286
Chem. Res. Toxicol. 1999, 12, 286-296
Opposite Behaviors of Reactive Metabolites of Tienilic Acid and Its Isomer toward Liver Proteins: Use of Specific Anti-Tienilic Acid-Protein Adduct Antibodies and the Possible Relationship with Different Hepatotoxic Effects of the Two Compounds Eric Bonierbale, Philippe Valadon, Catherine Pons, Bernard Desfosses, Patrick M. Dansette,* and Daniel Mansuy Laboratoire de Chimie et Biochimie Pharmacologiques et Toxicologiques, URA 400 CNRS, Universite´ Paris V, 45 rue des Saints-Pe` res, 75270 Paris Cedex 06, France Received June 10, 1998
Tienilic acid (TA) is responsible for an immune-mediated drug-induced hepatitis in humans, while its isomer (TAI) triggers a direct hepatitis in rats. In this study, we describe an immunological approach developed for studying the specificity of the covalent binding of these two compounds. For this purpose, two different coupling strategies were used to obtain TAcarrier protein conjugates. In the first strategy, the drug was linked through its carboxylic acid function to amine residues of carrier proteins (BSA-N-TA and casein-N-TA), while in the second strategy, the thiophene ring of TA was attached to proteins through a short 3-thiopropanoyl linker, the corresponding conjugates (BSA-S-5-TA and βLG-S-5-TA) thus preferentially presenting the 2,3-dichlorophenoxyacetic moiety of the drug for antibody recognition. The BSA-S-5-TA conjugate proved to be 30 times more immunogenic than BSAN-TA. Anti-TA-protein adduct antibodies were obtained after immunization of rabbits with BSA-S-5-TA (1/35000 titer against βLG-S-5-TA in ELISA). These antibodies strongly recognized the 2,3-dichlorophenoxyacetic moiety of TA but poorly the part of the drug engaged in the covalent binding with the proteins. This powerful tool was used in immunoblots to compare TA or TAI adduct formation in human liver microsomes as well as on microsomes from yeast expressing human liver cytochrome P450 2C9. TA displayed a highly specific covalent binding focused on P450 2C9 which is the main cytochrome P450 responsible for its hepatic activation in humans. On the contrary, TAI showed a nonspecific alkylation pattern, targeting many proteins upon metabolic activation. Nevertheless, this nonspecific covalent binding could be completely shifted to a thiol trapping agent like GSH. The difference in alkylation patterns for these two compounds is discussed with regard to their distinct toxicities. A relationship between the specific covalent binding of P450 2C9 by TA and the appearance of the highly specific anti-LKM2 autoantibodies (known to specifically recognize P450 2C9) in patients affected with TA-induced hepatitis is strongly suggested.
Introduction Covalent binding of reactive metabolites to liver cell proteins may trigger various cell disorders which can eventually lead to overt hepatotoxicity. Two different kinds of hepatotoxicities may result from alkylation of proteins: direct toxicity and immune-mediated toxicity. In the case of direct toxicity, the extent of liver lesions is clearly related to the administered dose; the higher the dose, the more severe the lesions and the higher their incidence. This kind of toxicity can usually be reproduced in laboratory animals. Therefore, compounds that produce such direct toxic effects should be easily detected with usual studies in laboratory animals. On the contrary, immune-mediated drug-targeted liver toxicities occur with a very low incidence (generally less than one case out of 10 000 patients), are not clearly related to the administered dose, are very dependent on the individual, * Author to whom correspondence should be addressed. Telephone: (+33) 01 42 86 21 91. Fax: (+33) 01 42 86 83 87. E-mail: dansette@ citi2.fr.
and are not reproduced in laboratory animals. Therefore, compounds leading to such toxicities have eluded detection during preclinical trials so far. Better knowledge of the mechanism leading to this kind of immune-mediated disorder would help (1) in implementing tests that are useful in identifying drugs at risk in the early stages of development and (2) in diagnosing unexplained hepatotoxicities in which a drug is the suspected triggering agent. Another crucial aspect of research and development of new drugs would be to establish a relationship between the molecular structure of a compound and its ability to cause either direct or immune-mediated hepatotoxic effects. Obviously, we are far from such a structure-hepatotoxicity relationship because of the complexity of the phenomena involved. In this context, it has been reported that two closely related compounds, a drug, tienilic acid [TA,1 ticrynafen, 2,3-dichloro-4-(2-thienylcarbonyl)phenoxyacetic acid], and its isomer [TAI, 2,3-dichloro-4-(3-thienylcarbonyl)phenoxyacetic acid], lead to an immune-mediated hepatitis in humans and to direct hepatotoxicity in rats.
10.1021/tx980136z CCC: $18.00 © 1999 American Chemical Society Published on Web 02/20/1999
Alkylation of Tienilic Acid and Its Isomer
Chem. Res. Toxicol., Vol. 12, No. 3, 1999 287
Figure 1. Structures of tienilic acid and its isomer and preparation of BSA-N-TA and casein-N-TA conjugates. Detailed methods for the synthesis are described in Materials and Methods.
