Specificity of in Vitro Covalent Binding of Tienilic Acid Metabolites to

Sylvaine Lecoeur,? Eric Bonierbale,* Dominique Challine,? Jean-Charles Gautier,t. Philippe Valadon,j Patrick M. Dansette,* Rachel Catinot,§ Francois ...
0 downloads 0 Views 3MB Size
Chem. Res. Toxicol. 1994, 7, 434-442

434

Specificity of in Vitro Covalent Binding of Tienilic Acid Metabolites to Human Liver Microsomes in Relationship to the Type of Hepatotoxicity: Comparison with Two Directly Hepatotoxic Drugs Sylvaine Lecoeur,? Eric Bonierbale,*Dominique Challine,? Jean-Charles Gautier,t Philippe Valadon,j Patrick M. Dansette,* Rachel Catinot,§ Francois Ballet,§ Daniel Mansuy,S and Philippe H. Beaune*lt Imtitut National de la Santk et de la Recherche Mkdicale U75, CHU Necker-Enfants-Malades, Uniuersitk Renk Descartes, F- 75730 Paris, France, URA 400 CNRS, Uniuersitk Renk Descartes, F-75270 Paris, France, and Dkpartement Skcuritk du M6dicament, Rhbne-Poulenc-Rorer, 3 digue d’Alfortville, 94140 Alfortville, France Received November 24,199P

In order to better understand the first steps leading to drug-induced immunoallergic hepatitis, we studied the target of anti-LKM2 autoantibodies appearing in tienilic acid-induced hepatitis, and the target of tienilic acid-reactive metabolites. It was identified as cytochrome P450 2C9, (P4502C9): indeed, anti-LKM2 specifically recognized P450 2C9, but none of the other P450s tested (including other 2C subfamily members, 2C8 and 2C18). Tienilic acid-reactive metabolite(s) specifically bound to P450 2C9, and experiments with yeast expressing active isolated P450s showed that P450 2C9 was responsible for tienilic acid-reactive metabolite(s) production. Results of qualitative and quantitative covalent binding of tienilic acid metabolitels) to human liver microsomes were then compared to those obtained with two drugs leading to direct toxic hepatitis, namely, acetaminophen and chloroform. Kinetic constants ( K m and Vm-) were measured, and the covalent binding profile of the metabolites to human liver microsomal proteins was studied. Tienilic acid had both the lowest K , and the highest covalent binding rate a t pharmacological doses. For acetaminophen and chloroform, several microsomal proteins were covalently bound, while covalent binding was highly specific for tienilic acid and dihydralazine, another drug leading to immunoallergic hepatitis. Although low numbers of drugs were tested, these results led us to think that there may exist a relationship between the specificity of covalent binding and the type of hepatotoxicity.

Introduction Liver is an important target for the toxic effects of drugs because of ita crucial role in the metabolism of xenobiotics. In general, two classes of drug-induced hepatotoxicity can be distinguished: that occurring systematically, as exemplified by acetaminophen overdose, and that involving idiosyncratic toxicity. Drugs can lead to idiosyncratic toxicity via a direct mechanism (toxicity of the drug or its metabolites) or through an abnormal immune response (1, 2). The phenomena leading to hepatotoxicity, and particularly to immunotoxicity, are not well elucidated. Upon metabolism, especially by P450,’the two classes of drugs may produce reactive intermediates able to alkylate cell macromolecules (4, 5). The formation of reactive metabolites is also involved in the first step of the postulated mechanism of autoimmune hepatitis (2,5,6). This mechanism involves several steps described in Figure 1(6): step 1,formation of reactive metabolites mainly by ~

*Addrew correspondence and reprint requeata to this author at INSERM U75, CHU Necker-Enfanta-Malades,156 rue de Vaugirard, F-75730 Paris Cedex 15, France. Telephone: 40 56 93 06. Fax: 40 61 55 82. t Institut Nationalde la SanG et de la Recherche Mbdicale,Universit.4 Ren6 Descartes. t URA 400 CNRS, Universit.4 Ren6 Descartes. I Rhbne-Poulenc-Rorer. 0 Abstract published in Advance ACS Abstracts, April 1, 1994. 1 Conventionsused in this article: The generic term ‘P450” is used to indicate a cytochrome P450. Individual forms of P450 are designated according to internationally accepted nomenclature (3).

