Covalent Binding of Sulfamethoxazole Reactive Metabolites to Human

Feb 29, 1996 - Comparative oxidative metabolic profiles of clomipramine in cats, rats and dogs: preliminary results from an in vitro study. C. LAINESS...
0 downloads 11 Views 234KB Size
Download PDF
500

Chem. Res. Toxicol. 1996, 9, 500-507

Covalent Binding of Sulfamethoxazole Reactive Metabolites to Human and Rat Liver Subcellular Fractions Assessed by Immunochemical Detection Alastair E. Cribb,*,† Cindy E. Nuss,† David W. Alberts,† Diane B. Lamphere,† Denis M. Grant,‡ Scott J. Grossman,† and Stephen P. Spielberg§ Departments of Safety Assessment and Worldwide Strategic Operations, Merck Research Laboratories, West Point, Pennsylvania 19486, and Division of Clinical Pharmacology, Hospital for Sick Children, Toronto, Ontario, Canada M5G 1X8 Received September 28, 1995X

Potentially serious idiosyncratic reactions associated with sulfamethoxazole (SMX) include systemic hypersensitivity reactions and hepatotoxicity. Covalent binding of SMX to proteins subsequent to its N-hydroxylation to form N4-hydroxysulfamethoxazole (SMX-HA) is thought to be involved in the pathogenesis of these reactions. A polyclonal antibody was elicited in rabbits against a SMX-keyhole limpet hemocyanin conjugate that recognized covalent protein adducts of SMX in microsomal protein and was used to characterize the covalent binding of SMX and its putative reactive metabolites to hepatic protein in vivo and in vitro. In vitro covalent binding of SMX to rat and human liver microsomal protein was NADPH-dependent, while binding of SMX-HA was not dependent on NADPH. SMX and SMX-HA produced similar patterns of covalent binding, with major protein targets in the region of 150, 100 (two bands), 70 (two bands), and 45-55 kDa. The pattern of covalent binding to human and rat liver microsomal protein was similar. Binding of SMX-HA was completely eliminated by GSH or by addition of cytosolic fractions and acetylcoenzyme A. The acetoxy metabolite of SMX also led to covalent binding, but it was primarily attributable to the formation of SMX-HA from acetoxySMX. In vivo exposure of rats to SMX did not result in detectable covalent binding by the methods employed. When rat liver slices were incubated with 2 mM SMX or 500 µM SMXHA, no toxicity was observed and yet covalent binding of SMX-HA to 130, 100, 70, and 55 kDa proteins could be detected. These results confirm that covalent binding of SMX occurs via the formation of SMX-HA and that covalent binding of SMX-HA in vitro results from its conversion to the more reactive nitroso metabolite. Acetylation of SMX-HA protected against its covalent binding. Further studies are required to determine how this in vitro covalent binding relates to in vivo covalent binding in humans and to either direct or immune-mediated cytotoxicity in SMX idiosyncratic drug reactions. Sulfamethoxazole (SMX)1 is associated with a variety of idiosyncratic toxicities, including hepatotoxicity and systemic hypersensitivity reactions (1). Formation of reactive metabolites capable of covalently binding to critical cellular macromolecules is thought to be crucial to the pathogenesis of idiosyncratic reactions (2, 3). Covalent drug-protein adducts may either directly kill cells or lead to the formation of immunogenic haptenprotein conjugates. Immunogens formed by covalent binding of reactive metabolites to cellular macromolecules could be presented to the immune system as a result of release from damaged cells (leading to an antibody response) or through some as yet undefined alteration in the normal processing of these macromolecules leading

to aberrant expression on the cell surface or altered epitope presentation in conjunction with major histocompatibility complex class I (MHCI) molecules (leading to a cell-mediated immune response). The resultant immune response may then be responsible for some or all of the clinical manifestations of the adverse reaction. Identification of critical target proteins altered by covalent adducts may help to unravel these various processes. Understanding the link between specific targets of covalent binding and pathogenesis of idiosyncratic reactions, including hepatotoxicity, will be facilitated by investigation of a range of structurally distinct compounds causing clinically similar syndromes.