TA, a uricosuric diuretic drug used in the treatment of hypertension, was launched in Europe in 1976, and then in the United States in 1979. It was withdrawn from the U.S. market in 1980 because of its hepatotoxic secondary effects (1). However, TA was found to be devoid of any direct hepatotoxic effects in usual laboratory animals. The low incidence and all the clinical features supported an immune mechanism for TA-induced hepatitis. In many patients suffering from hepatitis after administration of TA, a specific autoantibody, called antiLKM2 (anti-liver kidney microsomes) (2) and directed against the human liver cytochrome P450 2C9 (3), was detected. This antibody seems to be a specific marker of TA-induced hepatitis since it was not found in any other drug-induced or virus-triggered hepatitis. On the contrary, the isomer of tienilic acid, TAI (Figure 1), bearing the aroyl group at position 3 of the thiophene ring, which was contaminating the first drug batches in very small amounts, led to direct hepatotoxic effects in rats.2 A mechanism for the first steps involved in TA-induced hepatitis has been proposed (4, 5). It is based on the alkylation of P450 2C9, the human liver enzyme mainly responsible for TA metabolism (5), by a reactive metabolite of TA, and the appearance of autoantibodies directed against P450 2C9. Accordingly, recent results showed that TA acts as a mechanism-based inhibitor of P450 2C9 (7, 8). Enzyme inactivation seems to occur after covalent binding of about 1 mol of TA metabolite per mole of P450 2C9 (8). Interestingly enough, TAI, which only differs from TA by the position of the aroyl group on the thiophene ring (Figure 1), is also activated by P450 2C9 (9). This leads to covalent binding of its metabolites to liver microsomal 1 Abbreviations: ABTS, 2,2′-azino-bis(3′-ethylbenzothiazolesulfonic acid); anti-LKM2, anti-liver kidney microsomes type 2; BSA, bovine serum albumin; BSA-N-TA or casein-N-TA, carrier protein-TA conjugate in which TA is bound by its carboxylic end by an amide function to lysine residues of BSA or casein, respectively; BSA-S-5TA or βLG-S-5-TA, carrier protein-TA conjugate in which TA is bound by substitution of its thiophene ring by the sulfur atom of a thiopropanoyl linker to BSA or βLG, respectively first modified with SPDP; DETAPAC, diethylenetriaminepentaacetic acid; DTT, dithiothreitol; ELISA, enzyme-linked immunosorbent assay; GSH, glutathione; HRP, horseradish peroxidase; PAGE, polyacylamide gel electrophoresis; PB, phosphate buffer; PBS, phosphate-buffered saline; PBST, phosphate-buffered saline containing 0.1% Tween 20; SDS, sodium dodecyl sulfate; SPDP, N-succinimidyl 3-(2-pyridyldithio)propionate; βLG, β-lactoglobulin; TA, tienilic acid; TAI, tienilic acid isomer; TCA, trichloroacetic acid. 2 Personnal communication from J. P. Labeaune, Anphar-Rolland.
proteins that is more intense than that of TA (9, 10). However, contrary to TA, TAI does not act as a suicide substrate of P450 2C9 (8). This study was undertaken to elucidate the molecular origin of the different toxic effects caused by TA and TAI. For that purpose, antibodies able to specifically recognize covalent adducts between TA (or TAI) metabolites and proteins have been raised in rabbits against a carrier protein-TA conjugate. The use of these antibodies allowed us to locate the adducts formed between microsomal proteins and TA (or TAI) reactive metabolites after metabolism of TA and TAI by human liver microsomes and yeasts expressing human P450 2C9. This study shows that TAI metabolites covalently bind to liver proteins in a rather nonspecific manner, whereas TA metabolites covalently bind selectively to P450 2C9.