REACTIVE METABOLITEzM’

anti-protein y

J

Figure 1. Postulated initial events in drug-inducedautoimmune hepatitis. Step 1: Formation of reactive metabolites (M*) by protein y. Step 2: Covalent binding of M* on y. The complex behaves as a neoantigen. Step 3: Immunological response on production of autoantibodies. Step 4: The autoantibodies recognize complex M*-y and the native protein y.

cytochromes P450;step 2, covalent binding of metabolites on the protein(& generating the reactive metabolite(s), particularly P450 the complex protein-reactivemetabolite behaves as a neoantigen; step 3, immunological response, including the production of autoantibodies; step 4, recognition by the autoantibodies of the neoantigen and/or native P450 which produced the reactive metabolite. Steps 1, 2, and 4 have been clearly demonstrated for dihydralazine-inducedhepatitis2 (7,8): cytochrome P450 lA, responsible for the production of diliydralazinereactive metabolites, was specificallyalkylated by the metabolites.

0S93-22~~/94/2707-0434$04.50I0 0 1994 American Chemical Society

Chem. Res. Toxicol., Vol. 7, No. 3, 1994 436

Covalent Binding and Immunoallergic Hepatitis

Furthermore, anti-liver microsome antibodies present in sera of patients suffering from dihydralazine hepatitis specificallyrecognized P450 1A2 (8). A similar mechanism has been shown for tienilic acid (TA)3 (6, 9): anti-liver kidney microsome autoantibodies (anti-LKM2) characteristic of this kind of hepatitis, recognized P45Os 2C which were shown to metabolize the drug. The aim of the present work was as follows: (1) to determine the specificity of recognition of the autoantibodies, i.e., which P450 of the 2C subfamily was recognized by the anti-LKM2; (2) to evaluate the specificity of the covalent binding of tienilic acid-reactive metabolite@)on P450s; (3) to show that the target of the autoantibodies and the target of reactive metabolites were identical; and (4) to determine qualitative and quantitative covalent binding of tienilic acid and to compare these data with those obtained with two drugs leading to direct hepatotoxicity (acetaminophen and chloroform). Our results showed the following: (1)that anti-LKM2 recognized only P450 2C9, but neither 2C8 nor 2C18; (2) that the reactive metabolite@) were produced mainly by P450 2C9 which was their unique target; (3) covalent binding was highly specific in the case of immunotoxic drugs (tienilic acid and dihydralazineI2 (IO)but was not specific in the case of nonimmunotoxic drugs (acetaminophen and chloroform). These results led us to hypothesize that a relationship might exist between the specificity of the covalent binding and the type of hepatotoxicity. Data concerning these four drugs are discussed.

Materials and Methods Chemicals. Electrophoresis reagents were purchased from Serva Fine Biochemicals (Heidelberg, Germany), nitrocellulose sheets from Bio-Rad (Richmond, CA), peroxidase-conjugated anti-human and anti-rabbit immunoglobulins from Dako (Copenhagen, Denmark), luminol (ECL Western kit) and scintillation liquid (ACS 11)from Amersham (Buckinghamshire, U.K.), and NADP from Boehringer (Mannheim,Germany). Glass-fiber fiiter disks (ref. GF/B) were from Whatman International Ltd. (Maidstone, U.K.). Other reagents, all of analytical grade, were purchased either from Prolabo (Paris, France) or from Sigma (St. Louis, MO). [keto-WITienilic acid (56 mCi/mmol, HPLC purity 99 %) was purchased from Amersham (Buckinghamshire, U.K.) and was a generous gift from Novo Nordisk (Denmark); [W]chloroform (3.6 mCi/mmol, HPLC purity99%) and [G-*H]acetaminophen (1.3Ci/mmol,HPLCpurity98%)werefromNEN Research Products (Les Ulis, France). Human Sera. Six serum samples containing anti-LKM2 antibodies were obtained from patients suffering from TAinduced hepatitis. They were a generous gift of Dr. C. Andre (CHU Henri Mondor, Crete& France). Human Liver Microsomes. Human livers, obtained from kidney transplant donors in compliance with French law, were immediately frozen and stored at -80 OC. Microsomal fractions were prepared by differential centrifugation as previously described (II), frozen, and stored at -80 "C. Their cytochrome P450 contents were determined by differential spectroscopy (12), and the protein concentration was measured according to Lowry et al. (13) using bovine serum albumin as standard.

