Molecular mimicry of trifluoroacetylated human liver protein adducts by constitutive proteins and immunochemical evidence for its impairment in halothane hepatitis

Eur. J. Biochem. 210, 569 - 576 (1992) 0 FEnS 1992

Molecular mimicry of trifluoroacetylated human liver protein adducts by constitutive proteins and immunochemical evidence for its impairment in halothane hepatitis

JosefG7JT', Urs CHRISTEN1, Jorg HUWYLER', Maria BURGIN' and J. Gerald KENNA'

Department of Pharmacology, Biocenter of the University of Basel. Switzerland Department of Pharmacology and Toxicology, St. Mary's Hospital Medical School, London, England

(Received August 6 , 1992) - EJB 92 1140

A monospecific antibody (anti-CF3C0 antibody) was obtained by affinity chromatography on a N"-trifluoroacetyl-L-lysine (CF3CO-Lys) matrix of a rabbit polyclonal antiserum, directed against trifluoroacetylated protein adducts (CF,CO-proteins). The anti-CF3C0 antibody recognized distinct CF3CO-proteins on immunoblots of a liver biopsy obtained from a human individual 10 h after halothane anaesthesia. Cross-reactive proteins of 52 kDa and 64 kDa were recognized on immunoblots of livers obtained from human individuals not exposed to halothane. Recognition of both CF,CO-proteins and the 52-kDa and 64-kDa cross-reactive proteins was abolished in the presence of 1 mM CF3CO-Lys. Anti-CF3C0 antibody, affinity-adsorbed to the 52-kDa or the 64-kDa cross-reactive proteins of human liver, recognized the majority of target CF,CO-proteins on immumoblots of the human liver biopsy of an individual exposed to halothane. Liver biopsies of 5 out of 7 (71 %) patients with halothane hepatitis exhibited an absence or low amounts of immuno- recognizable 52-kDa and/or 64-kDa cross-reactive proteins. In contrast, of 22 control human individuals tested, all liver tissue samples were positive for the 52-kDa and/or the 64-kDa cross-reactive proteins. These data indicate that epitopes on the cross-reactive proteins of 52 kDa and 64 kDa of human liver bear strong immunochemical resemblance to epitopes on human liver CF,CO-proteins. Low-level expression of the cross-reactive proteins of 52 kDa and 64 kDa is discussed as one possible factor in human susceptibility to halothane hepatitis.

Halothane hepatitis is a severe form of hepatoxicity which occurs in between I in 30000 and 1 in 3000 patients exposed to the anaesthetic agent halothane [l]. A variety of clinical and laboratory findings suggest that halothane hepatitis has an immunological basis [2]. In particular, it has been shown that sera from patients with halothane hepatitis contain antibodies against a group of liver protein antigens which are trifluoroacetylated on lysine residues, as a consequence of interaction with the reactive CF,COCI metabolite of halothane, and which range in molecular mass from 54 kDa to 100 kDa [3,4]. The acyl halide CF3COCI is a major metabolite produced upon oxidative, cytochrome-P450-mediated metab- olism of halothane. Subcelluar fractionation studies have shown that, in livers from rabbits and rats exposed to halothane in vivo, the trifluoroacetylated protein antigens (so-

Correspondence to J. Gut, Department of Pharmacology, Bio- center of the University, Klingelbergstrasse 70, CH-4056 Bascl, Switzerland

Fax: +41 61267 2208. Ab6reviufion.s. Halothane, 2-bromo-2-chloro-l,1 J-trifluoroetha-

ne; CE',CO-RSA, trifluoroacetylated rabbit serum albumin; CF3CO- Lys, ~-trifluoroacetyl-L-lysinc; CF3CO-proteins, trifluoroacetylated protein adducts, i. e. any polypeptide carrying trifluoroacctylated amino acid residues without reference to the function or identity of the polypeptide; anti-CF,CO antibody, the monospecific TgC fraction obtained from the polyclonal antiserum raised against CF,CO-RSA through affinity purification on a CF3CO-Lys matrix.

called CF3CO-proteins, are concentrated in the rnicrosomal fraction [5, 61. CF,CO-proteins have also been detected on the surface of hepatocytes from halothane-treated rats [7] and in centrilobular sections from livers of halothane-exposed guinea-pigs [8]. Limited studies have indicated that similar, if not identical, CF,CO-proteins are present in livers of halothane-exposed human individuals [6, 91. Several of the CF,CO-proteins have been purified from livers of halothane- treated rats and characterized by amino acid sequence analysis and cDNA cloning. Antigens of 57, 59, 63, and 100 kDa have been identified as trifluoroacetylated forms of protein disulfide isomerase [lo], microsomal carboxylesterase [I I], the Ca2 +-binding protein calreticulum [12], and the stress protein ERp99 [13], respectively. The function of a further antigen (58 kDa) remains to be determined [14].

Current evidence indicates that all individuals produce CF,CO-proteins when exposed to halothane [6, 7, 15, 161. However, an antibody response to these adducts appears to bc restricted to the small subset of susceptible individuals who develop halothane hepatitis [4, 61. Among others, one possibility for the lack of immune responsiveness in normal individuals might be the existence of a repertoire of self- peptides which bear a strong structural resemblance to the trifluoroacetyl-modified protein epitopes produced in the liver following exposure to halothane. Such structural analogues might promote natural tolerance by selective thymic deletion [17 -201, and/or by induction of thymic or peripheral anergy

570

[21, 221, of maturing lymphocytes able to recognize epitopes that structurally mimic epitopes present on CF,CO-proteins.

