Evaluation of Hepatotoxic Potential of Drugs by Inhibition of Bile-Acid Transport in Cultured Primary Human Hepatocytes and Intact Rats

Vsevolod E. Kostrubsky*,1, Stephen C. Strom{dagger}, Janean Hanson*, Ellen Urda*, Kelly Rose{ddagger}, James Burliegh{ddagger}, Philip Zocharski§, Hongbo Cai{dagger}, Jacqueline F. Sinclair,|| and Jasminder Sahi{ddagger}

* Department of Drug Safety Evaluation, Pfizer Global Research and Development, Ann Arbor, Michigan 48105; {dagger} University of Pittsburgh Medical Center, Department of Pathology, Pittsburgh, Pennsylvania 15261; {ddagger} Departments of Pharmokinetics, Dynamics, and Metabolism and § Pharmaceutical Sciences, Prizer Global Research and Development, Ann Arbor, Michigan 48105; Veterans Administration Medical Center, White River Junction, Vermont 05009; and || Departments of Biochemistry and Pharmacology/Toxicology, Dartmouth Medical School, Hanover, New Hampshire 03756

Received July 2, 2003; accepted August 6, 2003


    ABSTRACT
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Inhibition of canalicular bile acid efflux by medications is associated with clinical liver toxicity, sometimes in the absence of major liver effects in experimental species. To predict the hepatotoxic potential of compounds in vitro and in vivo, we investigated the effect of clinical cholestatic agents on [3H]taurocholic acid transport in regular and collagen-sandwich cultured human hepatocytes. Hepatocytes established a well-developed canalicular network with bile acid accumulating in the canalicular lumen within 15 min of addition to cells. Removing Ca2+ and Mg2+ from the incubation buffer destroyed canalicular junctions, resulting in bile acid efflux into the incubation buffer. Canalicular transport was calculated based on the difference between the amount of bile acid effluxed into the Ca/Mg2+-free and regular buffers with linear efflux up to 10 min. Hepatocytes cultured in the nonsandwich configuration also transported taurocholic acid, but at 50% the rate in sandwiched cultures. Cyclosporin A, bosentan, CI-1034, glyburide, erythromycin estolate, and troleandomycin inhibited efflux in a concentration-dependent manner. In contrast, new generation macrolide antibiotics with lower incidence of clinical hepatotoxicity were much less potent inhibitors of efflux. An in vivo study was conducted whereby glyburide or CI-1034, administered iv to male rats, produced a 2.4-fold increase in rat total serum bile acids. A synergistic 6.8-fold increase in serum total bile acids was found when both drugs were delivered together. These results provide methods to evaluate inhibitory effects of potentially cholestatic compounds on bile-acid transport, and to rank compounds according to their hepatotoxic potential.

Key Words: cholestasis; preclinical; clinical; toxicity; hepatocytes; bile acids; macrolides; transport.


    INTRODUCTION
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Hepatic toxicity is a major factor for discontinuing the development of compounds in pharmaceutical preclinical development or Phase I clinical trials. Some compounds demonstrate liver toxicity while being tested in experimental animals, suggesting potential adverse clinical liver effects. In contrast, other compounds demonstrate only minor or no signs of hepatotoxicity in the animal species tested, yet cause an increase in serum liver enzymes in more than 10% of humans during early clinical trials (reviewed in Fattinger et al., 2001Go; Pessayre et al., 1985Go). Our study focused primarily on compounds that were not predicted to result in clinical hepatotoxicity, due to the unresponsiveness of the preclinical species.

The common characteristic of the drugs used in this study is the preferential elimination from the body via the biliary pathway, which accounts for at least 50% of total drug clearance. We and others have hypothesized that hepatotoxicity in humans taking these drugs may be associated with drug-mediated inhibition of active canalicular transport of bile components, including, but not limited to, bile acids (Kostrubsky et al., 2001Go; Stieger et al., 2000Go). These drugs are likely substrates for active liver transporter-mediated uptake and efflux into the bile canaliculi. The transporters that participate in biliary drug elimination also transport endogenous bile components. Therefore, there is a potential for mutual inhibition of drug and bile acid efflux from the liver, resulting in an increase in drug and bile acids retained in the liver over time. We hypothesized that compounds showing greater inhibitory potency for bile acid transport will have a greater risk of being hepatotoxic. To address this issue, we have used an in vitro model of cultured human hepatocytes with an extensive canalicular network. We developed a competition assay between the drug and radioactively labeled bile acid to test whether canalicular efflux of taurocholate can be inhibited in a concentration-dependent manner, and whether this inhibition would correlate with clinical hepatotoxicity. For the current study, we have used six macrolide antibiotics with both high and low incidences of clinical hepatotoxicity. Our data indicate that drugs with greater hepatotoxic risk are stronger inhibitors of taurocholate transport in cultured human hepatocytes.

Previously, Fattinger et al.(2001)Go observed an increase in rat serum bile acids after treatment with drugs that cause hepatotoxicity in clinic. These authors found a dose-dependent increase in rat total bile acids after drugs were administered, either alone or in combination. We investigated whether inhibitors of taurocholate transport in cultured hepatocytes would also cause an increase in rat serum bile acids. Since bile acids are increased in humans taking these drugs, and associate with clinical hepatotoxicity, the increase in rat serum bile acids might be predictive of hepatotoxicity in humans.


