Metabolic Activation and Immunochemical Localization of Liver

Benedetta C. Sallustio, Yvette C. DeGraaf, Josephine S. Weekley, and Philip C. ... Mark P. Grillo, Fengmei Hua, Charles G. Knutson, Joseph A. Ware, an...
0 downloads 0 Views 2MB Size
Chem. Res. Toxicol. 1994, 7, 575-582

575

Metabolic Activation and Immunochemical Localization of Liver Protein Adducts of the Nonsteroidal Anti-inflammatory Drug Diclofenac Sally J. Hargus,*vt**Hamid R. Amouzedeh,? Neil R. Pumford,tv§ Timothy G. Myers,tJl Stacie C. McCoy,+and Lance R. Pohlt Laboratory of Chemical Pharmacology, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Maryland 20892,and Department of Anesthesia, Georgetown University School of Medicine, Washington, D.C. 20007 Received March 15,1994'

Diclofenac is a nonsteroidal anti-inflammatory agent that is reported to cause serious hepatic injury in some patients. To investigate the possibility that protein adducts derived from reactive intermediates of diclofenac might be responsible for the hepatotoxicity produced by this drug, we recently developed polyclonal antisera that recognized protein adducts of diclofenac. In the present study, we have characterized further the diclofenac adducts in rat liver. Immunoblotting studies showed that diclofenac-labeled hepatic proteins were formed in a dose- and timedependent manner in rats given diclofenac. Subcellular fractionation of liver homogenates from diclofenac-treated rats showed that a BO-kDa microsomal protein and 110-, 140-, and 200-kDa plasma membrane proteins were labeled preferentially. Immunofluorescence studies of isolated hepatocytes and immunohistochemical analysis of liver slices from diclofenac-treated mice and rats confirmed that plasma membrane proteins were labeled by diclofenac metabolites and showed that the bile canalicular domain of the plasma membrane was a major site of diclofenac adduct formation. Additionally, we found that cytochrome P-450 and UDPglucuronosyltransferase, but not acyl-CoA synthase, catalyzed the formation of reactive intermediates of diclofenac that were bound covalently to proteins in vitro. The metabolites catalyzed by cytochrome P-450 in vitro were bound exclusively to a 50-kDa microsomal protein, even in the presence of albumin. In contrast, the 110-, 140-, and 200-kDa plasma membrane proteins as well as others appeared to be labeled when diclofenac was activated by UDPglucuronosyltransferase. Chemical stability studies of the adducts formed in vivo showed that they were unstable under basic conditions and suggested that the metabolites were bound to cellular proteins as ester linkages. These results demonstrated that plasma membrane proteins are major targets of diclofenac metabolites in vivo, and the presence of these adducts may be important in the mechanism of diclofenac hepatotoxicity.

Introduction Diclofenac is a nonsteroidal anti-inflammatory drug (NSAID)' that is prescribed for patients undergoing treatment for osteoarthritisor rheumatoid arthritis. About 15% of patients taking diclofenac experience an elevation in serum alanine aminotransferase levels, and a small number of patients have developed a severe form of hepatotoxicity that is fatal in some individuals (1-11). Although the mechanism of the severe form of diclofenac

* Correspondence and reprint requests should be addressed to this author at the Laboratory of Chemical Pharmacology, NHLBI, NIH, Building 10, Room 8N104, Bethesda, MD 20892. Telephone: 301-4964841; FAX 301-402-0171. t National Institutes of Health. t Georgetown University School of Medicine. Present address: Department of Pharmacology and Toxicology, University of Arkansas, Little Rock, AR 72205. I Present address: Laboratory of Molecular Pharmacology, National Cancer Institute, Bethesda, MD 20892. 0 Abstract published in Advance ACS Abstracts, June 15, 1994. 1 Abbreviations: NSAIDs, nonsteroidalanti-inflammatory drugs; SDSPAGE,sodium dodecyl sulfate-polyacrylamide gel electrophoresis;BSA, bovine serum albumin; RSA, rabbit serum albumin; TBS, Tris-buffered saline; FCS, fetal calf serum; BSS, balanced salt solution; DMEM, Dulbecco's modified Eagle's medium; PBS, phosphate-buffered saline; UGT, uridine diphosphate glucuronosyltransferase; DAG, diclofenac acylglucuronide; UDPGA, uridine diphosphate glucuronic acid; EDC, 1-ethyl-3-[(dimethylamino)propy1lcarbodiimide.

hepatotoxicity is not known, clinical evidence suggests that it may be due to an immune-mediated mechanism (2-8)or to a disruption of cellular function (9-11). In both cases, covalent modification of liver proteins may play an important role in the etiology of diclofenac hepatotoxicity (12).To study this possibility,we recently have raised a polyclonal antibody against diclofenac coupled through its carboxyl group to a carrier protein. The antiserum was used to detect protein adducts of 50, 70,110, and 140 kDa in liver homogenates of mice treated with diclofenac (13).Radiolabeleddiclofenachas also been shown to bind covalently to protein when incubated with cultured primary rat hepatocytes (14). In the present study we have characterized the subcellular localization of diclofenac adducts because the presence of adducts in specific subcellular fractions can lead to insight regarding the mechanism of toxicity. For example,protein adducts located on the plasma membrane may be recognized by the immune system or, alternatively, may affect important cellular processes. We have also investigated the enzyme systems involved in the bioactivation of diclofenac to reactive intermediates that covalently modify proteins. Identification of the enzymes that catalyze the formation of reactive intermediates of diclofenac may help to explain, at least in part, the

This article not subject to U S . Copyright. Published 1994 by the American Chemical Society

576 Chem. Res. Toxicol., Vol. 7,No. 4, 1994 idiosyncratic nature of diclofenac hepatotoxicity. For example, induction of cytochrome P-450sor other enzymes t h a t metabolize diclofenac may occur in some individuals. We found that diclofenac adducts with molecular masses of 110,140and 200 kDa were concentrated in the plasma membrane while a 50-kDa adduct was localized in the microsomal fraction of t h e liver. Results of the bioactivation studies indicated that UDP-glucuronosyltransferase (UGT; EC 2.4.1.17)catalyzed the formation of diclofenac metabolites that were bound selectively t o hepatic plasma membrane proteins, while cytochrome P-450scatalyzed the formation of metabolites that were bound t o a microsomal protein.