The concept of tolerance to self-epitopes which resemble epitopes present on CF3CO-proteins has been supported by recent experiments performed in this laboratory [16]. Studies undertaken with an affinity-purified monospecific anti- CF,CO antibody revealed the existence of constitutive protein antigens of 52 kDa and 64 kDa in homogenates prepared from liver, kidney, and heart tissue of rats that had never been exposed to halothane. Hapten inhibition studies and antibody cxchange experiments demonstrated that both proteins expressed self-epitopes which bore a strong immunochemical similarity to the CF,CO-proteins produced in the livers of halothane-treated rats.

In the present study, we have investigated whether molecular mimicry of CF3CO-proteins by epitopes present on constitutive proteins also occurs in humans. In addition, we analy7ed liver samples from seven patients afflicted with halothane hepatitis to investigate whether in such individuals expression of constitutive 'CF,CO-like' protein epitopes might be impaired.

EXPERIMENTAL PROCEDURES

Materials

W-Trifluoroacetyl-L-lysine (CF,CO-Lys) was purchased from Senn Chemicals (Dielsdorf, Switzerland). Rabbit serum albumin, phcnylmethylsulfonyl fluoride, and soybean trypsin/chymotrypsin inhibitor were obtained from Sigma (St. Louis, MO, USA). Goat anti-(rabbit IgG) - horseradish-peroxidase conjugate and Affi-Gel 102 amino-terminal agarose were from Bio-Rad (Richmond, CA, USA). Halothane was obtained from Halothane Laboratories (Hackcnsack, NJ, USA) and was distilled prior to use. The enhanced chemiluminescence detection system was obtained from Amersham ln- ternational (Amersham, U K). The monospecific anti-CF,CO antibody used in these experiments was prepared by affinity chromatography of an antiserum raised against trifluoroacetylated rabbit serum albumin (CF,CO-RSA) on an Affi-Gel 102 amino-terminal agarose column, to which CF,CO-Lys had bcen coupled [16]. Aliquots of the final preparation (0.1 mg IgG/mlj were stored at - 80°C and thawed only once.

Treatment of rats

Male Sprdgue-Dawley rats (250 - 300 g) were pretreated by three daily injections (intraperitoneally) of 80 mg/kg body mass of sodium phenobarbital in 10 mM Na2HP04, 3 mM KH2P04, 137 mM NaC1, pH 7.4 (buffer A). Halothane (10 mmol/kg body mass, as a 50% by vol. solution in sesame oil) was administered 18 h before sacrified. Control animals did not receive halothane. Rats were sacrified by decapitation. The livers were removed immediately and rinsed with ice-cold 50 mM Tris/HCl pH 7.4, containing 0.1 35 M NaCl, 0.5 mM phenylmethylsulfonyl fluoride and 60 pg/ml soybean trypsin/ chymotrypsin inhibitor; homogenates (1 : 5, massivol.) were prepared by disruption of the tissue by four strokes in a Potter- Elevehjem homogenizer. Aliquots were stored at ~ 80 "C.

Preparation of homogenates from human liver samples

Frozen (- 80 'C) human liver tissue, obtained from kidney donor individuals, was kindly provided by Prof. U. A. Meyer (Department of Pharmacology, Biocenter of the University,

Basel, Switzerland) and taken from thc human liver bank cstablished in his laboratory [23]. Human liver biopsy samples were obtained immediately post-mortem from six patients with halothane hepatitis who died from complication of fulminant hepatic failure, from one patient who died of cardiac failure following anaesthesia with halothane, and from three patients who died of cardiac failure following anaesthesia with agents other than halothanc. In addition, a portion of a diagnostic liver biopsy sample was obtained from a patient with halothane hepatitis, with the informed oral consent of the patient. All biopys samples were snap-frozen in liquid nitrogen and stored at - 70°C. Frozen human liver sections or human liver biopsy samples were thawed in 50 mM Tris/ HCl pH 7.4, containing 0.135 M NaCl, 0.5 M phenylmethylsulfonyl fluoride, and 60 pg/ml soybean trypsin/chymotrypsin inhibitor, and homogenates were prepared as described above.

Gel electrophoresis and immunoblotting

Liver homogenates (5 - 10 mg protein/ml) were mixed (1 : 1, by vol.) with dissociation buffer (12 mM Tris/HCl pH 6.8, 8% mass/vol. SDS, 20% glycerol, 4 mM EDTA, 40 mM dithiothreitolj and heated to 95'-C for 10 min. SDSjPAGE was performed [24] using a 4% stacking gel and a 10% separating gel. Protein loading was 250 pg/cni slot width. Electrophoresis was for 4 h at 30 mA/gel. Resolved proteins were transferred electrophoretically to nitrocellulose [25] at 360 V h using a transfer buffer comprising 15.6 mM Tris, 120 mM glycine and 20Y0 methanol (by vol.) at pH 8.3. After transfer, the nitrocellulose was stained with amido black for visualizing proteins, destained, and then blocked for 2 h at room temperature with buffer A containing 2% (massjvol.) dry milk powder and 0.02% (mass/vol.) Thimerosal (blocking solution). The nitrocellulose was cut into strips of 3 mm width which were used for antibody overlay. Incubation with anti- CF3C0 antibody (diluted 1 : 64 into 500 p1 blocking solution) was for 18 h at room temperature. After five 5-min washes with blocking solution, incubation with horseradish-peroxidase-conjugated goat anti-rabbit second antibody (diluted 1: 100 into 500 p1 blocking solution) was for 2 h at room temperature. One wash with blocking solution and three washes with buffer A for 5 min each were followed by a final wash with buffer A for 30 min. Visualization of antigenic proteins was by enhanced chemiluminescence detection, which was performed according to the manufacturer's protocol.

