Overlapping but distinct specificities of anti-liver-kidney microsome antibodies in autoimmune hepatitis type II and hepatitis C revealed by recombinant native CYP2D6 and novel peptide epitopes

Abstract

Anti-liver-kidney microsome antibodies (anti-LKM) occur in autoimmune hepatitis (AIH) type II and in a subset of patients with hepatitis C. Anti-LKM1 in AIH are directed against cytochrome P4502D6 (CYP2D6), but conflicting data exist concerning the specificity of anti-LKM in hepatitis C. The aim of this study was to evaluate binding specificities of anti-LKM antibodies in both diseases using novel test antigens as well as their inhibitory capacity on CYP2D6 enzyme activity. Sera from 22 patients with AIH type II and 17 patients with hepatitis C being anti-LKM-positive in the immunofluorescence test were investigated for binding to native recombinant CYP2D6 and liver microsomes by ELISA and immunoblotting, and to synthetic peptides covering the region 254–339 (254–273, 257–269, 270–294, 291–310, 307–324, 321–339, 373–389) as well as the novel peptide 196–218 by ELISA. Furthermore, all sera were tested for inhibition of CYP2D6-dependent bufuralol 1′-hydroxylase activity. Twenty of the 22 AIH type II sera (91%) and nine of the 17 hepatitis C sera (53%) were positive for CYP2D6 by ELISA and/or immunoblotting. The previously described major peptide epitope comprising CYP2D6 amino acids 257–269 was recognized by 16 of the 22 AIH sera but by only one hepatitis C serum. A further epitope, 196–218, could be defined for the first time as another immunodominant epitope for AIH because it was recognized by 15 of the 22 AIH (68%) but only three of the 17 hepatitis C sera (18%). With the exception of the peptide 254–273, the other peptides showed no significant reactivity. Analysing the inhibitory properties of anti-LKM antibodies it emerged that 95% of AIH sera and 88% of hepatitis C sera inhibited enzyme function. These data indicate that anti-LKM antibodies in AIH and hepatitis C react with CYP2D6, as shown by their inhibitory activity, and that besides the known epitope 257–269 a further immunodominant epitope exists on CYP2D6 which is recognized by sera from patients with AIH II but hardly by sera from patients with hepatitis C.

INTRODUCTION

Antibodies to liver-kidney microsomes (anti-LKM) were first detected by immunofluorescence test in a subgroup of patients with autoimmune hepatitis which primarily affects children (AIH type II) [1,2]. They were later labelled ‘anti-LKM1’ when it became evident that anti-LKM antibodies can also occur in patients with drug-induced hepatitis as well as in patients with hepatitis B/D co-infection (anti-LKM2 and anti-LKM3, respectively) [3–5]. Further studies revealed that sera from patients with anti-LKM⁺ AIH recognized a 50-kD protein in liver microsomes [6]. The 50-kD protein was identified as the microsomal cytochrome P450 protein, CYP2D6 [7–9]. Its enzymatic activity is potently inhibited by anti-LKM1 in vitro [7] but not in vivo [10]. The internal amino acid sequence 257–269 of CYP2D6 has been reported to represent the core of an immunodominant epitope recognized by most anti-LKM1, whereas several other sequences appear to be recognized at lower frequencies [11–13].

Anti-LKM antibodies are, however, also produced in the course of hepatitis C [14,15], but data whether they are also reacting with CYP2D6 have been conflicting [16–22]. From preliminary studies we had evidence that anti-LKM antibodies in sera from patients with AIH type II and hepatitis C may recognize different epitopes [19]. Furthermore, it could be shown that the major peptide epitope around amino acids 257–269 of CYP2D6 is predominantly recognized by sera from patients with AIH but not with hepatitis C [17,21].

The aim of the present study was therefore to investigate systematically antigen and epitope specificities of anti-LKM antibodies in both groups of patients. By using full-length catalytically active recombinant CYP2D6, anti-LKM-positive sera could be analysed for the first time with a test antigen containing all relevant linear and conformation-dependent epitopes.