Experimental Section Materials. Chemicals were purchased from the following commercial sources: sodium salt of diclofenac, insulin, phenol red, and hydrocortisone 21-hemisuccinatefrom Sigma (St.Louis, MO); BCA Protein Assay Reagent kit from Pierce Chemical (Rockford, IL); Percoll from Pharmacia LKB (Piscataway, NJ); collagenase (type 11)from Worthington (Freehold, NJ); goat antirabbit IgG (phycoerythrinconjugate)from Boehrhger Mannheim (Indianapolis, IN); Fluoromount G from Southern Biotechnology Associates (Birmingham, IL); goat anti-rabbit IgG (peroxidase conjugate) from TAG0 (Burlingame, CA); goat anti-rabbit IgG (alkaline phosphatase conjugate), Dulbecco’s modified Eagle’s medium (DMEM), penicillin G sodium, streptomycin sulfate, and amphotericin B from Gibco BRL (Grand Island, NY); Accurate Antibodies AS/AP Substrates from Bio/Can Scientific (Missisaugua, Ontario, Canada); Glycergel from Dako (Carpinteria, CA); type I rat tail collagen from Collaborative Biomedical Products (Bedford, MA); #1circular cover slips and histology tissue cassettes from PGC Scientifics (Gaithersburg, MD); Blotto from Advanced Biotechnologies Inc. (Columbia, MD); and enhanced chemiluminescence substrates from Amersham (Arlington Heights, IL). Diclofenac antisera and diclofenac-labeled rabbit serum albumin (RSA) were prepared as described previously (13). Immunoblotting Procedure. Proteins were separated by sodium dodecylsulfate-polyacrylamide gel electrophoresis(SDSPAGE), transferred to a nitrocellulose membrane, and immunoblotted with diclofenac antiserum according to published procedures (13),with the following modifications: unoccupied sites on nitrocellulosemembranes were blocked with 100% Blotto solution. Diclofenac antiserum and goat anti-rabbit IgG (peroxidase conjugate) were diluted 1 : l O OOO and 1:20 OOO, respectively, in 17% Blotto, and all immunoblot washes were carried out in 17% Blotto solution. Labeled proteins were detected using enhanced chemiluminescence according to the manufacturer’s instructions. Dose-Response and Time Studies. Male Sprague-Dawley rats (175-250 g, Taconic Farms, Germantown, NY) were given intraperitoneal injections of 1,10,50,100,or 200mg/kgdiclofenac (20 mg/mL stock solution dissolved in water) or 1mL of saline. Rats were killed 16 h postinjection for the dose-response study. For the time study, rats were given 200 mg/kg diclofenac and were killed 0,3,6,9, 12, or 24 h later. The livers were removed and placed in ice-cold buffer A (10 mM Tris-acetate, 250 mM sucrose, 1mM EDTA, pH 7.4). A minimum of 2 rats were used for each treatment group, and tissues from each treatment group were pooled before homogenization. Livers were minced and then homogenized in 4 volumes of ice-cold buffer A (PotterElvehjem, 6 strokes) to obtain total homogenate. Samples were frozen in liquid Nz and stored at -80 OC until they were subjected to immunoblot analysis with diclofenac antisera. Preparation of Rat Liver Subcellular Fractions. Male Sprague-Dawleyrats (175-250 g) were given diclofenac (200 mg/ kg) and killed by decapitation 16 h later. Liver homogenates were prepared as described above. Subcellular fractions were prepared using established methods (15). Briefly, centrifugation

