Fine specificity and subclasses of IgG anti-actin autoantibodies differ in health and disease

Journal of Autoimmunity 20 (2003) 333–344 www.elsevier.com/locate/issn/08968411

Fine specificity and subclasses of IgG anti-actin autoantibodies differ in health and disease A. Zamanou a, M. Samiotaki b, G. Panayotou b, L. Margaritis c, P. Lymberi a* a

Immunology Laboratory, Biochemistry Department, Hellenic Pasteur Institute, 127, Vas. Sofias Avenue, 11521 Athens, Greece b Biomedical Sciences Research Center “Alexander Fleming”, 166 72 Vari, Greece c Cell Biology and Biophysics Department, School of Biology, University of Athens, 15784 Athens, Greece Received 19 July 2002; revised 17 February 2003; accepted 27 February 2003

Abstract Current opinions suggest that autoantibodies occurring in autoimmune diseases are generated by B-cells which primarily produce polyspecific natural autoantibodies, through either polyclonal activation or specific antigen selection of these B-cells. In this study, we compared the immunological properties (polyspecificity, fine specificity and IgG subclasses) between natural anti-actin antibodies (N-AAA) and disease-associated AAA (D-AAA). IgG AAA from sera of healthy donors, patients with autoimmune hepatitis type 1 (AIH-1) and patients with primary biliary cirrhosis (PBC) were affinity-purified on actin immunoadsorbent and tested initially for polyspecificity against various cytoskeleton proteins by enzyme-linked immunosorbent assay (ELISA). Fine specificity was studied by Western blotting using proteolytic peptides of actin and by ELISA using synthetic 12mer peptides, spanning the 221–377 aa sequence of actin. Results showed that both N-AAA and D-AAA are polyspecific. Nevertheless, D-AAA from both diseases showed a specific reactivity pattern as compared to N-AAA, against the 16 kDa C-terminal (229–377 aa) proteolytic peptide of actin and more specifically against the P36 synthetic peptide (351–362 aa). Quantitation of AAA IgG subclasses revealed that IgG1 and IgG3 were specifically increased in D-AAA from AIH-1 and PBC, respectively, as compared to N-AAA. We conclude that D-AAA are differentiated from N-AAA in terms of fine specificity and IgG subclasses, probably through specific antigen selection of B-cells primarily producing N-AAA.  2003 Elsevier Science Ltd. All rights reserved. Keywords: Anti-actin autoantibodies; Autoimmune liver diseases; Epitopes; Peptides; IgG subclasses

1. Introduction A fundamental property of normal immune system is the production of natural polyspecific autoantibodies (NAbs), which are able to react with a variety of different antigens ranging from proteins to DNA and haptens [1–3]. The earliest question raised since the discovery and isolation of NAbs has been their relationship with disease-associated autoantibodies (DAbs) [4,5]. The main hypothesis, based on idiotypic and VHJHDH or VLJL gene similarities occuring between DAbs and NAbs, is that DAbs and their respective B-cell clones arise out of a population of B lymphocytes originally committed to produce NAbs [4,6–9]. The * Corresponding author. Tel.: +30-210-6478-808; fax: +30-210-6478-808. E-mail address: [email protected] (P. Lymberi).

production of DAbs during disease may result from either of the following mechanisms: (a) polyclonal activation of B lymphocytes primarily producing NAbs [6,10–12]; in this case, DAbs share common characteristics with NAbs (e.g. polyspecificity, common idiotypes) [6,7,10] and (b) specific antigen selection of B lymphocytes primarily producing NAbs [4,13–15]; in this case, some of the characteristics of DAbs are differentiated from those of NAbs (e.g. DAbs become monospecific, their fine specificity and/or their isotype change, somatic mutations are accumulated on their CDR3 sequences) [13,15,16]. A significantly high proportion of polyspecific NAbs possess anti-actin activity [17]. In addition, anti-actin antibodies (AAA) were the first anti-cytoskeleton autoantibodies described in liver diseases [11]. They are considered as a marker of autoimmune hepatitis type 1

0896-8411/03/$ - see front matter  2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0896-8411(03)00036-2