    MATERIALS AND METHODS
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Chemicals.
HMM (modified Williams E) culture medium, dexamethasone, and insulin were obtained from BioWhittaker (Walkersville, MD). Penicillin G/streptomycin was acquired from GIBCO Laboratories (Grand Island, NY). 3H-Taurocholic acid (2 Ci/mmol) was obtained from PerkinElmer Life Sciences (Boston, MA). Cyclosporin A (CyA), glyburide, troleandomycin (TAO), salicylic acid, and a bile-acid analysis kit were purchased from Sigma (St. Louis, MO). Cholyl-lysyl-fluorescein (CLF) was synthesized at Pfizer. CI-1034, an endothelin receptor antagonist was a discontinued Pfizer drug candidate, procured from material management (Pfizer Global R&D, Ann Arbor, MI). Type I (rat tail) collagen was purchased from BD Biosciences (Bedford, MA). All reagents used were of the highest chemical purity available.

Hepatocyte culture.
Human hepatocytes were prepared from livers not used for whole organ transplant within 24 h of procurement. Hepatocytes were isolated by a three-step collagenase perfusion technique as described previously (Strom et al., 1996Go) and plated at a cell density of 2 x 106 cells per well in 6-well plates previously coated with 0.2 mg/ml type I collagen. The isolated hepatocytes (approximately 92% purity) were maintained in HMM medium supplemented with 10-7 M dexamethasone, 10-7 M insulin, 100 units/ml penicillin G, 100 µg/ml streptomycin, and 10% bovine calf serum, and were kept at 37°C in a humidified incubator with 95% air/5% CO2. Cells were allowed to attach for 4–6 h. At this time the medium was replaced with serum-free medium and changed on a daily basis thereafter. Following 24–48 h in culture, medium was removed and the cells overlaid with a neutralized preparation of collagen (0.1 ml/well) at a final concentration of 1.5 mg/ml, as described previously, with modifications (LeCluyse et al., 1994Go; Liu et al., 1999aGo,bGo). Specifically, collagen stock solution on ice was diluted with cold HMM culture medium, and pH was adjusted to 7.4 using cold, sterile 0.2 N NaOH. The plate was tilted to spread collagen, and excess collagen on the sides of wells was pipetted to fill uncovered spots. After 60 min in a humidified incubator at 37°C, culture medium was added to the plates and changed every 24 h thereafter. Some plated cells were not overlaid for the duration of the experiment, for comparison of transport with sandwiched cells.

Inhibition of bile acid transport.
After 96–120 h in culture, to allow for canaliculi to develop, the culture medium was replaced with regular Hank’s balanced salt solution (HBSS). Following 10 min incubation, 1 µM [3H]taurocholic acid or 5 µM CLF, with or without test compound, at concentrations indicated in the figure legends, was added in standard HBSS buffer, and plates were incubated for 15 min at 37°C. Transport was stopped by removing buffer and washing cells three times with 2 ml of cold standard HBSS. Taurocholate efflux from canalicular spaces was initiated by adding, in a time-dependent sequence, 1 ml of standard HBSS or Ca/Mg2+-free HBSS and incubating at 37°C (Liu et al., 1999cGo). Removal of Ca/Mg2+ from the incubation buffer opens the tight junctions and releases the bile acid that has accumulated therein (City, 1992Go). Aliquots of media (100 ml) were harvested at the indicated time points and counted in a liquid scintillation counter. Cells were washed once with regular HBSS and harvested in 1 ml of 0.2 N NaOH/0.1% SDS. Aliquots of cell lysate were counted in a liquid scintillation counter to determine the amount of taurocholate retained by the cells. All transport activities were normalized per milligram of total protein (Lowry et al., 1951Go).

The quantity of bile acid that accumulated in canaliculi was defined by the difference in efflux of radioactive taurocholate in incubation buffers, in the absence and presence of Ca/Mg2+, and calculated at 10 min after addition of the corresponding buffer. The difference in amount of radioactivity between the two buffer conditions in the absence of inhibitor corresponded to a 100% taurocholic acid efflux in canaliculi. In the presence of an inhibitor, this difference became smaller and was used to calculate the percent inhibition of canalicular taurocholic acid efflux.

Inhibition of taurocholate cellular uptake was calculated based on the amount of radioactivity recovered from cell lysates, in the presence or absence of inhibitors incubated in standard HBSS buffer.

The results represent the data from seven separate hepatocyte cultures prepared from different donors. We observed that taurocholate transport was strongly present and produced little variation from culture to culture when plated hepatocytes formed a confluent monolayer with microscopically distinguishable bile canaliculi on the day of the experiment. In contrast, absolute values of taurocholate transport were not sufficient to study the effect of inhibitors if the cells were subconfluent and formed separate aggregates of cells.

Fluorescent images of canalicular-accumulated CLF were taken immediately after washing cells and adding regular or Ca/Mg2+-free buffer.