Hargus et al. of liver homogenate at 960s (10 min) yielded a pellet that was discarded; subsequent centrifugation of the supernatant yielded the mitochondrial pellet (7OOOg, 15 min), the microsomal pellet (100000g, 1h), and the cytosolic fraction (100000g supernatant). The mitochondrial and microsomal pellets were resuspended in buffer A and centrifuged again; pellets were then resuspended in 1volume of buffer A. The plasma membrane fraction was isolated using the method of Loten and Redshaw-Loten (16). Briefly, livers from treated or control rats were removed and homogenized in 4 volumes of buffer A (Douncehomogenizerwith loose-fitting pestle, 6 strokes). The homogenates were diluted further (25 mL of buffer A/g of liver) and centrifuged (1500g, 15 min). The pellets were resuspended in buffer A (7.5 mL/g of liver) and homogenized (Dounce, 3 strokes), and then Percoll solution (0.13 mL/mL of homogenate) and sucrose solution (2 M, 0.01 mL/mL of homogenate) were added. After centrifugation (35OOOg, 20 min), the cloudy layers containing the plasma membrane fraction were collected and diluted with buffer A (10 mL/g of liver); centrifugation (1500g, 15 min) of the diluted solution yielded plasma membrane pellets that were collected in a minimum volume of buffer, yielding 5.4 mg of protein/g of liver. Samples were frozen in liquid N2 and stored at -80 OC until they were subjected to immunoblot analysis with diclofenac antisera. Immunohistochemistry of Liver Sections. Immunohistochemistry of liver slices was done as described previously (17) with some modifications. Briefly, liver slices (2 mm thick) obtained from mice or rats given diclofenac (300 mg/kg ip and killed after 8 h or 200 mg/kg ip and killed after 12 h, respectively) were microwave-fixed in saline (2 min, 140 OF). The slices were individually placed in plastic histology cassettes (5 cassettes in a beaker containing 200 mL of saline) and then transferred to 70% ethanol after microwave fixation. The sliceswere embedded in paraffin, sectioned (5 pm), and mounted on poly(L-lysine)treated glass slides (American Histolabs, Gaithersburg, MD). All of the followingsteps were done at room temperature. Slides were placed in a slide staining dish, and paraffin was removed from the sections by washes with xylene (2 incubations for 5 min each), followed by washes for 2 min each with absolute ethanol, 70% ethanol, 35% ethanol, and Tris-buffered saline (TBS; 50 mM Tris, 150mM NaC1, pH 7.6). Nonspecific binding sites were blocked with 5% fetal calf serum (FCS) in TBS for 15min. After addition of diclofenac antiserum (1:4500 in TBS containing 2 5% FCS, 2 h), the slides were washed 3 times in TBS (2 min each). This was followed by the addition to the tissue of goat antirabbit IgG conjugated to alkaline phosphatase (diluted 1:500 in TBS containing 2% FCS). After 2 h, slides were washed 3 times in TBS (2 min each). Tissues were stained by the addition of substrate solution (Accurate Antibodies AS/AP) and incubated for 20 min, and after washing 3 times in TBS (2 min each) they were counterstained with Mayer’s Hematoxylin for 1 min and then rinsed in TBS. Cover slips were mounted with Dako Glycergel mounting medium and sealed with clear nail polish. Isolation of Primary Rat Hepatocytes and Immunofluorescence Detection of Diclofenac-Labeled Hepatocytes. Hepatocytes were isolated from a male Sprague-Dawleyrat (330 g) that was given diclofenac (200 mg/kg ip) 12 h before liver perfusion. The rat was anesthetized with xylazine (15 mg/kg) and ketamine (45 mg/kg), administered by intramuscular injection. The liver was perfused retrogradely with balanced salt solution (BSS; 142 mM NaCl, 6.7 mM KC1,lO mM HEPES, pH 7.4) through the suprahepatic inferior vena cava at 20 mL/min for 10 min, and then with BSS containing collagenase (type 11, 100 units/mL) and CaC12.2H20 (2.5 mM) for 15 min. The liver was transferred to a beaker, and hepatocytes were suspended in a solution containing BSS and 1.5% bovine serum albumin (BSA) (BSS-BSA). The suspension was filtered through a nylon filter (112 pm) and centrifuged at 50g for 2 min. The pellet was suspended in 20 mL of BSS-BSA,placed on 15mL of 50 % Percoll, and centrifuged at 130g for 10 min. The pellet containing viable cells (91% , determined by trypan blue exclusion) was washed with BSS once and resuspended in DMEM containing 8%FCS,

Diclofenac Adduct Formation penicillin G sodium (200 units/mL), streptomycin sulfate (200 pg/mL), amphotericin B (0.5 pg/mL), insulin (10 pg/mL), hydrocortisone 21-hemisuccinate (50pg/mL), phenol red (10 pg/ mL), and NaHC03 (24 mM). Cells were seeded a t 2 X lo6 cells in 25-cm2 flasks containing #1circular glass cover slips coated with type I rat tail collagen and incubated in 5% C02/95% air at 37 OC. The medium was changed after 2 and 4 h. At 5 h, the cover slips were removed from flasks, placed in 12-well plastic tissue culture dishes,and Tied for 10min in 2 mL of 3.7 5% formalin in phosphate-buffered saline (PBS, pH 7.4). After fixation, cover slips were washed with PBS three times. Cover slips were incubated for 15 min in 2 mL of 5% FBS in PBS and then were washed three times in PBS. Cover slips were incubated for 90 min with diclofenac antiserum (diluted 1:1000 in PBS containing 2% FCS), or with diclofenac antiserum that was preabsorbed with diclofenac (0.1 mM, 30 min, room temperature) in a humidified chamber. After incubation, cover slips were washed three times in PBS and then were incubated for 45 min with phycoerythrin-conjugated goat anti-rabbit IgG (1:40 dilution in PBS containing 2 % FCS) in a humidified chamber. Cover slips were washed three times in PBS and then were mounted on slides with Fluoromount-G. The edges of the cover slips were sealed with clear nail polish. Cells were observed (excitation X = 540 nm; emission X = 580 nm) and photographed with a Zeiss Microscope (Model Axiophot, Thornwood, NY). Metabolic Activation of Diclofenac in Vitro. All incubations were conducted in capped 20-mL glass scintillation vials, in a total volume of 1 mL, and contained 5 mg of homogenate or microsomal protein from untreated rats suspended in sodium phosphate (100mM, pH 7.4). In addition, the cytochrome P-450dependent reactions contained diclofenac (1 mM), NADH (5 mM), NADPH (5 mM), EDTA (2 mM), SKF525A (0.5 mM), and BSA (1 mg/mL) as noted; the UGT-dependent reactions contained diclofenac (1mM), MgC12 (1mM), UDP-glucuronic acid (UDPGA,5 mM), and &glucuronidase (5000units) as noted; the acyl-CoA synthase-dependent reactions contained diclofenac (1 mM), coenzyme A (2 mM), ATP (5 mM), MgC12 (5 mM), dithiothreitol(1 mM), and Triton X-100 (0.15%) as noted. All incubations took place in a metabolic shaking bath for 3 h a t 37 "C. Samples were removed and frozen immediately in liquid N2 and stored at -80 "C until they were analyzed by immunoblotting with diclofenac antisera. Chemical Stabilityof Diclofenac-LabeledProteins. The source of the protein adducts was total rat liver homogenate prepared 3 h after treatment with diclofenac (200 mg/kg) or diclofenac-labeledRSA. Incubations were conducted in 20-mL capped glass scintillation vials in a total volume of 2 mL and contained 200 mM sodium phosphate (pH 7.5 or lo), diclofenaclabeled RSA (2.5pg/mL), diclofenac-labeledrat liver homogenate (5 mg/mL), and hydroxylamine (170 mM) as noted. All incubations took place in a metabolic shaking bath at 37 "C. Samples were removed at 0 h and after 3 h, frozen in liquid N2, and stored at -80 "C until they were analyzed by immunoblotting with diclofenac antisera. Other Methods. Protein concentrations for all samples were determined with the BCA Protein Assay Reagent kit with BSA as the standard, according to the manufacturer's instructions.