* M. Bourdi, M. Tinel, P. H. Beaune, and D. Passayre. Interactions of Dihydralazine with Cytochromes P450 1 A A Possible Explanation for the Appearance of Anti-P450 1A2Autoantibodies. Article in revision for Mol. Pharmacol. 9 Abbreviations: TA, tienilic acid; anti-LKM2, anti-liver and kidney microsomalantibodies type 2;ECL, enhanced chemolumineecence;GST, glutathione S-transferase;anti-LM, anti-liver microsomalantibodies; SDS, sodium dodecyl sulfak, PAGE, polyacrylamide gel electrophoresis; PCR, polymerase chain reaction; IPTG,isopropyl B-D-thiogalactopyranoside; PBS, phosphate-buffered saline.

Antibodies. AntLP450 2C and 1A were rabbit polyclonal antibodies prepared in the laboratory against human P450 MP (14) and rat P450 1A1, respectively. Monoclonal anti-human P450 3A4 was described in Beaune et al. (15). Anti-human P450 2E1 was a generous gift of Dr. F. P. Guengerich (Vanderbilt University School of Medicine, Nashville, TN). They all recognized a single band or a major band in human liver microsomes, and they only recognized the corresponding P450 expressed in the yeast. Antibodies to TA were obtained in rabbits by injecting bovine serum albumin coupled with 5-chlorotienilic acid through an SH linker (16). They mainly recognized the (4-carbonyl-2,3-dichlorophenoxy)aceticmoiety of TA.4 These antibodies were used to develop blots of human liver microsomes incubated with TA and NADPH. Quantitative Covalent Binding. Covalent binding of reactive metabolites of [keto-14C]TA,[Wlchloroform, and [G-*H]acetaminophen was measured accordingto a previously described method (17,18). Human liver microsomes (50 pL corresponding to 0.2 nmol of P450) were incubated for 5 min with the radiochemical (seven concentrations from 0.1 to 100 pM), 1.5 mM NADP, 25 mM glucose 6-phosphate (G6P), and 0.25 unit of glucose-6-phosphate dehydrogenase (G6PD) in 150-pL fiial volume of buffer (100 mM Na2HPO4, 10 mM MgCla, 1 mM diethylenetriaminepentaaceticacid, pH 7.4). Covalent binding of each radiochemical tested was linear with respect to both protein concentration and time of incubation. Glass-fiber fiiter disks were previously dipped in 10% trichloroacetic acid. Aliquota (50 pL) of incubation mixtures were then loaded onto filter disks. Filters were washed twice by methanol and then once by ethyl acetate. Filters were dried and counted after addition of 5 mL of scintillation liquid. Results were obtained with the radiochemical used at increasing concentrations, from lo-' to lo-' M. For each point of concentration, two experiments in duplicate were performed, with or without an NADPHgenerating system. Km and .,V on human liver microsomes were graphically determined from double-reciprocal plots. Quantitative covalent binding of TA on yeast microsomes was performed with 0.1 mM TA (about 1pCi) and yeast microsomes (50 pmol of P450) in 150-pL final volume of buffer. Qualitative Covalent Binding. (A) Preparation of Samples. Human liver microsomes (1 nmol of P450) were incubated for 30 min with the radiochemical (about 1pCi), 1.5 mM NADP, 25 mM G6P, 2 units of G6PD in 1mL of the same buffer as described above. Proteins from the incubation mixture were precipitated by addition of 1 mL of 10% trichloroacetic acid and washed with 2 mL of various compounds: 5% trichloroacetic acid, ethyl acetate (twice),acetone, buffer; between each washing, a pellet was obtained by centrifugation for 10 min at 3000g. The final pellet was dissolved in 0.5 mL of solubilization buffer [30% glycerol, 1%sodium dodecyl sulfate (SDS), 0.2 M Tris-HC1,pH 6.8,0.01% pyronine]. Before electrophoresis, each sample was heated to 100 OC for 2 min with or without 7.5% mercaptuethanol. Yeast microsomes corresponding to 50 pmol of P450 were incubated with TA (about 1pCi); for control yeast, an equivalent amount of microsomal protein was used. (B) Electrophoresis and Immunoblotting. Polyacrylamide gel electrophoresis in the presence of SDS (SDS-PAGE) was performed according to the method of Laemmli (19) using 4 % stacking and 9 % separating gels. Proteins were electrotransferred (1h, 400 mA) to a nitrocellulose sheet (20). After transfer, part of the sheet corresponding to six lanes was cut every 2 mm perpendicular to the direction of migration, and each band was counted after addition of 5 mL of scintillator in a scintillation counter. The part of the sheet corresponding to the central lanes was kept for immunoblots and probed with anti-human or antirat P450 lA1-2,2C, 2E1, and 3A4. Not all P45Os could be tested for each sheet. Therefore, only those of interest were assayed in each experiment. Immunoblots were developed as previously described (15) with 4-chloro-1-naphthol or luminol (enhanced chemiluminescence ECL) as substrate (according to the manu4

