Multiple Adduction Reactions of Nitroso Sulfamethoxazole with

Apr 9, 2009 - These findings pose many other questions as to the identity of the ...... Toxicity: A Perspective Based on the Critical Examination of T...
0 downloads 0 Views 1MB Size
Download PDF
Chem. Res. Toxicol. 2009, 22, 937–948

937

Multiple Adduction Reactions of Nitroso Sulfamethoxazole with Cysteinyl Residues of Peptides and Proteins: Implications for Hapten Formation Hayley E. Callan,† Rosalind E. Jenkins,† James L. Maggs,† Sidonie N. Lavergne,† Stephen E. Clarke,‡ Dean J. Naisbitt,† and B. Kevin Park*,† MRC Centre for Drug Safety Science, Department of Pharmacology and Therapeutics, School of Biomedical Sciences, The UniVersity of LiVerpool, LiVerpool L69 3GE, and DMPK, GlaxoSmithKline, Ware, Hertfordshire SG12 0DP, United Kingdom ReceiVed January 29, 2009

Sulfamethoxazole (SMX) induces immunoallergic reactions that are thought to be a result of intracellular protein haptenation by its nitroso metabolite (SMX-NO mass, 267 amu). SMX-NO reacts with protein thiols in vitro, but the conjugates have not been defined chemically. The reactions of SMX-NO with glutathione (GSH), a synthetic peptide (DS3), and two model proteins, human GSH S-transferase π (GSTP) and serum albumin (HSA), were investigated by mass spectrometry. SMX-NO formed a semimercaptal (N-hydroxysulfenamide) conjugate with GSH that rearranged rapidly (1-5 min) to a sulfinamide. Reaction of SMX-NO with DS3 also yielded a sulfinamide adduct (mass increment, 267 amu) on the cysteine residue. GSTP was exclusively modified at the reactive Cys47 by SMX-NO and exhibited mass increments of 267, 283, and 299 amu, indicative of sulfinamide, N-hydroxysulfinamide, and N-hydroxysulfonamide adducts, respectively. HSA was modified at Cys34, forming only the N-hydroxysulfinamide adduct. HSA modification by SMX-NO under these conditions was confirmed with ELISA and immunoblotting with an antisulfonamide antibody. It is proposed that cysteine-linked N-hydroxysulfinamide and Nhydroxysulfonamide adducts of SMX are formed via the reaction of SMX-NO with cysteinyl sulfoxy acids. Evidence for a multistep assembly of model sulfonamide epitopes on GSH and polypeptides via hydrolyzable intermediates is also presented. In summary, novel, complex, and metastable haptenic structures have been identified on proteins exposed in vitro to the nitroso metabolite of SMX. Introduction The sulfonamide antibiotics, such as sulfamethoxazole (SMX;1 Figure 1), have been associated with numerous, idiosyncratic, adverse reactions in patients, and especially with severe necrolytic skin lesions (1-4). The incidence of these hypersensitivity reactions can be particularly high in HIVinfected patients (5). A number of hypotheses have been advanced to explain hypersensitivity reactions induced by the sulfonamides, incorporating elements of metabolic activation, metabolite-mediated cytotoxicity, and immunological mechanisms (5, 6). Sulfamethoxazole hydroxylamine (SMX-HA), a metabolite of SMX in humans and rats (7), and putative derivatives thereof are toxic to cells of the immune system in vitro (8, 9). SMX metabolites are thought to play additional roles in the pathogenesis of hypersensitivity reactions because they have been shown to activate dendritic cells (10) and to be immunogenic (11). The multiple derivatives of SMX that may form in vivo are shown in Figure 1. SMX can be hydroxylated at the N4 position by several enzymes, including CYP 2C9 (7, 10), CYP 2C8 (10), * To whom correspondence should be addressed. Tel: +44(0151)794 5559. Fax: +44(0151)794 5540. E-mail: [email protected]. † The University of Liverpool. ‡ GlaxoSmithKline. 1 Abbreviations: ACN, acetonitrile; DMSO, dimethyl sulfoxide; ELISA, enzyme-linked immunoadsorbant assay; GSH, glutathione; GSTP, glutathione S-transferase π; HSA, human serum albumin; SMX, sulfamethoxazole; SMX-HA, sulfamethoxazole hydroxylamine; SMX-NO, sulfamethoxazole nitroso.