Chem. Res. Toxicol., VoZ. 7,No. 4, 1994 577 -200 kDa -140 kDa -110 kDa

-50 kDa

0 1 10 50 100200

(mg/kg) Figure 1. Dose-dependent covalent modification of hepatic proteins in diclofenac-treated rats. Rats were given 0 (saline control), 1,10,50,100, or 200 mg/kg diclofenac intraperitoneally and killed after 16 h. Proteins from liver homogenates were separated by SDS-PAGE (100 pg/lane), transferred to nitrocellulose, and immunoblotted with anti-diclofenac sera.

-200 kDa -140 kDa .Irr-

g

Y n 3 l r u -

-110 kDa

Results

0 3 6 9 1 2 2 4 h Figure 2. Time-dependent covalent modification of hepatic proteins in diclofenac-treated rats. Rats were given 200 mg/kg diclofenac intraperitoneally and killed after 0,3,6,9,12, or 24 h. Proteins from liver homogenates were separated by SDSPAGE (100 pg/lane), transferred to nitrocellulose, and immunoblotted with anti-diclofenac sera.

Dose- and Time-Dependent Formation of Diclofenac-Labeled Proteins in Rat Liver. Liver homogenates were prepared from rats 16 hours after injection of diclofenac (1 mg/kg) and were immunoblotted with diclofenac antisera. Two major protein adducts with molecular masses of 110and 140 kDa were revealed (Figure 1). The intensity of the signal for the 110- and the 140kDa diclofenac-labeled proteins increased in a dosedependent manner, and additional labeled proteins were detected predominantly at 50 and 200 kDa in rats given 50,100, or 200 mg/kg diclofenac (Figure 1). The labeling

of the 50-kDa protein appeared to reach peak levels earlier and was eliminated from the liver more rapidly than were the labeling of the 110-,140-,and 200-kDa adducts (Figure 2). After 24 h, the 50-kDa adduct could not be detected in the liver homogenate. Subcellular Localization of Diclofenac-Protein Adducts. Immunoblotting studies revealed that the diclofenac-protein adducts with molecular masses of 110, 140, and 200 kDa were concentrated in the plasma membrane fraction of the liver, while the 50-kDa adduct was found primarily in the microsomal fraction (Figure

Hargus et al.

578 Chem. Res. ToxicoZ., VoZ. 7, No. 4, 1994

-200 kDa -140 kDa -110 kDa

-50 kDa

1 2 3 4

-H-

5 6

-m- -CyT-

7 8 9 1 0

- m C - -pM-

Figure 3. Subcellular localizationof diclofenac-labeledproteins. Rats were given 200 mg/kg diclofenac (lanes 2, 4, 6, 8, 10) or saline intraperitoneally (lanes 1,3,5,7,9) and killed 16 h later. Liver tissue was removed and homogenized, and subcellular fractions were prepared. Proteins from homogenates (H), mitochondrial pellets (MT), cytosolic fractions (CYT), microsomal pellets (MIC), or plasma membrane fractions (PM) were separated by SDS-PAGE (100pg/lane), transferred to nitrocellulose, and immunoblotted with anti-diclofenac sera.

3). Immunofluorescence studies of isolated hepatocytes from diclofenac-treated rats confirmed that diclofenaclabeled proteins were present in the plasma membrane and showed that at least a portion of these proteins was present on the surface of the cells (Figure 4A). In most cases, the fluorescence appeared to be concentrated in patches. In contrast, no surface fluorescence was observed (1)when the hepatocytes were incubated with preimmune sera in place of immune sera, (2) when immune sera were preincubated with diclofenac (0.1 mM) prior to the incubation with the cells, or (3) when diclofenac antisera were incubated with hepatocytes from control rats (results not shown). Immunohistochemistry of liver slices from diclofenac-treated mice (Figure 4B) or rats (results not shown) revealed that the protein adducts were found in the same area as other proteins localized in the bile canalicular domain of the liver parenchymal cell plasma membranes (18,19). A perivenous pattern of staining was observed in liver slices from diclofenac-treated mice, but a distinct regional pattern was not well-defined in liver slices from diclofenac-treated rats (results not shown). Bioactivationof Diclofenac. To determine whether the covalent binding of diclofenac to hepatic proteins could be mediated by metabolites formed from cytochrome P-450-, UGT-, or acyl-CoA synthase-dependentreactions, we incubated diclofenac with rat liver homogenates or microsomes in the presence or absence of cofactors needed for catalysis by these enzyme systems. The reaction mixtures then were analyzed for covalent adduct formation by immunoblotting with diclofenac antisera. A single protein adduct with molecular mass of 50 kDa was detected when liver homogenate was incubated with NADH and NADPH, the cofactors for cytochrome P-450 (20) (Figure 5, lane 2). No adducts were detected in the absence of these cofactors (Figure 5, lane l),and adduct formation was inhibited in incubations that contained 0.5 mM SKF525A, a known inhibitor of cytochrome P-450 reactions (20) (Figure 5, lane 7). When microsomal fractions