Materials and Methods Chemicals. TA, 5-chloro-TA, and TAI (11) were provided by Anphar-Rolland (Chilly-Mazarin, France). [14C]TA was prepared by both CEA (Saclay, France) and Amersham (Bucks, U.K.); [14C]-5-chloro-TA and [14C]TAI (all labeled in the keto group) were prepared by Amersham. Their radiochemical purity was checked by HPLC and found to be higher than 98%. Bovine serum albumin (BSA) and β-lactoglobulin (βLG) were from Sigma (St-Quentin Fallavier, France). βLG was acid treated. Yeast Transformation, Cell Culture, and Preparation of the Yeast Microsomal Fraction. Isolation and sequencing of human cDNA MP-4 (P450 2C9) clone have been previously reported (12-14). The DNA sequence was inserted into the pAAH5 vector (based on an ADH1 promoter and terminator) that contains the leu2 gene. Saccharomyces cerevisiae strain D12 [a, leu-, (cir+)] was used for expression. Details of protocols for construction of vectors were presented elsewhere (14-16). These transformed yeasts were cultured semianaerobically at 28 °C in synthetic minimal medium [0.67% (w/v) Difco yeast nitrogen base without amino acids and 2% (w/v) glucose] supplemented with amino acids (except leucine). Growth was monitored carefully by light scattering at 600 nm, and cultures were allowed to reach an A600 of 1.7-1.8 (about 5 × 107 cells/ mL), at which time the culture was stopped by quick refrigeration. Cells were harvested by centrifugation (3000 rpm, 5 min), pooled, and washed twice with 1/10 volume of distillated water. Yeast cell pellets were stored at -80 °C until they were processed for microsome preparation. Routinely, yeasts from 5 L of culture were pooled and processed simultaneously. Yeast microsomes were prepared as previously described (14, 17) using lyticase enzyme (Sigma) for cell wall digestion followed by sonication. The final microsomal pellet was homogenized in 50
288 Chem. Res. Toxicol., Vol. 12, No. 3, 1999 mM Tris-HCl buffer (pH 7.4) containing 1 mM EDTA and 20% glycerol (v/v), aliquoted, frozen under liquid N2, and stored at -80 °C until it was used. Human Liver Microsomes. Human liver was obtained from a donor after accidental death. Microsomes were prepared as previously described (18). Cytochrome b5 and P450 contents were determined by the method of Omura and Sato (19); the protein concentration was measured with the BCA protein assay (Pierce), using BSA as a standard. Rat Liver Microsomes. Male Sprague-Dawley rats (Iffa Credo) weighing 180-200 g were treated for 3 days by daily ip injections of clofibrate (500 mg/kg, 10% in corn oil) (10). After overnight fasting, animals were sacrified by cervical dislocation, livers removed, and microsomes prepared as previously described (18). Human Serum Containing Anti-LKM2 Antibodies. This serum was obtained from a patient (KUT) suffering from TAinduced hepatitis. Its anti-LKM2 antibody titer was 1/5000 (4). It was a generous gift from J.-C. Homberg (CHU Saint-Antoine, Paris, France). Incubation Conditions. TA or TAI was diluted in 0.1 M phosphate buffer (PB, pH 7.4) containing 1 mM diethylenetriaminepentaacetic acid (DETAPAC) and 8% glycerol with yeast microsomes and was incubated at 37 °C (28 °C with yeast microsomes) for 30 min. Incubation mixtures contained pAAH5/ P450 2C9 yeast (0.2 µM P450 2C9) or human liver or rat liver microsomes (both 1 µM P450), a saturating substrate concentration (100 µM [14C]TA, 56 Ci/mol, or [14C]TAI, 25 Ci/mol), 1 mM NADPH, and 5 mM GSH (when specified). Incubations were initiated, after a preincubation of 3 min for temperature equilibration, by adding NADPH to the mixture (t0). Quantitative analysis of TA or TAI covalent binding to proteins was determined (10, 20) at t0 and t30 by spotting 50 µL aliquots of the incubation mixture on glass-fiber filter disks (GFB) from Whatman. Filters were dipped into 10% trichloroacetic acid (TCA) to precipitate proteins and then dipped into two different baths of methanol and one of ethyl acetate to further remove unbound TA or TAI. Filters were further dried and counted for 1 min after addition of 2 mL of scintillation cocktail. Synthesis of Carrier Protein-TA Conjugates in Which TA Is Bound by Its Carboxylic End to Amine Functions of Lysine Residues: BSA-N-TA and Casein-N-TA. The original method used for the coupling reaction of acids was described by Erlanger (21). Radiolabeled TA (100 mg, 1 µCi) was dissolved in dioxane (2.85 mL) and tri-n-butylamine (0.07 mL). The solution was chilled at 10 °C, and isobutyl chlorocarbonate (0.04 mL) was added. The mixture was stirred at 4 °C for 20 min, and BSA or casein (400 mg) dissolved in dioxane/ water (v/v, 20 mL) and 1 N NaOH (0.4 mL) were poured into the solution. After 1 h, 1 N NaOH (0.19 mL) was added and the mixture allowed to stand for 2 h at 4 °C. The protein conjugate