P. Valadon et al., unpublished results.

Lecoeur et al.

436 Chem. Res. Toxicol., Vol. 7, No. 3,1994 1

A

2

3

4

1

2

3

2 3

4

4

2 3

2

4

3

4

2

3

4

2C18

antLP450 2C rabbit IgG

r

B

1

SAM

2

3

1

2

PIJ

3

1

2

BOV

MIG

3

1

2

3

DIC

1

2

3

P450 2C9

+ +

P450 2C8

anti-P450 2C rabbit IgG

PIJ

SAM

MIG

DIC

Figure 2. Recognition of P450 2C8, P450 2C9, and P450 2C18 expressed in yeast and in human liver by anti-LKM2 and by rabbit IgG human anti-P450 2C. Yeast microsomes were separated by electrophoresis and transferred to a nitrocellulose sheet. Immunoblots were performed with anti-LKMZ autoantibodiesof patients suffering from TA-induced hepatitis (diluted 1/1oooO);S A M ,PIJ, MIG, BOV, and DIC are the codes for different patient sera. Anti-human IgG labeled with peroxidase (diluted 1/2oooO) was used as the second antibody. Blots were stained with luminol. (A) Lane 1: Human liver microsomes corresponding to 10 mg of proteins. Lane 2: Control yeast microsomes. Lane 3: Yeast microsomes correspondingto 2.5 pmol of P450 2C9. Lane 4: Yeast microsomescorresponding to 5 pmol of P450 2C18. (B) Lane 1: Control yeast microsomes. Lane 2: Yeast microsomes corresponding to 2.5 pmol of P450 2C9. Lane 3: Yeast microsomes corresponding to 5 pmol of P450 2C8. facturer's recommendations). The radioactive count of each band enabled obtaining of a covalent binding profile which was compared with the location of P450 by antibodies. Yeast Preparation. Human liver P450 1Al(21), 1A2,52C8: 2C18,6 2D6 (22),3A4 (23, and 2C9 (24) cDNA-coding sequences were amplified by the polymerase chain reaction (PCR). When cDNAs were obtained from mRNA, the sequence was checked by full sequencing, and very few mutations were observeds (21). The P450 cDNAs were inserted in the yeast expression vector pV8 or pV60. These vectors are based on URA3 (pV8 and pV60) and ADEZ (pV60) selection markers and the GAL10-CYC1 promoter (21,25). They were transfected in the yeast Saccharomyces cereuisiae strain W303R. The P450 reductase gene of this strain was placed under the control of the hybrid promoter GAL10-CYC1 and consequently overexpressed P450 reductase when grown in the presence of galactose as carbon source. Microsomes were prepared according to methods previously described (26). Specific activities of yeast microsomes were respectively 27,48,39,32,87,12, and 100 pmol of P450/mg of protein, for P450 lAl,lA2,2C8, 2C9,2C18,2D6, and 3A4. The yeast microsomes were active on their characteristic substrates (27): 1A1, ethoxyresorufin; lA2, methoxyresorufin; 3A4, testosterone G@-hydroxylation;2D6, imipramine 2-hydroxylation. P450 2C9 and 2C18 were active on TA 5-hydroxylation [respectively 1.18 f 0.24 and 0.68 f 0.15 nmol/(mimnmol of P450); P450 2C8 was able to slowly metabolizebenu>[a]pyrene, but was unable to hydroxylate TA. The yeast microsomes were checked in immunoblots with anti-human P450 and were used for covalent binding of TA and for characterizing the anti-LKM2 antibodies. BacteriaPreparation. The human P450 2C9 cDNA-coding sequence (24) was amplified by PCR and inserted into bacterial expression vector pGex 1XT (Pharmacia, France). This vector is based on ampicillin resistance, a tac promotor inducible by isopropyl &D-thiogalactopyranoside (JPTG) and expresses a cloned gene as a fusion protein to glutathione S-transferase (GST) (28). Strain Escherichia coli JMlOl was transfected and grown ~

