Introduction

Top Abstract Introduction Methods Results Discussion References

Sulfonamide antimicrobials cause a delayed-onset, hypersensitivity-type syndrome characterized by fever, skin rash and multiorgan toxicity, including hepatotoxicity and blood dyscrasias, occurring 7 to 14 days after initiation of therapy (reviewed in Cribb et al., 1996b). This syndrome is frequently called the "sulfonamide hypersensitivity reaction" and will be referred to as such in this report. It appears to be clinically distinct from the IgE-mediated sulfonamide allergy (immediate hypersensitivity reaction), which is rapid in onset, accompanied by urticaria or symptoms of anaphylaxis and usually nonfebrile (Cribb et al., 1996b). Sulfonamides are also known to cause a similar delayed-onset hypersensitivity syndrome in dogs characterized by fever, polyarthritis, blood dyscrasias and hepatotoxicity (Cribb, 1989). The delayed onset, fever, rash and frequent eosinophilia observed in humans and dogs are consistent with an immune-mediated pathogenesis. However, drug-dependent antibodies or sensitized T cells have only rarely been demonstrated in patients with sulfonamide hypersensitivity reactions (Cribb et al., 1996b).

Drugs are thought to not be capable of directly functioning as an immunogen and eliciting an immune response because of their small size; they must first covalently bind to cellular proteins after metabolic activation to reactive intermediates (Park et al., 1987; Pohl et al., 1988). SMX, one of the most commonly used sulfonamide antimicrobials, is N-hydroxylated by cytochrome P450 (CYP2C6 in rats and CYP2C9 in humans) to an hydroxylamine (SMX-HA) (Cribb et al., 1995a; Cribb and Spielberg, 1992). SMX-HA is subsequently oxidized to a nitroso metabolite that is cytotoxic and binds to cellular proteins (Cribb et al., 1991; 1996a). If appropriately presented to the immune system, these drug/protein conjugates could elicit an immune response directed against the hapten (covalently bound drug), against hapten/protein conjugates or against the protein carrier (Park et al., 1987; Pohl et al., 1988; Kenna et al., 1988; Pumford et al., 1993).

Antibodies recognizing native ER proteins (not covalently modified by the causative drug) have been found in the sera of patients with halothane (Pumford et al., 1993), tienilic acid (Beaune et al., 1987) and dihydralazine hepatitis (Bourdi et al., 1990) and in aromatic anticonvulsant hypersensitivity reactions (Leeder et al., 1992). We therefore investigated whether patients with a clinical history of delayed-onset sulfonamide hypersensitivity reactions have similar antibodies in their sera. Furthermore, we made a preliminary identification of some of the proteins recognized by the antibodies in the patient sera.

	Methods

Top Abstract Introduction Methods Results Discussion References

Human sera. All samples were collected in accordance with the Hospital for Sick Children Scientific and Ethics Review Committees and after informed, written consent was obtained. Patients were drawn from primary care patients and from patients referred to the Division of Clinical Pharmacology and Toxicology, Hospital for Sick Children, for investigation of a clinical history of an adverse drug reaction. Group I subjects were healthy individuals (n = 11) chosen from students and staff of the Hospital for Sick Children who had no history of sulfonamide adverse drug reactions. Group II patients (n = 21) had a history consistent with delayed-onset sulfonamide hypersensitivity reactions (fever and skin rash with or without additional organ toxicity starting 7-14 days after initiation of therapy). Fifteen of these patients (group IIA) tested positive with the in vitro microsomal toxicity assay previously described (Riley et al., 1991; Shear et al., 1986), whereas 6 patients (group IIB) tested negative. The in vitro microsomal toxicity assay tests the susceptibility of mononuclear leukocytes isolated from the patient to the toxic effect of SMX-HA formed by mouse liver microsomes from SMX in the presence of NADPH (Riley et al. 1991; Shear et al. 1986). An increase in toxicity over base line of >7% (control population mean ± 3 S.D.) was considered positive. A positive microsomal toxicity assay has been shown to correlate with the occurrence of sulfonamide hypersensitivity reactions and increases the likelihood that a hypersensitivity reaction had in fact occurred in the patient. A negative assay does not eliminate a diagnosis of sulfonamide hypersensitivity. Group III patients (n = 18) had received a sulfonamide antimicrobial and experienced a clinical event not consistent with a delayed-onset sulfonamide hypersensitivity reaction (e.g., urticarial rash with an onset of <7 days, swelling, vomiting or diarrhea; Cribb et al., 1996b). All patients in this group tested negative in the in vitro microsomal toxicity assay with SMX. Five of the 18 patients had signs consistent with IgE-mediated sulfonamide allergy (e.g., urticarial rash with onset of <7 days, anaphylaxis), whereas the other patients experienced a variety of clinical events (e.g., swelling, vomiting or diarrhea).