* Author to whom correspondence should be addressed. † Department of Safety Assessment, Merck. ‡ Hospital for Sick Children. § Worldwide Strategic Operations, Merck. X Abstract published in Advance ACS Abstracts, February 1, 1996. 1 Abbreviations: AcetylCoA, acetyl coenzyme A; BSA, bovine serum albumin; ELISA, enzyme-linked immunosorbent assay; HOM, cellular homogenate; KLH, keyhole limpet hemocyanin; MHCI, major histocompatibility complex, class I; MIC, microsomal fraction; MIT, mitochondrial fraction; MF, plasma and nuclear membrane fraction; PBS, phosphate-buffered saline, pH 7.4 (KCl, 0.2 g/L; KH2PO4, 0.2 g/L; NaCl, 8 g/L; Na2PO4‚7H2O 2.16 g/L); P450, cytochrome P450; P9, pellet obtained at 9000g; S9, supernatant obtained at 9000g; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; SMX, sulfamethoxazole; SMX-HA, sulfamethoxazole hydroxylamine, N4hydroxysulfamethoxazole.

While mechanisms and risk factors for the immunological response may be similar between compounds, the metabolic risk factors will be distinct. An understanding of the links between bioactivation, detoxification, and covalent binding is therefore required for each compound. Investigation of the nature and quantity of covalent binding may be used as an indicator of metabolic bioactivation and to identify pathways involved in the detoxification of reactive metabolites. SMX is N-hydroxylated to a hydroxylamine (SMX-HA), which is further oxidized to nitroso and nitro metabolites as well as acetylated to form hydroxamic acid and acetoxy metabolites (4-6). The

0893-228x/96/2709-0500$12.00/0

© 1996 American Chemical Society

Covalent Binding of Sulfamethoxazole

oxidation of the hydroxylamine to its nitroso metabolite is spontaneous in aqueous buffers (5). Glutathione prevents the autoxidation to the nitroso metabolite and will react with nitrosoSMX to form a semimercaptal, which can be thiolytically cleaved to regenerate the hydroxylamine (5). Very low intracellular GSH appears to enhance the toxicity of the sulfonamide hydroxylamines (SMX-HA) (7), but depletion of cellular GSH does not appear to precede or to be linked with the cytotoxicity of SMX-HA (5, 8). Acetylation of SMX-HA was recently shown to produce acetoxySMX, which may also be a cytotoxic, covalent binding species involved in SMX toxicity (6). It has been proposed that a defect in reactive metabolite detoxification plays a role in susceptibility to sulfonamide hypersensitivity reactions, but the defect has not been identified (9). The objective of these studies was to identify potential protein targets of covalent adducts of SMX which may be involved in direct or immune-mediated toxicity. The role of specific metabolic pathways in modulating the observed covalent binding was also assessed.

Experimental Procedures Caution. The following chemicals are hazardous and should be handled carefully: acrylamide/bis(acrylamide) (neurotoxin), wear gloves and respiratory protection; sulfamethoxazole hydroxylamine, unknown risk to humans in synthetic form, handle appropriately (avoid skin contact, inhalation, and ingestion). Chemicals and Reagents. All routine chemicals were of the highest purity available from Sigma Chemical Co. (St. Louis, MO). Immunochemicals were purchased from Amersham Corp. (Arlington Heights, IL). Precast polyacrylamide electrophoresis minigels and associated buffers and apparatus were purchased from Bio-Rad Laboratories (Hercules, CA). Sulfamethoxazole hydroxylamine was purchased from Dalton Chemical Co. (Toronto, ON). AcetoxySMX was synthesized as previously described (6). Anti-P450 2C6 and anti-P450 2C11 antibodies were obtained from Gentest Corp. (Woburn, MA), and anti-P450 3A1 and anti-P450 2B1 antibodies were obtained from Human Biologicals, Inc. (Phoenix, AZ). Animal Treatments. All animal treatments were conducted in accordance with and after approval by the Merck Institutional Animal Care and Use Committee. Synthesis of Immunogen and Immunization Protocol. The immunogen for raising anti-SMX antibodies was prepared by a diazotization reaction. SMX (400 mg) was dissolved in 5 mL of 0.5 N HCl. This was mixed with 10 mL of 2% sodium nitrite and kept for 1 h at 4 °C. The mixture was brought to neutral pH by the addition of 1 N NaOH. Then 7.5 mL of the mixture was added to 5 mL of a 3 mg/mL solution of keyhole limpet hemocyanin (KLH) or to 5 mL of 4% bovine serum albumin (BSA) in 0.1 M borate buffer (pH 8.5). The mixtures were incubated overnight at 4 °C and dialyzed for 24 h against 100 volumes of phosphate-buffered saline (PBS: KCl, 0.2 g/L; KH2PO4 0.2 g/L; NaCl, 8 g/L; Na2PO4‚7H2O, 2.16 g/L; pH 7.4) changed three times. Two rabbits were immunized. The rabbits were initially immunized with 750 µg of immunogen (SMX-KLH conjugate) emulsified in Freund’s complete adjuvant in four subcutaneous and two intramuscular sites. Four and eight weeks later, the rabbits were immunized in four subcutaneous sites with the immunogen in incomplete Freund’s adjuvant. Two weeks after the final immunization, the rabbits were exsanguinated under pentobarbital anesthesia and sera prepared. In Vivo Exposure of Rats to SMX. Male CRL(CD)BR SD strain rats (Charles River Laboratories, Raleigh, NC) weighing 200-225 g received SMX once daily dissolved in corn oil by intraperitoneal injection according to four different schedules. Group one (n ) 3) received SMX at 50 mg/kg for 10 days and was killed 16 h after the last injection. Group two (n ) 2)