Antibody exchange immunochemistry

In antibody exchange experiments [26], anti-CF3C0 antibody (diluted 1 : 16 in 500 pI blocking solution, corresponding to 18.5 pg IgG) was adsorbed (18 h) to the cross-reactive proteins of 52 kDa and 64 kDa present on strips obtained from immunoblots of human liver. After five 5-min washes with blocking solution, the areas of the strips containing either the 52-kDa or the 64-kDa protein/anti-CF3C0 antibody complex were excised (disk of 3 x 10 mm each). Ten disks, constituting the sole source of anti-CF3C0 antibody, were co- incubated with one target strip, originating from an immunoblot of a human liver homogenate obtained from an individual 10 h post-halothane anaesthesia, in 800 p1 blocking solution for 18 h. After five 5-min washes with fresh blocking solution, the target strips were incubated with horseradish- peroxidase-conjugated second antibody (diluted 1 : 100) for 2 h, then washed once with blocking solution and three times with buffer A for 5 min each followed by a 30-min wash

57 1

A -205

-116 - 97.4

6 -205

-1 16 - 97.4

-66 -6 6

-4 5

-3 6

-2 9

-4 5

-36

-2 9

1 2 3 4 1 2 3 4

Fig. 1. Recognition of CF,CO-proteins and to 52-kDa and 64-kDa cross- reactive proteins in rat (A) and human (B) liver tissue. (A) The recognition of cross-reactive proteins and of CF3CO-proteins was analyzed on immunoblots of rat liver tissue. Lane 1, rat not exposed to halothane; lane 3, rat exposed to a single dose of halothane 18 h before the experiment; lanes 2 and 4, corresponding experiments in presence of 1 mM CF,CO-Lys. (B) Immunoblotting experiments with anti-CF,CO antibody were perrormed using human liver tissue. Lanc 1, liver ofa individual not exposed to halothane (KDL 34); lane 3, liver of patient 8 (see Table 1); lanes 2 and 4, corresponding experiments in presence of 1 mM CF3CO-Lys. Migration distances of proteins of known molecular mass (in kDa) are indicated.

in buffer A. CF,CO-proteins were visualized by enhanced chemiluminescence detection.

Other methods Protein was estimated by the Bio-Rad assay procedure

using bovine serum albumin as a standard. The extent of modification of NH2 residues in the CF3CO-RSA preparation was estimated according to Habeeb [27]; compared to unreacted rabbit serum albumin, about 90% of residues were modified in CF,CO-RSA. For analysis of amounts of antigenic protein, the films obtained after exposure of immunoblots to the enhanced chemiluminescence detection system were scanned and the volumes of signals quantitated using a computing densitometer (Molecular Dynamics 300A densitometer). The signals of the 52-kDa and the 64-kDa cross- reactive proteins present in liver homogenate KDL 34, present as a standard on each film, were used to normalize signals recorded from different films.

RESULTS Recognition of CF3CO-proteins and cross-reactive proteins of 52 kDa and 64 kDa in human liver

In earlier experiments [16], a monospecific anti-CF3C0 antibody was obtained by passing an anti-(CF,CO-RSA) serum, obtained from rabbits immunized with the model immunogen CF,CO-RSA [7, 161, through an affinity matrix which consisted of the hapten dcrivative CF,CO-Lys coupled to Affi-Gel102 amino-terminal agarose [ lh] . The anti-CF,CO antibody recognized a variety of CF,CO-proteins present in liver homogenates of rats exposed to a single dose of halothane 18 h before sacrifice (Fig. IA, lane 3). In addition, the anti-

CF,CO antibody recognized two constitutive, cross-reactive proteins of 52 kDa and 64 kDa, present in livers of rats that had not been treated with halothane (Fig. 1 A, lane 1). Recog- nition of both the CF,CO-proteins and the cross-reactive proteins of 52 kDa and 64 kDa by anti-CF,CO antibody was completely abolished in the presence of 1 mM CF,CO-Lys (Fig. 1 A, lanes 2 and 4, respectively). This finding is in keeping with values of the apparent half-maximal inhibitory constants of about 10 pM and 200 pM, respectively, for the recognition of the cross-reactive proteins of 52 kDa and 64 kDa and of CF,CO-proteins by anti-CF,CO antibody in the presence of

Immunoblotting studies of liver biopsy samples estab- lished that the monospecific anti-CF,CO antibody also recog- nizes CF,CO-proteins of human origin. Human CF3CO-proteins were detected in a liver biopsy sample obtained from a human individual who had died of cardiac complications 10 h after general anaesthesia under halothane (Table 1, patient 8). This sample contained a large number of proteins which were recognized by the anti-CF,CO antibody (Fig. 1 B, lane 3) and which had molecular masses very similar to those of the CF,CO-proteins expressed in livers of halothane-treated rats (Fig. 3 A, lane 3). Constitutive, cross-reactive proteins of 52 kDa and 64 kDa were detected when liver tissue obtained from a human kidney donor (patient KDL 34), who had not been exposed to halothane, was probed for reactivity with anti-CF3C0 antibody (Fig. 1 B, lane 1). CF3CO-Lys at 1 mM abolished recognition of both the CF,CO-proteins and the cross-reactive proteins of 52 kDa and 64 kDa by the anti- CFjCO antibody (Fig. 1 B, lanes 2 and 4, respectively).