Rat liver microsomes. Rats were treated in accordance with the Merck Institutional Animal Care and Use Committee Guidelines. Rat hepatic microsomes were prepared from freshly isolated livers of 200- to 225-g male CRL(CD)BR SD strain of rats (Charles River Laboratories, Raleigh, NC) that had been treated with 0.9% saline, dexamethasone (150 mg/kg/day in saline with 2% Tween 80), -naphthoflavone (80 mg/kg/day in corn oil) or phenobarbital (80 mg/kg/day in saline) intraperitoneally at 24-hr intervals for 72 hr as previously described (Cribb et al., 1995a). The protein concentration of the microsomal fractions was determined with the BioRad (Hercules, CA) D/C Protein Assay Kit.

Covalent binding of SMX-HA/nitroso-SMX to rat hepatic microsomes. SMX-HA (100µM) was incubated with hepatic microsomes (4 mg/ml) isolated from phenobarbital-pretreated rats for 30 min at 37°C in PBS, pH 7.4. This allows SMX-HA to spontaneously oxidize to nitroso-SMX, which covalently binds to protein (Cribb et al., 1991, 1996a). The reaction was stopped by the addition of Laemmli's buffer without reducing agents (Cribb et al., 1996a) and heated to 95°C for 3 min. Covalent modification of microsomal protein was confirmed by immunoblotting with anti-SMX antibodies (see below for methods and fig. 1). Immunodetectable adducts were located at ~45 to 55, 76 and 80 (previously designated as 70-kDa proteins in our laboratory), 96 and 100, and 150 kDa, which is consistent with previous findings in our laboratory (Cribb et al., 1996a). Several different preparations of covalently modified microsomes were used over the course of these studies and showed some variation in the degree of covalent modification of specific proteins, but the same proteins were consistently bound by SMX, as previously reported (Cribb et al., 1996a). Hepatic microsomes from phenobarbital pretreated rats were used because phenobarbital induces the cytochrome P450 involved in the bioactivation of SMX, CYP2C6 (Cribb et al., 1995a). Other cytochromes P450 are also induced.

View larger version (24K): [in this window] [in a new window]   Fig. 1.   Immunoblots of patient sera against native and SMX-modified hepatic microsomes from phenobarbital-pretreated rats. Immunoblotting was conducted with patient sera at a 1:1000 dilution against native () and SMX-HA modified (+) hepatic microsomes (50 mg) separated by SDS-PAGE (12% and 10% gels). Sera from control subjects (9 and 33) did not recognize native or SMX-modified microsomal proteins. Subject 3 is an example of a patient with a sulfonamide hypersensitivity reaction whose serum did not have antibodies against native or SMX-modified microsomal protein. Subject 13 was the only patient examined whose serum contained antibodies preferentially recognizing proteins modified by SMX-reactive metabolites. Sera from all other positive patients showed similar immunoreactivity toward control and SMX-HA-modified microsomes, as exemplified by patient 19 (55-kDa protein). For patients 45 (54-, 55- and 96-kDa proteins), 7 (55-kDa protein) and 25 (80-kDa protein), only immunoreactivity against native microsomes () is shown because result with SMX-modified and native microsomes were identical. See tables 1 and 2 for a summary of results, and Methods for a description of subject groups. Immunoblotting with anti-SMX antiserum (-SMX) to identify proteins modified by SMX and the molecular weight of major immunoreactive proteins are shown.