334

A. Zamanou et al. / Journal of Autoimmunity 20 (2003) 333–344

(AIH-1), as they are frequently detected in AIH-1 (in 50–90% of patients’ sera) as compared to viral hepatitis and healthy donors (0–18%) [18-21]. In addition, they have been associated with a more severe outcome of the disease [19]. AAA are also frequently detected in other autoimmune liver diseases, such as primary biliary cirrhosis (PBC) (44–80%) [22,23] and in lower titers in various non-liver diseases (autoimmune or infectious) [24–26]. However, little is known about the immunological properties of disease AAA (D-AAA), their relationship with natural AAA (N-AAA) and their role in pathogenesis [11,21]. It should be pointed out that AAA are directed against the ubiquitous cytoskeleton protein actin [11], which is highly conserved during evolution [27] and is therefore considered to represent a universal self antigen. Thus, studying the characteristics of these autoantibodies in health and disease, would add to the understanding of liver autoimmunity. The aim of the present study was to investigate the relationship between D-AAA and N-AAA, particularly whether the former result from N-AAA overproduction or from an antigendriven specific response. For this purpose, we compared the main immunological properties (i.e. polyspecificity, fine specificity and IgG subclasses) of affinity-purified IgG N-AAA with those of IgG D-AAA from AIH-1 and PBC.

2. Materials and methods 2.1. Sera Sera from 39 patients with AIH-1, 24 with PBC and 21 healthy donors (HD) were studied. Patients’ sera were provided by the Immunology Laboratories of Evangelismos and Red Cross Hospitals (Athens, Greece). Patients’ diagnosis was provided to us by the respective hospitals and was based on conventional clinical criteria and laboratory findings: (a) All AIH-1 sera were ASMA (anti-smooth muscle antibodies) positive (IFL titer R1:80, tested on stomach–kidney sections), 40% of these sera were also ANA positive (IFL titer R1:40, tested on Crithidia luciliae) but none was AMA (antimitochondrial antibodies) positive. AST (aspartate transferase)/ALT (alanine transferase) levels were elevated (200–500 IU/l) and so were IgG levels in 15/39 sera (2000–3500 mg/dl). Thirty of the 39 AIH-1 patients were women (30–60 years old). Patients fulfilled the classical histological criteria for AIH-1 (piecemeal necrosis and bridging necrosis). On the contrary, they had no sign of destruction of intrahepatic bile ducts, which suggests (in combination with the autoantibody levels) that AIH-1 patients did

not belong to the PBC/AIH-1 overlap syndrome (definite AIH-1). Finally, hepatitis caused by alcoholism, drugs or infection by hepatotropic viruses was excluded (anti-HBV and anti-HCV negative sera). (b) All PBC sera were AMA positive (IFL titer R1:160, M2 type, tested on stomach–kidney sections). Eighteen out of 23 were ASMA negative and only 5/23 were ASMA positive (IFL titer 1:40–1:80). They were characterized by hyperbilirubinemia (R2 mg/ dl) and elevated levels of ALP (alkaline phosphatase) (400–500 IU/l). Most patients (18/23) were women (40–60 years old). Patients fulfilled the classical histological criteria for PBC (intrahepatic bile duct destruction and cirrhosis). No history of alcoholism was mentioned for the patients while infection by hepatotropic viruses was excluded (anti-HBV and anti-HCV negative sera). (c) All healthy donors were AMA and ASMA negative, they had no history of HBV, HCV, HIV or other recent infection. They were of the same gender (70% women) and age (30–60 years old) as patients. 2.2. Antigens Human skeletal filamentous (F) actin, globular (G) actin and myosin were prepared according to described methods [28,29]. Denatured actin (A-actin) was prepared by boiling G-actin at 100(C for 20 min, at a concentration of 0.8 mg/ml in 0.1 M carbonate– bicarbonate buffer (cbc). To avoid rearrangement, actin was cooled at 0(C immediately after boiling. It was then diluted and immediately transferred in enzyme-linked immunosorbent assay (ELISA) plates and stored in a cold room (4(C). Desmin from chicken stomach was provided by Dr C. Vorgias (Department of Biochemistry and Molecular Biology, University of Athens). Bovine skeletal tropomyosin (TM) was purchased from Sigma (St. Louis, MO, USA). Human thyroglobulin (Tg) was purified from thyroid glands, as previously described [30]. The purity of all antigens was verified by sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) under reducing and non-reducing conditions. 2.3. Antibodies A mouse monoclonal AAA and a rabbit polyclonal AAA both specific for the C-terminal end of actin (367–377 aa), a rabbit polyclonal AAA raised against native G-actin, mouse monoclonal antibodies against tropomyosin, troponin,
-actinin and desmin (all from Sigma), as well as a rabbit polyclonal anti-myosin antibody raised in our laboratory [23], were used as controls in Western blotting (WB). Goat anti-human