Chem. Res. Toxicol., Vol. 9, No. 2, 1996 501 received 300 mg/kg and was killed 16 h after treatment. Group three (n ) 2) was pretreated with phenobarbital (80 mg/kg) for 3 days, treated with 300 mg of SMX/kg, and killed 16 h after receiving SMX. Group four (n ) 2) received 400 mg of SMS/kg and was killed 3 h after treatment. Each group had a corresponding control group that received an equal volume of corn oil. Additional rats were treated with 0.9% saline, dexamethasone (150 mg kg-1 day-1 in saline with 2% Tween 80), or phenobarbital (80 mg kg-1 day-1 in saline) for 3 days as described previously (10) to induce various cytochrome P450 (P450) enzymes for use in in vitro studies. Exposure of Rat Liver Slices to SMX. Rat liver slices were prepared and incubated in Waymouth’s MB752/1 media supplemented with gentamicin (85 µg/mL), fungizone (2.5 µg/ mL), insulin (1 µM), and dexamethasone (5 µM) at 37 °C in a 95/5% O2/CO2 atmosphere as described previously (11). Slices were exposed to vehicle (1% DMSO), SMX (2 mM), or SMX-HA (500 µM) for 4 or 24 h. Liver slices were harvested, rinsed, and frozen at -70 °C until analyzed by immunochemical detection. Toxicity was assessed after 24 h of incubation by lactate dehydrogenase release, cellular ATP (12), glutathione content (13), and total protein synthesis (11). Preparation of Subcellular Fractions. Subcellular fractions of rat liver were prepared from freshly isolated livers and from rat liver slices by differential centrifugation. Livers were homogenized in 2 volumes of ice-cold 0.05 M Tris/0.15 M KCl buffer (pH 7.4); each liver slice containing a total of approximately 2 mg of protein was homogenized in 500 µL of icecold PBS. The homogenate (HOM) was then centrifuged for 20 min at 400g. The resulting pellet containing nuclear and plasma membrane fractions was resuspended in PBS and referred to as the membrane fraction (MF). The supernatant obtained was centrifuged for 20 min at 15000g. The resultant pellet was resuspended in PBS and designated the mitochondrial fraction (MIT). In some cases, a single centrifugation was performed for 20 min at 9000g. The pellet obtained in this case was referred to as the P9 fraction and contained mitochondrial, plasma, and nuclear membranes. The supernatant obtained containing cytosol and microsomal fractions was referred to as the S9 fraction. The cytosolic fraction (CYT) was obtained by centrifuging the S9 fraction at 100000g for 60 min and harvesting the supernatant. The pellet was resuspended and centrifuged at 100000g for 60 min again, and the resulting pellet was resuspended in 0.25 M sucrose/5 mM HEPES (pH 7.4) buffer. This fraction was the microsomal fraction (MIC). Human liver microsomes and S9 fractions (20 mg/mL in 0.25 M sucrose) were obtained from Human Biologicals, Inc. The protein content of all fractions was determined using the Bio-Rad DC protein assay kit (Bio-Rad Laboratories). Recombinant N-Acetyltransferases. Recombinant Nacetyltransferases were expressed in Escherichia coli and cellular lysates prepared as has been described previously (14). In Vitro Covalent Binding Studies. Covalent binding studies were performed in vitro under a variety of conditions as specified in figure legends or in the text. All experiments were performed in duplicate, and each experimental condition was repeated at least twice on different days. Briefly, subcellular fractions were incubated with SMX or SMX-HA for various time periods at 37 °C in the presence or absence of cofactors. Reactions for covalent binding studies were terminated by the addition of at least an equal volume of Laemmli buffer [0.2 mg of bromophenol blue/mL; 2% SDS (w/v); 20% glycerol (v/v) in Tris buffer (pH 6.8) with or without 5% β-mercaptoethanol (v/ v)]. Samples were frozen at -20 °C until analysis or immediately heated to 95 °C for 4 min for loading on SDSpolyacrylamide gels (SDS-PAGE). For analytical studies of drug and metabolite concentrations, reactions were terminated by the addition of an equal volume of methanol. SMX and its metabolites were analyzed by HPLC as described previously (10). ELISA. Antisera were screened for specific antibodies by ELISA using the corresponding BSA-SMX conjugates as the solid phase. Ninety-six-well microtiter immunoplates were