Antibodies. Polyclonal antibodies recognizing SMX were elicited in rabbits against an SMX/keyhole limpet hemocyanin conjugate as previously described (Cribb et al., 1996a). Antibodies against the following proteins recognized by patients with halothane hepatitis were elicited in rabbits as previously described: anti-57 kDa (PDI) (17), anti-82 kDa (grp78; BiP) and anti-100 kDa (grp94) (Thomassen et al., 1990). Additional antibodies against stress proteins were obtained from StressGen Biotechnologies Corp. (Victoria, BC). SPA 827 is an anti-peptide monoclonal antibody against grp78 that recognizes grp94 and grp78 and a 40-kDa unidentified protein. SPA 890 is a polyclonal antibody against bovine PDI that recognizes human and rat PDI. Antibodies against cytochrome P450 were obtained as follows: anti-CYP2D6 was elicited in rabbits as previously described (Cribb et al., 1995a); anti-CYP3A, anti-CYP2B and anti-CYP1A were obtained from Human Biologics (Phoenix, AZ); anti-CYP2E1 was from Amersham Corp (Arlington Heights, IL) and anti-CYP2C6 was from Dacchei Pure Chemicals (Tokyo, Japan). All secondary antibodies were obtained from Amersham.

Immunoblotting. Immunoblotting was performed according to standard protocols (Cribb et al., 1995b, 1996a). When screening for the presence of anti-microsomal antibodies in the patient sera, native and SMX-modified rat hepatic microsomes (50 µg/lane) were separated by SDS-PAGE under nonreducing conditions on 16-cm 10% and 12% acrylamide gels and then transferred by semidry electroblotting to nitrocellulose membranes. Membranes were blocked for 1 hr with PBS containing 5% nonfat dry milk and 2% bovine serum albumin (blocking buffer) at room temperature. Immunoblotting was performed using a 1:1000 dilution of human serum. All antisera were added in an antibody dilution buffer (blocking buffer containing 0.1% Tween 20). Membranes were then incubated at 4°C overnight. After an initial rinse, the membranes were washed twice with PBS and 0.1% Tween 20 for 15 min. Horseradish peroxidase-linked anti-human Ig (recognizing IgA, IgM and IgG) was then added in antibody dilution buffer at a dilution of 1:5000. After incubation for 1 hr at room temperature, washing was repeated as described above. Bound IgG was visualized using enhanced chemiluminescence (ECL Kit; Amersham) and autoradiographic film. A positive response was defined as a detectable signal clearly above background that was reproducible on two or more immunoblots (unless quantity of serum limited analysis to one immunoblot).

Extraction of microsomal proteins with sodium deoxycholate. Peripheral ER proteins were extracted as previously described (Butler et al., 1992). Briefly, 1 ml of rat hepatic microsomes (40 mg/ml of protein) was resuspended in 4 ml of 10 mM Tris · HCl, pH 7.5, containing 0.1 mM EDTA and 0.1% (w/v) DOC and gently stirred for 1 hr at 4°C. After centrifugation (100,000 × g at 4°C for 2 hr), the supernatant was harvested, and the pellet was resuspended in 1 ml of PBS. Microsomes and extracted microsomes were diluted to 2 mg/ml in PBS. Aliquots of microsomes, extracted microsomes and supernatant were incubated with 100 mM SMX-HA for 30 min at 37°C.

Immunoprecipitation. Immunoprecipitation of microsomal proteins was performed as described (Amouzadeh and Pohl, 1995) with the following modifications: 100 ml of a 10% dilution of antiserum in Tris-casein buffer, 360 ml of PE buffer [Dulbecco's PBS, pH 7.2, 1 mM EDTA and 1% Nonidet P-40 (v/v)], and 40 ml of protein A-agarose beads (2.6 mg/ml) were mixed at 4°C for 1 hr in a 1.5-ml microcentrifuge tube on a rotary mixer. The tube was centrifuged at 8200 × g for 3 min at 4°C. The supernatant was removed, and the pellet was washed twice with PE buffer. Then, 500 mg of microsomal protein in 500 ml of PIB (PE buffer with 1 mM phenylmethylsulfonyl fluoride and 1 mM leupeptin) was added to the protein A-agarose bead pellet. The tubes were mixed overnight on a rotary mixer at 4°C. The agarose beads were collected by centrifugation as above, and the pellet was washed twice with PIB and three times with PE buffer. The pellet was resuspended in 120 ml of Laemmli's buffer with 5% -mercaptoethanol and heated to 95°C for 3 min. The tubes were then centrifuged at 8200 × g for 3 min. The supernatants (20 ml/lane for immunoblotting with anti-stress protein antibodies and 40 ml/lane for immunoblotting with the patient sera) were submitted to SDS-PAGE as described below.