were used as the source of cytochrome P-450, only the 50-kDa adduct was formed in the presence of NADH and NADPH, even when albumin was included in the incubations as an accessible target protein for reactive intermediates (results not shown). In contrast, when the cofactor for UGT, UDP-glucuronic acid (UDPGA) (20), was added to the reaction mixture, multiple proteins were modified by diclofenac (Figure 5, lane 3). No adducts were detected when @-glucuronidase,an enzyme that hydrolyzes 1-0-acylglucuronides (20),was included in the incubation mixture (Figure 5, lane 5). Protein adducts were not detected when diclofenac was incubated with liver microsomes in the presence of CoASH and ATP, cofactors of acyl-CoA synthase (21) (results not shown). Chemical Stability of Diclofenac-Protein Adducts. The 50-, 110-, and 140-kDa adducts of diclofenac in liver homogenates from diclofenac-treated rats were stable at pH 7.5 for 3 h at 37 "C, in the presence or absence of hydroxylamine (Figure 6, lanes 2 and 4). However, at pH 10, covalently bound diclofenac was apparently cleaved from these proteins (Figure 6, lane 6). Hydroxylamine had very little effecton the course of these reactions (Figure 6, lane 8). The 200-kDa adduct also underwent cleavage at pH 10, although this reaction could only be followed when the immunoblots were exposed to X-ray film for longer time periods (results not shown). In contrast, the diclofenac-RSA conjugate that was synthesized by incubating diclofenac with RSA in the presence of 1-ethyl-3[(dimethy1amino)propyllcarbodiimide (EDC)was cleaved almost completely at pH 10, but only in the presence of hydroxylamine (Figure 6, lane 8).

Discussion In a previous in vivo study, diclofenac adducts with molecular masses of 50,70,110, and 140kDa were detected in mouse liver homogenates by diclofenac antisera (13). In the present study we have found that the 50-, 110-,and 140-kDa proteins also are labeled selectively in liver homogenates of diclofenac-treatedrats. There were some differences in adduct formation observed between diclofenac-treated mice and rats. A 200-kDa adduct was found only in rat liver homogenates, while the 70-kDa adduct was found only in mice. The 70-kDaaadduct detected in mouse homogenates was possibly a contaminant that resulted from incomplete clearing of blood from the mouse liver when the mice were killed by cervical dislocation. In contrast, the rats were killed by decapitation, which allowed the liver to be exsanguinatedand thus minimized blood contamination in liver homogenate. In this regard, a 70-kDa adduct that comigrated with albumin has been detected in blood samples from rats and mice treated with diclofenac.2 There appeared to be three possible metabolic pathways that might lead to the covalent binding of diclofenac to liver proteins, based on studies done with compoundsthat are structurally related to diclofenac. The first pathway (Figure 7A) involves the conjugation of the carboxylic acid group of diclofenac to glucuronic acid, yielding diclofenac acyl glucuronide (DAG), a reaction that is catalyzed by the microsomal UGTs (22). Acyl glucuronides are known reactive intermediates, and previous studies have shown that they can bind covalently to serum albumin in vitro (23) and to plasma and tissue proteins in vivo (24-26). 2

Pumford and Pohl, unpublished results.

Chem. Res. Toxicol., Vol. 7, No. 4, 1994 579

Diclofenac Adduct Formation

Figure 4. (Panel A) Immunofluorescence detection of diclofenac-labeled proteins on the surface of primary cultured hepatocytes from a rat given 200 mg/kg diclofenac intraperitoneally 12 h before liver perfusion and hepatocyte isolation. After 5 h in culture, hepatocytes were fixed with formalin and incubated with anti-diclofenac sera and then incubated with phycoerythrin-conjugated goat anti-rabbit IgG. Magnification shown is 504X (reduced from original photograph at 630X magnification). (Panel B) Immunohistochemical staining of liver section from a diclofenac-treated mouse. The mouse was given 300 mg/kg diclofenac intraperitoneally and killed 8 h later. Slides of liver tissue sections were prepared and incubated with anti-diclofenac sera and then with alkaline phosphatase-conjugated goat anti-rabbit IgG. Magnification shown is 504X (reduced from original photograph at 630X magnification).

-140 kDa

-200kDa

-110kDa

-140kDa -110kDa

m n

&* l o

1 -50 kDa

1 2 3 4 5 6 7 8 9 Figure! 5. Cytochrome P-450- and UGT-dependent formation of diclofenac-labeled protein adducts in vitro. Rat liver homogenate from control rats was incubated with the following: diclofenac (lane 1);diclofenac, NADH, and NADPH (lane 2); diclofenac and UDPGA (lane 3); diclofenac, NADH, NADPH, and UDPGA (lane 4); diclofenac, UDPGA, and p-glucuronidase (lane 5); diclofenac, UDPGA, NADH, NADPH, and p-glucuronidase (lane 6); diclofenac, NADH, NADPH, and SKF525A (lane 7); and diclofenac, UDPGA, and SKF525A (lane 8). Lane 9 contained liver homogenate from rats treated with diclofenac (200 mg/kg, 3 h) for comparison with the in vitro incubations. Proteins were then separated by SDS-PAGE (100 pg/lane), transferred to nitrocellulose, and immunoblotted with antidiclofenac sera.