and myeloperoxidase (12), to generate SMX-HA. SMX-HA can then autoxidize to the nitroso-SMX derivative (SMX-NO). Autoxidation occurs via a postulated nitroxide radical intermediate and is hypothetically linked to the production of superoxide (13). SMX-NO is unstable in solution and yields both the oxidation product nitro-SMX and the azoxy and azo dimers (14). Reduction of SMX-NO to SMX-HA in mononuclear leucocytes occurs via reaction with glutathione (GSH) and ascorbate (9). Enzymatic reduction of SMX-NO has not been reported, but rodent liver cytosol contains reductases that are active against p-nitrosophenol (15). This enzymatic and nonenzymatic reduction allows for the possibility that N4-oxidation metabolites are capable of cycling, generating reactive oxygen species, and leading to the oxidation of GSH to glutathione disulfide (GSSG), and subsequently, to toxicity (11, 13, 14, 16-19). The stable conjugation of these metabolites with GSH has not been found to be a pathway for the elimination of reactive metabolites of SMX, although GSH at physiological concentrations does seem to reduce the toxicity and protein binding associated with SMX reactive metabolites in vitro (20, 21). It has been proposed that GSH is capable of inhibiting autoxidation, possibly via reduction of the nitroxide radical intermediate. In addition, GSH was found to react spontaneously with SMXNO in vitro to produce a semimercaptal conjugate, which can rearrange to a stable sulfinamide and can also undergo thiolytic cleavage to yield SMX-HA (13, 22). Studies with N-R-acetylL-lysine and alanine have not revealed any evidence of SMXNO reactivity with amine nucleophiles (22).

10.1021/tx900034r CCC: $40.75  2009 American Chemical Society Published on Web 04/09/2009

938

Chem. Res. Toxicol., Vol. 22, No. 5, 2009

Callan et al.

Figure 1. Metabolism of SMX to SMX-HA by CYP2C9/CYP2C8/MPO (7, 10, 12), reduction of SMX-HA to SMX by NADH cytochrome b5 reductase and cytochrome b5 (58), autoxidation of SMX-HA to sulfamethoxazole nitroso (SMX-NO) (13), nonenzymic conjugation of SMX-NO with thiols (13), and reduction of SMX-NO to SMX-HA by cytosolic reductases (15).

Covalent binding of SMX metabolites to proteins has been investigated extensively using immunochemical techniques (11, 14, 16-18, 20, 23, 24), but a systematic analysis of the chemistry of adduct formation is still lacking. In vitro modification of cellular proteins by SMX, SMX-HA, and/or SMX-NO has been detected using drug-specific antibodies in T-lymphocyte and hepatoma cell lines (24), keratinocytes (16), rat liver microsomes and liver slices (20), and a lymphoma cell line (23), with 5-6 proteins of essentially the same molecular weights being detected in each study. In metabolically active liver preparations, treatment with SMX or SMX-HA resulted in the same profile of modified proteins (20), suggesting that exposure to exogenous reactive metabolites or those generated intracellularly may elicit similar responses. None of the modified proteins has been identified, except for an immunoreactive band comigrating with β-actin in SMX-NO-treated leukocytes (9, 23). In addition, Cheng et al. have recently detected SMX-NO adducts on albumin and IgG in an in vivo animal model (25). It is possible that their consistent detection is a reflection of their relative abundance rather than selective haptenation by SMX metabolites. In vivo modification of serum proteins has been detected by immunoblotting both in hypersensitive patients and in tolerant individuals, with adduct formation being more

common in the former (26, 27). Similar results have been obtained in hypersensitive and tolerant dogs administered SMX. However, no adducts were found in dogs experimentally exposed to SMX-NO (28). Haptenated splenocyte, lymphocyte, and keratinocyte proteins were detected by flow cytometry following SMX-NO treatment of rats and following SMX and SMX-HA treatment of GSH depleted rats (11). However, none of the modified proteins has been characterized. The majority of the above studies used only antibody binding as an end point to determine the degree of covalent binding. The adductive reactions of SMX-NO with GSH clearly suggest that the primary protein adduct is likely to be an unstable cysteinyl semimercaptal (13, 22). Indeed, Cribb et al. (20) found that the covalent binding of SMX’s reactive metabolites to human and rat liver microsomes was eliminated entirely by β-mercaptoethanol. In lymphoma cells, however, SMX-HA formed nonlabile protein adducts (23), the stability of which was attributed to rearrangement of semimercaptals to sulfinamides under nonreducing intracellular conditions. Subsequently, it was shown that chemical thiolation of human serum albumin (HSA) produced a substantial enhancement of the protein’s haptenation by SMX-HA and SMX-NO in vitro (17). Nevertheless, while the cysteinyl sulfinamide structure has been favored

Nitroso Sulfamethoxazole Adducts of Cysteine

as a potential antigenic determinant of sulfonamides (17, 22, 23), our group has noted that although sulfinamides are stable in neutral aqueous solutions, they are likely to undergo hydrolysis to the parent amine under the acidic conditions of exogenous antigen processing (22). Here, the reactions of SMX-NO with GSH, a synthetic peptide, and the model proteins HSA and glutathione S-transferase π (GSTP) were characterized by MS. This study will help to define the chemical entities responsible for triggering sulfonamide hypersensitivity and to determine whether the specific protein target is critical in the initiation of adverse events, as suggested by the critical protein hypothesis (29).