	Results

Top Abstract Introduction Methods Results Discussion References

Subjects in the control group, group I (n = 11), showed no significant immunoreactivity toward native or SMX-modified rat hepatic microsomal protein by immunoblotting under the conditions used. Seventeen of 21 (81%) patients with delayed-onset sulfonamide hypersensitivity reactions (group II) had antibodies in their sera that recognized one or more native rat hepatic microsomal proteins. Three proteins of ~55, ~80 and ~96 kDa were the major antigenic targets (table 1; fig. 1). Sixty-seven percent (14 of 21) of the patients had antibodies recognizing the 55-kDa protein, 19% (4 of 21) had antibodies recognizing the 80-kDa protein and 14% (3 of 21) had antibodies recognizing the 96-kDa protein (table 2). One patient (patient 46) recognized an additional 30-kDa protein, and one patient (patient 45) recognized an additional protein of ~54 kDa.

The major proteins of 55, 80 and 96 kDa corresponded to proteins that were targets of covalent binding by SMX-reactive metabolites (fig. 1). However, only 1 patient (patient 13) had antibodies in the serum whose recognition of microsomal protein was modified by the presence of covalently bound SMX. Antibodies in the serum of patient 13 strongly recognized all proteins containing immunodetectable SMX adducts (as determined with rabbit anti-SMX antiserum) and also weakly recognized the native 55- and 80-kDa proteins in the absence of covalently bound SMX. The antibodies present in the serum of patient 13 also recognized SMX covalently linked to bovine serum albumin (by enzyme-linked immunosorbent assay and immunoblot; not shown), and immunoreactivity could be competed out with free SMX (not shown), confirming that the antibodies recognized the hapten specifically. None of the other patient sera in group II showed differential immunoreactivity between SMX-modified and native microsomal proteins. This is consistent with previous work in our laboratory in which we were unable to demonstrate IgM or IgG antibodies recognizing SMX in a group of patients with SMX hypersensitivity reactions using an enzyme-linked immunsorbent assay.¹ Two patients had received non-SMX-containing sulfonamide antimicrobials, and 5 had received an unspecified sulfonamide antimicrobial (table 1); it is possible that they had anti-sulfonamide antibodies that were not cross-reactive with SMX.

There was no difference in the incidence of anti-microsomal protein antibodies in patients with a positive in vitro microsomal toxicity assay and in patients with a negative in vitro microsomal toxicity assay (table 2). In addition to these patients, patient 2 (the sister of individual with a sulfonamide hypersensitivity reaction and a positive in vitro microsomal toxicity assay) had antibodies that recognized the 80-kDa protein and weakly recognized the 55-kDa protein.

In contrast, only 1 of 17 group III patients had an antibody response against microsomal proteins. This patient (patient 25; fig. 1) was 18 months old and had had diarrhea, vomiting, fever and rash 1 day after receiving sulfisoxazole acetyl/erythromycin and tested negative in the in vitro microsomal toxicity assay. The patient had antibodies recognizing the 80-kDa protein. Three patients in group III had received sulfisoxazole acetyl/erythromycin combination products and so may have experienced an erythromycin-mediated event. None of the patients in group III had IgG or IgM antibodies recognizing SMX-modified microsomal protein.

A series of experiments were undertaken to try to identify the 55-, 80- and 96-kDa proteins. Because of its molecular weight and previous reports of anti-cytochrome P450 antibodies in drug-induced hepatotoxicities (Beaune et al., 1987; Bourdi et al., 1990; Leeder et al., 1992), we hypothesized the 55-kDa protein might be a cytochrome P450. Serum recognizing the 55-kDa protein was immunoblotted against hepatic microsomes from control rats and those treated with the cytochrome P450 inducers phenobarbital, dexamethasone and -naphthoflavone (fig. 2). The recognition of the 55-kDa protein by the patient serum was not affected by treatment despite the fact that cytochrome P450 levels were clearly induced as indicated by immunoblotting with anti-cytochrome P450 antibodies (fig. 2). Furthermore, the 55-kDa protein did not comigrate with CYP2C6, CYP2B1/2, CYP2D, CYP2E1 or CYP3A, as confirmed by immunoblotting (not shown).

View larger version (32K): [in this window] [in a new window]   Fig. 2.   Cytochrome P450 inducers do not affect the expression of the 55-kDa protein. Hepatic microsomes from rats treated with saline (control), phenobarbital (PB), dexamethasone (DEX) or -naphthoflavone (BNF) were separated by SDS-PAGE (50 mg/lane) and immunoblotted with serum (patient 7) recognizing the 55-kDa protein. Immunoblotting was also performed with -CYP2C6, -CYP2B and -CYP3A antiserum to confirm induction of cytochrome P450.

