Immunochemical Identification of Mouse Hepatic Protein Adducts

Immunochemical Identification of Mouse Hepatic Protein Adducts Derived from the Nonsteroidal Anti-Inflammatory Drugs Diclofenac, Sulindac, and Ibuprof...
0 downloads 0 Views 517KB Size
546

Chem. Res. Toxicol. 1997, 10, 546-555

Immunochemical Identification of Mouse Hepatic Protein Adducts Derived from the Nonsteroidal Anti-Inflammatory Drugs Diclofenac, Sulindac, and Ibuprofen Lara T. Wade, J. Gerald Kenna,* and John Caldwell Pharmacology and Toxicology, Imperial College School of Medicine at St. Mary’s, Norfolk Place, London W2 1PG, U.K. Received August 29, 1996X

Reactive metabolite-modified hepatic protein adducts have been proposed to play important roles in the mechanism(s) of hepatotoxicity of nonsteroidal anti-inflammatory drugs (NSAIDs). In the present study, immunochemical techniques have been used to compare the patterns of drug-protein adducts expressed in livers of mice given single doses of one or other of three different NSAIDs. These were diclofenac and sulindac, which are widely used but potentially hepatotoxic drugs, and ibuprofen, which is considered to be nonhepatotoxic. Specific polyclonal antisera were produced by immunization of rabbits with conjugates prepared by coupling each of the NSAIDs to the carrier protein keyhole limpet hemocyanin. Immunoblotting studies revealed dose-dependent formation of major 110 kDa polypeptide adducts in livers from mice sacrificed 6 h after administration of single doses of either diclofenac (0-300 mg/kg) or sulindac (0-100 mg/kg). Lower levels of several other adducts, of 140 and 200 kDa, were also expressed in livers from these animals. In contrast, livers from mice treated with ibuprofen (0-200 mg/ kg) predominantly expressed a 60 kDa adduct and only relatively low levels of a 110 kDa adduct. The various adducts were shown by differential centrifugation to be concentrated in the nuclear fraction of liver homogenates. Those derived from diclofenac and sulindac were further localized, by Percoll density gradient centrifugation, to a subfraction which contained a high activity of the bile canalicular marker enzyme alkaline phosphatase. This suggests that they are concentrated in the bile canalicular domain of hepatocytes. The different patterns of adduct formation raise the possibility that formation of certain NSAID protein adducts, particularly 110 kDa adducts, has toxicological significance.

Introduction Hepatotoxicity, ranging from mild elevations in serum transaminases to severe hepatocellular and/or cholestatic injury, has been described in patients treated with many different nonsteroidal anti-inflammatory drugs (NSAIDs1) (1, 2). These adverse reactions are relatively rare but are an important clinical issue due to the widespread use of this class of drugs, the idiosyncratic nature of the toxicity, and the potential for progression to fulminant hepatic failure. Moreover, individual NSAIDs differ markedly in the incidence, severity, and type of liver injury that they produce (1, 2). Several NSAIDs have been withdrawn from the market because of severe hepatotoxicity, and many others remain in general use despite being associated with liver damage, while some appear to cause little hepatic injury (1-3). Currently, the mechanisms underlying NSAID-induced hepatotoxicity are unknown. The presence of rashes, * Address correspondence to this author. Telephone: 44 171 594 3881. Fax: 44 171 723 7535. E-mail: [email protected]. X Abstract published in Advance ACS Abstracts, April 15, 1997. 1 Abbreviations: NSAID, nonsteroidal anti-inflammatory drug; KLH, Keyhole limpet hemocyanin; EDC, 1-ethyl-3-[3-(dimethylamino)propyl]carbodiimide; TNBS, trinitrobenzenesulfonic acid; RSA, rabbit serum albumin; Thimerosal, mercury[(o-carboxyphenyl)thio]ethyl sodium salt; DMSO, dimethyl sulfoxide; ECL, enhanced chemiluminescence; ELISA, enzyme-linked immunosorbent assay; PBS, phosphate-buffered saline (137 mM sodium chloride, 8 mM disodium hydrogen phosphate, 1.5 mM potassium dihydrogen phosphate, and 2.7 mM potassium chloride (pH 7.4)); SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis; rt, room temperature (1821 °C); UGT, uridine diphosphate glucuronosyltransferase (E.C. 2.4.1.17); DPP IV, dipeptidyl-peptidase IV (E.C. 3.4.14.5).

S0893-228x(96)00153-1 CCC: $14.00

fever, and/or eosinophilia in some patients, which are indicative of hypersensitivity reactions, suggests that the immune system may be involved (4). Conversely, the absence of these markers in other patients raises the possibility that direct toxicities caused by the parent drugs and/or their metabolites could be responsible (5, 6). A common mechanism which might underlie either immune-mediated hepatotoxicity or direct toxicity is formation of NSAID-modified protein adducts (7). Metabolism of NSAIDs in vivo proceeds via several enzymic processes that may involve generation of reactive metabolites capable of binding to cellular macromolecules. Acyl glucuronides of NSAIDs have been identified which are chemically labile and have been shown to mediate binding of several NSAIDs to plasma proteins (8). Other potential routes of bioactivation of these drugs are via cytochrome P450 or acyl-CoA ligases (9, 10). Immunochemical approaches have been used recently by several groups to investigate the formation of hepatic protein adducts derived from the widely used NSAID diclofenac (11-15). Studies undertaken with specific polyclonal antisera, produced by immunization of rabbits with synthetic diclofenac-protein conjugates, revealed that diclofenac-protein adducts were expressed in livers of mice and rats treated with the drug in vivo and also in rat and human hepatocytes incubated with diclofenac in vitro (11-15). Immunoblotting studies indicated that many different hepatic proteins could become covalently modified by reactive metabolites derived from the drug, while in vitro investigations undertaken with rat liver subcellular fractions revealed that adduct formation © 1997 American Chemical Society

NSAID-Modified Hepatic Proteins

Figure 1. Chemical structures of diclofenac, sulindac, and ibuprofen.

could be catalyzed both by uridine diphosphate glucuronosyltransferases (UGTs) and by cytochrome P450 (11-15). Unfortunately, markedly different patterns of diclofenac adducts were detected in the immunoblotting studies undertaken in different laboratories, and no common pattern of adduct formation was observed (12, 14, 15). In the present investigations, immunochemical techniques have been used to compare the pattern of protein adducts produced in vivo in livers of mice treated with diclofenac with the patterns of adducts expressed in mice given sulindac or ibuprofen (Figure 1). Both diclofenac and sulindac are widely used NSAIDs which are associated with a significant incidence of hepatotoxicity (4, 6). In contrast, ibuprofen is generally considered to be the NSAID least likely to cause liver injury and is widely available as an over-the-counter medication (1, 16). The results we have obtained suggest that the potential of particular NSAIDs to elicit hepatotoxicity could be related to formation of certain hepatic protein adducts.