5 J.

C. Gautier and J. Coeme, unpubliihed resulta.

at 37 "C in Luria-Bertani medium containing 50 pg/mL ampicillin. After addition of 100 pM IPTG and growth for 2 h at 37 "C, cells were harvested by centrifugation. Pelleted cells were resuspended in phosphate-buffered saline (PBS), lysed by sonication and 10% Triton, and centrifuged at 12OOOg for 30 min. The pellet was resuspended in PBS. An aliquot of pellet and supernatant was checked in immunoblotswith anti-human P450 2C and was also used to characterize the anti-LKMZ autoantibodies.

Results Recognition of P450 2C9 by Anti-LKM2 Autoantibodies: Specificity of anti-LKM2. The specificity of recognition by autoantibodieswas first determined: all five sera from patients treated with TA and who suffered from autoimmune hepatitisrecognized P450 2C9 expressed in yeast; one of them (MIG) also very weakly recognized P450 2C18 expressed in yeast (Figure2A). The sera tested had different titers; however, they all recognized P450 2C9, but not P450 2C18 nor 2C8 (amountsof spectrally determined P450 2C8 and 2C18 loaded were twice those of P450 2C9) (Figure 2). The sera also recognized minor bands correspondingto yeast proteins. These bands were recognized equally well in both control and modified yeasts. The sera recognized the fusion protein involving 2C9, expressed in E. coli (Figure 3). This fusion protein was found in the pellet, indicating that it was hydrophobic and insoluble. This fusion protein gave proteolysis products, which were recognized by anti-LKM2 (Figure 3). Specificrecognitionof P450 2C9 by sera was confirmed by the fact that there existed no recognition of P450 2C9 by control serum, no recognition of non-IPTG-induced bacteria (i.e., not expressing the fusion protein), and no recognition of P450 lAl,lA2,2D6, and 3A4 inserted in yeast (Figure 3). In human liver microsomes, anti-LKM2

Chem. Res. Toxicol., Vol. 7, No. 3,1994 437

Covalent Binding and Immunoallergic Hepatitis

Table 1. Kinatic Parameters of Total Irrevereible Binding of Tienilic Acid, Chloroform, and Acetaminophen to Human Liver Microsomes. covalent binding: covalent binding: apparent Vlmxb 0.01 mM 1 mM Kmb (PM) [pmoV(min.mg)l [pmoV(min*mg)l [pmoV(min*mg)l tienilic acid 7f6 58 f 35 40f 18e 162 f 96f 112 f 58 78 f 41 chloroform 250 f 210 17 f 7 acetaminophen 9f7 0.24 f 0.17 0.24 f 0.14 13 f 7 a Conditions were as described in Materials and Methods, with incubations containing 0.1-100 mM radiochemicals, 0.2 nmol of cytochrome P450,and a NADPH-generating system in a final volume of 150 mL. Apparent Kmwas calculated by double-reciprocal plots of irreversible binding of radioactive tienilic acid, chloroform, and acetaminophen as a function of drug concentration. b Mean of 3 experiments on different livers f SD. c Mean of 8 experiments on different livers f SD. d Mean of 5 experiments on different livers f SD. Significantly different from covalent binding data of 0.01 mM acetaminophen and 0.01 mM chloroform, p < 0.01. f Significantly different from covalent binding data of 1 mM acetaminophen and 1 mM chloroform, p < 0.01. Data were analyzed statistically using distribution-freeMann and Whitney unpaired

t-test between covalent binding of tienilic acid and chloroform or acetaminophen.