PATIENTS AND METHODS

Patients

Thirty-nine patients with chronic liver disease were selected on the basis of a typical LKM pattern detected by indirect immunofluorescence test. Seventeen were anti-hepatitis C virus (HCV) and HCV-RNA-positive. The remaining 22 patients were negative for markers of hepatitis A, B, C and D, but had clinical and biochemical evidence for chronic hepatitis and histologically either chronic active hepatitis or liver cirrhosis. They were defined as AIH type II according to the recommendations of the international autoimmune hepatitis group [23]. Sera and clinical data were provided by colleagues from different hospitals in Germany. Serological tests had been performed before introduction of immunosuppressive therapy, and none of the patients had received interferon-alpha (IFN-α) therapy at that time. Demographic features of the patients are given in Table 1.

Table 1.

Age, sex and biochemical parameters in 22 patients with autoimmune hepatitis (AIH) type II and in 17 hepatitis C patients being anti-liver-kidney microsome antibody (LKM)-positive in the immunofluorescence test

One of the 22 patients with AIH type II (no. 16) and one of the 17 patients with hepatitis C (no. 22) were HGV-RNA-positive

Antigens

Microsomes were prepared from rat and human liver according to standard procedures. Recombinant human CYP2D6 was produced in Spodoptera frugiperda (Sf9) cells as recently described [24]. Insect cell membranes were isolated by ultracentrifugation and resuspended at a concentration of 2–5 mg/ml of 0.1 m sodium phosphate buffer pH 7.4. Cytochrome P450 content of the membranes was determined spectrophotometrically. Quantitative Western blotting with a MoAb (114/2) indicated that 80–90% of the recombinant CYP2D6 protein was expressed in the haeme-bound P450 form ([24] and Zanger et al., unpublished results). As control, membranes from uninfected Sf9 cells were used.

Peptides

Two synthetic peptides previously shown to react with anti-LKM1 [11] were synthesized, purified by high performance liquid chromatography (HPLC) and conjugated to bovine serum albumin (BSA) by Biotrend (Cologne, Germany). The amino acid sequences are: peptide 257–269, EHRMTWDPAQPPR; peptide 373–389, GMTHMTSRDIEVQGFRI.

Additional peptides previously reported by others to represent LKM1 epitopes were designed in a modified way to analyse systematically the region between amino acids 254 and 339. An overlap of four amino acids between consecutive peptides was introduced in order to avoid the potential loss of epitopes located in-between. The peptides tested in this region consisted of amino acids 254–273, 270–294, 291–310, 307–324, and 321–339. In a similar way, another modified peptide was designed for the region around amino acid 200 where Parez et al. recently reported that a peptide 200–214 was recognized by some hepatitis C sera [25]. These novel peptides were synthesized by MWG Biotech (Ebersberg, Germany), and also coupled to BSA. The amino sequences are: peptide 196–218, EYDDPRFLRLLDLAQEGLK EESG; peptide 254–273, LLTEHRMTWDPAQPPRDLTE; peptide 270–294, DLTEAFLAEMEKAKGNPESSFNDEN; peptide 291–310, NDENLRIVVADLFSAGMVTT; peptide 307–324, MVTTSTTLAWGLLLMILH; peptide 321–339, MILHPDVQRRVQQEIDDVI.

Immunofluorescence test

Anti-LKM-positive sera were identified by indirect immunofluorescence (IF) test using cryostat sections of rat liver, kidney, heart and stomach, and human thyroid as substrate. All 39 sera had titres above 1:80. There was no difference in anti-LKM titres between anti-HCV-positive and -negative patients (data not shown).

ELISA

ELISA was performed according to standard methodology [26] in a modification as described recently using flat-bottomed microtitre plates (Maxisorp, Nunc, Roskilde, Denmark) [27]. Human and rat liver microsomes were applied at a concentration of 10 μg/ml, insect cell membranes containing recombinant CYP2D6 and the membranes from uninfected Sf9 cells as well as the BSA-coupled peptides at a concentration of 1 μg/ml. These antigen concentrations had been determined by serial dilutions. Also optimal dilution of patients' sera had been determined by serial dilutions (from 1:100 to 1:10 000) and was found to be 1:1000 for all antigens. Bound antibodies were visualized using peroxidase-conjugated goat anti-human IgG and IgM antibodies (Dianova, Hamburg, Germany) at a dilution of 1:2000 and o-phenylenediamine as substrate. The reaction was stopped with 25% H₂SO₄ and absorption was measured at 495 nm. Results are shown as optical densities (OD) × 1000. All tests were performed in duplicates.