A. Zamanou et al. / Journal of Autoimmunity 20 (2003) 333–344

IgG (-specific) (Sigma) and monoclonal mouse antihuman IgG1, IgG2, IgG3 and IgG4 (1, 2, 3, 4 specific, respectively) (Calbiochem, San Diego, USA) were used in ELISA for the measurement of IgG and IgG subclasses. Alkaline phosphatase-labeled goat antihuman, anti-mouse and anti-rabbit IgG (-specific), peroxidase-labeled sheep anti-human  light chain and anti-human  light chain (all from Sigma) were used in ELISA and WB. 2.4. Immunoglobulins Human IgG was isolated from an individual healthy control serum on Protein-G-Sepharose column (Pharmacia, Uppsala, Sweden) and this preparation was used as reference for the quantitation of IgG and IgG subclasses of human AAA. The concentration of IgG and of IgG1, IgG2, IgG3, and IgG4 in the IgG preparation, was measured by nephelometry (commercial kit from Dade Behring S.A., Marburg, Germany). Human IgG from a pool of healthy donors (Sigma) was used for the isolation of AAA on actin immunoadsorbent which were subsequently used as a control in ELISA against P36. 2.5. Actin peptides 2.5.1. Proteolytic peptides Actin was cleaved into three main fragments of 26, 19 and 16 kDa by the proteolytic enzyme Staphylococcus aureus V8 protease (P2922, Sigma), according to a well established method [31]. The 26 and 19 kDa fragments contain the N-terminal end of actin (residues 1–228 aa and 1–169 aa, respectively), while the 16 kDa contains the C-terminal end of actin (residues 229–377 aa) [31]. Immediately after cleavage, sample buffer was added and the preparation was loaded on an SDS– polyacrylamide gel. 2.5.2. Synthetic peptides Sixteen synthetic 12mer peptides (P23–P38), overlapping by two residues and spanning the sequence 226– 377 aa of actin (i.e. the 16 kDa C-terminal fragment) were a kind gift from Prof. S. Avrameas. TgP41, a 20mer synthetic peptide of human thyroglobulin (residues 2651–2670 aa), was used as control. 2.6. WB with actin proteolytic fragments The proteolytic fragments of actin were loaded on an SDS–polyacrylamide gel (4% stacking gel and 12.5% separating gel). Electrophoresis was performed under reducing conditions. Briefly, 480 µg of cleaved actin were diluted in 1 sample buffer containing 10 mM Tris, 2% SDS, 5% -mercaptoethanol, 10% glycerol and 0.002 M bromophenol blue. The protein bands were electrotrans-