Experimental Procedures Caution: The following chemicals may stimulate allergic reactions in humans and should be handled carefully (SMX-NO). Materials. SMX-HA and SMX-NO were synthesized as described previously (22). Phosphate buffer was comprised of 13 mM KH2PO4 and 67.3 mM Na2PO4, pH 7.4. Rabbit anti-SMX antibody was a gift from Dr M. J. Rieder (University of Western Ontario) and was generated as outlined previously (24). Synthetic peptide DS3 (VLSPADKTNWGHEYRMFCQIG, >95% pure by HPLC) was prepared by Cambridge Research Biochemicals (Billingham, Cleveland, United Kingdom) and was a gift from Dr. C. K. Pease (Unilever Colworth, Bedfordshire, United Kingdom). Bacterial strains and plasmid vectors were produced as outlined previously (30). DTT was purchased from GE Healthcare (Buckinghamshire, United Kingdom). Sequencing-grade modified trypsin and iodoacetamide were purchased from Promega (Southampton, United Kingdom). Standard acetonitrile (ACN) and methanol were obtained from VWR International (Lutterworth, Leicestershire, United Kingdom), while HPLC grade solvents were obtained from Fisher Scientific (Leicestershire, United Kingdom). Dimedone (1,1-dimethyl-3,5-cyclohexanedione), H218O (97% 18O), and all other chemicals were purchased from Sigma-Aldrich (Poole, Dorset, United Kingdom). Western Blotting of Drug-Protein Adducts. Protein samples prepared in phosphate buffer were electrophoresed on 12% SDSpolyacrylamide gels under nonreducing conditions (Laemmli buffer; Bio-Rad, Hemel Hemstead, Hertfordshire, United Kingdom) without mercaptoethanol for 90 min and were transferred at 250 V and 4 °C to a polyvinylidene fluoride (PVDF) membrane over 1 h. Immunoblotting for drug-protein adducts was performed with antiSMX rabbit antiserum (1:500 dilution in phosphate buffer; incubated overnight at 4 °C). The antibody was generated as outlined previously (28, 31). Unbound antibody was removed by washing with PBS-Tween, and the membrane was incubated with peroxidase-conjugated antirabbit IgG antibody (1:250 dilution; 2 h at room temperature). The membrane was finally developed using a chemiluminescent substrate (Thermo Scientific, Cramlington, Northumberland, United Kingdom) and was scanned with a digital camera. SMX conjugated to keyhole limpet hemocyanin was used as a positive control, and HSA incubated with phosphate buffer served as a negative control. Immunoassay of Drug-Protein Adducts. Protein samples from incubations with SMX-NO were analyzed by enzyme-linked immunosorbant assay (ELISA). Ninety-six well microtiter plates were loaded with 10 µg protein/well and coated overnight at 4 °C. Anti-SMX rabbit antiserum (Panigen, Blanchardville, WI; 1:500 dilution; incubated for 4 h at room temperature) was used as the primary antibody followed by alkaline phosphatase-conjugated antirabbit IgG (1:1000 dilution; incubated for 1 h at room temperature) as the secondary antibody. Following addition of the alkaline phosphatase substrate for ELISA (Sigma-Aldrich), colorimetric readings were obtained at 450 nm on an MRX microtiterplate reader (Dynex Technologies, Worthing, West Sussex, United Kingdom). Reaction of SMX-NO with GSH. SMX-NO (final concentration, 5 mM) in dimethyl sulfoxide (DMSO; final concentration, 1%) was