Acyl glucuronides may bind covalently to proteins either by a transacylation mechanism (23,27,28) or by undergo-

2

3

4

5 6

-Didofenac-RSA 7 8

Figure 6. Chemical stability of diclofenac-labeled proteins. Diclofenac-RSA or liver homogenate from rats that were given 200 mg/kg diclofenac intraperitoneally and killed 3 h later was incubated without hydroxylamine (lanes 1, 2, 5, 6) or with hydroxylamine (lanes 3,4, 7, 8) at pH 7.5 (lanes 1-4) or pH 10 (lanes 5-8). Lanes 1,3,5, and 7 are samples at zero time; lanes 2,4,6, and 8 are samples after 3-h incubation a t 37 OC. Proteins were then separated by SDS-PAGE (100pg/lane for homogenate samples; 50 ng/lane for diclofenac-RSA), transferred to nitrocellulose, and immunoblotted with anti-diclofenac sera.

ing acyl migration and binding to the protein via Schiff base formation (28-30) (Figure 7A). The results of the present study have clearly shown that acyl glucuronide metabolites of diclofenac are formed in vitro and can bind covalently to hepatic proteins (Figure 5). This pathway of diclofenac activation appears to be responsible for the formation in vivo of the 110-, 140-, and 200-kDa plasma membrane adducts. First, it is known that diclofenac is metabolized in vivo in rats and humans to DAG (31). Second, some of the proteins labeled by the acyl glucuronide metabolites of diclofenac in liver homogenates in vitro had the same molecular masses as the 110-, 140-, and 200-kDa adducts formed in vivo (Figure 5, lane 9). Third, the adduct chemical stability studies were

580 Chem. Res. Toxicol., Vol. 7, No. 4, 1994

Hurgus et ul.

A

CI

OH

CI

UDPGT UDP-GA

Diclofenac-1-0-acyl glucuronide

Diclofenac

i

Acyl migration

H2N-Protein

-

protein-N

Diclofenac-2-0-acyl glucuronide

B

,

-

CI

P-450

Reactive Intermediates

Protein

Diclofenac-Protein Adducts

Diclofenac

C

CI

0JOH

Acyl-CoA synthase

C'

wsco*

CoASH CI

CI

Diclofenac HX-Protein

(X = S , 0, NH)

Figure 7. Possible pathways of diclofenac activation and covalent binding to proteins. (A) UDP-glucuronosyltransferase-dependent

pathway; (B)cytochrome P-450-dependentpathway; (C) acyl-CoA synthase dependent pathway. consistent with chemical bonds formed from the interaclinkage of diclofenac to the glucuronic acid moiety (Figure tion of an acyl glucuronide metabolite of diclofenac with 7A). In contrast, the diclofenac-RSA adduct, formed from the plasma membrane proteins. Since the adducts were the EDC-coupled reaction of diclofenac with RSA, was stable at neutral pH in the presence or absence of mainly bound as an amide linkage (34) because it was hydroxylamine, they did not appear to be acyl derivatives hydrolyzed almost completely at pH 10, but only in the of histidine, tyrosine, or cysteine (32,331. The instability presence of hydroxylamine. The additional adducts that of the adducts at pH 10 in the absence of hydroxylamine formed in vitro but were not detected in vivo may have suggested that the adducts were bound as ester linkages been formed as a result of the disruption of cellular (32, 33). This would be consistent with a direct transorganelles, which exposed binding sites that were sequesacylation of DAG to a threonine or serine residue, or tered in the intact cell. covalent binding after acyl migration and Schiff base The second possible pathway for metabolic activation formation, which would result in the formation of an ester of diclofenac involved cytochrome P-450 (Figure 7B). Our

Diclofenac Adduct Formation

results showed that cytochrome P-450-dependent metabolism of diclofenac could lead to the formation of the 50-kDa adduct formed in vivo (Figure 5). Unlike the acyl glucuronide metabolite of diclofenac, the reactive metabolites formed by cytochrome P-450 appeared to be very reactive because only one protein was covalently modified and the reactive intermediates could not be trapped by BSA (results not shown). It is likelythat diclofenac binds covalently to the cytochrome P-450 that activates it. The results of the present study do not allow us to draw conclusions regarding the isoform of cytochrome P-450 that may be labeled. It is known that the CYP2C family catalyzes the formation of 4’-OH-diclofenac in humans (35); however, the rat P-450 isozymes responsible for diclofenac hydroxylation have not been elucidated. Although the nature of the P-450-dependent reactive intermediates of diclofenac is not known, an arene oxide (36) or another intermediate may be formed during the 3’-, 4’-, or 5-hydroxylation of diclofenac (31). Perhaps an arene oxide intermediate reacts with a carboxyl side chain of an aspartate or glutamate residue on the protein, which would result in the formation of an ester bond. The chemical stability studies of the BO-kDa protein adduct formed in vivo indicated that the bond is unstable at pH 10 (Figure 6), which is consistent with an ester linkage. Our results are in contrast to the results of studies done with radiolabeled diclofenac in cultured hepatocytes, in which no cytochrome P-450-dependent adducts were detected (14). This discrepancy may be due to the use of different model systems, such as incubations with microsomes or homogenate vs incubations with cultured hepatocytes, or the use of different methods for detection of covalent adducts. A third pathway that seemed possible for the bioactivation of diclofenac is the formation of diclofenac acylCoA, which then could bind to proteins by transacylation (Figure 7C). MEDICA 16, nafenopin, and bezafibrate, all carboxylicacids, are thought to covalently modify proteins by this mechanism (21). However, no acyl-CoA synthasedependent diclofenac adduct formation was detected in the present study (results not shown). The presence of plasma membrane adducts of diclofenac may have toxicological significance. For example, in order for diclofenac to cause toxicity by an immune mechanism, the adducts must be accessible to the immune system. The immunofluorescence studies clearly showed that diclofenac adducts are exposed at the surface of hepatocytes (Figure 4A). On the other hand, adduct formation may lead to alteration of the activites of important plasma membrane proteins, possibly those in the bile canalicular domain where adducts of diclofenac have been detected (Figure 4B). In particular, the 110-kDa adduct may correspond to one of the bile canalicular proteins that have been identified in rat hepatocyte plasma membrane with molecular mass of 105-110 kDa. These proteins are important in bile transport (18,37-41) and cell adhesion (19, 42), and similar proteins may be present in human bile canaliculi. Since diclofenac-labeled hepatic proteins were detected at doses as low as 1mg/kg in rats (Figure 11, a dose that is comparable to the amount prescribed for patients (43,441,similar plasma membrane adducts could be formed in humans and may have a role in the hepatotoxicity produced by diclofenac. Our results differ from the results of a recently published study, which showed that adducts with apparent molecular masses of