335

ferred from the gel to nitrocellulose sheets (NC) (Hybond-C extra 45 µm, Amersham, UK) at 125 mA, overnight at 4(C. The transferred bands were reversibly stained by Ponceau S. Subsequently, NC strips (individually vertically cut from NC sheet) were incubated in Tris-buffered saline (TBS) containing 5% nonfat milk and 0.2% Tween-20, for 2 h at room temperature (RT), in order to block the residual binding sites. After three washings with TBS containing 0.5% Tween-20 and one with TBS, the strips were incubated with sera diluted 1:100 in TBS containing 1% milk and 0.2% Tween-20, for 2 h at RT. Strips were washed as above and then incubated with alkaline phosphatase conjugated anti-human IgG (2 µg/ml) for 90 min at RT. Visualization of reactive bands was performed using NBT-BCIP (Sigma) substrate buffer and reaction was stopped by TBS containing 20 mM ethylene diamine tetraacetic acid (EDTA). Alternatively, the NC strips were incubated (in the same buffer and for the same time as with human sera) with predetermined concentrations or dilutions of the control monoclonal and polyclonal antibodies: monoclonal anti-desmin antibody, as well as monoclonal and polyclonal anti-C-terminal actin peptide antibodies (2 µg/ml), monoclonal anti-tropomyosin, anti-troponin and anti-
-actinin (1:50 of ascite fluid), polyclonal AAA (1:10 of antiserum) and polyclonal anti-myosin (1:100 of antiserum). These strips were subsequently incubated with alkaline phosphataselabeled anti-mouse or anti-rabbit IgG (2 µg/ml) for 90 min at RT and visualization of reactive bands was performed as mentioned above. 2.7. Peptide sequence analysis Sequence analysis of the proteolytic fragments of actin was performed by nano-spray mass spectrometry. Fragments were separated in one-dimensional SDSPAGE. After visualization by silver staining, the bands corresponding to the unknown V8 proteolytic peptides (35 kDa) were excised and subjected to in-gel digestion with trypsin (modified sequence grade trypsin, Promega). Tryptic fragments were extracted from gel pieces and a final peptide preparation in 2% acetonitrile/ 0.1% formic acid was applied to a C-18 HPLC column (75 µm15 cm) on an UltiMate nano-HPLC system (LC-Packings, Netherlands). Peptides were separated with an acetonitrile gradient at 150 nl/min and analyzed on an ion trap mass spectrometer (LCQ-Deca, ThermoFinnigan, USA) with a nano-spray source. MS and MS/MS data were used to search the SwissProt database with the SEQUEST search engine. 2.8. Isolation of total IgG from serum Total IgG was purified on a protein G-Sepharose column (Pharmacia) from three AIH-1, three PBC and

336

A. Zamanou et al. / Journal of Autoimmunity 20 (2003) 333–344

three HD sera. The selection of these sera from the larger series was based on the following criteria: 1. AIH-1 sera were ASMA positive, AMA negative (definite AIH-1 by the above mentioned conventional criteria) and positive for IgG anti-G-actin antibodies by ELISA (OD405 nm>2cut-off value, defined by the mean OD405 nm HD+2.5 SD) [23]. 2. PBC sera were AMA positive of M2 type, ASMA negative (definite PBC by the above mentioned conventional criteria) and positive for IgG anti-Gactin antibodies by ELISA (OD405 nm>2cut-off value). 3. HD sera were ASMA/AMA negative and also negative for IgG anti-G-actin antibodies by ELISA (OD405 nm
small molecules (IMMULON II Dynex Technologies Inc., Chantilly, USA) were coated with peptides at a concentration of 10 µg/ml in 0.05 M cbc. Peptide-coated plates were dried overnight at 37(C and used immediately after drying. Free binding sites were blocked with PBS containing 5% bovine serum (PBS–BS) (150 µl/well) for 1 h at 37(C. Sera diluted 1:100 in PBS–BS–0.1% Tween were incubated (100 µl/well) with the coated plates for 90 min at 37(C and purified antibodies serially diluted from 20 to 0.3125 µg/ml were incubated overnight at 4(C. Antibody reactivity was detected using alkaline phosphatase-labeled anti-human IgG (at 1 µg/ml in PBS–BS–0.1% Tween, 100 µl/well). P-nitrophenyl phosphate was used as substrate and optical density (OD) was measured at 405 nm by a Dynatech MR 5000 multiskan spectrophotometer (Dynatech, Chantilly, VA, USA). For sera screening, a reference serum with OD corresponding to that of the cut-off point (i.e. mean OD value of HD sera+2.5 SD), was always included in each plate in order to directly define positive sera. A serum was regarded as positive if the obtained OD value was higher than that of the reference serum. In all ELISAs, a blank was used: at least four wells per plate were incubated only with the secondary antibody and the enzyme substrate. The OD value of blank ranged from 0.05 to 0.1, among the different assays. The OD values of sera and AAA mentioned in Section 3, are corrected for the respective blank value. 2.11. Competitive ELISA using immobilized P36 peptide In order to check the specificity of D-AAA against P36 peptide (sequence: LSTFQQMWITKQ or Leu-SerThr-Phe-Gln-Gln-Met-Trp-Ile-Thr-Lys-Gln), a competitive enzyme immunoassay procedure was developed according to a method described elsewhere [32]. The procedure was the same as for the non-competitive ELISA with the difference that D-AAA (at a dilution corresponding to 50% maximum binding), were preincubated with the inhibitors (soluble P36 peptide, G-actin and bovine serum albumin (BSA) at 12–0.1875 µmol/ml) for 2 h at 37(C before being added into P36-coated wells. The results were expressed as the concentration of inhibitor required to achieve a 50% inhibition (I50) of binding. In this assay, a blank was also used (as mentioned above) and before any calculation of I50, OD values were corrected for blank value. 2.12. ELISA for the quantitation of IgG class and subclasses Plates were coated with anti-isotype specific antibodies, anti-: 1 µg/ml in PBS; anti-1: 2 µg/ml in PBS; anti-3: 2 µg/ml in PBS; anti-2: 2 µg/ml in 0.1 M borate buffer pH 8.6 and anti-4: 2 µg/ml in 0.1 M borate buffer