Inhibition of bile acid transport in intact rats.
In vivo experiments were conducted in jugular vein precannulated Sprague-Dawley rats weighing approximately 300 g. Animals received powdered rodent chow (Purina certified chow; Purina, St. Louis, MO) and tap water by bottle ad libitum. All procedures involving animals were conducted in accordance with Guide for the Care and Use of Laboratory Animals and under a protocol approved by the Institutional Animal Care and Use Committee. Animals were fasted overnight prior to experiment. The drugs were administered, via the tail vein, to restrained animals, as follows: a single iv dose of vehicle-control glyburide (25 mg/kg), CI-1034 (25 mg/kg), or their combination, at 25 mg/kg each. The vehicle was a mixture of N,N-dimethylacetamide (DMA) and a 40% ß-cyclodextrin sulfobutyl ether sodium salt (SBECD) solution (w/v) prepared in 50 mM Tris(hydroxymethyl)aminomethane such that the total vehicle volume was composed of 5% DMA and 95% SBECD solution. Blood samples were collected from the jugular vein at indicated time-points, and serum was analyzed for total bile acids using a Hitachi 911 analyzer and Sigma reagents.


    RESULTS
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Effect of Collagen Overlay and Cation Removal on Canalicular Morphology
Human hepatocytes were cultured, either in sandwiched configuration or on regular collagen monolayers, for up to 7–8 days. As shown in Figure 1AGo, sandwiched hepatocytes maintained a three-dimensional shape and developed large canalicular tubular structures. In contrast, hepatocytes cultured on a collagen monolayer maintained a flat architecture and did not demonstrate developed canalicular networks of comparable size (Fig. 1BGo). Removal of the Ca/Mg2+ from the incubation buffer has been shown to disrupt canaliculi in sandwich-cultured rat hepatocytes without compromising their cellular membranes (Liu et al., 1999bGo). Consistent with this report and shown in Figure 2Go, the large canalicular tubular structures (Fig. 2AGo) shrunk when cells were incubated in Ca/Mg2+-free buffer (Fig. 2BGo).



View larger version (116K):
[in this window]
[in a new window]
 
FIG. 1. Effect of collagen overlay on bile canaliculi formation: Human hepatocytes prepared from the same donor cultured in collagen sandwich configuration (A) or on collagen monolayer (B) after 96 h in culture. The picture of intact cells was taken with a digital Nikon E995 camera attached to a Nikon Diaphot 300 microscope.

 


View larger version (102K):
[in this window]
[in a new window]
 
FIG. 2. Effect of Ca/Mg2+-free buffer on canalicular structures of cultured hepatocytes: Human hepatocytes prepared from a different donor from the one shown in Figure 1Go and cultured in sandwich configuration for 96 h developed canalicular thick tubular structures shown in A (arrows), and shrunk when a Ca/Mg2+-free buffer was added to the cells (B). The picture was taken with a digital camera a few min after replacement of the buffer.

 
To visualize the canalicular uptake and efflux of a compound under two different buffer conditions, we incubated cells with CLF, a fluorescent bile salt. CLF was added to cells in the regular buffer for 15 min. Extracellular CLF was rinsed out, and fluorescent microscopy was applied to demonstrate CLF accumulated inside the canalicular network (Fig. 3AGo). Replacement of regular CLF-containing buffer with Ca/Mg2+-free buffer resulted in no detectable fluorescence in canaliculi (Fig. 3DGo), suggesting leakage of CLF into the incubation media. We also investigated whether incubation of CLF together with CyA, an inhibitor of hepatobiliary transport, would prevent intracanalicular accumulation of CLF. No detectable CLF was found in canaliculi when both compounds were incubated together in regular cationic buffer for 15 min (Fig. 3BGo). Finally, coincubation of CLF with salicylic acid, a substrate for urinary and hepatic lateral transport, rather than biliary or apical elimination, was performed. As expected, no inhibitory effect of salicylic acid on CLF transport was detected (Fig. 3CGo).



View larger version (150K):
[in this window]
[in a new window]
 
FIG. 3. Effect of buffer conditions, cyclosporin A, and salicylate on CLF (cholyl-lysyl-fluorescein) transport. Human hepatocytes were cultured in sandwich configuration for 96 h. At the end of this time, cells received either 5 µM CLF alone (A, D) or in combination with 1 µM CyA (cyclosporin A) (B), or 1 µM salicylate (C) for 15 min. Cells were washed and fresh regular buffer was added in A, B, and C, whereas Ca/Mg2+-free buffer was added to D. Fluorescent images were taken immediately to detect CLF accumulation in canaliculi at 530 nm. Background fluorescence in round structures is due to dead cells trapped between the two layers of collagen.

 
To compare sandwich conditions with hepatocytes cultured on a regular collagen monolayer, we evaluated the capacity of hepatocytes to efflux taurocholic acid. As shown in Figure 4Go, sandwich-cultured hepatocytes had about 50% greater efflux of taurocholic acid after a 10-min incubation, under both regular and Ca/Mg2+-free buffer conditions. This suggests that a greater amount of taurocholate accumulated in canaliculi in sandwich cells compared with regular hepatocytes. This higher transport activity was also associated with more developed canaliculi in these cells (Figs. 1Go and 2Go).



View larger version (8K):
[in this window]
[in a new window]
 
FIG. 4. Effect of collagen overlay on taurocholate transport: Hepatocytes were prepared from different donors than shown in Figures 1Go–3Go and cultured either on a collagen monolayer (filled bars) or overlaid (open bars) with a neutralized preparation of collagen. After 96 h in culture, cells were treated with 1 µM [3H]taurocholic acid for 15 min. Cells were washed and regular or Ca/Mg2+-free buffers were added. Aliquots of media were harvested after 10 min incubation and counted in a liquid scintillation counter. Each value represents the mean of duplicate treatments from two different donors, with the range indicated by the vertical bars.