For each test, a standard curve was established using three marker sera with high, medium, and low antibody titres as well as a negative control. These sera were used as standard sera in all further tests. Under these conditions, normal values were determined by testing sera from 20 healthy controls against the different antigens. The mean of the OD (× 1000) of these sera + 3 s.d. was taken as upper limit of the normal range. Using this standardization, the normal range for rat liver microsomes and CYP2D6 was < 200, for the peptides 257–269 and 373–389 < 300, for the peptides 196–218 and 254–273 < 250, for the peptide 270–294 < 350, for the peptide 291–310 < 600, for the peptide 307–324 < 550, and for the peptide 321–339 < 400.

Western blotting

All sera were tested by Western blotting following standard procedures using human and rat liver microsomes (25 μg per lane) and recombinant CYP2D6 (1 μg per lane). After transfer, nitrocellulose sheets were incubated with patient sera at a dilution of 1:100, and bound antibodies were visualized with peroxidase conjugates of goat anti-human IgG (Dako, Hamburg, Germany) and 3-amino-9-ethylcarbazole as substrate.

Inhibition of enzyme activity

CYP2D6 enzyme activity was determined as described [24] with reconstituted Sf9 membranes containing 1.0 pmol of P450 and 1 pmol of purified rat NADPH:P450 oxidoreductase. Duplicate incubations with 100 μm (+) − bufuralol as substrate were performed with 1 mm NADPH in a final volume of 0.1 ml of 0.1 m sodium phosphate buffer pH 7.4. Following incubation at 37°C for 30 min, 1’-OH-bufuralol was quantified in HClO₄ using an HPLC assay [28]. The enzymatic activity (100% control activity) measured in this assay was 0.37 ± 0.02 nmol of 1’-hydroxybufuralol formed per assay, corresponding to a turnover number of 12.3/min. For immunoinhibition, sera were added at a final dilution of at least 1:50 and preincubated at room temperature for 15 min before the addition of the substrate. The resulting activity was calculated in percentage of control activity. Sera were considered positive if they inhibited the reaction by at least 80% at a maximal dilution of 1:50.

Hepatitis serology

Anti-HCV antibodies were detected by a second generation ELISA (Ortho Diagnostics, Raritan, NJ) and recombinant immunoblot assay (RIBA; Chiron Corp., Emeryville, CA)). Furthermore, all sera were tested for HCV-RNA by a nested reverse transcribed polymerase chain reaction (PCR) as recently described [29]. Hepatitis B serology was performed by the Abbott IMX-assay, anti-hepatitis delta virus antibodies were determined by a test kit obtained from Sorin (Turin, Italy). In all sera HGV-RNA was determined by PCR as recently described using four published primers (G8, G9, G11, GBV-C.h1) [30,31].

Immunoglobulins

Quantitative serum immunoglobulins were measured by nephelometry (Beckmann, Munich, Germay) (normal values: IgG 800–1800 mg/dl; IgA 93–445 mg/dl; IgM 65–280 mg/dl).

Statistical analysis

Statistical analysis was performed using Student's t-test; P < 0.05 was considered statistically significant. The Spearman rank correlation coefficient r_S was used to describe correlations.

RESULTS

Reactivity of anti-LKM-positive sera (as defined by IF test) with liver microsomes and recombinant native CYP2D6 in ELISA and Western blotting

Patients' sera were first tested by ELISA and Western blotting for reactivity against microsomes derived from rat (RLM) and human liver (HLM) as well as the recombinant native CYP2D6. The individual results are given in Table 2.

Table 2.

Age, sex and results obtained with sera from patients with anti-liver-kidney microsome antibody (LKM)-positive autoimmune hepatitis (AIH) and hepatitis C using different antigens in different methods

a. Patients with AIH type II (n = 22)

b. Patients with hepatities (n = 17)

*Normal values for RLM and CYP2D6: OD × 1000 < 200; for the peptides 257–269 and 196–218: OD × 1000 < 300. †Inhibition of CYP2D6 enzymatic activity in pmol CYP2D6 inhibited per μl serum.