502 Chem. Res. Toxicol., Vol. 9, No. 2, 1996

Cribb et al.

Figure 1. Expected orientation of adducts of SMX to protein. coated with 20 ng of BSA-SMX conjugate or straight BSA in 100 µL of PBS overnight at 4 °C. Wells were washed four times with 0.05% Tween 20 in PBS and excess binding sites blocked with 5% fetal calf serum in PBS (100 µL for 1 h). Various dilutions of antisera were made in 2% fetal calf serum-PBS (v/v), and 100 µL was added to each well for 2 h at room temperature. The wells were washed four times with PBSTween, and then a 1/3000 dilution of alkaline phosphataselinked anti-rabbit IgG antibody in fetal calf serum-PBS was added to each well for 1 h (100 µL). The wells were washed as above, and then 100 µL of a p-nitrophenol phosphate solution (phosphatase reagent 104, Sigma) in 50 mM carbonate buffer with 0.02% MgCl2 (w/v) (pH 9.6) was added to each well. After a 30-min color development, absorbance at 405 nm was determined using an automated microplate reader. For competitive inhibition studies, drugs were preincubated with antisera for 1 h before performing the ELISA. Immunoblotting. Proteins (25 µg) were resolved on 7.5% or 12% precast SDS-PAGE minigels (Bio-Rad Laboratories) or on 16 cm full-size gels of 7.5, 10, or 12%. Unless otherwise specified, gels were run under nonreducing conditions. Protein was then transferred to nitrocellulose by semidry electroblotting (Trans-Blot apparatus, Bio-Rad Laboratories). Immunoblotting was performed according to standard protocols (15) using a 1/1000 dilution of anti-SMX antiserum (Tokyo, Japan). Briefly, membranes were blocked for 1 h with PBS containing 5% nonfat dry milk and 2% BSA (blocking buffer). The antiserum was then added in blocking buffer containing 0.1% Tween 20 and incubated at room temperature for 2 h. The membranes were washed for 30 min with PBS/0.1% Tween with three changes of media. An anti-rabbit IgG horseradish peroxidase conjugate (1/5000 dilution; Amersham) was used as a secondary antibody and was added in the same buffer as the antisera for 1 h at room temperature. After washing as above, bound IgG was visualized using the chemiluminescent ECL reagent (Amersham). The blots were exposed to autoradiographic film for various periods of time as required to obtain a suitable exposure.

Results The orientation of the synthetic immunogen and the proposed covalent binding species of SMX are shown in Figure 1. The rabbits developed high titers toward SMX as determined by ELISA (Figure 2), and antiserum obtained from rabbit 39356 was used in all experiments reported here. The antiserum did not cross-react with BSA or other drug-protein conjugates (e.g. diclofenacBSA) on ELISA, and binding could be inhibited by SMX but not sulfamethazine (Figure 2), consistent with recognition of the N1-substituent and not the p-aminobenzene group.

Figure 2. (Top) Detection of antibodies to SMX in serum of rabbit immunized with SMX-KLH conjugate by ELISA using BSA or BSA-SMX as the solid phase antigen. (Bottom) Competitive inhibition of binding of anti-SMX serum antibodies to SMX-BSA (1/20000 dilution of antisera) by sulfamethoxazole but not sulfamethazine.