Chem. Res. Toxicol., Vol. 22, No. 5, 2009 939 incubated with GSH at a molar ratio of 1:1 in either phosphate buffer or water (H216O or H218O) at room temperature. Aliquots (10 µL) were removed at intervals between 1 and 60 min for analysis by LC-MS. Modification of DS3 Peptide by SMX-NO. DS3 peptide (0.5 mM) was prepared in phosphate buffer immediately before addition of equimolar SMX-NO in DMSO (final concentration, 1%). Some incubations also contained either GSH or DTT (standard molar ratio DS3:thiol, 1:1). The solutions were incubated in the dark at room temperature (the peptide is photolabile) for various times before analysis by MS. DS3 and SMX-NO were also incubated in water (H216O or H218O). Trapping of Peptide Sulfenic Acid Intermediate. DS3 peptide (0.5 mM), SMX-NO dissolved in DMSO, and dimedone (final molar ratio, 1:1:3) in phosphate buffer were incubated in the dark at 37 °C. Aliquots (10-20 µL) were removed for immediate analysis by LC-MS. Expression, Isolation, and Modification of His-Tagged GSTP. The cDNA for human GSTP was cloned into the His-tag expression vector pET-15b (Novagen, Madison, WI) and expressed as previously described (30). Briefly, transformed Escherichia coli strain BL21 cells were grown under carbenicillin selection and induced by the addition of 2 mM isopropyl β-D-1-thiogalactopyranoside. The proteins were released by sonication using a Soniprep 150 (MSE, London, United Kingdom). The supernatant was recovered following centrifugation at 14000g and 4 °C for 10 min. Nickel-bead isolation was carried out according to Jenkins et al. (32). GSTP was eluted from the beads using 200 mM histidine before 60 µg (0.1 mM) was incubated with 1 mM SMX-NO in DMSO (final concentration, 1%) for 5-60 min at room temperature. The protein was precipitated with 9 volumes of ice-cold methanol and was recovered by centrifugation (14000g, 4 °C for 10 min) before solubilization in 20 µL of ammonium bicarbonate buffer (50 mM, pH 7.4). Samples were digested with trypsin (3 µg) overnight at 37 °C. Hydrolysates were desalted with C18 ZipTips (Millipore, Billerica, MA) according to the manufacturer’s instructions before resuspension of the peptides in 20 µL of 5% ACN/ 0.05% trifluoroacetic acid (TFA). Aliquots (1-2 µL) were then analyzed by LC-MS/MS. Modification of HSA by SMX-NO. HSA (150 µM) in phosphate buffer was incubated with SMX-NO (final concentration, 1.5 mM) dissolved in DMSO (final concentration, 1%) for up to 1 h at room temperature. Samples (50 µL) were reduced by incubation with DTT (1 mM, 15 min) at 55 °C and alkylated by incubation with iodoacetamide (55 mM, 15 min) at room temperature, before the protein was precipitated with 9 volumes of ice-cold methanol. Protein was recovered by centrifugation (14000g, 4 °C for 10 min) and was resuspended in 100 µL of ammonium bicarbonate buffer (50 mM, pH 7.4), and 50 µg was digested with trypsin (3 µg) at 37 °C overnight. The resulting hydrolysates were desalted with C18 ZipTips (Millipore) according to the manufacturer’s instructions before analysis by LC-MS/MS. The highest ratio of SMX-NO to HSA (10:1) was the lowest ratio that allowed consistent identification of SMX-NO-modified tryptic peptides on the mass spectrometer employed. While this ratio is not physiologically relevant, the data obtained enabled structural characterization of the nitroso-protein adducts. Immunochemical assays conducted at more physiological ratios provided quantitative data on protein modification. Trapping of HSA Sulfenic Acid Using Dimedone. HSA (0.15 mM) in water was incubated with a range of concentrations of dimedone (1.5-120 mM) for 10 min at 37 °C. Samples were prepared for MS analysis as outlined above. LC-MS. Except for incubations containing peptide DS3, which were clarified by centrifugation to sediment a light precipitate, solutions were analyzed without further processing. Aliquots of incubations (10 µL) and supernatants (10-27 µL) were chromatographed at room temperature on a Prodigy 5 µm ODS-2 column (150 mm × 4.6 mm i.d.; Phenomenex, Macclesfield, Cheshire, United Kingdom) by gradient elution with ACN/0.1% (v/v) formic acid: 10-50% over 30 min for incubations of SMX-NO with GSH; 10-35% over 20 min followed by 35-65% over 5 min for all

940

Chem. Res. Toxicol., Vol. 22, No. 5, 2009

Callan et al.

Figure 2. Concentration-dependent binding of SMX-NO to GSTP (A) and HSA (C). Binding was shown to be instantaneous to both proteins and was not significantly altered for up to 120 min (B and D). Western blots of modified protein probed with an antisulfonamide antibody support the ELISA data (E).

other incubations. The flow rate was 0.9 mL/min. The HPLC system consisted of two Jasco PU980 pumps (Jasco UK, Great Dunmow, Essex, United Kingdom) and a Jasco HG-980-30 mixing module. Eluted compounds were monitored at 254 nm with a Jasco UV975 spectrophotometer. UV peak areas were determined by the integration facility in MassLynx 3.5 software (Waters Corp., Manchester, United Kingdom). Eluate split-flow to the LC-MS interface was ca. 40 µL/min. A Quattro II MS (Waters Corp.) fitted with the standard coaxial electrospray source was operated using nitrogen as the nebulizing and drying gas. The interface temperature was 80 °C; electrospray capillary voltage, 3.8 kV; standard cone voltage, 30 V. The instrument was set up for full-scanning acquisitions in the positive-ion mode: m/z 50-1050 with a scan time of 5 s. For nano-LC-MS/MS analysis of tryptic digests, aliquots (0.5-1 µL) were delivered into a QSTAR Pulsar i hybrid mass spectrometer (Applied Biosystems, Foster City, CA) by automated in-line LC (integrated LCPackings HPLC system; 5 mm C18 nanoprecolumn and 75 µm × 15 cm C18 PepMap column; Dionex, Camberley, Surrey, United Kingdom) via a nanoelectrospray source head and 10 µm inner diameter PicoTip emitter (New Objective, Woburn, MA). A gradient from 5% ACN/0.05% TFA (v/v) to 48% ACN/ 0.05% TFA (v/v) in 60 min was applied at a flow rate of 300 nL/ min, and MS and MS/MS spectra were acquired automatically in positive-ion mode using information-dependent aquisition (IDA) (Analyst, Applied Biosystems). Database searching was performed using ProteinPilot version 2 (Applied Biosystems) against the latest version of the SwissProt database, with the confidence level set to 90% and with each of the potential drug adducts set as a high probability user-defined modification of cysteine. Direct Infusion MS/MS. For direct infusion of samples into a nanospray source, samples were diluted 1:20 to 1:200 in 50% methanol/1% formic acid (v/v). Aliquots (4 µL) were loaded into

a conductive needle pulled to a 1 µm i.d. tip (EconoTip, New Objective) and infused at a flow rate of approximately 40 nL/min. The QSTAR was operated in positive-ion mode with Q1 set to unit resolution. Survey scans of 1 s were acquired for m/z 400-2000, with the needle voltage being ramped to 1700 V to initiate the ion current, and the voltage was then reduced to 900-1200 V to maintain the signal for data acquisition. Data were accumulated for 30-60 s. Product ion spectra were acquired for specific parent ions using an m/z range of 65-2000 and with the collision energies being increased stepwise until optimal fragmentation was achieved. Spectra were again accumulated for 30-60 s.