Experimental Procedures Materials. The following compounds were obtained from Sigma (U.K.): ibuprofen (R-methyl-4-(2-methylpropyl)benzeneacetic acid) sodium salt, sodium diclofenac (2-[(2,6-dichlorophenyl)amino]benzeneacetic acid), sulindac (cis-5-fluoro-2-methyl1-[p-(methylsulfinyl)benzylidene]indene-3-acetic acid), 1-ethyl3-[3-(dimethylamino)propyl]carbodiimide (EDC), 2,4,6-trinitrobenzenesulfonic acid (TNBS), Freund’s complete adjuvant, Freund’s incomplete adjuvant, mercury [(o-carboxyphenyl)thio]ethyl sodium salt (Thimerosal), tricaprylin, dimethyl sulfoxide (DMSO), Percoll, Keyhole limpet hemocyanin (KLH), and rabbit serum albumin (RSA). Caution: EDC and TNBS are irritant to skin and eyes, and TNBS and Thimerosal are toxic by ingestion and inhalation; these chemicals must be handled using appropriate safety precautions. Nitrocellulose (Schliecher and Schuell; Protran, 0.45 µM) was obtained from Anderman and Co. Ltd. (U.K.). Electrophoresis grade reagents were obtained from BioRad (U.K.). Affinity-isolated goat anti-rabbit immunoglobulins conjugated to horseradish peroxidase and to alkaline phosphatase and rat monoclonal antibody conjugated to mouse CD71 (transferrin receptor) were obtained from Serotec (U.K.). Horseradish peroxidase conjugated to affinityisolated goat anti-rat IgG was obtained from Sigma (U.K.). Enhanced chemiluminescence (ECL) reagents were ECL Western blotting reagent from Amersham plc (U. K.) and SuperSignal from Pierce & Warriner Ltd. (U.K.). All other reagents used were of analytical grade and were obtained from BDH (U.K.). Synthesis and Characterization of NSAID-Modified Protein Conjugates. Ibuprofen, sodium diclofenac, and sulindac were covalently coupled to KLH and RSA by a two-step conjugation reaction, using EDC, as described by Pumford et al. (11). The conjugates were dialyzed extensively against distilled water, then lyophilized, and stored at 4 °C.

Chem. Res. Toxicol., Vol. 10, No. 5, 1997 547 To estimate hapten densities, the numbers of free lysine residues on the RSA-diclofenac and RSA-ibuprofen conjugates were quantitated using a modification of the method described by Habeeb (17). The conjugates were dissolved in 0.1 M sodium bicarbonate/carbonate buffer pH 9 (0.5-1 mg of protein /mL) and then incubated in the dark with 0.1% TNBS (final concentration 0.033%) for 2 h at 40 °C. The reaction was stopped by the addition of 10% sodium dodecyl sulfate (SDS) (final concentration 2.5%), followed by 1 M HCl (final concentration 0.06 M), and the absorbance read at 340 nm. Quantification was achieved by use of standard curves prepared from unconjugated RSA and glycine. This procedure could not be applied to the KLH-sulindac conjugate because of marked absorbance of light by sulindac at 340 nm. In fact, sulindac was found to have UV absorption maxima in aqueous solution at 226, 283, and 330 nm, and the extinction coefficient at 330 nm was determined experimentally to be 33.8 mg-1 mL-1. Consequently, the hapten density of the KLH-sulindac conjugate was estimated by UV spectroscopy, at 330 nm, using this value. Estimation of the hapten densities of the KLH conjugates was not possible because these were very poorly soluble in aqueous solution. Immunization Protocol. Female New Zealand white rabbits (2.5-3 kg; Froxfield, U.K.) were each immunized with 250 µg of KLH-NSAID conjugate, suspended in 1 mL of a 50% (v/ v) emulsion of Freund’s complete adjuvant in phosphatebuffered saline (PBS), by subcutaneous injection at 6-8 sites along the back. Two rabbits were immunized with each of the KLH-NSAID conjugates. Each animal received booster injections of the corresponding KLH-NSAID conjugate (250 µg suspended in 1 mL of a 50% (v/v) emulsion of Freund’s incomplete adjuvant in PBS) after 4 and 8 weeks. Peripheral blood was collected from ear veins 3 weeks after boosting and at weekly intervals thereafter. Serum was prepared and stored at -20 °C. Enzyme-Linked Immunosorbent Assay (ELISA). This was undertaken essentially as described by Kenna et al. (18). Test antigens (RSA-drug conjugates or unconjugated RSA at 15 µg/mL in PBS, 0.1 mL/well) were incubated overnight in 96well microtiter plates, at 4 °C. Plates were washed in wash buffer (154 mM NaCl, 10 mM Tris-HCl, pH 7.6, 0.5% (w/v) casein, 0.02% (w/v) Thimerosal) using four cycles of an automated plate washer (S8/12, Flow Laboratories, U.K.). Serial dilutions of antisera were prepared in PBS and added to plates (0.1 mL/well), which were incubated for 3 h at rt. Plates were washed as before and incubated for 2 h with 0.1 mL/well horseradish peroxidase-conjugated goat anti-rabbit IgG (1:1000 dilution in PBS). Plates were then washed with wash buffer followed by PBS. Peroxidase activity was determined using the colorimetric substrate o-phenylenediamine (19). Dosing of Mice. Female CD1 mice (20-25 g; Charles River, U.K.) were given diclofenac (100, 200, or 300 mg/kg in H2O), ibuprofen (50, 100, or 200 mg/kg in H2O), or sulindac (25, 50, or 100 mg/kg in 10% DMSO/tricaprylin) by ip injection (10 µL/g of body weight, n ) 4 or 5/dose group). Control mice received ip injections of equivalent volumes of the appropriate vehicle (H2O or 10% DMSO/tricaprylin). Six hours after dosing, the animals were sacrificed by cervical dislocation and the livers were removed. Subcellular Fractionation of Mouse Livers. The following steps were carried out on ice, using ice cold buffers. Freshly isolated livers from each dose group were pooled, weighed, and minced in five volumes of sucrose buffer (0.25 M sucrose, 15 mM Tris-HCl, 0.1 mM EDTA, pH 6.8), then homogenized (Potter-Elvehjem homogenizer, 1000 rpm, 10 strokes), and filtered through two layers of muslin. In initial studies, combined nuclear/mitochondrial (10000g pellet, 16 min), microsomal (100000g pellet, 75 min), and cytosolic (100000g supernatant, 75 min) fractions were prepared by differential centrifugation (see Figure 4). In subsequent studies, nuclear (1000g pellet, 10 min), mitochondrial (10000g pellet, 16 min), microsomal (100000g pellet, 75 min), and cytosolic (100000g supernatant, 75 min) fractions were prepared by sequential centrifugation of the crude whole homogenate as described by Touster et al. (20). Each pellet was washed twice, by resuspending in buffer using a Dounce homogenizer (loose fitting

548 Chem. Res. Toxicol., Vol. 10, No. 5, 1997

Wade et al.