The CF,CO-proteins and the cross-reactive proteins of 52 kDa and 64 kDa were not recognized by preimmune rabbit serum, nor was there any peroxidase reaction when immunoblots of human liver proteins were developed with either horseradish-peroxidase-coupled goat anti-rabbit second antibody or peroxidase substrate alone (data not shown). These findings exclude the possibility that the recognition of human liver CF,CO-proteins and of the cross-reactive proteins of 52 kDa and 64 kDa was a consequence of trivial, antigen- non-specific interactions. Additional experiments revealed that CF,CO-Lys did not cause a general, antigen-non-specific inhibition of human antigen ~ antibody interactions. Jmmunoblots of homogenates of human liver biopsies were developed with a combination of anti-CF,CO antibody and a polyclonal antibody directed against human microsomal epoxide hydrolase. Recognition of CF,CO-proteins by anti- CF3C0 antibody ceased at about 100 pM CF3CO-Lys; in contrast, no inhibition of recognition of microsomal epoxide hydrolase by anti-(epoxide hydrolase) antibody was observed even at 50 mM CF,CO-Lys (not shown).

CFXCO-Lys [16].

Antibody exchange immunochemistry

Disks of nitrocellulose with antigen/antibody complexes consisting of anti-CF3C0 antibody bound to either the 52-kDa protein (Fig. 2A, lane 1) or the 64-kDa protein (Fig. 2B, lane 1) of human liver as the sole source of anti- CF3C0 antibody were incubated with target strips which contained human liver CF,CO-proteins derived from patient 8 (Table 1). Most of these CF,CO-proteins were recognized by anti-CF3C0 antibodies that spontaneously exchanged from the 52-kDa or the 64-kDa proteins (Fig. 2A, lane 2, and Fig. 2B, lane 2, respectively). The presence of 1 mM CF3CO- Lys during the antibody exchange step abolished recognition of the target CF3CO-proteins by the exchanged antibodies

572

1 2 3 4

-6 6 J

-45

-3 6

-2 9

1 2 3 4

Fig. 2. Antibody exchange immunochemistry in human liver tissue. Hu- man kidney donor liver tissue homogenate (KDL 34) was subjected to SDS/PAGE and immunoblotting. Ten source disks with anti-CF3C0 antibody affinity-adsorbed to either the 52-kDa (A, lane 1) or the 64-kDa cross-reactivc protein (B, lane I) were co-incubated with one target strip. Lanes 2, the target was the liver homogenate obtained from patient 8 (Table I); lanes 3, same as lanes 2, but 1 mM CF3CO- Lys was included in the antibody cxchange step. In lanes 4, CF,CO- protcins present in the liver biopsy of patient 8 (Table 1) were detected with anti-CF,CO antibody not aKinity-adsorbed to eithcr cross-reactive protein.

(Fig. 2A, lane 3, and Fig. 2B, lane 3). No reactivity towards target CF,CO-proteins was observed in control experiments in which naked nitrocellulose disks which had been incubated with anti-CF3C0 antibody were used as source disks (not shown).

Quantitation of the 52-kDa and 64-kDa proteins in human liver From a human liver bank [23], 19 livers were selected at

random an screened on immunoblots for reactivity with anti- CF3C0 antibody. None of these liver samples were from patients that had been exposed to halothane. To compare the amounts of the cross-reactive proteins of 52 kDa and 64 kDa present on the immunoblots of each individual quantitatively, the results obtained were analyzed by densitometry. Quite appreciable inter-individual variability on the amounts of immunorecognizdble 52-kDa and the 64-kDa proteins was observed, although at least one of the proteins was detected in each of the 19 livers (Fig. 3 A, lanes 1 - 10, and Fig. 5, group A). Recognition of the two cross-reactive proteins by anti- CF3C0 antibody was inhibited by 1 mM CF,CO-Lys in all cases (Fig. 3A, lancs 1 - 10). The amounts of the 52-kDa cross-reactive protein were not correlated to the amounts of the 64-kDa cross-reactive protein (Fig. 3 A, lanes 1 - 10, and Fig. 5, group A). Analysis of liver biopsy samples obtained from three patients who died of post-operative cardiac complications shortly after anaesthesia with drugs other than halothane (Table 1, patients 9 - 11) revealed similar amounts of the 52-kDa and the 64-kDa cross-reactive proteins (Fig. 4A, lanes 9-11, and Fig. 5, group B). Detection of the cross-reactive proteins in the liver biopsy sample from the patient who died of post-operative complications 10 h after

Table 1. Clinical data of patients. The patient groups were as follows: (A) patients with halothane hepatitis; (€3) patient with cardiac complications following halothane anaesthesia; (C) patients with cardiac complications following anaesthesia with drugs other than halothane. The anaesthctic agents used were halothane (€I), etomidate/N2O2 (Et/N202) and isofluorane. Abbreviations: F, female; M, male; wk, weeks; yr, ycars; mo, months; d, days; -, data not available, PT, prothrombin time; AST, aspartate aminotransfcrase.

Patient Age/ Typeof No./intervals An- PT (time AST Grade of Outcome Time Antibodieb to

Group no. general agent longed) pathy from detectable in sex surgery of previous aesthetic pro- encephalo interval protein adducts

anacsthesia surgcry serum to death

A 1 2 3

4

5

6

7 B 8 C 9

10

1 1

57/F 57/F 58/F

3/M

62/F

34/F

53/F

47/M

6/M 91/M

6S/F

hysterectomy D + C excision o i metatarsal head h ypospadias

cataract

laparoscop y

carniotom y CABG MVR evacuate subdural haematoma repair of VSD

S

213 wk, 26 yr H 86 1/6 wk H 154 3/6mo, 7 yr, H 24 18 yr

2/l mo, 2 mo H 73

3/14 mo, 4 yr, H 53 12 yr S / 3 mo, 19 mo, H 1 Yr, 4 Yr, 5 Yr 1/17 d H

1

-

H -

- Et/N202 ~ isoflurane

IU/L

900 870

> 2400

960

1860

442

-

4 died 4 died 4 died

3 died

4 died

0 survived a

4 died

died

died died

died

6 d 23 d 33 d

18 d

28 d

17 d

1 0 h

24h 48 h

24 h

no yes (76 kDa) yes(100 + 76 kDa)

yes (100 + 76 kDa) yes (57 kDa)

yes (100 kDa)

no

a Biopsied at day 54.