Results Immunochemical Data. Western blotting and ELISA revealed concentration-dependent modification of GSTP and HSA when incubated with SMX-NO (Figure 2). Time-course experiments showed that modification of the proteins was effectively instantaneous and did not significantly alter over a 2 h time period (Figure 2). The molar ratios of protein:drug used in the immunochemical analyses varied from 400000:1 to 15:1 and were much lower than those involved in the proteomic analysis. These immunochemical ratios are more physiologically relevant than those in the proteomic studies but give no qualitative information as to the nature of the adduct. Reaction of SMX-NO with GSH. After 1 min in phosphate buffer at 37 °C, the reaction of GSH and SMX-NO (molar ratio, 1:1) yielded principally SMX-HA (Figure 3a) but also two products (ratio of areas of λ254 absorbance peaks, 5.0) in the mass chromatogram for m/z 575 (Figure 3b), corresponding to [M + H]+ for an additive conjugate of the reactants. The latter

Nitroso Sulfamethoxazole Adducts of Cysteine

Figure 3. UV chromatogram and LC-MS mass chromatogram showing SMX-HA and conjugate products of the spontaneous reaction of SMXNO (5 mM) with GSH (molar ratio, 1:1) in aqueous solution, pH 7.4, at 37 °C for 1 min. (A) UV chromatogram (254 nm). (B) Mass chromatogram for [M + H]+ (m/z 575) of the sulfinamide and unstable semimercaptal conjugate products.

were assigned previously as the sulfinamide and semimercaptal (N-hydroxysulfenamide) conjugates (Figure 1), on the basis of their relative stability at pH 7.4 (13) and the sulfinamide’s 1H NMR spectrum (22). These assignments were confirmed here by fragmentation under electrospray conditions to m/z 322 ([GSO]+) and m/z 270 ([SMX·NOH + 1]+), respectively. The only other product observed was the azoxy (tR 34 min; [M + H]+ at m/z 519) noted previously as a degradation product of SMX-NO in aqueous solution (14). After 5 min, the Nhydroxysulfenamide was no longer detectable by either UV absorbance or mass spectrometry. The sulfinamide was stable under these conditions for at least 60 min. Reaction Profile of GSH and SMX-NO in Water. When GSH and SMX-NO were incubated in water [16O], a marked increase in the amount of free SMX (m/z 254) as compared to the reaction in phosphate buffer (pH 7.4) was observed after 15 min. There was also a peak for the GSH N-hydroxy sulfonamide (m/z 607), both of which suggest increased, acidcatalyzed hydrolysis of the sulfinamide (Figure 1). A single peak (tR 4.7 min) in the mass chromatogram for m/z 340 was ascribed to the [M + 1]+ of GSH sulfinic acid, the other expected product of sulfinamide hydrolysis. There was still a trace peak for the semimercaptal, eluting immediately after the N-hydroxy sulfonamide. The sulfinamide of GSH (m/z 575) and SMX-HA were also found after 15 min. The relative abundances of SMX, SMXHA, and the GSH adducts, determined from the areas of their UV (254 nm) peaks, were 1:0.47:0.82, respectively. After 2 h, when the N-hydroxy sulfonamide was the only quantifiable GSH adduct, the corresponding figures were 1:0.40:0.15. The disproportionate formation of SMX and SMX-HA, and especially the abundant production of SMX-HA in the absence of excess GSH, suggests both hydroxylamine and amine might be formed from sulfinamide under these conditions, although the mechanism is obscure. When the reaction was monitored for [18O] incorporation from H218O, the same products were observed, with the sulfinamide decreasing and the N-hydroxy sulfonamide increasing in relative amount over time. Conversion of sulfinamide to N-hydroxy sulfonamide went to completion between 1 and 2 h. The mass increments observed were in accordance with the incorporation of the heavy oxygen, so that the sulfinamide [M + 1]+ ion increased from m/z 575 to 577 and