We observed previously that the hepatic microsomal proteins of ~55, ~80 and ~100 kDa covalently modified by SMX-reactive metabolites (Cribb et al., 1996a) comigrated with proteins that were targets of covalent binding by the trifluoroacetyl chloride reactive metabolite of halothane (Pohl et al., 1988).² These proteins were thought to correspond to the luminal ER proteins PDI, grp78 and grp94, respectively, based on their migration pattern on SDS-PAGE and the knowledge that stress proteins are abundant microsomal proteins. Therefore, hepatic microsomes were immunoblotted with the anti-sera SPA 827 (anti-grp78/grp94) and SPA 890 (anti-PDI) to determine whether PDI, grp78 and grp94 comigrated with the proteins recognized by the patient sera and with proteins covalently modified by SMX-reactive metabolites (fig. 3). The 55-kDa and 80-kDa proteins recognized by the patient sera comigrated with PDI and grp78, respectively, and with proteins modified by SMX. Although the protein migrating at ~100 kDa covalently modified by SMX-reactive metabolites comigrated with grp94, the native 96-kDa protein recognized by the patient sera did not comigrate with grp94.

View larger version (30K): [in this window] [in a new window]   Fig. 3.   Comigration of 55- and 80-kDa proteins with PDI and grp78. Left, hepatic microsomes were separated by SDS-PAGE (12%), transferred to nitrocellulose and immunoblotted with patient sera recognizing the 70-kDa (patient 2, lane 2) or the 55-kDa (patient 19, lane 19) protein and with antibodies against PDI (SPA 890, lane 890) and grp78/grp94 (SPA 827, lane 827). Right, immunoblotting was performed as described above except 10% gels and patient 45, whose serum recognized 54/55- and 96-kDa proteins, were used. The 55-kDa protein comigrates with PDI, but the 96-kDa protein does not comigrate with grp94. The migration of proteins covalently modified by SMX is also shown using -SMX antisera (lane SMX). The 100-kDa protein covalently modified by SMX comigrates with grp94, whereas the 96-kDa protein recognized by patient 45 does not. Although not clearly visible in this immunoblot, a 96-kDa protein is also a target for covalentl modification by SMX-reactive metabolites and appears to comigrate with the 96-kDa protein recognized by patient antibodies.

To confirm the identities of the 55- and 80-kDa proteins recognized by the patient sera as PDI and grp 78, respectively, microsomes were treated with 0.1% DOC, a procedure known to extract luminal stress proteins from microsomal membranes (Martin et al., 1993a, 1993b). The original microsomal preparation, the microsomal pellet after extraction and the DOC extract were then incubated with SMX-HA to identify targets of covalent binding. The proteins were separated by SDS-PAGE, transferred to nitrocellulose and immunoblotted with serum from patient 13, which recognized the native 55- and 80-kDa proteins as well as covalently bound SMX. Treatment with DOC extracted the 55-kDa, 76/80-kDa (two bands), 96/100-kDa (two bands) and 150-kDa proteins that were covalently modified by SMX (fig. 4, top). Extraction of PDI, grp78 and grp94 by treatment with DOC was also confirmed by immunoblotting (fig. 4, bottom). Thus, the 55-, 80- and 96-kDa proteins recognized by the patient serum and the 76- and 100-kDa proteins modified by SMX were extracted by DOC, which is consistent with their being loosely adherent ER proteins.

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Extraction of proteins modified by SMX-HA and recognized by patient sera from rat liver microsomes by 0.1% DOC. Top, liver microsomes before extraction (lane 1), after extraction (lane 2) and the extract (lane 3) were incubated with SMX-HA (100 mM), separated by SDS-PAGE (50 mg/lane), transferred to nitrocellulose and immunoblotted with serum from patient 13 that recognized the 55- and 80-kDa native proteins as well as covalently bound SMX metabolites. Treatment with DOC removed the 55-, 76/80-, 96/100- and 150-kDa proteins from the microsomes. The 45- to 55-kDa proteins thought to be cytochrome P450 were not extracted by this treatment. The molecular weights of the major SMX-modified proteins are shown. Bottom, treatment with DOC also extracted PDI (immunoblotted with antibody SPA 890, labeled as 890) and grp78 and grp94 (immunoblotted with SPA827, labeled as 827).