Figure 2. Recognition of NSAID-modified RSA conjugates by antisera raised against KLH-NSAID conjugates. Serial dilutions of antisera from rabbits immunized with KLH-diclofenac (panel A), KLH-sulindac (panel B), and KLH-ibuprofen (panel C) were tested by ELISA for the presence of antibodies which recognized RSA-diclofenac (]), RSA-sulindac (^), RSA-ibuprofen (O), or RSA alone (x), which had been been immobilized on wells of microtiter plates. pestle) followed by recentrifugation. At each step, the supernatants decanted from the “crude” pellets and those obtained from the subsequent wash steps were combined and then used in the next centrifugation step. The pelleted subcellular fractions were resuspended in one volume of sucrose buffer and stored at -70 °C. The nuclear fraction was further separated by Percoll density gradient centrifugation, as described by Prpic’ et al. (21). Freshly prepared nuclear fractions were diluted to 6% (w/v) in sucrose buffer and then mixed with Percoll (final Percoll concentration 11.9%). After centrifugation at 30000g for 45 min, three bands of material were evident, and these were harvested using Pasteur pipets. Percoll was removed by diluting the fractions with sucrose buffer and then resedimenting the subcellular components by centrifugation at 10000g for 20 min. The final pellets were resuspended in one volume of fresh sucrose buffer and stored at -70 °C. SDS-PAGE and Immunoblotting. Protein samples (2-4 mg of protein/mL) were solubilized by boiling for 2 min in buffer comprising 8% (w/v) SDS, 20% (v/v) glycerol, 0.002% (w/v) bromophenol blue, 125 mM Tris-HCl (pH 6.8), and 40 mM dithiothreitol. SDS-PAGE was carried out using a minislab gel apparatus (2050 midget electrophoresis unit, Hoeffer Instruments, San Francisco, CA) with 3% stacking/8% resolving gels, using the discontinuous buffer system described by Laemmli (22). Gels were prerun for 20 min at 30 mA/minigel before loading samples; then samples were loaded (normally at 50 µg/ well) and run at 30 mA/minigel, at 15 °C, for approximately 75 min. Electrophoretic transfer of resolved proteins to nitrocellulose was carried out at 4 °C, using transfer buffer which contained 15.7 mM Tris, 120 mM glycine (pH 8.3), and 20% (v/ v) methanol, for 1 h at 200 V using a 2051 multiblot electrophoretic transfer unit from Pharmacia. Nitrocellulose was either stained for protein for 2 min using 0.1% amido black 10B in 45% (v/v) methanol, 10% (v/v) acetic acid and then destained using 70% (v/v) methanol, 2% (v/v) acetic acid or used for antibody development. Prior to antibody development, the nitrocellulose was blocked for 18 h at 4 °C by incubation in 154 mM NaCl, 10 mM TrisHCl (pH 7.6), 2.5% (w/v) casein, and 0.02% (w/v) Thimerosal. Subsequent steps were carried out at rt with continuous shaking. The blocked nitrocellulose was incubated for 3 h with primary antibody diluted in wash buffer. Unbound antibodies were removed by washing for 5 min in wash buffer containing 0.5% (v/v) Triton X-100 and 0.1% (w/v) SDS followed by two 10 min washes with wash buffer alone. The nitrocellulose was then incubated for 2 h with horseradish peroxidase-conjugated or alkaline phosphatase-conjugated secondary antibody diluted in wash buffer (see figure legends for dilutions). Blots were finally washed with wash buffer (four 10 min washes) followed by 0.1% (w/v) Tween 20 in 154 mM NaCl, 50 mM Tris-HCl (pH 7.6) (four 10 min washes). In initial studies using alkaline phosphataseconjugated goat anti-rabbit IgG, blots were developed using

Western Blue stabilized substrate (Promega Corp., Madison, WI; see Figures 4 and 6). In subsequent studies, horseradish peroxidase-conjugated secondary antibodies were used and bound antibodies were visualized by ECL, using ECL Western blotting reagent from Amersham (see Figures 5 and 10) or SuperSignal from Pierce & Warriner (see Figures 7 and 8). Densitometric scanning was undertaken using a CS-930 dual wavelength scanner coupled to a DR-2 data recorder, from Shimadzu (Kyoto, Japan). Other Methods. Protein determinations were carried out using the BCA protein assay reagent kit (Pierce). Alkaline phosphatase activity was determined using a Sigma diagnostic kit (code no. 104-LS) as described by the manufacturers. NADPH cytochrome c reductase activity was measured spectrophotometrically, according to the method of Williams et al. (23).

Results Characterization of Rabbit Antisera. Polyclonal antisera were raised by immunizing rabbits with conjugates prepared by coupling diclofenac, sulindac, or ibuprofen to KLH (2 animals/KLH drug conjugate) via a twostep carbodiimide reaction (see Experimental Procedures). These were shown by ELISA to contain antibodies which recognized conjugates prepared by coupling the respective NSAIDs to the carrier protein RSA (Figure 2). The antisera exhibited varying degrees of cross-reactivity with the other NSAID-RSA test antigens, but none recognized RSA alone (Figure 2). Both of the anti(KLH-ibuprofen) antisera exhibited a markedly greater degree of recognition of RSA-ibuprofen than RSA-diclofenac or RSAsulindac (Figure 2C). In contrast, antibodies in the sera from rabbits immunized with either KLH-diclofenac or KLH-sulindac bound to the three RSA-NSAID conjugates to similar extents (Figure 2A,B). Results obtained with only one antiserum from each pair of immunized rabbits are shown in Figure 2. The KLH conjugates were very poorly soluble in aqueous soultion, which precluded determination of their hapten densities. To estimate the hapten densities of the RSA-diclofenac and RSA-ibuprofen conjugates, the numbers of free amino groups were determined using TNBS (see Experimental Procedures). The values obtained were 49.6 and 49, respectively, whereas unconjugated RSA contained 60 free amino groups (which is consistent with the known lysine content of the protein). Thus it can be estimated that the approximate hapten densities were 10.4 mol of diclofenac/mol of RSA and 11

NSAID-Modified Hepatic Proteins

Chem. Res. Toxicol., Vol. 10, No. 5, 1997 549

Figure 3. ELISA inhibition studies. Diluted antisera were incubated overnight with various compounds and then analyzed by ELISA for recognition of RSA-NSAID conjugates. The compounds tested were RSA-diclofenac (]), RSA-sulindac (^), RSA-ibuprofen (O), RSA (x), diclofenac (open plus), sulindac (*), and ibuprofen (+). The concentrations of the various RSA-NSAID conjugates were adjusted to take account of their deduced hapten densities (10-11 mol of NSAID/mol of protein; see Results) and were between 1 mg of protein/mL and 1 ng of protein/mL, and equivalent concentrations of RSA were used. Panel A: Recognition of RSA-diclofenac by anti(KLH-diclofenac) antiserum (1:2000 dilution). Panel B: Recognition of RSA-sulindac by anti(KLH-sulindac) antiserum (1: 2000 dilution). Panel C: Recognition of RSA-ibuprofen by anti(KLH-ibuprofen) antiserum (1:10000 dilution).