was extensively dialyzed against H2O and the precipitate discarded after centrifugation. The number of TA molecules bound per protein was determined with radioactivity. These two protein-drug conjugates were termed BSA-N-TA and caseinN-TA since the coupling reaction led to the formation of an amide bound between the protein and TA. Coupling of the Thiol Function of Mercaptoethanol with Thiophene Ring Position 5 of 5-Chloro TA: Synthesis of Compound 1. In this reaction, the chlorine atom of 5-chloro-TA was substituted with the thiol group of mercaptoethanol. First, 5-chloro-TA (200 mg) was dissolved in 0.1 M deoxygenated carbonate buffer (pH 10.3, 50 mL). Then mercaptoethanol (100 µL) was added to the solution and the substitution allowed to occur for 6 days at 50 °C under an argon atmosphere. Products were precipitated with 2 N HCl and extracted with ethyl acetate. After washing with water and brine, the product was purified by silica gel TLC (90/10/1 ethyl acetate/methanol/acetic acid); Rf ) 0.18. A fraction of 1 obtained was methylated with CH2N2 in ether and subsequently analyzed by mass spectrometry and 1H NMR: CIMS (NH3) 438 [5% (M + NH4+)], 421 [100% (M + H+)], 377 [4% (M + H+ - C2H4O)],
Bonierbale et al. 362 [2% (M + H+ - CO2methyl)], 345 [6% (M + H+ - C2H6SO)]; 1H NMR (CDCl3) δ 3.17 (t, J ) 6 Hz, 2H, SCH2), 3.82 (s, 3H, OCH3), 3.88 (q, J ) 12, 6 Hz, 2H, HOCH2), 4.78 (s, 2H, OCH2), 6.78 (d, J ) 8.5 Hz, 1H, H-2′), 7.02 (d, J ) 4 Hz, 1H, H-4), 7.22 (d, J ) 4 Hz, 1H, H-3), 7.28 (d, J ) 8.5 Hz, 1H, H-3′); UV λmax 364 nm ( ) 15 300 M-1 cm-1). Synthesis of Carrier Protein-TA Conjugates in Which TA Is Bound by Its Thiophene Ring to Proteins through a Linker: BSA-S-5-TA and βLG-S-5-TA. The substitution reaction described in the previous section was used to bind TA to BSA and βLG. First, these two proteins were thiolated as described by Axe´n (22) to obtain a large number of thiol groups that are able to substitute for the chlorine of 5-chloro-TA. BSA or βLG (35 mg) was dissolved in 0.1 M deoxygenated PB (pH 7.5, 4 mL). N-Succinimidyl 3-(2-pyridyldithio)propionate (SPDP, 4.65 mg) dissolved in methanol (300 µL) was added to activate lysine residues of the proteins (30 lysines/βLG and 59 lysines/BSA) for 1 h with slow agitation under an argon atmosphere. Disulfide bridges were reduced for 1 h by dithiothreitol (DTT, 38.5 mg) dissolved in 0.1 M deoxygenated PB (pH 7.5, 2.5 mL). The mixture was loaded onto a gel filtration column (Trisacryl GF 0.5 M gel). Elution was performed with 0.1 M deoxygenated PB (pH 6) containing 0.1 M NaCl and 0.1 mM DTT at a flow rate of 60 mL/h. Fractions of 1 mL were collected and checked for absorbance at 280 nm. Those containing the protein were combined (around 12 mL); the thiol contents were estimated by the Ellman reaction using DTT as a standard and the elution buffer as a blank. The pH of fractions containing the thiolated protein was adjusted to 10.3 with 1 M potassium carbonate buffer. [14C]-5-Chloro-TA (7.6 mg, 0.1 µCi/µmol), dissolved in water (1 mL) containing a few crystals of Trizma base, was added to the solution. The mixture was agitated slowly in the dark at 40 °C under an argon atmosphere for 4 days. Then the pH was adjusted to 8 with 0.1 M ammonium acetate buffer (pH 5.5). Free thiol residues on the proteins were blocked by iodoacetamid (30 mg) for 30 min at 40 °C. Extensive dialysis against a 0.1 M ammonium acetate buffer (pH 5.5) was then performed. The mixture was loaded onto a gel filtration column (Sephadex G25 superfine). Elution of the protein-drug conjugate was performed with 0.1 M ammonium acetate buffer (pH 5.5) at a flow rate of 20 mL/h. Fractions of 1 mL were collected and checked for absorbance at 280 and 355 nm for the protein and protein-drug conjugate detection, respectively. Fractions with a constant A280/A355 ratio were collected, and the proteindrug conjugate was dialyzed against 0.05 M PB (pH 7). TA covalent binding was assayed by spectrophotometry (using an of 15 300 M-1 cm-1 at 364 nm of compound 1) and radioactivity. Aliquots were stored at -20 °C until they were used. These protein-drug conjugates were respectively termed BSA-S-5TA and βLG-S-5-TA since the coupling reaction involved sulfur substitution on thiophene ring position 5 of TA. Rabbit Immunization and Antisera Preparation. Polyclonal anti-BSA-S-5-TA or anti-BSA-N-TA sera were obtained from three female New Zealand white rabbits immunized with the respective protein-drug conjugate. A preliminary bleed was taken prior to immunization in each case. For the initial injection, conjugates (1 mg) were dissolved in 1.2 mL of saline containing a few crystals of Trizma base. These solutions were emulsified (v/v) with Freund’s complete adjuvant (and with Freund’s incomplete adjuvant for booster