Finally, when PDI and grp78 were immunoprecipitated from liver microsomes with specific antibodies (anti-57 kDa for PDI and anti-82 kDa for grp78) and the immunoprecipitated proteins analyzed by immunoblotting with patient sera and anti-PDI or anti-grp78 sera (fig. 5), it was found that immunoprecipitated PDI was recognized by patient sera recognizing the 55-kDa protein and grp78 was recognized by patient sera recognizing the 80-kDa protein. However, immunoprecipitated grp94 (with anti-100-kDa antiserum) did not comigrate with the 96-kDa protein recognized by patient sera (data not shown).

View larger version (22K): [in this window] [in a new window]   Fig. 5.   Recognition of immunoprecipitated grp78 and PDI by patient sera. Top, grp78 was immunoprecipitated with the anti-82-kDa serum, resolved on a 10% 16-cm SDS-PAGE along with hepatic microsomes and immunoblotted with either anti-grp78 serum (-82-kDa) or patient serum recognizing the 80-kDa protein (2). Bottom, PDI was immunoprecipitated with the anti-57-kDa serum, resolved on a 7.5% 16-cm SDS-PAGE along with rat hepatic microsomes and immunoblotted with either anti-PDI (-57-kDa) serum or patient sera recognizing the 55-kDa protein (1 and 19). The diffuse band below the 57-kDa PDI represents immunoprecipitated IgG fraction. IP, immunoprecipitated protein; MIC, rat hepatic microsomes.

	Discussion

Top Abstract Introduction Methods Results Discussion References

Sulfonamide antimicrobials have been associated with a variety of adverse events (Cribb et al., 1996b). We have studied patients who have experienced what are referred to as sulfonamide hypersensitivity reactions; these are delayed-onset systemic reactions characterized by fever, skin rash, blood dyscrasias and multiorgan toxicity occurring 7 to 14 days after the start of therapy. Some patients will experience toxicity associated predominantly with one organ (i.e., dermatopathy, hepatotoxicity or agranulocytosis). Although the clinical signs of delayed sulfonamide hypersensitivity reactions are consistent with an immune-mediated pathogenesis, direct evidence in support of this hypothesis has been sparse. The delayed-onset sulfonamide hypersensitivity reaction is distinct from the immediate hypersensitivity reactions or IgE-mediated sulfonamide allergies that have also been reported (Cribb et al., 1996b). Clinically, the immediate hypersensitivity reaction or sulfonamide allergy is characterized by a rapid onset of symptoms of anaphylaxis (usually immediate) and/or urticaria (usually within 1 to 2 days). These individuals have been shown to have anti-SMX IgE antibodies and do not appear to share the same biochemical risk factors as individuals with sulfonamide hypersensitivity reactions (reviewed in Cribb et al., 1996b). Failure to distinguish clinically between these two different syndromes has led to some confusion in examination of the pathogenesis of these reactions.

We have previously shown (Cribb et al., 1996a) in vitro that oxidation of SMX to SMX-HA and, subsequently, nitroso-SMX leads to covalent binding to specific microsomal proteins. Using native and SMX-modified hepatic microsomes from rats pretreated with phenobarbital as the solid phase in an immunoblotting assay, we screened sera from patients with delayed-onset sulfonamide hypersensitivity reactions for the presence of antibodies directed against the hapten (SMX), hapten/protein conjugates or native microsomal proteins. The occurrence of anti-microsomal protein antibodies has been demonstrated in the sera of patients experiencing several different drug-induced toxicities: halothane hepatitis (Pohl et al. 1988), tienilic acid hepatitis (Beaune et al., 1987), dihydralazine hepatitis (Bourdi et al., 1990) and aromatic anticonvulsant hypersensitivity syndromes (Leeder et al., 1992). In the latter three, antibodies are directed against native cytochrome P450 proteins, whereas patients with halothane hepatitis have antibodies against multiple trifluoroacetylated microsomal proteins as well as against native proteins (Kenna et al., 1988; Pohl et al., 1988; Pumford et al., 1993).