mol of ibuprofen/mol of RSA. The TNBS assay could not be applied to the sulindac conjugate because the drug interfered with the assay. However, spectral studies (see Experimental Procedures) indicated that the hapten density of RSA-sulindac was also 11 mol of NSAID/mol of RSA. The abilities of the RSA-NSAID conjugates, unconjugated RSA, and unconjugated NSAIDs to inhibit recognition by the antisera of the respective RSA conjugates were analyzed. In no case was antibody binding inhibited by unconjugated RSA, which demonstrates that the antisera did not recognize the carrier protein alone. As expected from the ELISA binding studies (Figure 2), recognition of RSA-diclofenac by the anti(KLH-diclofenac) antiserum was inhibited equipotently by RSAdiclofenac, RSA-sulindac, and RSA-ibuprofen, implying recognition of an immunochemically cross-reactive epitope (Figure 3A). Interestingly, this antibody reactivity was not inhibited by either of the three unconjugated drugs (Figure 3A). This demonstrates that the affinity of interaction between the antibodies and the free drugs is lower than the affinity of interaction with the RSA conjugates. This could mean that an important component of the epitope recognized is the amide linkage between the NSAIDs and RSA and/or portions of the lysine side chain on the protein carrier. Alternatively, perhaps the multivalent nature of the RSA conjugates greatly amplifies a relatively low-affinity interaction between the antibodies and the NSAIDs. Recognition of RSA-sulindac by the anti(KLH-sulindac) antiserum was also inhibited far more efficiently by RSA-sulindac than by unconjugated sulindac (Figure 3B). However, binding of this antibody was not inhibited by RSA-diclofenac or RSA-sulindac, which is surprising in view of the extensive recognition of these conjugates observed in the ELISA binding studies (Figure 2B). Similarly, binding of anti(KLH-ibuprofen) antiserum to RSA-ibuprofen was inhibited efficiently by RSA-ibuprofen but not by RSA-diclofenac or RSA-ibuprofen (Figure 3C). Presumably the anti(KLH-sulindac) and anti(KLHibuprofen) antisera each contain more than one popula-

tion of antibodies to NSAID adducts, having different affinities and specificities. Recognition of RSA-ibuprofen by anti(KLH-ibuprofen) antiserum was inhibited very efficiently by unconjugated ibuprofen but not by the other unconjugated drugs (Figure 3B), implying that this antiserum contains antibodies that recognize the drug with both high affinity and a high degree of specificity. Detection of NSAID-Modified Mouse Hepatic Proteins. Initial immunoblotting studies were undertaken using crude liver subcellular fractions (10000g and 100000g pellets and the 100000g supernatants). These revealed that the rabbit antisera recognized several novel polypeptide antigens which were expressed in a dosedependent manner in livers of mice given diclofenac, sulindac, or ibuprofen but not in livers from control animals (Figure 4). The novel antigens were detected in the combined nuclear/mitochondrial fraction (10000g pellet) but not in the other subcellular fractions. Antisera from rabbits immunized with KLH-diclofenac or KLHsulindac recognized major drug-induced 110 kDa polypeptide antigens, which were expressed in livers of mice treated with diclofenac and sulindac, respectively (Figure 4A,B). In addition, these antisera recognized lower levels of drug-induced 140 kDa antigens which were expressed in livers from diclofenac- and sulindac-treated mice and of a drug-induced 200 kDa antigen that was detected in livers from diclofenac-treated mice (Figure 4A,B). In contrast, antiserum to KLH-ibuprofen recognized a 60 kDa drug-induced antigen which was expressed in livers of mice treated with ibuprofen (Figure 4C). It should be noted that all polyacrylamide gels were prerun before loading of samples (see Experimental Procedures). Initial experiments established that this was essential for consistent and reproducible detection of the 110 kDa diclofenac-induced antigen. The distribution of the NSAID-induced polypeptide antigens was further investigated using nuclear, mitochondrial, microsomal, and cytosolic fractions which were prepared by quantitative subcellular fractionation of mouse liver homogenates. All of the novel antigens were concentrated in the nuclear fraction (i.e., the low-speed pellet) (Figure 5A-C). These studies were undertaken

550 Chem. Res. Toxicol., Vol. 10, No. 5, 1997

Wade et al.

Figure 4. Detection by immunoblotting of novel protein adducts in livers of mice treated with NSAIDs. Panel A: 10000g pellets from livers of mice given diclofenac (at 0, 100, 200, and 300 mg/kg), probed using anti(KLH-diclofenac) antiserum (1:5000 dilution). Panel B: 10000g pellets from livers of mice given sulindac (0, 25, 50, and 100 mg/kg), probed using anti(KLH-sulindac) antiserum (1:2500 dilution). Panel C: 10000g pellets from livers of mice given ibuprofen (0, 50, 100, and 200 mg/kg), probed using anti(KLHibuprofen) antiserum (1:1000 dilution). Bound antibodies were detected using an alkaline phosphatase-conjugated secondary antibody (1:2000) and visualized using Western Blue.

Figure 5. Subcellular distribution of NSAID adducts in mouse liver. Whole homogenates (Hom), nuclear (Nuc), mitochondrial (Mit), microsomal (Mic), and cytosolic (Cyt) fractions were prepared from livers of mice treated ip with NSAIDs (T) or an equivalent dose of vehicle alone (C). Panels A-C: Detection of NSAID adducts. (A) Fractions from diclofenac-treated mice (200 mg/kg), probed with anti(KLH-diclofenac) antiserum (1:10000 dilution). (B) Fractions from sulindac-treated mice (100 mg/kg), probed with anti(KLHsulindac) antiserum (1:10000 dilution). (C) Fractions from ibuprofen-treated mice (200 mg/kg), probed with anti(KLH-ibuprofen) antiserum (1:5000 dilution). Bound antibodies were detected using a horseradish peroxidase-conjugated secondary antibody (1: 10000) followed by ECL using Western blotting reagent from Amersham. Panels D, E: Activities of alkaline phosphatase (D) and NADPH cytochrome c reductase (E) in the mouse liver subcellular fractions. Panel F: Immunoblotting analysis of the distibution of the transferrin receptor. The primary antibody was a monoclonal rat anti-mouse CD71 antibody (1:500 dilution), and bound antibody was detected using a horseradish peroxidase-conjugated goat anti-rat IgG (1:2000 dilution) followed by ECL using Western blotting reagent from Amersham.

using a more sensitive immunochemical detection system (ECL), which revealed that the anti(KLH-ibuprofen) antiserum recognized comparatively low levels of a 110 kDa ibuprofen-induced antigen, which also was expressed in the nuclear fraction (Figure 5C). Inhibition studies verified that the novel hepatic polypeptide antigens recognized by the various antisera in immunoblots were NSAID-modified protein adducts. Recognition of the diclofenac-induced antigens by anti(KLH-diclofenac) antiserum was inhibited by RSAdiclofenac but not by RSA, in a concentration-dependent manner over the range 1-100 µg/mL (Figure 6A). Furthermore, recognition of these antigens was inhibited

very efficiently by very low concentrations (100 nM) of unconjugated diclofenac (Figure 6A). Similarly, the respective RSA-NSAID conjugates and free drugs inhibited recognition of the ibuprofen- and sulindac-induced antigens by anti(KLH-ibuprofen) and anti(KLH-sulindac) antisera, respectively (Figure 6B,C). The ability of the unconjugated drugs to inhibit recognition of the hepatic adducts by the antibodies was surprising, in view of the results obtained when synthetic RSA-NSAID conjugates were used as test antigens in ELISAs (Figures 2 and 3), and so further specificity studies were undertaken. The anti(KLH-diclofenac) antiserum did not recognize the NSAID-derived adducts