573

+ + + + + + + i- + + A

64 kDac 52 kDa-

1 2 3 4 5 6 7 8 9 10

B

1 2 3 4 5 6 7 0 9 10

Fig. 3. Expression of the 52-kDa and 64-kDa cross-reactive proteins in human liver tissue. Homogenates of human liver were subjected to SDS/ PAGE followed by immunoblot analysis. (A) Rccognition of the cross-reactive proteins of 52 kDa and 64 kDa by anti-CF,CO antibody was analyzed in the absence or the prescnce (+) of 1 mM CF,CO-Lys. Samples 1 - 10 were randomly selected from the human liver bank [23] and correspond to KDL 25, 27, 28,29, 30, 31, 32, 34, 36, and 37. (R) In lanes 1 - 10, human liver homogenates of randomly selected samples corresponding to KDL 1,4, 5, 7, 9. 10, 11, 12, 13, and 14 were examined for the levels of microsomal cpoxidc hydrolase (SO kDa) using anti- (human epoxide hydrolase) antibody.

+ + + + + + + + + + + A

64 kDa- 52 kDa-

1 2 3 4 5 6 7 8 9 10 11

B

50 kDa-

1 2 3 4 5 6 7 8 9 10 11

Fig. 4. Expression of the 52-kDa and 64-kDa cross-reactive proteins in liver biopsies of halothane hepatitis patients. Homogenates of human liver biopsies were subjected to SDSjPAGE followed by immunoblot analysis. (A) Recognition of the cross-reactive proteins of 52 kDa and 64 kDa by anti-CF,CO antibody was assaycd in the absence or the presence ( + ) of 1 mM CF,CO-Lys. Lanes I - 1 1 correspond to the liver biopsies obtained from patients 1 , 2. 3, 4. 5 , 6, 7, 8, 9, 10, and 11 (Table 1). (B) In lanes 3 -11, the liver biopsies of patients 1-13 (Table 1) were examined for the levels of microsomal epoxide hydrolase (SO kna) using anti-(human cpoxide hydrolase) antibody.

anaesthesia with halothane (Table 1, patient 8) was not possible due to the presence of CF,CO-proteins in the sample, as discussed above (Fig. 1 D, lane 3, and Fig. 4A, lane 8).

Quantitation of the 52-kDa and 64-kDa proteins in livers of patients with halothane hepatitis

Liver biopsies from seven patients with halothane hepatitis (Table 1) were available for testing. None of the biopsy samples exhibited patterns of immunoreactivity with anti- CF3C0 antibody which were comparable to that seen with the liver sample from the patient who had died of post-operative complications 10 h after anaesthesia with halothane (Table 1, patient 8), or with liver homogenates obtained liom rats 18 h after exposure to halothane (compare Fig. 4A, lanes 1-7, with Fig. 4A, lane 8, and Fig. 1 C, lane 3, respectively). All of the liver samples from patients with halothane hepatitis were obtained many d a y after halothane exposure (Table I), by which time presumably the CF3CO-protcins had disappeared due to protein turnover as shown in rat liver, where persistence of CF3CO-proteins was < 10 days [9,28,29]. Immunoreactive

proteins of 52 kDa and/or 64 kDa were detected in five of the seven samples, and recognition of these proteins by the anti- CF3C0 antibody was abolished in the presence of 1 mM CF3CO-Lys (Fig. 4A, lanes 1 -7). In two of the liver samples (Fig. 4A, lanes 3 and 5 ) the cross-reactive proteins of 52 kDa and 64 kDa were detected in amounts typically found in the control livers (Fig. 5 ; compare patients 3 and 5 of group C with group A). However, in three other samples, the two cross-reactive proteins were present in very low amounts only (Fig. 4A, lanes 4, 6, and 7), while in a further two samples no immuno-reactive proteins were detected (Fig. 4A, lanes 1 and 2). Overall, the constitutive 52-kDa and 64-kDa proteins cross-reactive with anti-CF,CO antibody were immunochemically detectable at abnormally low levels in liver biopsy samples from five of the seven patients with halothane hepatitis (71 % ; Fig. 5 , group C).

The amounts of cross-reactive proteins present in the livers of patients afflicted with halothane hepatitis were not correlated with the severity of liver damage sustained by these patients. Thus, patient 6, who did not develop encephalopathy and who survived (Table I) , exhibited very low amounts of

5 74

A - m u .I ‘ 6 @ @

6 U u

2 4 0 0 n I

m 3 m

m m a 0 0

a L1 a m

C -

Fig. 5. Quantitation of the amounts of the 52-kDa and 64-kDa cross- reactive proteins in human liver tissue samples. Human liver tissue homogenates of (A) 19 randomly selected kidney donor individuals (KDL 1, 4, 5 , 7, 9, 10, 12, 13, 14, 25, 27, 28, 29, 30, 31, 32, 34, 36, and 37); (B) Liver biopsy hornogenates of three control individuals (patients 9, 10, and 11; Table 1) and (C) seven individuals with halothane hepatitis (patients 1, 2, 3, 4, 5, 6 , and 7; Table 1) werc quantitatively analyzed on immunoblots for the amounts of immunodetectable 52-kna (0) and 64-kDa (B) cross-reactive proteins by scanning densitometry as described in Experimental Pro- cedures.