Chem. Res. Toxicol., Vol. 22, No. 5, 2009 941

the N-hydroxy sulfonamide ion increased from m/z 607 to 611 (Figure 4A,B). Modification of DS3 Peptide by SMX-NO. DS3 peptide was incubated with SMX-NO at a molar ratio of 1:1. Peptide alone, from LC-MS analysis, was stable in phosphate buffer at room temperature for at least 105 min. Samples were incubated for time periods ranging from 5 min to 2 h before analysis by LC-MS. The results indicated that as the time of incubation increased, the relative proportion of modified peptide increased. The modification represented a mass increase of 267 amu (4+, m/z 681, and 3+, m/z 907), which would correspond to formation of the semimercaptal or the sulfinamide adduct. Peaks corresponding to the peptide disulfide (5+, m/z 981) as well as peptide sulfenic acid (3+, m/z 823) and unmodified peptide (3+, m/z 818) were also present in these incubations (Figure 5A). Use of dimedone as a trapping agent confirmed that the peak at m/z 823 was the sulfenic acid ion of the peptide: When peptide, SMX-NO, and dimedone were coincubated, a triply charged ion at m/z 864 (Figure 5B) was observed, corresponding to a peptide mass increase of 138 amu and representing the formation of a peptide-dimedone adduct. This product was absent from incubations containing SMX-NO and dimedone alone. Reaction Profile of DS3 with SMX-NO in Water. When DS3 and SMX-NO were incubated in water, not only the peptide-sulfinamide conjugate at m/z 907 (3+) but also the N-hydroxy sulfonamide at m/z 918 were observed. The same product profile was observed after 3.5 h in both [16O] and [18O] water, with mass increments corresponding to the incorporation of two heavy oxygen atoms (Figure 4C-F) (m/z 689-690 for the 4+ ion and m/z 918-919 for the 3+ ion). Modification of DS3 Peptide in the Presence of GSH and DTT. DS3 peptide was incubated with SMX-NO (1:10) in the presence of different molar ratios of GSH. Glutathionylated peptide was observed (3+, m/z 919.76) when peptide was incubated with GSH alone. When peptide, SMX-NO, and GSH were coincubated, both the glutathionylated (3+, m/z 919.76) and the SMX-NO (3+, m/z 907.09) modified peptide were observed (Figure 6B). Increasing the ratio of GSH to peptide within the incubations from 1:1 to 10:1 diminished the proportion of peptide-SMX-NO adducts and increased the formation of the GSH-SMX-NO conjugate. LC-MS analysis showed that the addition of GSH to the incubations dramatically decreased the intensity of the nitroso-modified peptide peaks after 5 min. Peaks for SMX-HA, dipeptide (3+, m/z 981), azoxy, and unmodified peptide (3+, m/z 818) were still present. After 60 min of incubation with excess (5 mM) GSH, peptide-nitroso adduct peaks were clearly still identifiable in the spectra. When peptide, SMX-NO, and DTT were coincubated, unmodified peptide was observed as well as the azoxy and SMX-HA forms of the drug, indicating a loss of modified peptide after addition of DTT. However, high resolution and sensitive analysis of the samples on the Q-TOF instrument revealed the dipeptide (3+, m/z 981) and a low intensity triply charged ion at m/z 907 (+267 adduct) corresponding to sulfinamide-modified peptide, as well as the unmodified peptide (3+, m/z 818). The modified peptide was still detectable after incubation for 2 h with excess DTT (Figure 6C). Modification of GSTP. GSTP (0.1 mM) was incubated with SMX-NO (1 mM) for 5-60 min. Samples were methanol precipitated to remove unbound drug before trypsin digestion. LC-MS/MS analysis revealed exclusive modification of cysteine residues within the protein when a 10:1 ratio of drug to protein was used. Modification of the Cys47-containing peptide 45-54 (ASCLYGQLPK) was observed after incubation for 15 min.

942

Chem. Res. Toxicol., Vol. 22, No. 5, 2009

Callan et al.

Figure 4. Mass spectra from reactions of (A) GSH with SMX-NO in [16O] water for 1 min with peaks at m/z 575 for semimercaptal and sulfinamide and m/z 607 for N-hydroxy sulfonamide (direct infusion analysis). (B) GSH with SMX-NO in [18O] water for 1 min showing partial incorporation of [18O] into sulfinamide adduct, masses m/z 575 and 577, as well as partial incorporation into the N-hydroxy sufonamide adduct, m/z 609 and 611 (direct infusion analysis). (C and E) DS3 with SMX-NO in [16O] water showing sulfinamide adduct peaks at m/z 680 and 907 (E) and N-hydroxy sulfonamide peaks at m/z 689 and 918 (C). (D) DS3 with SMX-NO in [18O] water showing heavy oxygen incorporation into the sulfinamide adduct peaks at m/z 681 and 908, N-hydroxy sulfonamide peaks at m/z 690 and 919, as well as the peak at m/z 824 for peptide sulfenic acid. (F) DS3 with SMX-NO in [18O] water showing sulfinamide adduct peaks at m/z 681, N-hydroxy sulfonamide peaks at m/z 690 and 919 with heavy oxygen incorporation, as well as a sulfinamide adduct peak at m/z 907 with no heavy oxygen incorporation.