 
Taurocholate Efflux
We next investigated the time-dependent efflux of taurocholic acid from canaliculi into the incubation buffer under two different buffer conditions. In regular buffer, linear transporter-mediated efflux of taurocholic acid was observed for 60 min. This efflux was limited by the transporter activity. In contrast, removal of Ca/Mg2+ would be expected to result in a large increase in efflux of bile acid from the canaliculi into the incubation buffer, due to destruction of canalicular tubular structures. A 3-fold increase in efflux, by 10 min, was observed under Ca/Mg2+-free buffer conditions (Fig. 5AGo). No further increase in taurocholic acid efflux was detected after a 20-min incubation, suggesting that all the accumulated bile acid was eliminated from the cells by this time. The difference between the amount of taurocholate detected in the absence and presence of Ca/Mg2+ measured at 10 min was used to represent the amount of taurocholate accumulated in canaliculi. The decrease in this difference, caused by an inhibitor of bile-acid transport, represents the inhibitory effect of compound on biliary export.



View larger version (20K):
[in this window]
[in a new window]
 
FIG. 5. Efflux of taurocholate from cultured hepatocytes: After 96 h in sandwiched culture, cells were treated with 1 µM [3H]taurocholic acid alone (A) for 15 min. In B, hepatocytes prepared from a different donor were treated with 1 µM [3H]taurocholic acid alone or in combination with 1 µM cyclosporin A for 15 min. Cells were washed and taurocholate efflux from canalicular spaces was initiated by adding 1 ml regular (closed circles, squares, triangles) or Ca/Mg2+-free (open circles, squares, triangles) buffer and incubated at 37°C. Aliquots of media were harvested at indicated time points and counted in a liquid scintillation counter. Each value represents the mean of duplicate treatments, with the range indicated by the vertical bars.

 
Effect of Inhibitors on Taurocholate Efflux
To investigate the effect of inhibitor on taurocholate efflux, we used CyA, a potent inhibitor of bile acid transport (Stieger et al., 2000Go), in sandwich-cultured hepatocytes. Coincubation of taurocholic acid and CyA would be expected to decrease taurocholate efflux. As shown in Figure 5BGo, CyA at 1 µM produced a dramatic decrease in taurocholate efflux (open triangles vs. open squares) causing a 60% inhibition of efflux after a 10-min incubation. In the presence of inhibitor, there was a linear increase in taurocholate efflux up to 10 min, suggesting that intracellular taurocholate was not limited for transport into the canaliculi.

The overall effect of different compounds on canalicular efflux, measured in two hepatocyte cultures, is shown in Figures 6AGo and 6BGo. Efflux at 100% is shown by an arrow in the control cells. A decrease in difference between the two buffer conditions, in the presence and absence of potential inhibitors, represents the percent inhibition compared to untreated hepatocytes, as summarized in Figure 6BGo. At the concentrations tested, salicylic acid did not inhibit transport of taurocholic acid. In contrast, CyA, CI-1034, bosentan, glyburide, and TAO, drugs that are eliminated preferentially via bile and clinically known to cause liver toxicity, inhibited taurocholate efflux. CyA, CI-1034, and glyburide caused a concentration-dependent inhibition of taurocholate efflux. CyA at 10 µM completely inhibited taurocholate efflux, since no difference in taurocholate efflux was observed between the two buffers.



View larger version (22K):
[in this window]
[in a new window]
 
FIG. 6. Inhibition of taurocholate efflux by drugs: Sandwich cultured hepatocytes prepared from two donors were treated with 1 µM [3H]taurocholic acid, alone or in combination with different potential inhibitors, for 15 min at concentrations indicated in the figure. Cells were washed and regular (A, filled bars) or Ca/Mg2+-free (A, open bars) buffers were added. Aliquots of media were harvested after 10 min incubation and counted in a liquid scintillation counter. (A) The absolute decrease in taurocholate efflux in both buffers; (B) the percentage of taurocholate efflux of control cells, calculated as the difference between Ca/Mg2+-free and regular buffer. Hepatocytes in B were prepared from a different donor than shown in A. Concentrations are given in µM. CI-1034, glyburide, and bosentan share some structural similarities and have a sulfonamide link. Each value represents the mean of duplicate treatments with the range indicated by the vertical bars.

 
Effect of Inhibitors on Taurocholate Uptake
We also investigated whether the inhibitors that blocked taurocholate efflux would decrease taurocholate uptake in hepatocytes when measured at 10 min after addition of the regular buffer. Hepatocytes and the aliquots of the buffer were harvested, and radioactivity of taurocholate was counted. Since canalicular structures were preserved in this experiment, the taurocholate detected in hepatocytes represented the total amount present in cells and canalicular spaces. Figures 7AGo and 7BGo show that CyA, CI-1034, glyburide, and bosentan inhibited taurocholate uptake in a concentration-dependent manner. In contrast, TAO, while similar to other compounds in causing inhibition of efflux, had no effect on taurocholate uptake at concentrations of 10 µM and 50 µM. TAO, at 100 µM, caused a 50% decrease in taurocholate uptake. Salicylic acid did not inhibit efflux nor decrease uptake of taurocholate at any concentration tested.