the cross-reactive proteins (Fig. 4A, lane 6, and Fig. 5, patient 6 of group C). In contrast, both patient 3 and patient 5 exhibited normal levels ot‘ the cross-reactive proteins (Fig. 4A, lanes 3 and 5, and Fig. 5, patients 3 and 5 of group C), yet both patients died from complications of fulminant hepatic failure and both patients exhibited very large elevations in levels of plasma AST, indicating massive hepatic necrosis (Table 1). In view of this, the presence of abnormally low amounts of the Cross-reactive proteins of 52 kDa and 64 kDa is unlikely to have been due to occurrence of hepatitis per se. Furthermore, studies performed using an antibody specific for human microsomal epoxide hydrolase excluded the possibility of a generalized. antigen-non-specific defect in expression of hepatic proteins in the livers of the patients. All of the samples from patients with halothane hepatitis contained significant levels of immunoreactive epoxide hydrolase (Fig. 4B, lanes 1 -7), which were similar to the levels expressed in liver samples from control individuals (Fig. 3 B, lanes 1 - 10, and Fig. 4B, lanes 8 - 11).

DISCUSSION

In the present study, we have shown that an affnity- purified monospecific anti-CF3C0 antibody recognized constitutive proteins of 52 kDa and 64 kDa that were present on immunoblots of liver homogenates from human individuals who had not been exposed to halothane or its metabolites. Inhibition studies revealed that recognition of these 52-kDa and 64-kDa constitutive proteins by the anti-CF,CO antibody was inhibited competitively by the hapten derivative CF3CO- Lys. Furthermore, antibody exchange experiments demonstrated that anti-CF,CO antibody which had been affinity- adsorbed onto the 52-kDa or thc 64-kDa cross-reactive proteins recognized CF3CO-proteins present in the liver from a

human individual anaesthesized with halothane. In view of these immunochemical data, we conclude that epitopes on the 52-kDa and the 64-kDa cross-reactive proteins present in human liver exhibit a strong immunochelmical similarity to epitopes on human liver CF,CO-proteins.

The nature of the epitopes on the cross-reactive proteins of 52 kDa and 64 kDa to which the anti-CF3C0 antibody binds is not known in detail at present. Treatment of CF3CO- proteins with 1 M piperidine has been shown to readily cleave the trifluoroacetyl groups from proteins and thereby abolish binding of anti-CF,CO antibodies [3 , 61. However, when liver homogenates from rats not treated with halothane were incubated with 1 mM piperidine, recognition of the 52-kDa and the 64-kDa cross-reactive proteins was much less affected [16]. Thus, in the case of these two constitutive proteins, the epitopes recognized by the anti-CF3C0 antibody seem not to comprise covalently bound trifluoroacetyl moieties [16] ; instead, moieties with a different structure appear to be involved. Very recently, the 64-kDa cross-reactive protein has been immunoaffinity-purified in this laboratory from the heart of rats not treated with halothane and identified as the E2 subunit of the pyruvate dehydrogenase multienzyme complex (Christen, U., Jeno, P., and Gut, J., unpublished results). Lipoic acid, the prosthetic group of the E2 subunit protein, exhibited properties largely identical to those of the hapten derivative CF3CO-Lys in competitive immunoblotting experiments performed with both the E2 subunit and the 52- kDa cross-reactive protein, and CF3CO-proteins. This suggests that lipoic acid is the principle involved in molecular mimicry of CF3CO-proteins by the 64-kDa protein.

A marked variability in the levels of expression of the constitutive proteins of 52 kDa and 64 k:Da in liver tissue obtained from a panel of 22 human kidney donors was observed; levels of expression of the 52-kDa protein did not correlate with levels of expression of th’e 64-kDa protein. Analysis of liver tissue from 7 patients with halothane hepatitis further demonstrated that these constitutive proteins were expressed at abnormally low levels in 5 individuals (71 %). In contrast, all of the liver samples from patients with halothane hepatitis exhibited normal levels of expression of the protein epoxide hydrolase (Fig. 4B, lanes 1 - 7).

The low levels of the cross-reactive proteins of 52 kDa and 64 kDa detected in five of the patients’ liver samples were highly unlikely to have been a consequence of ‘down-regulation’ of the proteins caused by circulating antibodies against CF,CO-proteins, or by halothane itself. Although patients 3 and 5 exhibited serum antibodies against CF,CO-proteins of 100 kDa and 76 kDa (Table l), the amounts of the 52-kDa and 64-kDa cross-reactive proteins present in the liver samples of these patients were well within the range observed in normal human individuals (Fig. 5, group C compa:red to group A). In addition, the 52-kDa and 64-kDa cross-reactive proteins were expressed at abnormally low levels in patients 1 and 7, neither of whom exhibited detectable serum antibodies against CF,CO-proteins (Fig. 5 and Table 1). Moreover, the subcellu- lar localization of at least the 64-kDa protein (the E2 subunit of the pyruvate dehydrogenase complex, which is situated on the inner mitochondria1 membrane) should preclude the formation of complexes of the 64-kDa protein with serum antibodies. It is extremely improbable that halothane caused a direct down-regulation in levels of expression of 52-kDa and 64-kDa the constitutive proteins, because of thc long time intervals between exposure of patients to halothane and collec- tion of liver samples (at least 17 days for 6 of the patients, and 6 days for the remaining patient). Studies of rats exposed to

575

halothane have shown that expression of the 52-kDa and 64-kDa proteins in liver, heart, and kidney 10 days after halothane exposure were identical to the levels of expression in tissues of unexposed animals [16, 301. Unequivocal detection of these constitutive proteins in livers of halothane-exposed rats at earlier time points was not possible because of the presence of CF,CO-proteins [16]. However, the levels of immunodetectable 52-kDa and 64-kDa constitutive proteins expressed in rat heart remained constant at 6 h, 12 h, 18 h, 90 h, and 10 days after halothane exposure [30], while in the kidney, a down-regulation of the two proteins might have occurred for a period of about 18 h (Huwyler, J., and Gut, J., unpublished j.