‡ Combined from analyses of recombinant CYP2D6, HLM, and RLM with size in kD.

§Sera reacted also with the peptide 373–389.

*Normal values for RLM and CYP2D6: OD × 1000 < 200; for the peptides 257–269 and 196–218: OD × 1000 < 300.

†Inhibition of CYP2D6 enzymatic activity in pmol CYP2D6 inhibited per μl serum.

‡Combined from analyses of recombinant CYP2D6, HLM, and RLM with size in kD.

§Patient may suffer from an overlap syndrome between hepatitis C and AIH type 2.

Twenty-one (95%) of the 22 AIH type II sera reacted with RLM, and 20 of them (91% of all sera) also with recombinant CYP2D6 (Tables 2a, 3).

Table 3.

Incidence of antibodies of the IgG type to rat and human liver microsomes, CYP2D6 and overlapping peptides in patients with autoimmune hepatitis (AIH) type II and hepatitis C as measured by ELISA

Of the 17 anti-LKM⁺ hepatitis C sera, 15 (88%) were positive with RLM and nine (53%) with recombinant CYP2D6 by ELISA and five showed the typical 50-kD determinant representing CYP2D6 in the Western blot (Tables 2b, 3).

Only 13 (59%) of the AIH type II and one (6%) of the hepatitis C sera reacted in the ELISA with HLM, and ODs were generally lower than with RLM, presumably because of the low abundance of CYP2D6 in human liver [32].

Identification of the peptide 196–218 as a further immunodominant epitope for AIH type II

Several synthetic peptides representing CYP2D6 internal amino acid sequences were previously reported by other groups to be recognized by anti-LKM-positive sera [11–13,21,25]. The major epitope appeared to be located within amino acids 257–269. Indeed, 16 (73%) of all 22 AIH sera recognized this peptide (Table 3), and the values correlated to the ELISA titres for recombinant CYP2D6 (AIH: r_S = 0.76, P < 0.001). In contrast, only one of the 17 hepatitis C sera (no. 29) reacted strongly with this peptide (Tables 2b, 3).

Analysis of the peptides 254–273, 270–294, 291–310, 321–339 by ELISA revealed that, with the exception of peptide 254–273 which reflected the results obtained with the shorter peptide 257–269, none of the other peptides was recognized by more than two sera of either the AIH or the hepatitis C group (Table 3).

However, applying the newly designed peptide comprising amino acids 196–218, 15 (68%) of the 22 AIH sera but only three (18%) of the 17 hepatitis C sera were positive (Table 3). Of these three hepatitis C patients, only one (no. 14) had high anti-peptide 196–218 antibody titres, while the other two patients were just beyond the threshold level. In AIH but not in hepatitis C the values correlated well with CYP2D6 antibody titres (Fig. 1a,b). Antibody titres to the peptide 196–218 hardly correlated with the titres for peptide 257–269 (AIH: r_S = 0.54, P < 0.05; hepatitis C: r_S = – 0.4, P > 0.05; data not shown).

Fig. 1.

Comparison of antibody activity using the bovine serum albumin (BSA)-coupled peptide 196–218 and recombinant native CYP2D6 in the ELISA. (a) Patients with autoimmune hepatitis (AIH) type II (n = 22). (b) Patients with anti-liver-kidney microsome antibody (LKM)-positive hepatitis C (n = 17). Units on axes are relative optical densities (OD) × 1000. The Spearman rank correlation coefficients r_S are given and outstanding individual sera are indicated by their number.

Immunoinhibition of cytochrome P450 enzyme activity by anti-LKM⁺ sera

Inhibition of CYP2D6 enzyme activity by individual sera was also determined to assess in a different way the presence of antibodies against CYP2D6. The same recombinant membrane preparation was used as for the ELISA.

The same 20 sera from AIH patients positive for CYP2D6 by ELISA also strongly inhibited enzyme activity.