Chem. Res. Toxicol., Vol. 7, No. 4, 1994 581

60 and 80 kDa were detected immunochemically in liver homogenates from diclofenac-treated rats (45). However, adducts of 50, 60, 80, and 126 kDa were detected in homogenates from rat hepatocytes incubated in the presence of diclofenac (45). Some of these proteins may be the same ones detected in the present study. The discrepancy in the in vivo results may be due to differences in antisera specificities. Studies are currently under way to determine the identity and function of the diclofenac-bound proteins detected in the present study. Additional information regarding the mechanism of diclofenac hepatotoxicity may allow prediction of adverse reactions to NSAIDs and other drugs that contain a carboxylic acid group in susceptible individuals.

Acknowledgment. We thank Dr. James Gillette for reviewing the manuscript and Mr. John George for technical assistance. T.G.M. was supported by a grant from NIGMS.

References Zimmerman, H. J. (1990) Update of hepatotoxicity due to classes of drugs in common clinical use: Non-steroidal anti-inflammatory drugs, antibiotics, antihypertensives, and cardiac and psychotropic agents. Semin. Liver Dis. 10,322-338. 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. Salama, A., Gottache, B., and Mueller-Eckhardt, C. (1991) Autoantibodies and drug- or metabolite-dependent antibodies in patienta with diclofenac-induced immune haemolysis. Br. J.Haematol. 77, 546-549. Salama, A., Schutz, B., Kiefel, V., Breithaupt, H., and MuellerEckhardt, C. (1989) Immune-mediated agranulocytosis related to drugs and their metabolites: Mode of sensitization and heterogeneity of antibodies. Br. J.Haematol. 72, 127-132. Epstein, M., Vickars, L., and Stein, H. (1990) Diclofenac induced immune thrombocytopenia. J. Rheumatol. 17, 1403-1404. Schapira, D., Bassan, L., Nahir, A. M., and Scharf, Y. (1986) Diclofenac-induced hepatotoxicity. Postgrad. Med. J. 62, 63-65. Breen, E. G., McNicholl, J., Cosgrove, E., McCabe, J., and Stevens, F. M. (1986)Fatalhepatitisassociatdwithdiclofenac. Gut 27,13901393. Sallie,R. W., McKenzie,T., Reed, W. D.,Quinlan, M. F., and Shilkin, K. B. (1991) Diclofenac hepatitis. A u t . N.Z . J. Med. 21, 251-255. Helfgott, S. M., Sandberg-Cook,J., Zakim, D., and Nestler, J. (1990) Diclofenac-associatedhepatotoxicity. J.Am. Med. Assoc. 264,2662. Iveson, T. J., Ryley, N. G., Kelly, P. M., Trowell, J. M., McGee, J. O., and Chapman, R. W. (1990) Diclofenac associated hepatitis. J. Hepatol. 10, 85-89. Purcell, P., Henry, D., and Melville, G. (1991) Diclofenac hepatitis. Gut 32, 1381-1385. Zimmerman, H. J. (1978) Chapter 5, Classificationsof hepatotoxins and mechanisms of toxicity. In Hepatotoxicity, Appleton, Century, Crofts, New York. Pumford, N. R., Myers, T. G., Davila, J. C., Highet, R. J., and Pohl, L. R. (1993) Immunochemical detection of liver protein adducta of the nonsteroidal anti-inflammatory drug diclofenac. Chem. Res. Toxicol. 6, 147-150. Kretz-Rommel, A., and Boelsterli, U. A. (1993) Diclofenac covalent protein binding is dependent on acyl glucuronide formation and is inversely related to P-450-mediated acute cell injury in cultured rat hepatocytes. Toxicol. Appl. Pharmacol. 120, 155-161. Hogeboom, G. H. (1955) Fractionation of cellular componenta of animal tissues: General method for isolation of liver cell componenta. Methods Enzymol. 1, 16-19. Loten, E. G., and Redshaw-Loten, J. C. (1986) Preparation of rat liver plasma membranes in a high yield. Anal. Biochem. 154,183185. Pumford, N. R., Martin, B. M., Thomassen, D., Burris, J. A., Kenna, J. G., Martin, J. L., and Pohl, L. R. (1993) Serum antibodies from halothane hepatitis patients react with the rat endoplasmicreticulum protein ERp72. Chem. Res. Toxicol. 6, 609-615. McCaughan, G., Wickson, J. E., Creswick, P. F., and Gorrell, M. D. (1990) Identification of the bile canalicular cell surface molecule GPllO as the ectopeptidase dipeptidyl peptidase IV An analysis

Hargus et al.