There is ample evidence suggesting that halothane-induced hepatitis has an immunological basis, implicating an immune response to trifluoroacetylated proteins in the mechanism of liver damage [ I , 2, 31, 321. We and other investigators have observed that all rats treated with halothane produce CF3CO- proteins, but that these animals do not mount an appreciable anti-(CF,CO-protein) antibody response and do not sustain immunochemical liver damage [28,33]. Indeed, although tran- sient production of antibodies reactive with the CF3C0 group have been observed in response to halothane treatment in rabbits [34] and guinea-pigs [35], so far a true animal model of halothane hepatitis is lacking [15]. Similarly, it appears that all humans exposed to halothane produce CF,CO-proteins, whereas only a small subset of susceptible individuals mount an immune response and develop halothane hepatitis [33].

Why then does an immune response to the trifluoroacetyl- modified proteins not occur in all halothane-exposed human individuals? One possibility for the lack of immune responsiveness of individuals cxposcd to halothane could be immunological tolerance towards CF,CO-proteins. immunological sclf/non-self-discrimination is thought to occur in the thymus and in the periphery and to involve elimination (clonal deletion) or functional silencing (anergyj of clones of maturing T-cells whose surface receptors can bind with high affinities to self-peptides presented in association with molecules of the major histocompatibility complex (MHC) [17 -201. In addition, ectotopic presentation of antigen by ‘non-pro- fessional antigen presenting cells’ may lead to an anergic state, in which unprimed T-cells become refractory to stimulation by antigen-presenting cells [21, 221. Perhaps one or more of these mechanisms render normal individuals, who express epitopes that immunochemically mimic CF,CO-proteins, tol- erant to epitopes on CF,CO-proteins. Conversely, abnormally low levels of expression of such epitopes might render individuals susceptible to an immune response towards CF,CO-proteins. Although far from proof, the data presented hcre support this line of reasoning for two reasons. First, molecular mimicry of CF,CO-proteins by constitutive proteins exists in the normal population. Second, a fraction of patients afflicted with halothane hepatitis appears to express abnormally low levels of these constitutive proteins.

If the hypothesis is correct, one must explain why the levels of the 52-kDa and 64-kDa cross-reactive proteins detected in liver biopsies from two of the seven patients with halothane hepatitis were within the normal range. Misdiagnosis of these patients is highly improbable, especially since antibodies to CF3CO-proteins were present in sera from both patients (Table 1). One explanation might be that the presumed immunological tolerance does not cover all immunogenic epitopes expressed on CF3CO-proteins. Studies of the nature of the epitopes recognized by patients antibodies have indicated that they are complex and comprise a combination of

thc trifluoroacetyl group plus essential structural features of the protein carriers [3]. Whereas the anti-CF,CO antibody might recognize the trifluoroacetyl group bound to any protein, irrespective of the nature of the protein carrier, antibodies from patients with halothane hepatitis are specific for particular liver CF,CO-proteins [3]. In the future it will be important to determine whether the concept of tolerance via epitope mimicry applies to these complex ‘CF,CO-carrier protein’ epitopes also.

The experiments reported here have raised the possibility of an association between susceptibility to halothane hepatitis and the extent of molecular mimicry of CF,CO-proteins by epitopes present on constitutive proteins. However, formal proof that this hypothesis is correct will require further investi- gation, as does the role of the 52-kDa and 64-kDa proteins in idiosyncratic toxicities of compounds that are structural analogues of halothane. These include the anaesthetics enflurane, isoflurane, and compounds such as the refrigerant 2,2-dichloro-l ,l,i-trifluoroethane, all of which produce protein adducts that are immunochemically very similar to CF3CO-proteins [36 - 391. Compounds of the latter type will replace chlorofluorocarbons as refrigerants, solvents and foam-blowing agents due to their low ozone-depleting poten- tial [40] and man will increasingly be exposed to them in the near future, both occupationally and environmentally.

We thank Professor U. A. Meycr (Dept. of Pharmacology, Bio- center, Basel) for providing human liver samples and a polyclonal antibody against human microsomal epoxide hydrolase. This work was supported by the Swiss National Science Foundation (grant 3- 109.0.88) and the help of the Roche Research Foundation. Josef Gut i s the recipient of a START Research Career Development Award (3- 018.0.87) of the Swiss National Science Foundation. J. Gerald Kenna was supported by an Advanced Training Fellowship in Toxicology and a Project Grant, both from Wcllcome Trust, UK.

REFERENCES

134.

Rev. Pharmucol. Toxicol. 28, 367 - 387.

Pharrnacol. Exp. Ther. 24.7, 1103 - 1209.

Exp. Ther. 242,133 - 140.