NSAID-Modified Hepatic Proteins

Figure 6. Recognition of NSAID-modified hepatic protein antigens by anti(KLH-NSAID) antisera is inhibited by the respective NSAID-RSA conjugates and by the free drugs. Diluted antisera were preincubated overnight with either RSANSAID conjugates, RSA alone, or free NSAIDs and then tested by immunoblotting for recognition of antigens expressed in nuclear fractions from NSAID-treated (T) or control (C) mice. Panel A: Fractions from diclofenac-treated mice (200 mg/kg), probed with anti(KLH-diclofenac) antiserum (1:2500). Panel B: Fractions from sulindac-treated mice (100 mg/kg), probed with anti(KLH-sulindac) antiserum (1:2000). Panel C: Fractions from ibuprofen-treated mice (200 mg/kg), probed with anti(KLH-ibuprofen) antiserum (1:2000). Bound antibodies were detected using alkaline phosphatase-conjugated secondary antibody (1:2000) and visualized using Western Blue.

expressed in livers of mice treated with sulindac or ibuprofen (Figure 7A). Similarly, the anti(KLH-sulindac) antiserum did not recognize adducts expressed in livers of mice given diclofenac or ibuprofen (Figure 7B), and the anti(KLH-ibuprofen) antiserum did not recognize adducts expressed in livers of mice given diclofenac or sulindac (Figure 7C). The adduct signal evident in Figure 7 (and Figure 8) is amplified markedly compared to that evident in Figure 5 (and Figure 9) because a superior ECL detection reagent, obtained from a different commercial source, was used to visualize bound antibodies (see figure legends for details). Recognition by anti(KLH-diclofenac) antiserum of adducts expressed in livers of diclofenac-treated rats was not inhibited by unconjugated ibuprofen or sulindac, at 10 µM, or by the NSAIDs tolmetin and fenoprofen but was inhibited partially by 10 µM indomethacin (Figure 8A, 51% inhibition by indomethacin, assessed by densitometry; see Figure 9 for additional chemical structures). None of the NSAIDs other than sulindac inhibited recognition of the sulindac-modified hepatic adducts by anti(KLH-sulindac) antiserum (Figure 8B). Recognition of ibupofenmodified hepatic adducts was inhibited most markedly by ibuprofen but was also reduced markedly in the presence of 10 µM concentrations of the other NSAIDS (Figure 8C, ibuprofen 96% inhibition, tolmetin 88%, indomethacin 32%, fenoprofen 82%, sulindac 53%, diclofenac 60%, assessed by densitometry). Subcellular Locations of NSAID-Modified Proteins. The subcellular distributions of the NSAID-

Chem. Res. Toxicol., Vol. 10, No. 5, 1997 551

modified proteins adducts (Figure 5A-C) were very similar to that of alkaline phosphatase (Figure 5D), which is a marker enzyme known to be localized to the bile canalicular plasma membrane (24). In contrast, both NADPH cytochrome c reductase activity, an endoplasmic reticulum marker (20), and the transferrin receptor, which is located in the sinusoidal plasma membrane (25, 26), were concentrated in the microsomal fraction (Figure 10E,F). Nuclear pellets prepared from livers of control mice and also diclofenac-, sulindac-, or ibuprofen-treated mice were further fractionated by Percoll density gradient centrifugation, which yielded four distinct subfractions. Immunoblotting studies showed that the adducts derived from both diclofenac and sulindac (110, 140, and 200 kDa) were enriched in two of the subfractions (F2 and F3) and were present at the greatest concentration in the latter (Figure 10A). The 60 kDa adduct derived from ibuprofen was also detected in the F3 subfraction, although no apparent enrichment of this adduct was achieved by the Percoll gradient centrifugation (Figure 10C). This may be contrasted with the very marked enrichment in expression of the 110 kDa ibuprofen adduct which was evident in the F3 subfraction (Figure 10C). The distribution of alkaline phosphatase activity in the Percoll subfractions was very similar to that of the 110, 140, and 200 kDa NSAID adducts, i.e., an enrichment in activity in the F2 subfractions with even greater levels present in the F3 subfractions (Figure 10D). The transferrin receptor was not detectable by immunoblotting in any of the subfractions (data not shown). NADPH-dependent cytochrome c reductase activity was enriched to a minor extent in the F1 subfraction and to a lesser extent in the F2 and F3 subfractions (Figure 10E).

Discussion The present immunochemical studies have shown that livers from mice treated ip with single doses of the widely used NSAID’s diclofenac and sulindac express similar patterns of covalently modified protein adducts and that these differ markedly from the pattern of adducts expressed in livers of mice treated with ibuprofen. Thus, the major adduct detected in livers of mice given diclofenac or sulindac exhibited an apparent molecular mass of 110 kDa, as determined by SDS PAGE, and lower levels of other adducts (most notably 140 and 200 kDa adducts) derived from these drugs were also observed. In contrast, the major adduct detected in livers of mice given ibuprofen exhibited an apparent molecular mass of 60 kDa, and in these animals only low levels of a 110 kDa adduct were evident. Although all of the NSAID adducts were concentrated in the nuclear fraction, as prepared by differential centrifugation, this fraction is known to contain large plasma membrane sheets derived from the bile canalicular domain of hepatocytes, in addition to nuclei (27). Separation of the nuclear fraction by Percoll density gradient centrifugation indicated that the 110, 140, and 200 kDa adducts derived from diclofenac and sulindac were concentrated in a subfraction derived from the canalicular region of the hepatocyte plasma membrane, as indeed was the 110 kDa ibuprofen adduct. Interestingly, no similar enrichment was observed for the 60 kDa adduct derived from ibuprofen, which raises the possibility that this adduct has a different subcellular location. In addition, whereas ELISA studies undertaken using synthetic RSA conjugates indicated marked immunochemical cross-reactivity

552 Chem. Res. Toxicol., Vol. 10, No. 5, 1997

Wade et al.

Figure 7. Lack of immunochemical cross-reactivity between the NSAID-modified hepatic protein adducts detected by immunoblotting. The anti(KLH-NSAID) antisera (1:20000 dilution) were each tested by immunoblotting for expression of adducts expressed in nuclear fractions from livers of mice treated ip with either diclofenac, ibuprofen, or sulindac, as indicated (T; see legend to Figure 5 for doses of drugs administered), or with vehicles alone (C). Bound antibodies were detected using a horseradish peroxidase-conjugated secondary antibody (1:10000) followed by ECL development with SuperSignal reagent from Pierce & Warriner.

between NSAID adducts, no such cross-reactivity between the adducts expressed in vivo in drug-treated mice could be detected. Three other groups of workers have used immunochemical approaches to identify and characterize protein adducts derived from diclofenac. Pumford and coworkers reported identification of a range of different protein adducts in liver homogenates from mice treated ip with diclofenac, the most abundant of which exhibited molecular masses of 50 and 70 kDa (11). Subsequently, immunohistochemical studies undertaken by the same group revealed that, in livers of diclofenac-treated rats, the adducts were localized to the bile canalicular plasma membrane of hepatocytes, while immuoblotting studies of the same livers indicated that the major adducts included 200, 140, and 110 kDa adducts that were concentrated in a purified plasma membrane fraction and a 50 kDa microsomal adduct (12). In contrast, studies undertaken by Kretz-Rommel and Boelsterli have reported immunochemical identification of 50, 60, 80, and 126 kDa adducts, which were expressed in cultured rat hepatocytes exposed to diclofenac in vitro, and of 60 and 80 kDa adducts expressed in vivo in livers of rats given diclofenac for 4 days, at a dose of 30 mg/kg/day (14). Furthermore, Gil and co-workers have reported detection of a major 60 kDa adduct which was generated in vitro when rat and human hepatocytes were cultured with diclofenac (15). The pattern of diclofenac adducts observed in the present study is most similar to the pattern of adducts reported by Pumford and co-workers (11, 12), although we have been unable to identify the 50 and 70 kDa adducts identified by these earlier workers. Since we used doses of diclofenac and a route of administration that corresponded to those employed in these previous studies (while the doses of ibuprofen and sulindac were adjusted to fall below the reported LD50s of the compounds (28, 29)), the similarities are reassuring and unsurprising. Perhaps we could not detect the 50 and 70 kDa adducts because the specificities of our antisera