582 Chem. Res. Toxicol., Vol. 7,No. 4, 1994 by tissue distribution, purification, and N-terminal amino acid sequence. Hepatology 11,534-544. (19) Mowery, J., and Hixson, D C. (1991) Detection of cell-CAM 105 in pericanalicular domain of the rat hepatocyte plasma membrane. Hepatology 13,47-56. (20) S i p , I . G.,andGandolfi,A. J. (1991)BiotransformationofToxicants. In Casarett and Doull’s Toxicology: The Basic Science of Poisono, 4th Edition (Casarett, L. J., and Doull, J., Eds.) pp 88-126, Pergammon Press, New York. (21) Hertz, R., and Bar-Tana, J. (1988) The acylation of proteins by xenobioticamphipathic carboxylicacids in cultured rat hepatocytes. Biochem. J. 254,39-44. (22) Magdalou, J., Chajes, V., Lafaurie, C., and Siest, G. (1990) Glucuronidation of 2-arylpropionic acids pirprofen, flurbiprofen, and ibuprofen by liver microsomes. Drug Metab. Dkpos. 18, 692-697. (23) Van Breeman, R. B., and Fenselau, C. (1985) Acylation of albumin by 1-0-acyl glucuronides. Drug Metab. Dispos. 13,318-320. (24) Smith, P. C., McDonagh, A. F., and Benet, L. Z. (1986) Irreversible binding of zomepirac to plasma protein in vitro and in vivo. J.Clin. Znuest. 77,934-939. (25) Sallustio, B. C., Knights, K. M., Roberts, B. J., and Zacest, R. (1991) Zn vivo covalent binding of clofibric acid to human plasma proteins and rat liver proteins. Biochem. Pharmacol. 42,1421-1425. (26) Dickinson, R. G., and King, A. R. (1993) Studies on the reactivity of acyl glucuronides-IV. Covalent binding of diflunisal to tissues of the rat. Biochem. Pharmacol. 45, 1043-1047. (27) Faed, E. M. (1984) Properties of acyl glucuronides: Implications for studies of the pharmacokinetics and metabolism of acidic drugs. Drug Metab. Reo. 15, 1213-1249. (28) Spahn-Langguth, H., and Benet, L. Z. (1992) Acyl glucuronides revisited Is the glucuronidation proceas a toxification as well as a detoxification mechanism? Drug Metab. Reo. 24, 5-47. (29) Ding, A,, Ojingwa, J. C., McDonagh, A. F., Burlingame, A. L., and Benet, L. Z. (1993)Evidence for covalent binding of acyl glucuronides to serum albumin via an imine mechanism as revealed by tandem mass spectrometry. Proc. Natl. Acad. Sci. U.S.A. 90, 3797-3801. (30) Grubb, N., Weil, A., and Caldwell, J. (1993) Studies on the in oitro reactivity of clofibryl and fenofibryl glucuronides. Evidence for protein binding viaa Schiff’s base mechanism.Biochem.Pharmacol. 46,357-364. (31) Stierlii, H., and Faigle,J. W. (1979) Biotransformationof diclofenac sodium (Voltaren) in animals and in man. 11. Quantitative deter-

(32) (33) (34) (35) (36) (37) (38) (39) (40) (41)

(42)

(43) (44) (45)

mination of the unchanged drug and principal phenolic metabolites, in urine and bile. Xenobiotica 9, 611-621. Riordan, J. F., and Vallee, B. L. (1972) Methods Enzymol. 25,494499. BGnning, P., Holmquist, B., and Riordan, J. F. (1978) Functional residues at the active site of angiotensin convertingenzyme.Biochem. Biophys. Res. Commun. 83,1442-1449. Balls, A. K., and Wood, H. N. (1956) Acetyl chymotrypsin and its reaction with ethanol. J.Biol. Chem. 219, 245-256. Leemann, T., Transon, C., and Dayer, P. (1993) CytochromeP 4 5 h (CYPPC): A major monooxygenase catalyzing diclofenac 4’-hydroxylation in human liver. Life Sci. 52, 29-34. Korzekwa, K., Trager, W., Gouterman, M., Spangler, D., and Loew, G. H. (1985) Cytochrome P450 mediated aromatic oxidation: A theoretical study. J. Am. Chem. SOC. 107,4273-4279. Vore, M. (1993)Canalicular transport: Discoveryof ATP-dependent mechanisms. Toxicol. Appl. Pharmacol. 1 IS, 2-7. Zimniak, P., and Awasthi, Y. C. (1993) ATP-dependent transport systems for organic anions. Hepatology 17, 330-339. Arias, I. M., Che, M., Gatmaitan, Z., Leveille, C., Nishida, T., and St. Pierre, M. (1993) The biology of the bile canaliculus, 1993. Hepatology 17, 318-329. Nathanson, M. H., andBoyer, J. L. (1991)Mechanismsandregulation of bile secretion. Hepatology 14, 551-566. Miiller, M., Ishikawa, T., Berger, U., Kliinemann, C., Lucka, L., Schreyer, A., Kannicht, K., Ruetter, W., Kurz, G., and Keppler, D. (1991) ATP-dependent transport of taurocholate across the hepatocyte canalicular membrane mediated by a 110 kDa glycoprotein binding ATP and bile salt. J. Biol. Chem. 266,18920-18926. Stamataglou, S. C., Rui-Chang, G., Mills, G., Butters, T. D., Zaidi, F., and Hughes, R. C. (1990) Identification of a novel glycoprotein (AGpllO) involved in interactions of rat liver parenchymal cells with fibronectin. J. Cell Biol. 111, 2117-2127. Caldwell, J. R. (1986) Efficacy and safety of diclofenac sodium in rheumatoid arthritis. Am. J.Med. 80 (Suppl. 4B), 43-47. Physician’s Desk Reference (1991) pp 1024-1027, Edward R. Barnhart, Medical Economics Co., Inc., Oradell, NJ. Kretz-Rommel, A., and Boelsterli, U. A. (1994) Selective protein adducts to membrane proteins in cultured rat hepatocytes exposed to diclofenac: Radiochemical and immunochemical analysis. Mol. Pharmacol. 45, 237-244.