A. Zamanou et al. / Journal of Autoimmunity 20 (2003) 333–344

337

Table 1 Reactivity of affinity-purified natural and disease-associated AAA against the panel of antigens Anti-actin antibodies

Antigens F-actin

a

N-AAA.1 N-AAA.2 N-AAA.3 D-AAA.1b D-AAA.2 D-AAA.3 D-AAA.4 D-AAA.5 D-AAA.6

c

+++ + +++ + + +++ +++ ++ ++

G-actin

A-actin

Myosin

Tropomyosin

Desmin

Tubulin

Thyroglobulin

+++ ++ +++ +++ +++ +++ +++ ++ +++

+++ +++ +++ +++ ++ +++ +++ +++ +++

+ + +++ ++ ++

+ +++ + +

+ +++ + ++ ++ +++ +++ +

-

+ + + -

Purified AAA (at 10 µg/ml) were incubated with ELISA plates coated with cytoskeleton antigens and control thyroglobulin. Their binding to the coated plates was revealed using an alkaline phosphatase-anti-human IgG. OD405 nm values were converted as follows:
: <0.5, +: 0.5–1.0, ++: 1.0–1.5, +++: >1.5. a N-AAA.1, N-AAA.2, N-AAA.3: natural anti-actin antibodies purified from respective sera of healthy controls. b D-AAA.1–3: anti-actin antibodies purified from AIH-1 sera, D-AAA.4–6: anti-actin antibodies purified from PBC sera.

pH 8.3. Plates were saturated with 0.5% gelatin–PBS for 1 h at 37(C. Serial dilutions of purified AAA and a reference human IgG preparation were incubated with the coated plates for 1 h at 37(C. The starting reference dilution was 320 ng/ml for IgG, 4 µg/ml for IgG1, 8 µg/ml for IgG2, 250 ng/ml for IgG3 and 1 µg/ml for IgG4. Bound IgG and IgG subclasses were revealed using an alkaline phosphatase-labeled anti-human IgG conjugate (1 µg/ml) (for the IgG, IgG1, IgG2 and IgG3 assays) or peroxidase-labeled anti-human  and  chain conjugates (1 µg/ml each) (for the IgG4 assay). P-nitrophenyl phosphate or O-phenylenediamine (OPD)+H2O2 were used as substrates of alkaline phosphatase and peroxidase and OD was measured at 405 and 490 nm, respectively. The concentration of IgG antibodies and subclasses was calculated using the respective standard curves. In this assay a blank was also used (as mentioned above) and before any calculation of IgG concentration, OD values were corrected for blank value. 2.13. