be measured.

Influence of ACU154 and PKI166 on mrp2- and bsep-Mediated Transport into CMVs. The influence of ACU154 and PKI166 on the uptake of model compounds into CMVs was assessed, using [³H]EG as a substrate for mrp2 and [³H]TCA as a substrate for bsep. ATP- dependent uptake of [³H]EG into CMVs from Wistar rats was inhibited by ACU154 (Fig. 9A), whereas the presence of PKI166 increased the uptake (Fig. 8A). No ATP-dependent uptake of [³H]EG into CMVs from TR^- rats was evident, confirming the absence of functional mrp2 in theses vesicles (Figs. 8B and 9B).

The ATP-dependent uptake of [³H]TCA occurred as expected into CMVs from Wistar and TR^- rats. This uptake was clearly inhibited by 1 μM ACU154 in the case of CMVs prepared from Wistar rats containing functional mrp2; however, it was not inhibited in the case of CMVs prepared from TR^- rats (Fig. 9, C and D). A similar effect was evident at 100 μM PKI166, but much less pronounced (Fig. 8, C and D).

Discussion

In preclinical absorption, distribution, metabolism, elimination studies with [¹⁴C]PKI166, the O-glucuronide [¹⁴C]ACU154 was found in high concentrations in rat bile and was detected in mouse but not in rat and dog plasma. In patients, daily doses of 200 mg of PKI166 resulted in maximal plasma concentrations of up to 0.5 μg/ml PKI166; the plasma exposure to ACU154 was in many patients up to 5-fold higher than the exposure to PKI166 and, in some patients, even over 10-fold. Biliary excretion of ACU154 was the main route of PKI166 elimination in rat. In TR^- rats a slower elimination of [¹⁴C]PKI166 from plasma and a shift toward urinary elimination was observed; in these mrp2-deficient rats, the urinary excretion amounted to about 43% compared with only 5% in Wistar rats (Novartis, unpublished results). The in vitro results of the present study help to understand 1) transport processes contributing to the hepatic elimination of PKI166 and 2) interaction of PKI166 and its main metabolite with some pivotal transport proteins in liver.

The analysis of the PKI166 flux through MDCKII cells indicated that PKI166 crossed these kidney-derived cells easily in both directions, with some vectorial transport toward the apical side. In control transfected MDCKII cells, the flux of PKI166 toward the apical side was about 30-fold higher than that of its glucuronide ACU154 (Figs. 2A and 3A); however, the expression of neither OATP2 nor MRP2 did increase this flux (Fig. 2). ACU154, which is formed from PKI166 in liver cells, is more hydrophilic and bulky, explaining its slower flux through MDCKII cells. Thus, ACU154 depends on active transport for efficient excretion into bile. ACU154 was shown to be a substrate of human MRP2 (Fig. 3) as well as rat mrp2 (Fig. 6). This finding confirms the in vivo results obtained in rats, which indicated that mrp2 activity is critical for the hepatic elimination of PKI166. MRP2/mrp2 transports a wide range of glutathione, glucuronate, and sulfate conjugates and is overall the main transporter for excretion of conjugates from hepatocytes into bile (Oude Elferink et al., 1995). In addition ACU154 was shown to be a substrate of OATP2, by measurement of the transcellular transport as well as the uptake into OATP2-transfected MDCKII cells. ACU154 was found at high plasma concentrations in PKI166-treated patients, and its uptake into hepatocytes possibly occurs via OATP2.

To analyze a possible interaction of PKI166 and ACU154 with hepatic transport of other drugs and endogenous substances, EG and TCA were used as well characterized substrates of crucial liver transport proteins. EG is efficiently transported by OATP2 as well as MRP2/mrp2, and TCA is a substrate of bsep (Akita et al., 2001; Sasaki et al., 2002).

PKI166 inhibited the transcellular transport of [³H]EG via double-transfected MDCKII cells as well as the uptake of [³H]EG into OATP2-expressing cells in a concentration-dependent way. This indicates an inhibition of the [³H]EG transport by OATP2. However, the ATP-dependent uptake of [³H]EG into CMVs via mrp2 was enhanced by PKI166 (Fig. 8, A and B). Modulation of the MRP2/mrp2 activity by a variety of different compounds has been shown in vitro (Niinuma et al., 1997; Bodo et al., 2003; Zelcer et al., 2003). Niinuma et al. (1997) described an increased affinity of rat mrp2 for S-(2,4-dinitrophenyl)-glutathione in the presence of sulfate conjugates, whereas the corresponding glucuronides reduced the uptake and decreased the affinity. Bodo et al. (2003) reported that a number of different compounds modulate the activity of both MRP2 and MRP3, reducing transport of EG by MRP3 while stimulating its transport by MRP2. These in vitro findings indicate the potential for a considerable allosteric modulation of the MRP2/mrp2 activity. Zelcer et al. (2003) proposed that MRP2 contains two similar but nonidentical binding sites, one for the transport and one for the modulation of the transport activity (Zelcer et al., 2003). PKI166 was here identified as a stimulator of mrp2 activity toward EG, whereas its glucuronide ACU154 reduced this activity either by direct competition or an allosteric mechanism. In an earlier in vivo study in rat, increased biliary EG excretion was induced by ursodeoxycholate-3,7-disulfate (Sano et al., 1993), indicating that the reported in vitro findings might have in vivo relevance.

ACU154 inhibited the ATP-dependent uptake of [³H]EG and, even more pronouncedly, that of [³H]TCA into CMVs prepared from Wistar rats (Fig. 9, A and C). Comparison with CMVs prepared from mrp2-deficient TR^- rats revealed that the inhibition of [³H]TCA uptake was dependent on the presence of functional mrp2 (Fig. 9). The latter finding suggests that TCA transport by bsep is not inhibited by cytoplasmic ACU154 but depends on the presence of mrp2, which allows ACU154 to get efficiently to the bile side of hepatocytes. Mrp2-dependent inhibition of bsep activity has been described, e.g., for EG in vitro (Stieger et al., 2000) and in vivo (Sano et al., 1993; Huang et al., 2000), and for two sulfate-conjugated bile acids that were identified as mrp2 substrates (Akita et al., 2001). Sano et al. (1993) found that the cholestatic effect of EG that is evident in Sprague-Dawley rats was absent in mrp2-deficient Eisai hyperbilirubinuria rats, in which the biliary excretion of EG was delayed; Huang et al. (2000) obtained similar results with Wistar and mrp2-deficient TR^- rats. The relevance of the observed inhibition of bsep activity by ACU154 for the in vivo situation remains to be determined.

In conclusion OATP2 and MRP2 were identified as transport proteins involved in ACU154 transport via hepatocytes into bile, demonstrating the usefulness of double-transfected cell lines (Sasaki et al., 2002). For PKI166, a high basal-to-apical flux was observed in MDCKII cells, indicating involvement of endogenous vectorial transport systems and an overall good permeation through the cells. Both PKI166 and its O-glucuronide ACU154 influenced the activity of pivotal liver transport proteins, which might have an impact on the homeostasis in liver cells at a high exposure (Fig. 10).