Fifteen of the 17 hepatitis C sera inhibited CYP2D6 activity. Interestingly, 10 of them did not recognize the 50-kD determinant in the Western blot, indicating that they may be directed against conformational epitopes (Table 2). Inhibition of enzyme activity correlated well with the antibody titres to CYP2D6 in the ELISA in AIH type II and hepatitis C (P < 0.01) (Fig. 2). No correlation was found when antibody titres to the peptide 257–269 and enzyme inhibition activity were compared (AIH II: r^S = 0.43, P > 0.05; hepatitis C: r_S = 0.14, P > 0.05). With respect to the peptide 196–218, there was a correlation with the inhibitory activity only in patients with AIH (r_S = 0.66, P < 0.01), but not in patients with hepatitis C (r_S = 0.1, P > 0.05) (Fig. 3).

Fig. 2.

Inhibition of CYP2D6 enzyme activity in relation to antibody activity towards this enzyme (measured by ELISA). (a) Patients with autoimmune hepatitis (AIH) type II (n = 22). (b) Patients with anti-liver-kidney microsome antibody (LKM)-positive hepatitis C (n = 17). Units on the abscissa are relative optical densities (OD) × 1000, on oridinate inhibition of CYP2D6 enzyme activity is given in pmol CYP2D6 inhibited per μl serum. The Spearman rank correlation coefficients r_S are given.

Fig. 3.

Inhibition of CYP2D6 enzyme activity in relation to antibody activity towards the peptide 196–218 (measured by ELISA). (a) Patients with autoimmune hepatitis (AIH) type II (n = 22). (b) Patients with anti-liver-kidney microsome antibody (LKM)-positive hepatitis C (n = 17). Units on abscissa are relative optical densities (OD) × 1000, on ordinate inhibition of CYP2D6 enzyme activity is given in pmol CYP2D6 inhibited per μl serum. The Spearman rank correlation coefficients r_S are given.

DISCUSSION

This study confirms and extends previous reports on the fine specificity of anti-LKM antibodies in patients with AIH type II and hepatitis C in a large series of patients. Thus, 20 of the 22 patients with AIH as well as nine of the 17 hepatitis C patients were positive with CYP2D6 by ELISA and 20 and 15, respectively, inhibited enzyme activity. Applying the peptide 257–269, previously defined as a major immunodominant epitope [13], in the ELISA, only one of the 17 sera from patients with hepatitis C was positive in contrast to 16 of the 22 sera from patients with AIH, thereby confirming its high diagnostic specificity. Why the sensitivity in detecting AIH-specific anti-LKM1 antibodies was even decreased (50%) using the seven amino acids larger peptide 254–273 remains unclear, but these data would fit the observations made by Jurado et al. [33], who also found a positive reaction in only 27% of 11 patients with AIH type II when applying the peptide 254–271.

However, considering the fact that 27% of the AIH patients were anti-peptide 257–269-negative, further epitopes seemed to be involved in anti-LKM1 reactivity in AIH. We could define a novel immunodominant epitope outside the region 254–339. Thus, we designed a peptide for the region around amino acid 200 comprising the amino acids 196–218. Interestingly, we found with this peptide a strong reaction with AIH type II sera (15 of the 22 sera = 68%), while only three of the 17 (18%) HCV sera were positive. These observations are in contrast to findings by Parez et al., who reported that some hepatitis C but no AIH sera reacted with a peptide 200–214 [25]. These discrepancies may be explained by the fact that the authors used in their study only one serum from a patient with AIH type II as a control. Furthermore, the longer peptide 196–218 used in our study presumably preserved the epitope recognized by most AIH sera.

In contrast to previous studies, in which for the detection of anti-LKM antibodies recombinant CYP2D6 had been applied as partial or full length fusion proteins expressed in Escherichia coli or as in vitro translation products [16–18,20–22,34,35], we used for the first time as test antigen an enzymatically and conformationally intact protein preparation of CYP2D6 [24]. Because cytochromes P450 require special conditions in order to be produced as intact haemoproteins in heterologous expression systems, it can be assumed that the preparations previously used to characterize anti-LKM antibodies were apoproteins devoid of haeme and therefore lacking some conformation-dependent epitopes. The CYP2D6 protein used in this study was obtained by recombinant baculovirus infection of Sf9 insect cells under conditions which favour haeme incorporation, and had been extensively characterized with respect to functional properties [24]. Spectroscopic analysis revealed that about 80–90% of the total expressed protein was in the correctly folded holoprotein form. This antigen also allowed analysis of the enzyme-inhibitory properties of anti-LKM antibodies [7,10].