View larger version (23K):
[in this window]
[in a new window]
 
FIG. 7. Inhibition of taurocholate uptake by drugs: Sandwich cultured hepatocytes were treated with 1 µM [3H]taurocholic acid, alone or in combination with different potential inhibitors, for 15 min at concentrations indicated in the figure. Cells were washed and fresh regular buffer was added. Aliquots of media were harvested after a 10-min incubation. Cells were harvested in 1 ml of 0.2 N NaOH/0.1% SDS. Aliquots of cell lysate (filled bars) and media (open bars) were counted in a liquid scintillation counter to determine the amount of taurocholate retained by the cells. (A) The absolute decrease in taurocholate uptake in total cell lysate (intracellular and canalicular amounts), which is also reflected by the decreased amount of taurocholate released in the regular media. (B) The percent uptake of control cells calculated as the decrease in total intracellular taurocholate. Concentrations are given in µM. Each value represents the mean of duplicate treatments, with the range indicated by the vertical bars.

 
Inhibition of Taurocholate Transport by Macrolide Antibiotics
To rank compounds according to their inhibitory potency within the given therapeutic area, we compared the effect on taurocholate transport of six macrolide antibiotics in order to determine if macrolides with high incidence of clinical cholestatic liver injury are also potent inhibitors of the hepatobiliary transport of taurocholate. As shown in Figure 8Go, erythromycin estolate, followed by TAO, were the most potent inhibitors of taurocholate efflux. In contrast, the new generation macrolides, roxithromycin, spiramycin, and telithromycin, as well as erythromycin base, were less-potent inhibitors. This ranking corresponds to numerous clinical reports on erythromycin estolate and TAO-induced liver toxicity (reviewed in Stricker and Spoelstra, 1985Go; Ticktin and Zimmerman, 1962Go). Erythromycin base and roxithromycin did not inhibit efflux, and telithromycin and spiramycin inhibited efflux by about 40% at 100 µM.



View larger version (20K):
[in this window]
[in a new window]
 
FIG. 8. Effect of macrolide antibiotics on taurocholate efflux: After 96 h in sandwich culture, cells were treated with 1 mM [3H]taurocholic acid alone or in combination with indicated concentrations of macrolides for 15 min. Cells were washed and regular or Ca/Mg2+-free buffers were added. Aliquots of media were harvested after a 10-min incubation and counted in a liquid scintillation counter. Each value represents the mean of duplicate treatments of cells prepared from two to four different donors, with the range indicated by the vertical bars.

 
Effect of Glyburide and CI-1034 on Rat Total Serum Bile Acids
To see if our in vitro results would correlate with in vivo, we investigated, in intact rats, whether drugs that demonstrated strong inhibitory effects on taurocholate transport in cultured hepatocytes also increase rat serum total bile acids, due to competition with endogenous bile acids for biliary efflux and hepatic uptake. As shown in Figure 9Go, there was a 2.4-fold increase in total serum bile acids at 10 min after iv administration of CI-1034 or glyburide, at 25 mg/kg each. A synergistic 6.8-fold increase in total serum bile acids was found when both drugs were delivered together. The increase in serum bile acids was a fast response to injected drugs and had decreased to almost control level by 90 min after drug administration.



View larger version (16K):
[in this window]
[in a new window]
 
FIG. 9. Effect of glyburide and CI-1034 on rat serum bile acids: A single iv dose of vehicle control, glyburide (25 mg/kg), CI-1034 (25 mg/kg), or their combination at 25 mg/kg each was administered to rats via the tail vein. Blood samples (200 ml) were collected from the jugular vein at indicated time points, and serum was analyzed for total bile acids using a Hitachi 911 analyzer.

 