Figure 3. Specificity of anti-SMX serum in recognizing covalent adducts of SMX to microsomal protein. Rat liver microsomal protein (2 mg/mL) was incubated with no drug, SMX (1 mM), SMX-HA (200 µM), or SMX-HA (200 µM) with 1 mM GSH for 2 h at 37 °C. Microsomal proteins were separated by SDS-PAGE (20 µg/lane) on 12% gels, transferred to nitrocellulose, and probed with anti-SMX serum (1/1000 dilution). In some incubations, the antibodies were competed out with SMX or BSASMX conjugates as indicated.

The antiserum was strongly immunoreactive with rat liver microsomes that had been incubated with SMX-HA at 37 °C (Figure 3). Immunoreactivity of the antiserum toward this material could be completely blocked by preincubation of the antiserum with SMX or SMX-BSA conjugates (Figure 3). The antiserum did not recognize

Covalent Binding of Sulfamethoxazole

any immunoreactive material when SMX alone was incubated with microsomal protein or when SMX-HA was incubated with liver microsomes in the presence of 1 mM GSH. Elimination of a boiling step prior to performing SDS-PAGE electrophoresis did not greatly affect the immunoreactivity toward microsomal protein incubated with SMX-HA (not shown). The addition of reducing agents (dithiothreitol or β-mercaptoethanol) to the loading buffer eliminated the covalent binding (not shown). Therefore, all gels were run under nonreducing conditions. The failure of SMX in the absence of NADPH to produce an immunodetectable adduct to the microsomal protein and loss of immunoreactivity in the presence of GSH indicate that simple intercalation of either SMX or SMX-HA into the microsomal protein was not responsible for the observed binding. Recognition by the anti-SMX antiserum was therefore taken to represent covalently bound drug. SMX was administered to rats in a variety of regimes as described under Experimental Procedures to identify in vivo targets of covalent binding in the liver. Despite extensive immunoblotting studies under a variety of SDS-PAGE conditions and loading of up to 100 µg of protein per lane, no covalent binding of SMX to liver proteins in any of the subcellular fractions was conclusively demonstrated under any of the in vivo exposure conditions employed. Western blotting is a semiquantitative technique and the limits of detection of covalently bound SMX will be dependent on the quantity of the target protein present in a subcellular fraction, the total amount of bound SMX to proteins in a given range, the epitope density on the target protein, the amino acid that is modified, and the availability of the covalently bound SMX to the antibody. With no specific knowledge of the target proteins it is not possible to quantify the limits of detection. Therefore, to confirm the potential for SMX reactive metabolites to covalently bind to protein and our ability to detect covalently bound material under conditions in which it would be expected to occur, a series of in vitro experiments were performed. When SMX and 1 mM NADPH were incubated with phenobarbital-treated rat or human liver microsomes, covalent binding of SMX to microsomal protein occurred as demonstrated by increased immunoreactivity of the anti-SMX antiserum (Figure 4). Formation of SMX-HA during the course of the incubation was documented by HPLC analysis (see Figure 4 legend). Phenobarbitaltreated rat liver microsomes were employed because we have previously shown greater turnover to SMX-HA compared to untreated rats (10). P450 2C6, the form oxidizing SMX (10), is induced by phenobarbital pretreatment. When SMX-HA was incubated with microsomal protein, as noted above, extensive covalent binding occurred. This covalent binding was diminished by the presence of NADPH (1 mM) (Figure 4). The pattern of covalent binding observed with SMX as the starting compound was essentially identical to that observed with SMX-HA in the presence or absence of NADPH (Figure 4). Covalent binding of SMX-HA to human and rat liver microsomal protein produced similar patterns of covalent binding to proteins (Figure 4). The experiments with subcellular fractions described below were therefore carried out with SMX-HA in the presence or absence of 1 mM NADPH as indicated. The covalent binding and cytotoxicity of SMX and SMX-HA were further investigated in rat liver slices. SMX (2 mM) and SMX-HA (500 µM) were incubated with

Chem. Res. Toxicol., Vol. 9, No. 2, 1996 503

Figure 4. Effect of NADPH on covalent binding of SMX and SMX-HA to liver microsomal protein. Human (2 mg/mL) and phenobarbital-induced rat (2 mg/mL) liver microsomes were incubated for 30 min with SMX-HA (25 µM) or for 120 min with SMX (1 mM) in the presence and absence of NADPH (NADPHgenerating system producing 1 mM NADPH). Rat liver microsomes generated a final concentration of 12 µM SMX-HA, and human liver microsomes generated a final concentration of 4.5 µM SMX-HA. Microsomal proteins were separated on 7.5% gels and probed with anti-SMX serum as described in Figure 3.