The peptide’s mass increased from 1078.6 (2+, m/z 540.3) to 1345.4 (2+, m/z 673.7), corresponding to an increase of 267 amu (Figure 7B). This suggested formation of either the semimercaptal or the sulfinamide adduct of SMX-NO. A second modification, of 283 amu, was observed in the 60 min sample, implying formation of the N-hydroxysulfinamide adduct (Figure 7C). A third mass addition, of 299 amu, was also observed in the 60 min sample (Figure 7D), with the mass increase suggesting incorporation of the N-hydroxysulfonamide adduct structure. The oxidation states of cysteine 47 were also investigated prior to SMX-NO addition. Cysteine 47 existed as a thiol (2+, m/z 540.3) and as the sulfenic (2+, m/z 548.3), sulfinic (2+ m/z 556.3), and sulfonic acids (2+, m/z 564.3) (Figure 7A,E-G). Modification of HSA. HSA (150 µM) was incubated with SMX-NO (1.5 mM) for 5-60 min. Modification of the Cys34containing peptide 21-41 (ALVLIAFAQYLQQCPFEDHVK)

was observed after incubation for 60 min: The peptide’s mass increased from 2432.3 (2489.3 for carboxamidomethylated peptide; 3+, m/z 830.8) to 2715.3 (3+, m/z 906.1), corresponding to an increase of 283 amu (Figure 8B). This suggested the formation of an N-hydroxysulfinamide adduct. Trapping of HSA Sulfenic Acid by Dimedone. Dimedone (120 mM) was incubated with HSA (0.15 mM) at 37 °C. After 10 min, modification of the Cys34-containing tryptic peptide 21-41 (ALVLIAFAQYLQQCPFEDHVK) was observed, resulting in a mass increase of 138 amu from 2432.3 (3+, m/z 811.8) to 2570.3 (3+, m/z 857.9), corresponding to a dimedone adduct. The MS/MS spectrum was weak, but it did confirm dimedone modification of cysteine 34 (Figure 8C).

Discussion SMX is a bacteriostatic sulfonamide antibiotic that is commonly used in combination with trimethoprim for the treatment

Nitroso Sulfamethoxazole Adducts of Cysteine

Chem. Res. Toxicol., Vol. 22, No. 5, 2009 943

Figure 5. Mass chromatogram and spectrum from LC-MS analyses of incubations of synthetic peptide DS3 (0.5 mM; monoisotopic mass, 2451) and SMX-NO (molar ratio, 1:1) in the presence and absence of dimedone, respectively. (A) The peaks at m/z 981 ([M + 5H]5+), m/z 907 ([M + 3H]3+), m/z 823 ([M + 3H]3+), and m/z 818 ([M + 3H]3+) were attributed to DS3 dipeptide, the peptide conjugate of SMX-NO (mass increment, 267), the peptide sulfinic acid, and unmodified DS3, respectively. (B) Putative peptide-dimedone adduct m/z 864.

Figure 6. Direct infusion nanospray mass spectra showing products of the reaction of (A) SMX-NO (0.5 mM) with synthetic peptide DS3 (molar ratio, 1:1). Peaks at m/z 818.08, 907.08, and 981.19 were attributed to unmodified peptide, the peptide-SMX-NO conjugate, and the disulfide of the peptide, respectively. (B) SMX-NO (0.5 mM) with synthetic peptide DS3 and GSH; peaks were as in panel A except m/z 919.76, which corresponded to glutathionylated DS3 peptide. (C) SMX-NO (0.5 mM) with synthetic peptide DS3 and DTT; the SMX-NO-modified DS3 peak (m/z 907.41) persisted.

of Staphylococcal, Streptococcal, and Escherichial infections. The drug combination is frequently prescribed as both a palliative and a prophylactic against pneumonia, but the incidence of severe hypersensitivity reactions, particularly in HIV patients, limits its clinical use (33). It is believed that HIV patients display altered cellular metabolism of the drug, as they have a decreased capacity to reduce the thiol-reactive nitroso intermediate back to the less reactive hydroxylamine (34, 35). Drug-modified protein has been detected in human plasma (27), but the precise nature of chemical adducts on proteins has not been fully characterized. In this study, we found that the chemistry of hapten formation was more complex than anticipated and involves reactions not only with cysteine per se but also with the higher oxidation forms of the amino acid.

Previous studies have taken the reactions of GSH with SMXNO as the simplest model of sulfonamide haptenation, although the predominant reaction in the presence of physiological levels of GSH is reduction back to the hydroxylamine (13). In the presence of equimolar levels of GSH and SMX-NO at pH 7.4, a labile semimercaptal adduct is formed that rapidly rearranges to the sulfinamide (13, 22). Here, incubations of GSH with equimolar SMX-NO also resulted in reduction and in the immediate appearance of the semimercaptal adduct, with rapid rearrangement to the more stable sulfinamide (Figure 3). The conversion of the semimercaptal to the sulfinamide adduct is believed to proceed via dissociative cleavage of the hydroxyl group, yielding a resonance-stabilized cationic intermediate, which can react with a molecule of solvent water, giving the

944

Chem. Res. Toxicol., Vol. 22, No. 5, 2009

Callan et al.