differ from the specificity of the antisera used by Pumford and co-workers. Certainly, the discrepancies between ELISA and immunoblotting results observed in the present studies indicate that the antisera we have produced contain multiple populations of antibodies, which apparently recognize quite distinct epitopes expressed on NSAID-protein adducts. Variability between immunochemical detection methods may be another contributory factor. We have observed that detection of the 110 kDa diclofenac adduct is markedly improved by using prerun minigels (where gels are electrophoresed for 20 min, at 30 mA/gel, before loading of samples) and also by exclusion of detergent from the immunoblotting steps. This implies that the 110 kDa adduct derived from diclofenac is labile under certain electrophoretic and immunoblotting conditions. The different patterns of adducts identified by other workers (14, 15) may well reflect the different methods used to generate the adducts (i.e., in vitro vs in vivo) and/or differences between patterns of adducts formed following single vs chronic dosing. Diclofenac, sulindac, and ibuprofen are extensively glucuronidated in vivo by UGTs (30-32), and glucuronidation has been shown to mediate covalent binding of ibuprofen to plasma proteins (33) and of diclofenac to hepatic proteins in vitro (12, 13, 34). This process is likely to be responsible for formation of the diclofenac and sulindac adducts detected in the present studies, since Hargus et al. have presented evidence implicating UGT-dependent glucuronidation in formation of 110, 140, and 200 kDa diclofenac adducts in vitro in rat liver homogenates (12). Whether UGTs are responsible for formation of the 60 kDa ibuprofen adduct is less clear. The reactivity of acyl glucuronides, as assessed by their susceptibility to hydrolysis and rearrangement, varies markedly between NSAIDs and can be strongly correlated to their ability to form adducts with plasma proteins in vitro (35). It has been postulated that this reactivity is at least partly dependent on the degree of substitution at the carbon R to the carboxylic acid group

NSAID-Modified Hepatic Proteins

Chem. Res. Toxicol., Vol. 10, No. 5, 1997 553

Figure 9. Chemical structures of tolmetin, indomethacin, and fenoprofen.

Figure 8. Immunoblot inhibition studies. Antisera were preincubated overnight with the various NSAIDs indicated (each at 10 µg/mL) and then tested by immunoblotting for recognition of NSAID-modified protein adducts expressed in nuclear fractions from NSAID-treated (T; see legend to Figure 5 for doses) or control (C) mice. Panel A: Recognition of diclofenac adducts by anti(KLH-diclofenac) antiserum. Panel B: Recognition of sulindac adducts by anti(KLH-sulindac) antiserum. Panel C: Recognition of ibuprofen adducts by anti(KLH-ibuprofen) antiserum. Immunoblots were developed as described in the legend to Figure 7.

(35). This would imply that acyl glucuronides of propionic acids, such as ibuprofen, should be less reactive than those of acetic acids, such as diclofenac and sulindac (Figure 1). The influence on chemical reactivity exerted by this type of structural difference has been demonstrated by Castillo et al. (33). The acyl glucuronide of ibufenac, an analogue of ibuprofen which lacks the methyl substituent on the R carbon and was withdrawn from the market due to hepatotoxicity, was found to be more reactive and to lead to greater covalent modification of plasma proteins in vivo and in vitro than the corresponding metabolite of ibuprofen (33). Formation of reactive acyl-CoA thioesters by acyl-CoA ligases is another potential mechanism of adduct generation. Acyl-CoA intermediates have been implicated in the chiral inversion of ibuprofen, and formation of hybrid triglycerides has been well documented (36, 37). Although this pathway has not been shown to mediate covalent modification of proteins by NSAIDs, it has been

implicated in alkylation of proteins by several other xenobiotic carboxylic acids (10). Alternatively, adduct formation might involve cytochrome P450-mediated oxidation, which has been described for both diclofenac and ibuprofen (38) and which has been implicated in formation of a 50 kDa diclofenac adduct in rat liver homogenates in vitro (12). In the future it will be important to determine whether these metabolic pathways play a role in formation of any of the adducts we have detected. Whereas the enzymes which could bioactivate NSAIDs are located either in the endoplasmic reticulum (UGTs, P450s, some CoA ligases (39, 40)) or in mitochondria (other CoA ligases (39, 41)), the subcellular fractionation studies have established that the NSAID adducts are localized in (an)other subcellular compartment(s). This implies that adduct formation is a highly selective process which involves specific interactions of reactive metabolite(s) with the target proteins, as does the restricted number of NSAID-modified hepatic adducts formed. Such interactions could arise because the target proteins play specific roles in the disposition, elimination, and/or biotransformation of the reactive species. If so, then differences in one or more of these processes could be important determinants of the patterns of adducts derived from particular NSAIDs. In rats, diclofenac and sulindac glucuronides are excreted predominantly into the bile (31, 42), whereas elimination of ibuprofen conjugates occurs predominantly via the urine (32). The proteins involved in excretion of these metabolites could therefore be potential targets for covalent modification. Two proteins responsible for the transport of organic anions across the canalicular plasma membrane are the 110 kDa ATP-dependent bile acid transporter (43, 44) and the 200 kDa multispecific organic anion transporter (45, 46). Perhaps the 110 and 200 kDa diclofenac- and sulindac-modified hepatic protein adducts that we have identified, and have found to be concentrated in a membrane subfraction apparently derived from the bile canalicular plasma membrane, correspond to modified forms of these proteins. A recent immunochemical investigation of protein adduct formation in rats treated iv with the NSAIDs diflusinal and zomepirac (47) is consistent with this interpretation. Both these drugs were shown to produce major 110, 140, and 200 kDa hepatic protein adducts in vivo, and like diclofenac and sulindac, in rats they form acyl glucuronides which are excreted into the bile (47). However, Hargus et al. have recently reported purification of a 110 kDa adduct from livers of diclofenac-treated rats (48). This was found, by amino acid sequence analysis,

554 Chem. Res. Toxicol., Vol. 10, No. 5, 1997

Wade et al.

Figure 10. Distribution of NSAID adducts in nuclear subfractions prepared by Percoll density gradient centrifugation. Subfractions (F1-4) were prepared by Percoll gradient centrifugation from nuclear (Nuc) fractions obtained from livers of control (C) mice and also from livers of mice treated (T) with NSAIDs. Panels A-C: Detection by immunoblotting of NSAID adducts expressed in fractions from mice treated with diclofenac (200 mg/kg) (A), sulindac (100 mg/kg) (B), or ibuprofen (200 mg/kg) (C). Experimental conditions were as described in the legend to Figure 4. Sample loading was 25 µg of protein/lane. Panels D, E: Activities of alkaline phosphatase (D) and NADPH cytochrome c reductase (E) in the fractions.

to correspond to the abundant bile canalicular plasma membrane protein dipeptidyl-peptidase IV (DPP IV). Moreover, the activity of this enzyme was shown to be reduced by 22% in livers of rats sacrificed 16 h after ip administration of a single dose of diclofenac (200 mg/kg) when compared to the the activity of DPP IV in control animals. In view of these findings, it was suggested that inhibition of the activity of this enzyme could play a role in the hepatotoxicity of diclofenac. In the future, it will be important to determine whether modified forms of DPP IV correspond to major hepatic protein adducts derived from other hepatotoxic NSAIDs and to investigate whether the 110 kDa adducts we and others have identified could also include other, much less abundant target proteins, such as the ATP-dependent bile acid transporter (43, 44). It is interesting to note that diclofenac and sulindac have been reported to cause hepatotoxicity in humans (4, 6), as has diflusinal (1). Moreover, zomepirac was associated with abnormal levels of serum transaminases, suggesting possible hepatotoxicity, before it was withdrawn from the market due to severe anaphylaxis (1), whereas ibuprofen is considered to be nonhepatotoxic. This suggests that the hepatotoxicities of certain NSAIDs could be related to formation of particular patterns of protein adducts. If the major adducts derived from diclofenac indeed include biliary transporters, one possible mechanism of toxicity might be interference with bile formation. This could explain, at least in part, the cholestatic liver injury observed in certain patients treated with NSAIDs (1, 2). It is also possible that immune processes play an important toxicological role, as has been demonstrated for several other groups of drugs (49). In support of this hypothesis, recent studies

have shown in vitro cytotoxic killing of mouse hepatocytes bearing diclofenac adducts by splenocytes from mice treated with the drug in vivo (50). Once the nature of the NSAID metabolite-modified target proteins has been determined, it should be possible to investigate these and other possibilities. In practice, since the clinical features and patterns of liver injury observed in patients treated with these drugs are complex (1-7), multiple processes are likely to be involved.

Acknowledgment. This work was supported by the Wellcome Trust.

References (1) Katz, L. M., and Love, P. Y. (1992) NSAIDs and the liver. In Therapeutic applications of NSAIDs: Subpopulations and new formulations (Famaey, J. P., and Paulus, H. E., Eds.) pp 247263, Marcel Dekker, New York. (2) Zimmerman, H. J. (1994) Hepatic injury associated with nonsteroidal anti-infammatory drugs. In Nonsteroidal antiinflammatory drugs. Mechanisms and clinical use (Lewis, A. J., and Furst, D. E., Eds.) pp 171-194, Marcel Dekker Inc., New York. (3) Lewis, J. H. (1984) Hepatic toxicity of nonsteroidal anti-inflammatory drugs. Clin. Pharmacol. 3,128-135. (4) Tarzi, E. M., Harter, J. G., Zimmerman, H. J., Ishak, K. G., and Eaton, R. A. (1993) Sulindac-associated hepatic injury: Analysis of 91 cases reported to the Food and Drug Administration. Gastroenterology 104, 569-574. (5) Scully, L. J., Clarke, D., and Barr, R. J. (1993) Diclofenac induced hepatitis. 3 Cases with features of autoimmune chronic active hepatitis. Digest. Dis. Sci. 38, 744-751. (6) Banks, A. T., Zimmerman, H. J., Ishak, K. G., and Harter, J. G. (1995) Diclofenac-associated hepatotoxicity: Analysis of 180 cases reported to the food and drug administration as adverse reactions. Hepatology 22, 820-827. (7) Boelsterli, U. A., Zimmerman, H. J., and Kretz-Rommel, A. (1995) Idiosyncratic liver toxicity of nonsteroidal antiinflammatory drugs: Molecular mechanisms and pathology. Crit. Rev. Toxicol. 25, 207-235.

NSAID-Modified Hepatic Proteins (8) Spahn-Langguth, H., and Benet, L. Z. (1992) Acyl glucuronides revisited: Is the glucuronidation process a toxification as well as a detoxification mechanism. Drug Metab. Rev. 24, 5-47. (9) Lecoeur, S., Bonierbale, E., Challine, D., Gautier, J.-G., Valadon, P., Dansette, P. M., Catinot, R., Ballet, F., Mansuy, D., and Beaune, P. H. (1994) Specificity of in vitro covalent binding of tienilic acid metabolites to human liver microsomes in relationship to the type of hepatotoxicity: Comparison with two directly hepatotoxic drugs. Chem. Res. Toxicol. 7, 434-442. (10) Hertz, R., and Bar-Tana, J. (1988) The acylation of proteins by xenobiotic amphipathic carboxylic acids in cultured rat hepatocytes. Biochem. J. 254, 39-44. (11) Pumford, N. R., Myers, T. G., Davila, J. C., Highet, R. J., and Pohl, L. R. (1993) Immunochemical detection of liver protein adducts of the nonsteroidal anti-inflammatory drug diclofenac. Chem. Res. Toxicol. 6, 147-150. (12) Hargus, S. J., Amouzedeh, H. R., Pumford, N. R., Myers, T. G., McCoy, S. C., and Pohl, L. R. (1994) Metabolic activation and immunochemical localization of liver-protein adducts of the nonsteroidal anti-inflammatory drug diclofenac. Chem. Res. Toxicol. 7, 575-582. (13) Kretz-Rommel, A., and Boelsterli, U. A. (1993) Diclofenac covalent protein binding is dependent on acyl glucuronide formation and is inversely related to P450-mediated acute cell injury in cultured rat hepatocytes. Toxicol. Appl. Pharmacol. 120, 155-161. (14) Kretz-Rommel, A., and Boelsterli, U. A. (1993) Selective protein adducts to membrane proteins in cultured rat hepatocytes exposed to diclofenac: radiochemical and immunochemical analysis. Mol. Pharmacol. 45, 237-244. (15) Gil, M. L., Ramirez, M. C., Terenico, M. C., and Castell, J. V. (1995) Immunochemical detection of protein adducts in cultured human hepatocytes exposed to diclofenac. Biochim. Biophys. Acta 1272, 140-146. (16) (1994) Relative safety of oral non-aspirin NSAIDs. Curr. Prob. Pharmacovigilance 20, 9-11. (17) Habeeb, A. F. S. A. (1966) Determination of free amino groups in proteins by trinitrobenzene sulfonic acid. Anal. Biochem. 14, 328336. (18) Kenna, J. G., Neuberger, J., and Williams, R. (1984) An enzymelinked immunosorbent assay for detection of antibodies against halothane altered hepatocyte antigens. J. Immunol. Methods 75, 3-14. (19) Voller, A., Bidwell, D. E., and Bartlett, A. (1979) The enzymelinked immunosorbent assay (ELISA). Dynatech Europe, U.K. (20) Touster, O., Aronson, N. N., Dulaney, J. T., and Hendrickson, H. (1970) Isolation of rat liver plasma membranes. Use of nucleotide pyrophosphatase and phosphodiesterase I as marker enzymes. J. Cell Biol. 47, 604-618. (21) Prpic, V., Green, K. C., Blackmore, P. F., and Exton, J. H. (1984) Vassopressin-, angiotensin II-, and R1-adrenergic-induced inhibition of Ca2+ transport by rat liver plasma membrane vesicles. J. Biol. Chem. 259, 1382-1385. (22) LaemmLi, U. K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680685. (23) Williams, C. H., and Kamin, H. (1962) Microsomal triphosphopyridine nucleotide-cytochrome c reductase of liver. J. Biol. Chem. 237, 587-595. (24) Ali, N., Aligue, R., and Evans, W. H. (1990) Highly purified bile canalicular vesicles and lateral plasma membranes isolated from rat liver on Nycodenz gradients. Biochemical and immunolocalization studies. Biochem. J. 271, 185-192. (25) Sciot, R., Verhoevon, G., Van Eyken, P., Cailleau, J., and Desmet, V. J. (1990) Transferrin receptor expression in rat liver: Immunohistochemical and biochemical analysis of the effect of age and iron storage. Hepatology 11, 416-427. (26) Levine, D. S., and Woods, J. W. (1990) Immunolocalization of transferrin and transferrin receptor in mouse small intestinal absorptive cell. J. Histochem. Cytochem. 38 (6), 851-858. (27) Evans, W. H. (1980) A biochemical dissection of the functional polarity of the plasma membrane of the hepatocyte. Biochim. Biophys. Acta 604, 27-64. (28) Davies, E. F., and Avery, G. S. (1971) Ibuprofen: A review of its pharmacological properties and therapeutic efficacy in rheumatic disorders. Drugs 2, 416-446. (29) Lenga, R. E., (1988) The Sigma-Aldrich Library of Chemical Safety Data, p 3208, Sigma-Aldrich Corp. (30) Riess, W., Stielin, H., Degen, P., Faigle, W., Gerardin, A., Moppert, J., Sallman, A., Schmid, K., Schweizer, A., Sulc, M., Theobald,