    DISCUSSION
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
In this work, we focused on compounds that are primarily (>50%) eliminated from the body via bile. The interaction at the level of hepatobiliary transporters between drug and bile acid may result in inhibition of bile-acid transport, and be associated with clinical liver toxicity. Animals are often resistant to liver toxicity (as detected by increases in liver enzymes and histological changes) by these drugs and are not always accurate predictors for humans. The purpose of this study was to set up in vitro and in vivo systems to study inhibition of bile acid transport by known hepatobiliary substrates, and to compare our results with clinical liver effects caused by those compounds. We showed that human hepatocytes are able to efflux bile acid with linear secretion of taurocholate for up to 60 min in the presence of Ca/Mg2+-containing buffer. Removal of cations caused a rapid efflux of bile acid accumulated in canaliculi, with linear output limited to 10 min. We also found a three-fold increase in the bile salt export pump (BSEP) mRNA expression in cells treated with 50 µM chenodeoxycholic acid for 48 h (data not shown), demonstrating the response of this major transporter. Transport was greater in sandwiched hepatocytes, likely due to greater transporter activity in developed canaliculi. The structural changes in rat and human hepatocytes cultured in sandwiched configuration, including normal distribution of actin filaments and microtubules, electron-microscopic analysis of bile canaliculi, and gap-junction localization have been described previously (Dunn et al., 1991Go; Hamilton et al., 2001Go; LeCluyse et al., 1994Go). We detected accumulation of fluorescent bile acid in human canaliculi 15 min after addition to the cells, a response similar to that described with rat canaliculi (Liu et al., 1999bGo). However, fluorescent bile acid did not accumulate in cells that were coincubated with CLF and CyA, suggesting that CyA inhibited transport of CLF. Removal of Ca/Mg2+ from the incubation buffer has been demonstrated to cause selective disruption of junctional complexes, resulting in leaky canaliculi (Liu et al., 1999bGo; Lora et al., 1997Go). We used this feature to study the effect of inhibitors on the efflux of bile acid. The amount of bile acid collected in canaliculi reflects the activity of the transport system. Disruption of canalicular junctions results in complete release of accumulated bile acid into the buffer. This release was linear for 10 min and, therefore, was not limited by intracellular bile acid availability within this time. The difference between bile acid released in the absence and in the presence of Ca/Mg2+ represents the amount (or 100% in control cells) of taurocholate accumulated in canaliculi. Coincubation of bile acid with CyA, bosentan, glyburide, CI-1034, or TAO decreased the release of taurocholate in a concentration-dependent manner. The same inhibitors also decreased the cellular uptake of bile acids but to a different extent. For example, TAO did not decrease bile acid cellular uptake up to 100 µM, whereas CyA, CI-1034, glyburide, and bosentan prevented bile acid uptake at lower concentrations, suggesting selectivity of TAO toward inhibition of efflux transporters. Since drugs could affect the bile-acid uptake transporter, it is important to measure the taurocholate efflux rate based on the linear portion of the efflux plot (Fig. 5BGo). Bile acid efflux is determined by its canalicular level and should not be limited by its intracellular availability.

The rationale for the present study comes from the understanding that some drugs are substrates for hepatic active biliary transport and are eliminated via the bile. Clinical adverse liver effects, including cholestasis, are associated with inhibition of biliary transport. Compounds with molecular structures that make them likely candidates for biliary elimination can inhibit bile acid transport and be potentially hepatotoxic. Drugs, including CI-1034, bosentan, TAO, glyburide, and erythromycin estolate, are known to cause hepatic dysfunction in humans. Bosentan causes an increase in serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) in 10–11% of patients at levels at least 3 times the upper limit of normal, and in 4% of patients at levels greater than 8 times the upper limit of normal (NDA, 2001Go). In addition, Fattinger et al. (2001)Go reported that bosentan-induced cholestatic liver injury developed within the first 4 weeks of treatment, with increased serum bile acid levels preceding increased ALT levels by 5 to 7 days. The same group reported administration of glyburide as a risk factor, because concomitant bosentan and glyburide treatment increased the incidence of liver injury to 29%. They also found an additive response on rat serum bile acids when both drugs were coadministered intravenously. We found a similar response with CI-1034 when it was combined with glyburide (Fig. 9Go). In a clinical Phase 1 study, CI-1034 was administered only to a small number of patients before being discontinued from development. Similarly to bosentan, CI-1034 increased ALT, AST, and serum bile acids at between 2 and 7 weeks of treatment.

TAO and erythromycin estolate are drugs with known hepatotoxic potential in humans. Serum aminotransferase elevations with erythromycin estolate have been reported in up to 38% of patients, with another study reporting an increase in about 10% of patients and a cholestatic pattern of injury developing two weeks after the beginning of therapy (reviewed in Stricker and Spoelstra 1985Go). Changes appear in most cases within 5 to 15 days after starting treatment. In agreement with many clinical reports of cholestasis, in our study, erythromycin estolate was found to be a potent inhibitor of taurocholate efflux (Fig. 8Go). In addition, erythromycin estolate has been shown to reduce bile acid concentrations and bile flow in isolated rat liver, effects that were not observed to the same extent with erythromycin base (Gaeta et al., 1985Go). We also found that erythromycin base, in contrast to erythromycin estolate, was ineffective in inhibiting taurocholate transport. It also does not have the same reported clinical hepatotoxicity as erythromycin estolate. Similar to erythromycin estolate, TAO has a high reported rate of liver toxicity increasing serum aminotransferase in 30% and producing jaundice in 4% of patients, typically after about two weeks of treatment (Ticktin and Zimmerman, 1962Go). TAO followed erythromycin estolate in inhibiting taurocholate efflux in our system (Fig. 8Go). In contrast to the high frequency of reported clinical hepatotoxicity after treatment with erythromycin estolate and TAO, new-generation macrolides are less hepatotoxic. Although roxithromycin, spiramycin, and telithromycin have clinical records of cholestatic hepatitis (Denie et al., 1992Go; Easton-Carter et al., 2001Go; U.S. FDA, 2003Go; Zuazo et al., 1997Go), they have less frequent incidences of liver injury. Similarly, in sandwiched human hepatocytes, roxithromycin did not inhibit efflux of taurocholate, and it increased liver aminotransferase in <1% of patients (product information). Both telithromycin and spiramycin inhibited taurocholate efflux only at 100 µM. Recent clinical data on telithromycin indicates that it increased ALT > 3 times the upper limit of normal in approximately 1% of patients at any post-therapy time point (FDA AIDAC Meeting, 2003).