In both disorders, AIH type II and hepatitis C, anti-LKM antibodies were found to strongly inhibit CYP2D6 activity. Since the peptide epitopes 256–269 and 196–218 are specifically recognized by AIH but not by hepatitis C sera, it is unlikely that antibodies against these peptides mediate enzyme inhibition. Indeed, when we preabsorbed patients' sera with these peptides, the inhibitory activity of the sera was not altered (data not shown). Preliminary studies isolating antibodies bound to peptide 196–218 indicated that these antibodies inhibited CYP2D6 enzyme activity only up to 20% (Zanger, unpublished observation). Therefore, one has to postulate that other antibodies are responsible for the inhibitory effect. Duclos-Vallée et al. [36] made similar observations. The suggested non-identity of inhibitory and peptide-binding antibodies would also explain why attempts to raise inhibitory antibodies against CYP2D6 by immunizing animals with synthetic peptides covering sequences around amino acids 257–269 were either not successful [36] or resulted in comparably very low inhibitory titres [37].

In a further attempt, we systematically analysed the region between amino acids 254 and 339, designing additional peptides which differed to some extent from those previously used by other authors. Thus, in order to avoid the potential loss of epitopes located in-between, an overlap of four amino acids between consecutive peptides was introduced. This region, however, seems not to contain any further immunodominant epitopes, because none of these peptides was recognized by more than three sera of either the AIH type II or the hepatitis C group. Of course, the existence of conformation-dependent epitopes in the region 254–339 cannot be excluded by these experiments. With the peptide 373–389, previously reported to be another immunodominant epitope [13], we observed a positive reaction only with two sera from patients with AIH type II. In a recent study by Choudhuri et al. [38] it was shown that sera from patients with AIH type II with associated endocrinopathies reacted with the epitope 321–351 on CYP2D6. In our series we only applied a shorter peptide (321–339), and this was recognized only by one serum from a patient with hepatitis C without evidence for other organ-specific autoimmune disorders. The region 341–427 reported to be recognized by some hepatitis C sera [39] has not been analysed in this study.

Manns et al. [10] found that there is a sequence similarity to HCV and herpes simplex virus (HSV)-1 within the 33 amino acid sequence of CYP2D6 identified in their study. We therefore also performed a similarity search for the segment 196–218 in Genbank but did not find significant similarities to other proteins than cytochrome P450s (data not shown). HGV as cause of this disease can be excluded with certainty because only two of the 39 anti-LKM⁺ patients were HGV-RNA-positive.

In the present study only one patient with hepatitis C (no. 29) had high antibody titres to the AIH type II-related peptide 257–269. One could argue that this patient, a 13-year-old female, suffers from an association of two liver disorders, namely AIH and hepatitis C. The patient responded well to immunosuppressive therapy with steroids and azathioprine, further suggesting an autoimmune condition. Patient no. 14, the only patient with hepatitis C and high antibody titres to the peptide 196–218 (and CYP2D6), was also a young female who responded well to immunosuppressive therapy. High level antibodies directed against the peptides 257–269 or 196–218 may therefore be a diagnostic tool for the identification of patients with overlap syndromes of AIH type II and hepatitis C. It is probably only in these patients that IFN therapy may be dangerous, leading to an exacerbation of the autoimmune process [40]. Since this ‘overlap’ condition is very rare, this would explain why several authors did not observe an exacerbation of the disease after treatment of anti-LKM⁺ hepatitis C patients with IFN [41–44].

In conclusion, our data, based on a large number of anti-LKM⁺ patients, confirm the previously shown specificity of the epitope 257–269 for AIH type II but add new information on the existence of a further linear immunodominant epitope in the region 196–218. Our observations also substantiate previous suggestions [11–13,17,18] that anti-LKM antibodies express different activities, being directed against linear epitopes as well as conformational epitopes as shown by their property to inhibit enzyme function. They also imply that the presence of antibodies to the two linear immunodominant epitopes 257–269 and 196–218 in patients with hepatitis C may be taken as evidence for a true overlap syndrome.