Figure 5. Covalent binding of SMX and SMX-HA to rat liver slices. Rat liver slices were exposed to SMX (2 mM) or SMXHA (500 µM) for 4 and 24 h. Slices were rinsed to remove free drug and homogenized in PBS, and subcellular fractions were prepared for immunoblotting as described under Experimental Procedures. Representative blots from one of three experiments performed in duplicate are shown. Blots of microsomal fractions were not performed because insufficient material was obtained. Twenty micrograms of total homogenate (HOM) and approximately 10 µg of other subcellular fractions [pellet (P9) and supernatant (S9) at 9000g and cytosol (CYT)] were loaded in each lane.

rat liver slices for 4 or 24 h. Slice homogenates and subcellular fractions were immunoblotted for the presence of covalently bound material (Figure 5). A major band of covalently bound material was detected at 130 kDa when SMX-HA was incubated with rat liver slices and was equally intense at 4 and 24 h. Subcellular fractionation demonstrated that this protein was mainly localized in the membrane fraction. A similar band was detected faintly in one of three experiments with SMX. Additional faint bands were observed at approximately 70, 55, and 35 kDa proteins with both SMX-HA and SMX. These bands were most evident in the HOM, P9, and S9 fractions. No evidence of cytotoxicity was present in any of three separate experiments with either compound. Thus, the immunochemical detection system employed

504 Chem. Res. Toxicol., Vol. 9, No. 2, 1996

Figure 6. Determination of molecular weight of SMX-HA covalent adducts. Liver microsomes from a phenobarbitaltreated rat (2 mg/mL) were incubated with SMX-HA (100 µM) with and without 1 mM NADPH for 30 min. The reaction was stopped by the addition of an equal volume of Laemmli buffer. Twenty-five micrograms of microsomal protein was separated on 16 cm 12% and 10% SDS-polyacrylamide gel along with biotinylated MW markers. The approximate MWs of the identified proteins are shown on the right. For the 10% gel, only microsomes incubated with SMX-HA in the presence of NADPH are shown.

was capable of detecting small amounts of covalently bound material derived presumably from the reactive intermediates of SMX and SMX-HA. Additional experiments were carried out to identify metabolic pathways which may modify the covalent binding of SMX reactive metabolites, to further identify the ultimate covalent binding species, and to determine the subcellular location and size of the major proteins covalently modified during incubation with SMX-HA. Several major microsomal proteins were adducted during the NADPH-dependent bioactivation of SMX or by SMX-HA. The majority of covalent binding occurred to proteins in the regions of 45-55, 70 (two bands), 100 (two bands), and 150 kDa (Figures 4 and 6) in both human and rat liver microsomes. Staining of the SDSPAGE gels with Coomassie blue indicated that these proteins corresponded to quantitatively important microsomal proteins (not shown). The proteins surrounding 50 kDa corresponded to the region of P450 as confirmed by immunoblotting with antibodies against P450 2C6 (recognized P450 2C6 and P450 2C11 in rats and P450 2C proteins in humans), P450 3A, and P450 2B (not shown). However, identification of specific P450 as targets for covalent binding was not possible due to the presence of several immunodetectable bands. The effect of P450 on covalent binding of SMX-HA was investigated by the use of specific P450 inhibitors and microsomes prepared from rats pretreated with inducers of P450. Covalent binding of SMX-HA to rat liver microsomes prepared from saline-, phenobarbital-, or dexamethasone-treated rats was similar in the absence of NADPH. However, in the presence of NADPH, cova-

Cribb et al.