The inhibitory potencies of individual compounds within a given therapeutic area need to be evaluated in conjunction with the projected human therapeutic plasma concentration. Higher plasma concentrations will likely be associated with greater levels of the drug in liver. When the drug has a high potency of inhibition of bile-acid transport in the described systems, this compound will be expected to have a greater probability of clinically adverse liver effects. For bosentan, CI-1034, TAO, and all tested macrolides, the human plasma concentrations are above 1 µg/ml (1–12 µg/ml). This is similar to the serum concentration of bile acids, which are concentrated in bile at 1000 times the concentration detected in serum. In contrast, the plasma concentration for glyburide is about 0.1 µg/ml and is associated with a low incidence of hepatic injury, despite the potent inhibition of bile-acid transport. Yet, a few reports have demonstrated cholestatic hepatitis with glyburide, an effect that may be attributed to the ability of glyburide to accumulate in liver and to be a potent inhibitor of bile-acid transport (Kelner et al., 1969Go; Stricker and Spoelstra, 1985Go; Wongpaitoon et al., 1981Go). Similarly, high hepatic concentrations (150 times greater than in the serum) were also reported in rats for erythromycin (Lee et al., 1953Go), and roxithromycin, an oxime derivative of erythromycin that has an elimination half-life of about six times that of erythromycin, also reached high tissue concentrations (Puri and Lassman, 1987Go). CyA is a potent inhibitor of Bsep (Stieger et al., 2000Go), as well as of taurocholate transport in our system (Fig. 6Go). In agreement with in vitro data, treatment of transplant patients with CyA resulted in 2- to 3-fold increases in total serum bile acids and this correlated well with blood levels of CyA (Tripodi et al., 2002Go). At the same time, CyA has a lower incidence of hepatotoxicity in comparison to nephrotoxicity (Klintmalm et al., 1981Go). CyA is pharmacologically active at plasma concentrations less than 1 µg/ml (0.1–0.4 µg/ml) and has a low hepatic extraction ratio, likely maintaining a low level of drug in the liver, a factor contributing to the low frequency of hepatotoxicity.

The fact that both glyburide and CI-1034 increased rat serum total bile acids after iv administration suggests their cholestatic potential by acting on the hepatobiliary transporters. Fattinger et al., 2001Go, observed a similar response to glyburide and bosentan. These authors found a dose-dependent increase in rat total bile acids after drugs were administered, either alone or in combination. Since bile acids are increased in humans taking these drugs and associate with clinical hepatotoxicity, the increase in rat total bile acids might be predictive of hepatotoxicity in humans.

In summary, cultured human hepatocytes transport bile acid into and out of cells. This transport can be inhibited, in a dose-dependent fashion, by compounds that are substrates for biliary elimination. The potency of this inhibition by a drug is likely to be a reflection of its in vivo inhibitory effect on bile-acid transport and could therefore be associated with adverse clinical liver effects. Even though rats may be resistant to liver toxicity induced by some drugs, a transient increase in serum bile acids combined with inhibitory effects on bile-acid efflux in sandwich cultured human hepatocytes may be better predictors of the cholestatic potential of compounds in humans. The evaluation of compounds from a given therapeutic area, using our proposed strategy, will help rank compounds according to their hepatotoxic potential.


    ACKNOWLEDGMENTS
 
We would like to thank Dr. Donna Banks for assistance in conducting experiments with fluorescent bile acid and Ms. Karen Metzler for technical support with the in vivo study. This work was supported in part by grants GM60346 and DK92310 (SS) to S.C.S.


    NOTES
 
1 To whom correspondence should be addressed. Fax: (734) 622-3478. E-mail: vsevolod.kostrubsky{at}pfizer.com. Back


    REFERENCES
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
City, S. (1992). Protein kinase inhibitors prevent junction dissociation induced by low extracellular calcium in MDCK epithelial cells. J. Cell Biol. 117, 169–178.[Abstract]

Denie, C., Henrion, J., Schapira, M., Schimitz, A., and Heller, F. R. (1992). Spiramycin-induced cholestatic hepatitis. J. Hepatol. 16, 386.[ISI][Medline]

Dunn, J., Tompkins, R., and Yarmush, M. (1991). Long-term in vitro function of adult hepatocytes in a collagen-sandwich configuration. Biotechnol. Prog. 7, 237–245.[ISI][Medline]

Easton-Carter, B., Hardikar, W., and Smith, A. L. (2001). Possible roxithromycin-induced fulminant hepatic failure in a child. Pharmacotherapy 21, 867–870.[ISI][Medline]

Fattinger, K., Funk, C., Pantze, M., Weber, C., Reichen, J., Stieger, B., and Meier, P. (2001). The endothelin antagonist bosentan inhibits the canalicular bile salt export pump: A potential mechanism for hepatic adverse reactions. Clin. Pharmacol. Ther. 69, 223–231.[CrossRef][ISI][Medline]

Gaeta, G. B., Utili, R., Adinolfi, L. E., Abernathy, C. O., and Giusti, G. (1985). Characterization of the effects of erythromycin estolate and erythromycin base on the excretory function of the isolated rat liver. Toxicol. Appl. Pharmacol. 80, 185–192.[ISI][Medline]