Figure 7. MS/MS spectra showing SMX-NO-modified, cysteine 47 containing tryptic peptide of GSTP. Modifications of 267 (B), 283 (C), and 299 (D) mass units correspond to the sulfinamide, N-hydroxy sulfinamide, and the N-hydroxy sulfonamide adducts, respectively. Figures E-G show representative MS/MS spectra of the sulfenic, sulfinic, and sulfonic acids of cysteine 47, respectively.

sulfinamide (36). Although the SMX-NO/GSH sulfinamide was stable at pH 7.4 for at least 60 min, sulfinamides are known to

undergo acid-catalyzed hydrolysis that cleaves the S-N bond and releases the corresponding amine and sulfinic acid (37, 38).

Nitroso Sulfamethoxazole Adducts of Cysteine

Chem. Res. Toxicol., Vol. 22, No. 5, 2009 945

Figure 8. MS/MS spectra showing the unmodified (A) and SMX-NO-modified (B) cysteine 34 containing tryptic peptide of HSA. Modification with an adduct of 283 mass units suggests the N-hydroxy sulfinamide adduct. Also shown is the conjugate of the sulfenic acid of cysteine 34 with dimedone (C).

Indeed, hydrolytic release of arylamines (39) and generation of cysteine sulfinic acid residues (40) are indirect evidence that the conjugates formed by the reaction of nitrosoarenes and proteins include sulfinamides. When the reaction of GSH and SMX-NO was carried out at pH 6.4, there was a large increase in the proportion of SMX detected, as well as peaks corresponding to GSH-N-hydroxy sulfonamide, SMX-HA, and the GSH sulfinamide. The slightly acidic conditions promoted hydrolysis of the sulfinamide, yielding the amine and the sulfinic acid of GSH, and resulting in a marked decrease in the azoxy dimer. Previously published work by Umemoto et al. (41) reported a N-hydroxysulfonamide adduct resulting from reaction of an arylnitroso compound and GSH sulfinic acid, supporting our findings. These data demonstrate the possibility that multiple haptenic structures may be derived from the reaction of SMXNO with cellular protein thiols. It should be noted that although a pH of 6.4 would be highly acidic in vivo, the intracellular environment of exogenous antigen processing might be sufficiently acidic to hydrolyze sulfinamide conjugates (22). However, for the purpose of these particular experiments, the use of acidic reaction conditions enabled more detailed characterization of one of the hypothetical mechanisms for the generation of N-hydroxysulfonamide adducts. The experiments with GSH and heavy water support the hydrolysis hypothesis: Figure 4A shows that after 1 min of incubation in the presence of H2[16O] a peak at m/z 575 corresponds to the isomeric semimercaptal and sulfinamide. However, Figure 4B shows that after 1 min of incubation in the presence of H2[18O], there are two GSH conjugates with m/z of 575 and 577, which are thought to correspond to the semimercaptal and sulfinamide [18O], respectively. On prolonged incubation, a peak with m/z 611 is also apparent, corresponding to the N-hydroxysulfonamide [18O]2 (Figure 4B). To investigate the reaction mechanism further and to address the possibility that amino acids other than Cys may be modified by SMX-NO, a synthetic peptide (DS3) comprising each of the common amino acids was employed. When the reaction was

performed at pH7.4, no reduction of SMX-NO to SMX-HA was observed following incubation with the peptide, in marked contrast to the reaction with GSH. MS analysis revealed the formation of the isobaric semimercaptal or sulfinamide adducts of SMX (3+, m/z 907), and the detection of peptide sulfenic acid (3+, m/z 823) and trace amounts of N-hydroxy sulfonamide (data not shown). There was no evidence for modification of any amino acid other than Cys. At pH 6.4, as observed for the reactions of GSH with SMX-NO, hydrolysis to the peptide cysteine sulfinic acid is promoted, leading to enhanced reaction with a second molecule of SMX-NO and N-hydroxy sulfonamide formation (3+, m/z 918) (Figure 4C), although the sulfinic acid intermediate was not invariably detected. Experiments employing heavy water revealed that the oxygen atoms in the N-hydroxy sulfonamide once again were derived predominantly from the solvent, giving rise to a triply charged ion of m/z 919, although there remained a significant proportion of the [16O]containing conjugate at m/z 918 (3+). To assess the susceptibility of the peptide-adduct bond to reductive cleavage by thiols, incubations were performed in which either GSH or DTT was added to the reaction. The reductant was added to the reaction mixture either prior to SMXNO, at the same time as the nitroso, or 1-120 min after the nitroso. For the GSH coincubations, the order of addition of reagents did not affect the profile of peaks observed, which included the sulfinamides of GSH and DS3 (data not shown) and glutathionylated DS3 (Figure 6B). Increasing the ratio of GSH diminished the relative levels of the drug-modified DS3 ions and increased the intensity of the glutathionlylated peptide derivative but did not completely remove the sulfinamide. Similarly, DTT added at any stage during the incubation resulted in a reduction in the relative levels of the sulfinamide, but even where an excess of DTT (5 mM) was present, weak ions corresponding to the sulfinamide and dipeptide were still apparent (Figure 6C). These results suggest that both thiols are acting to scavenge available SMX-NO rather than reductively cleaving the very rapidly (