Figure 3. Recognition by anti-LKM2 of P 4 W in human liver, P45Os expressed in yeast and in IPTG-induced bacteria. AntiLKM2 (serum PIJ) was diluted 1/1oooO. Immunoblots were performed with luminol. Lanes 1and 2: Pellet and supernatant of bacteria transformed with unmodified pGEX 1XT. Lanes 3 and 4: Pellet and supernatant of IPTG-induced bacteria transformed with P450 2C9 inserted in pGEX 1XT. Lanes 5 and 6: Pellet and supernatant of noninduced bacteria transformed with P450 2C9 inserted in pGEX 1XT. Lane 7: Human liver microsomes corresponding to 10 mg of proteins. Lanes 8-12: Yeast microsomes corresponding to 5 pmol of P450 2C18,2C9, 3A4, 1A1, and 1A2, respectively. Lane 13: Control yeast microsomes.

recognized one band, with the same molecular weight as that in yeast expressing P450 2C9; none of the anti-LKM2 tested recognized a lower band in the P450 2C8,2C18, or 2C19 region. Rabbit IgG anti-human P450 2C were also tested (not shown): they recognized both P450 2C9 and 2C18 expressed in the yeast, the P450 2C9 fusion protein expressed in induced bacteria, and one major band in human liver microsomes. This band corresponded to P450 2C9 (the same level of migration on electrophoresis). Rabbit IgG anti-human P450 2C weakly recognized human liver P450 2C18 and 2C8 on immunoblot: this was not surprising, since these P450s are probably expressed at lower levels in human liver microsomes (29,30). QuantitativeCovalent Binding. Before testing the specificity of covalent binding of TA-reactive metabolites, the kinetic parameters of TA covalent binding to human liver microsomes were determined and compared with data obtained for two drugs which lead to direct hepatotoxicity, acetaminophen and chloroform. Some of these results have already been published (17,31), but we sought to measure the kinetic values using the same human liver microsomes and the same methodology. As shown in Table 1,the K m of P450 toward TA was low (7 pM), while Vwas about 60 pmol/ (min-mg). In comparison, chloroform presented a higher apparent K m (250 pM) and thus lower affinity, with a fairly high V- [112 pmol/(min.mg)]. Acetaminophen presented a lower level of covalent binding [ V m m = 0.24 pmol/(min.mg)] with low K m (9 hM), indicating good affinity of P450 for this drug. K m and

V- presented in the table were determined with low concentrationsof substrates. For acetaminophen and TA, double-reciprocal plots at a higher concentration indicated a second K m and V-, higher than that of Table 1. For chloroform, only one Kmand one V- were found whatever the concentration used. Therefore, in order to better compare these compounds, their respective covalent bindings were evaluated at two concentrations (Table 1): 10 pM in the range of the low K m high affinity for acetaminophen and TA, and 1mM in saturation or nearsaturation conditions for the three compounds. In these two situations, covalent binding was higher for TA. It is of interest that the Km of TA (7 pM) was in the range of the plasma concentration (10-13 pM for a 250-mg oral dose; 32). Values obtained were in good agreement with those found in the literature (17,31,33). QualitativeCovalent Binding. Qualitative covalent binding was assessed in order to identify the microsomal proteins to which the reactive metabolites were bound. After separation of microsomal proteins by SDS-PAGE, transfer to a nitrocellulose sheet, and counting of bands, peaks of radioactivity were visualized, corresponding to protein(& to which radioactive metabolites were bound. In the case of TA (Figure 4A), one very major radioactive peak was well individualized and comigrated with the band recognized by rabbit anti-human P450 2C sera. This was consistent with binding of reactive metabolites to one protein, namely, the P450 which produced them. It is quite likely that this P450 was 2C9, since P45Os 2C8,2C18 (see Figure 2), and 2C19 (34)migrate in SDS-PAGE at a lower molecular weight, clearly different from the top of the radioactive peak. Covalent binding obtained with acetaminophen(Figure 5 ) showed several individualized peaks of radioactivity, indicating that microsomal binding of metabolites was nonspecific. These results were in good agreement with the literature: several cytosolic and microsomal proteins were bound by oxidative metabolites (35,36).Acetaminophen (0.8 pM) was used in order to load the same amount of radioactivity (1pCi) on the electrophoresis gel for the three drugs tested. However, qualitative covalent binding experimentswith 16pM acetaminophen (a concentration similar to that used for TA) were performed, and the profde of radioactivity bound to microsomal proteins was quite similar to that of Figure 5 (data not shown) but different from that obtained with TA. In comparison to TA, chloroform showed covalent binding to several proteins (Figure 6). Radioactivity which specifically bound to proteins comigrating with P450 was much higher for TA (22-24% of total bound radioactivity in the two experi-