injections) (23). Intradermal injections consisted of 250 µg of conjugate/rabbit. Blood was collected (35 mL) from the rear marginal ear vein 2 weeks after injections were performed (23, 24). Serum was separated and stored at -40 °C until it was used. Titers of the two different kind of sera were determined with enzyme-linked immunosorbent assays (ELISAs) against their respective antigens (procedures are described in detail in the next section). Sera Evaluation. The ability of sera to recognize TAI metabolite-protein adducts was tested with ELISAs. Plates were coated for 1 night at 4 °C with 0.05 M carbonate/ bicarbonate buffer (pH 9.5) containing TAI (20 pmol of TAI/well) bound to rat liver microsomes (see Incubation Conditions for
Alkylation of Tienilic Acid and Its Isomer details). Control wells were coated with the same amount of protein from incubations carried out in the absence of NADPH. Wells were then washed once with phosphate-buffered saline (PBS). The blocking step was carried out for 30 min at 37 °C with PBS containing 0.5% gelatine. After three washes with PBS containing 0.1% Tween 20 (PBST), sera diluted in PBST containing 0.5% gelatine were added and the mixtures incubated for 2 h at 4 °C. Five washes were then performed, and horseradish peroxidase-conjugated secondary antiserum (diluted 1/500 in PBST containing 0.5% gelatine) was added for 1 h at 37 °C. Wells were washed five times before the peroxidase activity with 2,2′-azino-bis(3′-ethylbenzothiazolesulfonic acid) (ABTS) as a substrate was determined. Plates were further read at 405 nm. Preparation of the Anti-TA-Protein Adduct Immunoglobulin Fraction: Titer Determination and Specificity. The immunoglobulin fraction of three pooled sera derived from the rabbit immunized with the BSA-S-5-TA conjugate and which gave the highest level of recognition of TAI metaboliteprotein adducts was subsequently obtained with a 40% saturated ammonium sulfate precipitation at 4 °C (25). The pellet was solubilized in PBS. After extensive dialysis, samples were ultrafiltrated and stored at -40 °C until they were used. For titer determination, ELISA plates were coated with 47.5 ng of βLG or βLG-TA (corresponding to 15 pmol of TA/well). For inhibition studies, anti-TA-protein adduct antibodies (1/25600 final dilution in PBST containing 0.5% gelatine) were incubated in glass tubes for 1 night at 4 °C with serial dilutions of potential inhibitors before being added to the wells (26). Localization by Immunoblotting of TA and TAI Metabolite-Protein Adducts after Incubations of TA or TAI with Microsomes of Yeasts Expressing Human Liver P450 2C9 and with Human Liver Microsomes. After incubation of TA or TAI with microsomes for 30 min, as described in Incubation Conditions, proteins were precipitated with 2 mL of 10% TCA and were extensively washed twice with 5% TCA, three times with ethyl acetate, once with ethanol and acetone, and finally with 2 mL of 0.1 M potassium phosphate buffer (pH 7.4) containing 1 mM DETAPAC. Samples were centrifuged at 3000 rpm between each washing step. Proteins were solubilized at 37 °C for 1 night with 100 µL of 1% sodium dodecyl sulfate (SDS). Protein concentrations of samples were determined with the BCA Protein Assay (Pierce), using BSA as a standard. Samples were diluted 1/1 (v/v) with denaturation buffer [0.2 M Tris-HCl (pH 6.8), 1% SDS (w/v), 30% glycerol (v/v), and 0.01% pyronine (w/v)] and heated to 95 °C for 5 min. SDS-polyacylamide gel electrophoresis (SDS-PAGE) was performed according to the method of Laemmli (24), using 3.5% stacking and 9% separation gels. The protein load was 20 µg/well (using a 20well comb, which was 15 mm thick). Electrophoresis was carried out for 4 h at 30 mA/gel. Proteins were electrophoretically transferred to nitrocellulose for 2 h at 200 mA using a transfer buffer [25 mM Tris, 190 mM glycine, and 20% methanol (v/v)]. After transfer, nitrocellulose was stained with Ponceau red. The blocking step was carried out for 1 h with PBST containing 1% (w/v) poly(vinylpyrrolidone). Incubation with the first antibody (diluted in the blocking solution) was carried out for 1 h. After four washes with PBST, incubation of the horseradish peroxidase-conjugated secondary antiserum (diluted in the blocking solution) was carried out for 1 h followed by six washes. Peroxidase activity was developed for 1 min with luminol (ECL, Amersham) as a substrate and revealed on Hyperpaper (Amersham). In some experiments, a radiochemical analysis of TA and TAI covalently bound to proteins was performed by loading the nitrocellulose sheets into the InstantImager (Packard). Radioactivity was then acquired for 15 h. The gray density on the images is directly proportional to the number of desintegrations emitted by radiolabeled TA or TAI bound to the proteins.