Eighty-one percent (17 of 21) of patients with clinical signs consistent with a sulfonamide hypersensitivity reaction had antibodies against one or more microsomal proteins. Only one patient (patient 13) had antibodies that recognized the SMX hapten. Three patients in which antibodies were not detected had their serum samples collected >1 year after the reaction had occurred. The antibodies may have been eliminated or were of too low a concentration to be detected by the assay that was used. Only 1 patient who had received sulfonamides and had a clinical event not consistent with a sulfonamide hypersensitivity reaction had antibodies against a microsomal protein. This patient was a 18-month-old child who experienced diarrhea, vomiting, fever and rash 1 day after receiving sulfisoxazole acetyl/erythromycin and tested negative in the in vitro microsomal toxicity assay. The patient had antibodies recognizing the 80-kDa protein. None of the control individuals, several of whom were known to have taken sulfonamides (SMX and sulfamethazine) uneventfully in the past, had detectable antibodies. Similarly, an additional 40 control subjects screened in other studies by similar immunoblotting techniques did not show these antibody responses (Leeder et al., 1992).³

Although SMX was not a necessary part of the epitope recognized by antibodies in patient sera, the proteins recognized by patient antibodies migrated with proteins that were targets in vitro of covalent binding by nitroso-SMX, the reactive metabolite of SMX formed through oxidation to SMX-HA (Cribb et al., 1996a). This suggests that modification of protein or proteins by SMX may be involved in the process or processes by which the proteins initially become immunogenic or are made accessible to the immune system. Identification of the protein targets is necessary for the elucidation of these processes.

The 55-kDa protein recognized by the patient sera was initially postulated to be a cytochrome P450. However, expression of the 55-kDa protein recognized by patient sera was not altered by inducers of cytochrome P450 and it did not comigrate with any of the major subfamilies of cytochrome P450 that we examined. The 55-kDa protein did, however, comigrate with the 57-kDa PDI and was extracted from microsomal membranes by 0.1% DOC in a manner similar to PDI. Immunoprecipitated PDI comigrated with the 55-kDa protein and was recognized by antibodies against the 55-kDa protein found in patient sera, further supporting that PDI was the 55-kDa antigenic target. Similar experimental evidence led to the conclusion that the 80-kDa protein recognized by the patient sera was grp78.

In contrast, the 96-kDa protein recognized by sera of 3 patients did not comigrate with grp94, although it, too, was extracted from hepatic microsomes by DOC. The 100-kDa protein covalently modified by SMX-reactive metabolites did, however, comigrate with grp94 and may in fact be this protein. It has been reported that different forms of grp94 exist with different immunorecognition sites and with different apparent molecular weights on SDS-PAGE (Feldweg and Srivastava, 1995). Thus, it is still possible that the 96-kDa protein recognized by the patient sera is a form of grp94.

Stress proteins/molecular chaperones are known to play important roles in protecting cells against chemical and physical stress, to help process and fold nascent and denatured proteins and to play a role in antigen presentation (Jacquier-Sarlin and Polla, 1994; Polla et al., 1993). The cytosolic heat shock proteins, which show immunological cross-reactivity with some of the microsomal stress proteins, are highly immunogenic, targets of immunological responses against bacterial pathogens and postulated to play important roles in allergy and autoimmunity (Polla et al., 1993). Because of the marked conservation of these families of proteins across species, the possibility that immunological cross-reactivity could lead to autoimmune disease or susceptibility to hypersensitivity reactions as a result of molecular mimicry (with, for example, bacterial proteins) is an attractive hypothesis. It must also be taken into account that such potential immunological cross-reactivity prevents us from concluding definitively that the proteins recognized in rat hepatic microsomes are in fact the proteins against which the antibody response is directed in vivo in humans.

Nevertheless, these proteins are also targets in human and rat liver microsomes for covalent binding of SMX-reactive metabolites in vitro (Cribb et al., 1996a). Stress proteins have been shown in many instances to be necessary for cell survival (Jacquier-Sarlin and Polla, 1994; Polla et al., 1993). The possibility that covalent binding to these proteins leads to a direct cytotoxicity must be explored. Whether antibodies against these or related proteins play a critical role in the pathogenesis of sulfonamide hypersensitivity reactions or whether the antibody response is an epiphenomenon is not known. These patients, although having circulating antibodies, do not currently have any clinical signs. Covalent binding of SMX, which does not appear to be part of the epitope, may alter either the expression, function or immunological presentation of the target protein, leading to the occurrence of an immune response.

Although these patients have antibodies that recognize at least three major proteins in rat liver hepatic microsomes, it is not yet known whether the target proteins recognized in rat liver are the same as those target proteins recognized in human liver or skin (a major target organ), nor is it known whether antigenic targets exist in other subcellular fractions or tissues. Nevertheless, the presence of antibodies against the 55-, 80- or 96-kDa rat liver microsomal protein may be a useful diagnostic aid.