Hamilton, G., Jolley, S., Gilbert, D., Coon, D., Barros, S., and LeCluyse, E. (2001). Regulation of cell morphology and cytochrome P450 expression in human hepatocytes by extracellular matrix and cell-cell interactions. Cell Tissue Res. 306, 85–99.[CrossRef][ISI][Medline]

Kelner, H. M., Christ, O., Rupp, W., and Heptner, W. (1969). Resoption, verteilung, und ausscheidung nach gabe von 14C-markiertem HB 419 an kaninchen, ratten und hunde. Arzneimittelforschung 19, 1388–1400.[Medline]

Klintmalm, G., Iwatsuki, S., and Starzl, T. (1981). Cyclosporin A hepatotoxicity in 66 renal allograft recipients. Transplantation 32, 488–489.[ISI][Medline]

Kostrubsky, V., Vore, M., Kindt, E., Burliegh, J., Rogers, K., Peter, G., Altrogge, D., and Sinz, M. (2001). The effect of troglitazone biliary excretion on metabolite distribution and cholestasis in transporter-deficient rats. Drug Metab. Dispos. 29, 1561–1556.[Abstract/Free Full Text]

LeCluyse, E., Audus, K., and Hochman, J. (1994). Formation of extensive canalicular networks by rat hepatocytes cultured in collagen-sandwich configuration. Am. J. Physiol. (Cell) 35, C1764–C1774.

Lee, C. C., Anderson, R. C., and Chen, K. K. (1953). Tissue distribution of erythromycin in rats. Antibiot. Chemother. 3, 920–924.[ISI]

Liu, X., Chism, J., LeCluyse, E., Brouwer, K., and Brouwer, K. (1999a). Correlation of biliary excretion in sandwich-cultured rat hepatocytes and in vivo in rats. Drug Metab. Dispos. 6(27), 637–644.

Liu, X., LeCluyse, E., Brouwer, K., Gan, L., Lemasters, J., Steiger, B., Meier, P., and Brouwer, K. (1999b). Biliary excretion in primary rat hepatocytes cultured in a collagen-sandwich configuration. Am. J. Physiol. Gastrointest. Liver Physiol. 40, G12–G21.

Liu, X., LeCluyse, E., Brouwer, K., Lightfoot, R., Lee, J., and Brouwer, K. (1999c). Use of Ca2+ modulation to evaluate biliary excretion in sandwich-cultured rat hepatocytes. J. Pharmacol. Exp. Ther. 289, 1592–1599.[Abstract/Free Full Text]

Lora, L., Mazzon, E., Martines, D., Fries, W., Muraca, M., A., d’Odorico, A., Naccarato, R., and Citi, S. (1997). Hepatocyte tight-junctional permeability is increased in rat experimental colitis. Gastroenterology 113, 1347–1354.[ISI][Medline]

Lowry, O., Rosebrough, N., Farr, A., and Randall, R. (1951). Protein measurement with the Folin phenol reagent. J. Biol. Chem. 193, 265–275.[Free Full Text]

New Drug Application (NDA). (2001). Medical review and efficacy conclusions. Food and Drug Administration, July 6, 2001. 21, 290.

Pessayre, D., Larrey, D., Funck-Brentano, C., and Benhamou, J. (1985). Drug interactions and hepatitis produced by some macrolide antibiotics. J. Antimicrob. Chemother. 16, 181–194.[ISI][Medline]

Puri, S. K., and Lassman, H. B. (1987). Roxithromycin: A pharmacokinetic review of a macrolide. J. Antimicr. Chem. 39, 660–668.

Stieger, B., Fattinger, K., Madon, J., Kullak-Ublick, G., and Meier, P. (2000). Drug- and estrogen-induced cholestasis through inhibition of the hepatocellular bile salt export pump (bsep) of rat liver. Gastroenterology 118, 422–430.[ISI][Medline]

Stricker, B. H., and Spoelstra, P. (1985). Drug-Induced Hepatic Injury (M. Dukes, Ed.), pp. 153–155. Elsevier, Amsterdam.

Strom, S., Pisarov, L., Dorko, K., Thompson, M., Schuetz, J., and Schuetz, E. (1996). Use of human hepatocytes to study P450 gene induction. Methods Enzymol. 272, 388–401.[ISI][Medline]

Ticktin, H. E., and Zimmerman, H. J. (1962). Hepatic dysfunction and jaundice in patients receiving triacetyl oleandomycin. N. Engl. J. Med. 267, 964–968.[ISI]

Tripodi, V., Nunez, M., Carducci, C., Mamianetti, A., and Carrento, A. (2002). Total serum bile acids in renal transplanted patients receiving cyclosporin A. Clin. Nephrol. 58, 350–355.[ISI][Medline]

U.S. Food and Drug Administration (U.S. FDA). Center for drug evaluation and research. Anti-infective drug advisory committee (AIDAC) meeting. Gaithersburg, MD, January 8, 2003.

Wongpaitoon, V., Mills, P., Russell, R., and Patrick, R. (1981). Intrahepatic cholestasis and cutaneous bullae associated with glibenclamide therapy. Postgrad. Med. J. 57, 244–246.[Abstract]

Zuazo, J. A., Revuelta, C., Perez Alvarez, J. C., Lopez Morante, A., and Saez-Royuela, F. (1997). Acute cholestatic hepatitis induced by spiramycin. Gastroenterol. Hepatol. 20, 474–475.