Chem. Res. Toxicol., Vol. 12, No. 3, 1999 289
Results and Discussion Preparation of Antibodies That Are Able To Recognize TA (or TAI) Metabolite-Protein Adducts. (1) Strategies for Synthesis of Carrier Protein-TA Conjugates. Accumulated evidence for the metabolism of tienilic acid (TA) or its isomer (TAI) demonstrated that these two compounds are responsible for the formation of protein adducts (5, 7-10). This covalent binding was assessed by quantification of radiolabeled TA or TAI metabolites bound to proteins. The key point in the comprehension of TA-induced hepatitis mechanisms was to determine the specificity of its covalent binding quantified with radioactivity. Indeed, under the hapten hypothesis, alkylation of specific proteins could explain the onset of an immune response triggered by the immunogenicity acquired by these modified proteins. Thus, a qualitative identification of target proteins alkylated by TA was clearly required. For this reason, we tried to obtain antibodies that are able to recognize TA covalent binding to microsomal proteins. For this purpose, we developed antibodies raised against carrier protein-TA conjugates by using two different coupling strategies. For each strategy, TA was bound to two different proteins to obtain two conjugates. One conjugate was used for immunization of rabbits to obtain anti-TA-protein adduct antibodies. The other was used as coating antigen in the ELISA to assess the specificity of the antibodies. The first coupling strategy consisted of linking TA by its carboxylic end to amine residues of lysine groups present on carrier proteins (BSA or casein) (Figure 1). These protein-drug conjugates preferentially present the thiophene ring and the ketone function of TA for recognition by antibodies. They were termed BSA-N-TA and casein-N-TA since the coupling was leading to amide bonds between TA and proteins. The second coupling strategy is derived from our knowledge of TA and TAI metabolism. Actually, covalent binding of TAI was demonstrated to be dependent on metabolic activation of its thiophene ring (5). This activation gives a highly electrophilic thiophene sulfoxide which reacts with nucleophiles present on proteins. Similarly, metabolic activation of TA is thought to generate protein adducts in which TA metabolites are bound to proteins through their thiophene rings. Thus, the second strategy consisted of synthesizing protein-TA conjugates structurally similar to TA metabolite-protein adducts formed during metabolic activation of the drug. Thus, we assumed that antibodies that are able to recognize the dichlorophenoxyacetic part of TA or TAI would be able to detect TA or TAI metabolite-protein adducts. The second coupling strategy relied on the possibility of substituting the chlorine atom of 5-chloro-TA with thiol groups. By using this reaction, we first showed that we could bind mercaptoethanol to the thiophene ring of TA by substituting the chlorine atom of the 5-chloro-TA with the thiol function of mercaptoethanol (Figure 2A). The product of this reaction was fully characterized by mass spectrometry and 1H NMR (see Materials and Methods). The same reaction was used to bind proteins to the thiophene ring of TA through a short linker derived from SPDP [Nsuccinimidyl 3-(2-pyridyldithio)propionate]. The role of SPDP is to increase the number of thiol groups on proteins by transforming the -amino groups of lysines into amidopropanethiols. These thiols can then be sub-
290 Chem. Res. Toxicol., Vol. 12, No. 3, 1999
Bonierbale et al.
Figure 2. (A) Coupling reaction between mercaptoethanol and 5-chloro-TA, i.e., synthesis of compound 1. (B) Preparation of BSAS-5-TA and βLG-S-5-TA conjugates. Lysine residues of BSA and βLG were first activated by SPDP to obtain thiol groups that are able to be substituted for the chlorine atom of 5-chloro-TA. Detailed methods for the syntheses are described in Materials and Methods.
stituted for the chlorine atom of 5-chloro-TA (Figure 2B). Conjugates resulting from binding of proteins to position 5 of the TA thiophene ring through a thiopropanoyl linker were termed BSA-S-5-TA and βLG-S-5-TA. These conjugates preferentially present the 2,3-dichlorophenoxyacetic part of TA for antibody recognition. (2) Synthesis of Carrier Protein-TA Conjugates. For carrier protein-TA conjugates in which TA is bound by its carboxylic end to amine functions of lysine residues (BSA-N-TA and casein-N-TA), we determined the number of TA molecules bound to BSA or casein by using the radiolabeled drug during the synthesis. Thus, we bound 20 and 9 TA molecules to BSA and casein, respectively. Consequently, the yield of lysine activation by TA was 34% for BSA (300 nmol/mg of protein, 59 lysines/BSA) and 60% for casein (324 nmol/mg of protein, 15 lysine residues/casein R). For carrier protein-TA conjugates in which TA is bound by its thiophene ring to proteins through a linker (BSA-S-5-TA and βLG-S-5-TA), as mentioned before, the second strategy of coupling between TA and carrier proteins is based on the ability of thiol groups to be substituted for the chlorine atom of 5-chloro-TA. Consequently, we first had to increase the number of thiol groups present on these proteins. This was done by activating lysine residues of BSA and βLG with a bifunctional linker consisting of SPDP. The number of thiol groups added to proteins by this method was subsequently dosed by the Ellman reaction. Thus, we were able to bind 32 and 20 thiols groups on BSA and βLG, respectively. Consequently, yields of lysine activation by SPDP were 54% for BSA (59 lysines/BSA) and 67% for βLG (30 lysines/βLG). Then these thiolated proteins were allowed to react with radiolabeled 5-chloroTA. The number of TA molecules bound per protein was further determined in two ways: radioactivity and UV spectroscopy. This latter determination was possible because the binding of TA to proteins resulted in the appearance of a strong and characteristic absorption signal at 364 nm similar to that of compound 1. Thus,