Figure 7. (Top) Effect of P450 inducers on covalent binding of SMX-HA to rat liver microsomes. Microsomes (2 mg/mL) from saline-treated (Sal), phenobarbital-treated (Pb), and dexamethasone-treated (Dex) rats were incubated with 100 µM SMXHA for 30 min in the presence or absence of 1 mM NADPH (+/ -). Immunoblotting was performed as in Figure 3 on 7.5% gels. (Bottom) Effect of P450 inhibitors on covalent binding of SMXHA to human liver microsomes. Human liver microsomes (2 mg/ mL) were preincubated with NADPH alone (Con) or with sulfaphenazole (SPZ; 100 µM) or with troleandomycin (TAO; 100 µM). Then SMX-HA (100 µM) was added for 2 h at 37 °C. Microsomes were immunoblotted as in Figure 3 on 7.5% gels.

lent binding to dexamethasone-treated rat liver microsomes was considerably less than that observed in the other microsomes (Figure 7). HPLC analysis revealed that approximately 30% of the SMX-HA was metabolized back to SMX and to additional unidentified polar metabolites (perhaps 5′-hydroxysulfamethoxazole derivative) by dexamethasone-treated rat liver microsomes. Phenobarbital-treated microsomes reduced 38% of the SMX-HA to SMX compared to less than 5% by salinetreated microsomes. SMX-HA wa not stabilized during the incubation and so could not be measured accurately. The P450 2C9 inhibitor sulfaphenazole and the P450 3A inhibitor troleandomycin had very little effect on the covalent binding of SMX-HA to human liver microsomes in the presence of NADPH (Figure 7). The addition of acetylcoenzyme A and cytosolic fractions to rat liver microsomes nearly completely eliminated the covalent binding associated with SMX-HA, whereas acetylcoenzyme A or cytosolic fractions alone had no significant effect (Figure 8). Similarly, addition of acetylcoenzyme A to human S9 fractions markedly diminished covalent binding associated with SMX-HA (Figure 8). Addition of UDP-glucuronic acid or adenosine

Covalent Binding of Sulfamethoxazole

Figure 8. (Top) Effect of N-acetyltransferase activity on covalent binding of SMX-HA to microsomal protein. SMX-HA (100 µM) was incubated with rat hepatic microsomes (2 mg/ mL) in the presence of NADPH for 2 h (microsomes only). Cytosol (2 mg/mL), acetylcoenzyme A (AcetylCoA; 400 µM), or GSH (1 mM) was included as indicated. Immunoblotting was performed as in Figure 3 on 7.5% gels. (Bottom) Effect of N-acetyltransferase on covalent binding of SMX-HA to human S9 fractions. SMX-HA (100 µM) in the presence of NADPH was incubated with human S9 fractions (4 mg/mL) and acetylcoenzyme A (400 µM) or GSH (1 mM) as indicated for 120 min. AcetoxySMX (100 µM) was incubated with human liver microsomes (2 mg/mL) with or without 1 mM GSH for 120 min. No acetoxySMX was detectable at the end of the incubation period. Microsomes incubated with 100 µM SMX-HA for 120 min are shown for comparison. Immunoblotting was performed as in Figure 3 on 12% gels.

3′-phosphate 5′-phosphosulfate and cytosol had no detectable effect on the quality or quantity of covalent binding associated with SMX-HA (data not shown). Incubation of synthetic acetoxySMX with hepatic microsomes was also associated with covalent binding. There was a diffuse pattern of binding associated with higher molecular weight proteins and, with extended incubation times, adduction of proteins in a similar pattern to that seen with SMX-HA (Figure 8). HPLC analysis demonstrated complete conversion of acetoxySMX to SMX-HA and SMX during the course of the incubation. In an attempt to minimize the role of secondary metabolism of acetoxySMX in decreasing covalent binding of acetylated metabolites, lysates from Escherichia coli overexpressing the N-acetyltransferases NAT1 and NAT2 were incubated with SMX-HA in the presence and absence of acetylcoenzyme A. As was observed with rat liver fractions, covalent binding associated with SMX-HA was markedly diminished by the addition of acetylcoenzyme A to the incubations (Figure 9). To assess the potential binding of SMX-HA associated metabolites to other subcellular fractions, SMX-HA (100 µM) was incubated with various rat liver subcellular fractions (5 mg/mL) in the presence of 1 mM NADPH and covalent binding determined by immunoblotting (Figure 10). Surprisingly little covalent binding occurred to cytosolic and mitochondrial fractions, although extensive

Chem. Res. Toxicol., Vol. 9, No. 2, 1996 505

Figure 9. Covalent binding of SMX-HA to E. coli lysates overexpressing NAT1 and NAT2 enzymes in the presence and absence of acetylcoenzyme A. E. coli lysates (2 mg/mL) were incubated with SMX-HA (100 µM) with and without 400 µM acetylcoenzyme A for 30 min. Immunoblotting was performed as described under Experimental Procedures. The dark background bands (arrows) were present in all blots of E. coli lysates and may represent E. coli antibodies present in the rabbit in which antiserum to SMX was elicited.