Chem. Res. Toxicol., Vol. 10, No. 5, 1997 555

(31)

(32) (33)

(34)

(35)

(36) (37) (38) (39)

(40)

(41) (42)

(43) (44)

(45)

(46)

(47) (48)

(49)

(50)

W., and Wagner, J. (1978) Pharmacokinetics and metabolism of the anti-inflammatory agent Voltaren. Scand. J. Rhuematol. Suppl. 22,17-29. Hucker, H. B., Stauffer, S. C., White, S. D., Rhodes, R. E., Arison, B. H., Umbenhauer, E. R., Bower, R. J., and McMahon, F. G. (1973) Physiological disposition and metabolic fate of a new antiinflammatory agent, cis-5-fluoro-2-methyl-1-[p-(methylsulfinyl)benzylidenyl]-indene-3-acetic acid in the rat, dog, Rhesus monkey, and man. Drug Metab. Dispos. 1, 721-736. Mills, R. F. N., Adams, S. S., Cliffe, E. E., Dickinson, W., and Nicholson, J. S. (1973) The metabolism of ibuprofen. Xenobiotica 3, 589-598. Castillo, M., and Smith, P. C. (1995) Disposition and reactivity of ibuprofen and ibufenac acyl glucuronides in vivo in the Rhesus monkey and in vitro with human serum albumin. Drug Metab. Dispos. 23, 566-572. Kretz-Rommel, A., and Boelsterli, U. A. (1994) Mechanism of covalent adduct formation of diclofenac in rat hepatic microsomal proteins. Retention of the glucuronic acid moiety in the adduct. Drug Metab. Dispos. 22, 956-961. Benet, L. Z., Spahn-Langguth, H., Iwakawa, S., Volland, C., Mizuma, T., Mayer, S., Mutschler, E., and Lin, E. T. (1993) Predictability of the covalent binding of acidic drugs in man. Life Sci. 53, 141-146. Hutt, A. J., and Caldwell, J. (1983) The metabolic chiral inversion of 2-arylproprionic acidssa novel route with pharmacological consequences. J. Pharm. Pharmacol. 35, 693-704. Caldwell, J., and Marsh, M. V. (1983) Interrelationships between xenobiotic metabolism and lipid biosynthesis. Biochem. Pharmacol. 32, 1667-1672. Leeman, T., Transon, C., and Dayer, P. (1993) Cytochrome P450TB (CYP2C): A major monoxygenase catalyzing diclofenac 4'-hydroxylation in human liver. Life Sci. 52, 29-34. Meijer, D. K. F. (1989) Transport and metabolism in the hepatobiliary system. In Handbook of Physiology. The gastrointestinal system (Schulz, S. G., Forte, J. G., and Rauner, B. B., Eds.) pp 717-758, Oxford University Press, New York. Knights, K. M., and Jones, M. E. (1992) Inhibition kinetics of hepatic microsomal long chain fatty acid-CoA ligase by 2-aryl proprionic acid non-steroidal anti-inflammatory drugs. Biochem. Pharmacol. 43, 1465-1471. Tracey, T. S., Wirthwein, D. P., and Hall, S. D. (1993) Metabolic inversion of (R)-ibuprofen. Formation of ibuprofenyl-Coenzyme A. Drug Metab. Dispos. 21, 114-120. Stierlen, H., and Faigle, J. W. (1979) Biotransformation of diclofenac sodium (Voltaren) in animals and man. II Quantitative determination of the unchanged drug and principal phenolic metabolites in urine and bile. Xenobiotica 9, 611-621. Sippel, C. J., Suchy, F. J., Ananthanarayanan, M., and Perimutter, D. H. (1993) The rat liver ecto-ATPase is also a canalicular bile acid transport protein. J. Biol. Chem. 268, 2083-2091. Ruetz, S., Fricker, G., Hugentobler, G., Winterhalter, K., Kurz, G., and Meier, P. J. (1987) Isolation and characterisation of the putative canalicular bile salt transport system of rat liver. J. Biol. Chem. 262, 11324-11330. Oude Elferink, R. P. J., Meijer, D. K. F., Kuipers, F., Jansen, P. L. M., Groen, A. K., and Groothuis, G. M. M. (1995) Hepatobiliary secretion of organic compounds; molecular mechanisms of membrane transport. Biochim. Biophys. Acta 1241, 215-268. Paulusma, C. C., Bosma, P. J., Zaman, G. J. R., Bakker, C. T. M., Otter, M., Scheffer, G. L., Scheper, R. J., Borst, P., and Oude Elferink, R. P. J. (1996) Congenital jaundice in rats with a mutation in a multidrug resistence-associated protein gene. Science 271, 1126-1128. Bailey, M. J., and Dickinson, R. G. (1996) Chemical and immunochemical comparison of protein adduct formation of carboxylate drugs in rat liver and plasma. Chem. Res. Toxicol. 9, 659-666. Hargus, S. J., Martin, B. M., George, J. W., and Pohl, L. R. (1995) Covalent modification of rat liver dipeptidyl-peptidase IV (CD26) by the nonsteroidal anti-inflammatory drug diclofenac. Chem. Res. Toxicol. 8, 993-996. Boelsterli, U. A. (1993) Specific targets of covalent drug-protein interactions in hepatocytes and their toxicological significance in drug-induced liver injury. Drug Metab. Rev. 25 (4), 395-451. Kretz-Rommel, A., and Boelsterli, U. A. (1995) Cytotoxic activity of T-cells and non-T cells from diclofenac-immunized mice against cultured syngeneic hepatocytes exposed to diclofenac. Hepatology 22, 213-222.

TX960153T