using the of compound 1 or radioactivity, we found that 22 and 11 molecules of TA were bound per BSA and βLG, respectively. (3) Preparation of Anti-TA-Protein Adduct Antibodies. Three rabbits were treated with 250 µg of either BSA-N-TA or BSA-5-S-TA. Since the number of TA molecules bound to BSA was nearly the same in both conjugates, we could determine which was more efficient in generating antibodies specific for TA chemically modified proteins. We monitored this specificity with ELISAs by following the recognition of casein-NTA when immunization was performed with BSA-NTA and recognition of βLG-S-5-TA when immunization was carried out with BSA-S-5-TA. Both BSA-N-TA and BSA-S-5-TA conjugates were immunogenic and elicited antibodies reactive toward their corresponding coating antigens. We selected the two rabbits that gave the higher level of recognition of either casein-N-TA or βLG-S-5-TA depending on the conjugate used for immunization. Their titers against casein-N-TA or βLG-S-5-TA were 1/3000 and >1/100000, respectively. This result indicated that BSA-S-5-TA was at least 30 times more immunogenic than BSA-N-TA. Then we tested the ability of our two different sera to recognize TAI metabolite-protein adducts. In this test, ELISA plates were coated with rat liver microsomes which have been incubated with TAI under conditions suitable for generating TAI metabolite-protein adducts.The typical S-shaped titration curves obtained show that sera obtained by BSA-N-TA immunization elicited a titer of 1/1200 against these TAI metabolite-protein adducts while sera obtained by BSA-S-5-TA immunization gave a titer of 1/128000. Consequently, immunization with the BSA-S-5-TA conjugate elicited antibodies 100 times more efficient in recognizing TAI metabolite-protein adducts. For this reason, we used the immunoglobulin fraction isolated by an ammonium sulfate precipitation from the sera generated by BSA-S-5-TA immunization in our further studies. This immunoglobulin fraction was further termed anti-TA-protein adduct. Titration of anti-
Alkylation of Tienilic Acid and Its Isomer
Chem. Res. Toxicol., Vol. 12, No. 3, 1999 291
Figure 3. Structures of compounds tested as competitive inhibitors for TA or TAI metabolite-protein adduct recognition by antiTA-protein adduct antibodies in ELISAs.
TA-protein adduct antibodies against βLG-S-5-TA was performed using serial dilutions of anti-TA-protein adduct antibodies ranging from 1/800 to 1/204800. This was done after we determinated that the best quantity of βLG-S-5-TA for coating ELISA plates was 47.5 ng of protein (providing 15 pmol of TA/well). In this case, no recognition of native βLG was observed even with an anti-TA-protein adduct dilution of 1/800. The titer against βLG-S-5-TA was 1/35000. (4) Characterization of Antibody Specificity. Compounds shown in Figure 3 and differing from TA in just small structural determinants were assayed as putative inhibitors of the antibody-antigen reaction between antiTA-protein adduct antibodies and βLG-S-5-TA. This was done by preincubating anti-TA-protein adduct antibodies (diluted 1/25000) with serial dilutions of the analogues for 1 night at 4 °C before adding them to the wells. For this anti-TA-protein adduct dilution, inhibition of the antibody-antigen reaction resulted in a decrease of the ELISA signal as a function of inhibitor concentration. The linear part of the curve allowed an estimation of the IC50. Figure 4 shows the results of experiments with TA, TAI, 1, ethacrynic acid (2), and 2,3dichlorophenoxyacetic acid (3) as putative inhibitors. The more potent and complete inhibition was observed with 1 (30 times more powerful as an inhibitor than TA). This was predictable since the coupling between TA and either the conjugate used for immunization or mercaptoethanol was identical. TA was at least 3 times more effective as a competitor for βLG-S-5-TA than was TAI. 2 was a weaker inhibitor, and 3 was not able to decrease significantly the level of recognition of βLG-S-5-TA by antiTA-protein adduct antibodies at concentrations ranging from 10-9 to 10-3 M. The same kind of inhibitions were performed with a coating consisting of rat microsomes which have been incubated with TAI under conditions suitable for generating TAI metabolite-protein adducts. Results expressed as IC50 are shown in Table 1. At high concentrations, all tested compounds completely inhibited
Figure 4. Competitive inhibition of βLG-S-5-TA recognition by anti-TA-protein adduct antibodies in ELISAs. Wells were coated with 47.5 ng of βLG-5-S-TA (15 pmol of TA/well). AntiTA-protein adduct antibodies (1/25000) were incubated with serial dilutions of different TA analogues overnight at 4 °C and were then added to wells.