Figure 10. Covalent binding of SMX to subcellular fractions of rat liver. S9, membrane, cytosol, and mitochondrial fractions of rat liver (5 mg/mL) were incubated with SMX-HA (100 µM) in the presence of 1 mM NADPH for 30 min; 20 µg of protein was then separated on 16 cm 10% SDS-PAGE and immunoblotted with anti-SMX serum as described in Figure 3.

covalent binding occurred in the plasma/nuclear membrane and S9 fractions.

Discussion It is generally accepted that generation of reactive metabolites and subsequent covalent binding are necessary steps in the pathogenesis of sulfonamide hypersensitivity reactions. The formation of SMX-HA is the

506 Chem. Res. Toxicol., Vol. 9, No. 2, 1996

critical first step in the generation of the reactive metabolites mediating these reactions. However, the steps linking SMX-HA with the clinical expression of toxicity have not been elucidated. The relative roles of direct cytotoxicity and immune-mediated toxicity in precipitating the clinical syndrome have not been established nor has the ultimate reactive metabolite been conclusively identified. The aim of these studies was to further characterize the metabolism and reactivity of the SMX metabolites as they relate to binding to potential target proteins. In contrast to previous studies with halothane (2) and diclofenac (16), no significant covalent binding of SMX was observed in rats in vivo. Manipulation of exposure conditions (e.g. higher doses, glutathione depletion) may sufficiently imbalance bioactivation and detoxification reactions to an extent allowing covalent binding to occur, but it appears that SMX is different from diclofenac and halothane in its propensity to form covalent adducts in the rat. Maximum plasma concentrations of SMX can reach 1.5-2 mM in patients on high-dose trimethoprimsulfamethoxazole (20 and 100 mg kg-1 day-1, respectively). Incubation of SMX and SMX-HA with rat liver slices at high concentrations did lead to the formation of covalent adducts which could be detected, indicating that the method employed was able to detect low-level covalent binding. The most highly bound protein by SMXHA appeared to be a plasma membrane protein. The other proteins identified were only poorly detected, but they did correspond in size to proteins that were adducted in in vitro incubations with microsomal protein. Their greater detection in S9 fractions than in cytosolic fractions suggests they were microsomal proteins, but we did not demonstrate them directly in that fraction due to the small amount of microsomal protein obtained. These findings strongly suggest that under normal circumstances SMX does not cause extensive covalent binding in the rat, a species in which sulfamethoxazole hypersensitivity reactions or overt cytotoxicity does not occur. We examined only covalent binding to liver protein, a known target organ in SMX hypersensitivity reactions. However, other tissues, particularly the skin, may be more commonly involved, and it is possible that covalent binding may occur in those tissues in the absence of binding to liver protein. Halothane is known to form significant covalent adducts in human liver during normal exposure and in the absence of toxicity or manifestations of hypersensitivity reactions (18). The extent to which SMX forms covalent adducts in vivo in humans is unknown, although a recent report identifying a sulfamethoxazole-protein conjugate in the plasma of individuals taking SMX (19) suggests that this does occur. It is also possible that significant covalent binding of SMX reactive metabolites may only occur in vivo in patients with genetic or environmental alterations in metabolism or cellular defense pathways which allow accumulation of toxic, reactive metabolites. The patterns of covalent binding to human and rat microsomal proteins associated with SMX and SMX-HA in the presence of NADPH were the same, supporting a role for SMX-HA in forming SMX-dependent covalent adducts. As has been previously reported for procainamide hydroxylamine (16), NADPH markedly diminished the covalent binding associated with SMX-HA and GSH completely eliminated the binding. The effect of NADPH on binding of SMX-HA is through both decreased autoxidation of SMX-HA to nitrosoSMX (5) and reduction of

Cribb et al.

SMX-HA to SMX (10). The stabilizing effect of NADPH is most apparent in microsomes from humans and untreated rats, in which reduction of SMX-HA to SMX is minimal (