438 Chem. Res. Toxicol., Vol. 7, No.3, 1994

Lecoeut et at. protein migration

C

P 450 2C

a

i

2o ol

-

94-

67

43-

30-

0 0

20

10

30

40

50

protein migration

B T p - " t

1

--

2

1

2

3

Figure 4. Covalent binding of [ W J T A to human liver microsomes: covalently bound radioactivity and immunoblots. Conditions were as described in Materialsand Methods using human liver microsomes (1nmol of P450) and 18pM [WJTA. (A) Covalent binding of [W]TA to human liver microsomes. Three covalent binding profiles on different livers were carried out, and one is shown in this figure. (0 - 0 )With the NADPH-generatingsystem; (0-- -0)without the NADPH-generating system. Arrow indicates migration of human liver P450 2C on panel A. (B) Immunoblot of human liver microsomes incubated with 18 pM TA and NADPH (lane 1) or without NADPH (lane 2) was developed with anti-TA antibodies and with 4-chloronaphthol. (C) Immunoblots of human liver microsomes incubated with 24 pM TA and NADPH (lane 1)or without NADPH (lane 2) were developed with anti-TA rabbit antibodies (lanes 1and 2) or with anti-LKM2 (lane 3) and developed with luninol.

1

0

P450 2E

ia

10

20

30

40

50

band number

,-= , 0

band number 10

20

30

protein migration

40

50 c

protein migration

Figure 5. Covalent binding of [aHJacetaminophen to human liver microsomes: covalently bound radioactivity. Conditions were as described in Materials and Methods using human liver microsomes (1 nmol of P450) and 0.8 pM acetaminophen. Immunoblotwas developed with anti-human P450 2E1 (diluted 1/1OOO) and anti-rat 1A1-2 (diluted 1/500) (result not shown). P450 2E1 and 1A were chosen because of their role in the metabolization of the drug (16). Three covalent binding profiles on different livers were performed, and a typical one is shown in this figure. (+) With the NADPH-generating system; ( 8 ) without the NADPH-generating system.

microsomes: covalently bound radioactivity. Conditions were as described in Materials and Methods using human liver microsomes (1nmol of P450) and 0.15 mM chloroform. Immunoblot was developed with anti-human P450 2E1 (diluted l/lW), since this P450 was shown to metabolize chloroform and anti-rat 1A1-2 (diluted 1/500). Three covalent binding profiles on different livers were performed, and a typical one is shown. (+) With the NADPH-generating system; ( 8 )without the NADPHgenerating system.

menta performed) than for chloroform (643.5%) or acetaminophen (24%). Covalent Binding of TA-Reactive Metabolites to Yeast Microsomes Expressing Isolated P450s. To establish the role of P4502C9 as a producer of TA-reactive metabolite(s) and as a target of covalent binding, several yeasts were used, each of which expressed a single cloned P450. Data corresponding to covalent binding of TA to different human P450expressed in the yeast are reported in Table 2. P450 2C9 was the main P450 producing TAreactive metabolite(s) [358f 17pmol/(min-nmolof P450)l;

P450 lAl,lA2, and 2C18 metabolized TA to a far lesser extent [respectively, 77,44, and 39 f 6 pmol/(min*nmol of P450)l. P450 3A4 did not produce any significant detectable reactive metabolite. Evaluation of the P450content in human liver (37)led to an estimate of the role of each tested P450 in the metabolism of TA (Table 2). P4502C9 and 2C18contents in human liver were calculated from data found in the literature, i.e., about 55% and 12%, respectively, of the total P4502C content (38). The total P4502C content in human liver was estimated to be 0.125 nmol/mg of

Figure 6. Covalent binding of [l~]chloroformto human liver

Chem. Res. Toxicol., Vol. 7, No. 3, 1994 439

Covalent Binding and Immunoallergic Hepatitis

Table 2. Covalent Binding of Tienilic Acid to Yeast a n d H u m a n Microsomes. covalent bindin8 [pmol/(min.nmol of P45O)l human liver control P450 1 A l P450 1A2 P450 2C9 P450 2D6 P450 3A4 P450 2C18 P450 2C8

calcd covalent binding [pmol/(min.mg of protein)]

human liver P450 concn (nmol of P450/mg of microsomal protein)

98

0.49 f O . l l c

48

77

0.0025d 0.025d 0.075’ 0.02tjd 0.25d 0.015* not determined

0.2 1.1 27

co.1 44 358 f 17 6

co.1

39 f 6 8f1

calcd % of covalent b i n d i d

0.7 3.8 92.8

0.15

0.5