L. R. (1985) Mol. Pharmacol. 28,468-414.

1. National Halothane Study (1966) J. Am. Med. Assoc. 197, 121 -

2. Pohl, L. K., Satoh, H.; Christ, D. D. &Kenna, J. G. (1988) Annu.

3. Kenna, J. G., Satoh, H., Christ, D. D. & Pohl, L. R. (1Y88) .I.

4. Kenna, J. G.. Neuberger, J. & Williams, R. (1987) J . Pharmacol.

5. Satoh, H., Gillette, J . R., Davies, H. W., Schulick, R. D. & Pohl,

6. Kenna, J. G., Ncubcrger, J . & Williams, R . (1988) Hepatology “3,

7. Satoh, H., Fukuda, Y., Anderson, D. K., Ferrans, V. J., Gillettc, J. R. & Pohl, L. R. (1985) J . Pharmacol. Exp. Ther. 233,857- 862.

8. Hubbard, A. K., Roth. T. P., Schuman, S. & Gandolfi, A. J. (1989) Clin. Exp. Immunol. 76, 422-427.

9. Pohl, R. L., Kenna, J. G.. Satoh, H., Christ, D. I). & Martin, J. L. (1989) Drug Metah. Rev. 20,203-211.

10. Kenna, J . G., Martin, J. L. &Pohl, L. R. (1990) Eur. J. Pharmacol. 183, 1139-1140.

11. Satoh, H., Martin, B. M., Schulik, A. H.; Christ, D. D., Kenna, J. G. & Pohl, L. R. (1Y89) Proc. Natl Acad. Sci. USA 86, 322- 326.

12. Butler, L. E., Thorn en, D., Martin, J. L., Martin, B. M., Kenna, J. G. & Pohl, L. R . (1992) Chem. Res. Toxicol. 5 , 406-410.

13. Thomassen, D., Martin, B. M.. Martin, J. L., Puniford. N. R. & Pohl, L. R. (1990) Eur. J . Pharmacol. I83, 11 38 - 11 39.

14. Martin, J. L., Pumford, N. R., LaRosa, A. C., Martin, B. M., Gonzaga, TI. M. S., Beaven, M. A. & Pohl, L. R. (1991) Bio- chem. Bioplzys. Res. Commun. 178, 679 - 685.

1635-1641.

576

15. Neuberger, J. M., Kenna, J. G. & Williams, R. (1987) Ini. J .

16. Christen, U., Biirgin, M. & Gut, J. (1991) Mol. PharmacoZ. 40,

17. Kappler, J. W., Roehm, N. & Marrack, P. (1978) Cell 4Y, 273-

18. Kisielow, P., Bliithmann, H.. Stacrz, U. D., Steinmetz, M. & von

19. Kappler, J. W., Staerz, U., White, J . & Marrack, P. C. (1988)

20. Ramsdell, F. & Fowlkes, B. J. (1990) Science 248, 1342-1348. 21. Schwartz, R. H. (1990) Science248, 1349-1356. 22. Sprcnt, J., Gao, E.-K. & Webb, S. R. (1990) Science 248, 1357-

23. Meicr, P. J., Miiller, H. K., Dick, B. & Meyer, U. A. (1983)

24. Laemmli, U. K. (1970) Nature 227,680-685. 25. Towbin, H., Staehelin, T. & Gordon, J. (1979) Proc. Natl Acad.

26. Hammarhack, J. A. & Vallee, R . B. (1990) J . Biol. Chem. 265,

27. Habeeb, A. F. S. A. (1966) Anal. Biochern. 14, 328-336. 28. Christen, U., Biirgin, M. &Gut, J. (1991) Biochem. Biophys. Ras.

Zmmunopharrnac. 9, 123- 131.

390 -400.

280.

Boehmer, H. (1988) Nature 333, 742- 746.

Nature 332, 3 5 - 40.

1363.

Gastroenterology 85,682 - 692.

Sci. USA 76,4350-4354.

12 763 - 12 766.

Commun. 175,256 - 262.

29. Kenna, J. G., Martin, J. L., Satoh, H. & Pohl, L. R. (1990) Drug.

30. Huwyler, J. & Gut, J. (1992) Biochem. Biophys. Res. Commun.

31. Mieli-Vergani, G., Vergani, D., Trcdger, J . M., Eddleston, A. L. W. F., Davis, M. & Williams, R. ( 3 980) J. Clin. Lab. Immunol.

32. Neuberger, J., Mieli-Vergani, G., Tredger, J. M., Davis, N. &

33. Kenna, J. G. (1991) Biochem. Soc. Trans. 19, 191 -195. 34. Callis, A. H., Brooks, S. D., Roth, T. P., Gandolfi, A. J. & Brown,

B. R. (1987) Clin. Exp. Imrnunol. 67, 343-351. 35. Siadal-Pajouh, M., Hubbard, A. K., Rolh, T. P. & Gandolfi, A.

J. (1987) Anesth. Analg. 66, 1209-1214. 36. Christ, D. D., Kenna, J. G., Kammerer, W., Satoh, H. & Pohl.

L. R. (1988) Anesthesiology 69, 833 - 838. 37. Zimmermann, H. (1991) Heputology 13, 1252- 1253. 38. Harris, J. W., Pohl, L. R., Martin, J. L. & Anders, M. W. (1991)

Proc. Nut2 Acud. Sci. USA 88, 1407- 1410. 39. Huwyler, J., Aeschlimann, D., Christen, U. &Gut, J. (1992) Eur.

J. Biochem. 207, 229 - 238. 40. Manzer, L. E. (1990) Science 24Y, 31 -35.

Metah. Disp. 18, 788 -793.

184, 1344-1349.

4 , 49-51.

Williams, R. (1981) Gut 22, 669-672.

Documents

Molecular mimicry of trifluoroacetylated human liver protein adducts by constitutive proteins and immunochemical evidence for its impairment in halothane hepatitis