the recognition of TAI metabolite-protein adducts. 1 was the best inhibitor of all compounds tested. Analogues carrying an aromatic thiophene ring and the 2,3-dichlorophenoxyacetic acid moiety (TA, TAI, and 4) were potent inhibitors exhibiting an IC50 averaging to 18.5 nM. Compounds 2 and 5 were good inhibitors. On the contrary, 3 was 4400 less efficient at inhibiting the signal than TA. Indeed, the more analogues differed from TA, the more their ability to compete with TAI metaboliteprotein adducts decreased. IC50 values of compounds differing from each other only by one structural determinant were compared. This allowed us to further characterize epitopes of TA or TAI
292 Chem. Res. Toxicol., Vol. 12, No. 3, 1999
Bonierbale et al.
Table 1. Competitive Inhibitors of TAIa and TAb Metabolite-Protein Adducts for Anti-TA-Protein Adduct Recognition in ELISAsc inhibitor IC50 (nM)a IC50 (nM)b inhibitor IC50 (nM)a IC50 (nM)b 1 TA TAI 4 2 5
2.3 7.5 17 31 50 150
4.5 4.5 13.5
8 6 7 3 9
1400 8800 10600 32800 460000
4600 29700 4600
a Rat liver microsomes were incubated with TAI to generate TAI metabolite-protein adducts. Then the amount of proteins corresponding to 20 pmol of bound TAI was coated onto each well of ELISA plates. Anti-TA-protein adduct dilution of 1/20000. See Materials and Methods for details. b Rat liver microsomes were incubated with TA to generate TA metabolite-protein adducts. Then the amount of proteins equivalent to 20 pmol of bound TA was coated onto each well of ELISA plates. Anti-TA-protein adduct dilution of 1/5000. c The concentrations of TA or TAI metabolite-protein adducts that were used for coating gave the best sensibility and the lowest background. Anti-S-5-TA dilutions used are around titers against TA or TAI metabolite-protein adducts. Values expressed as IC50 are means of two determinations.
metabolite-protein adducts recognized by anti-TAprotein adduct antibodies (26). Thus, TA was 1200 times more potent at inhibiting the recognition of TAI metabolite-protein adducts by anti-TA-protein adduct antibodies than 6. This result meant that the carboxyl group contributed to the 1200-fold increase in the level of recognition by the anti-TA-protein adduct antibodies. The effect of the loss of the two chlorine atoms was seen by comparing the IC50 values of 7 and TA. This comparison indicated that chlorine atoms contributed to the 1400fold increase in the level of recognition of TAI metaboliteprotein adducts by anti-TA-protein adduct antibodies. Similarly, the ketone function was also an important structural determinant recognized by anti-TA-protein adduct antibodies. This was concluded by the comparison of the IC50 values of 8 and TA on one hand and those of 3 and 5 on the other. Thus, the reduction of the ketone function or its suppression decreased the level of recognition of TAI metabolite-protein adducts by 200 times on average. On the other hand, the thiophene ring was not essential for anti-TA-protein adduct antibody recognition. This was concluded by comparison of the IC50 values of 2 and TA on one hand and the IC50 values of TA and 1 on the other. When the coating consisted of rat liver microsomes incubated with TA under conditions suitable for generating TA metabolite-protein adducts, TA and TAI gave the same IC50. This result indicated that the dichlorophenoxyacetic acid end was the major immunogenic part of TA since anti-TA-protein adduct antibodies were unable to distinguish TA from TAI in this assay. Altogether, these results show that recognition by the anti-BSA-S-5-TA antibodies is stronger for the dichlorophenoxyacetic acid part of TA which is not engaged in the covalent binding with proteins. Our mode of preparation of the BSA-S-5-TA antigen should lead to a mixture of BSA adducts with different BSA lysines selectively alkylated by the same TA-derived hapten (see Figure 2). This would result in the presence of this hapten in slightly different microenvironments due to different locations of the lysines in BSA and, consequently, to slightly different epitopes for the elicitation of antibodies. Thus, our antibodies generated against BSA-S-5-TA would be able to recognize not a single epitope but a
Table 2. Covalent Binding of Tienilic Acid (CBTA) and Its Isomer (CBTAI) to Microsomal Proteinsa type or source of microsome pAAH5/CYP2C9 with NADPH and without GSH pAAH5/CYP2C9 with NADPH and with GSH human liver with NADPH and without GSH human liver with NADPH and with GSH
CBTA (nmol of CBTAI (nmol of TA/nmol of P450) TAI/nmol of P450) 3.0 ( 0.1
72.5 ( 3.8
1.0 ( 0.1
7.0 ( 0.4
1.6 ( 0.1
14.6 ( 0.6
0.6 ( 0.1
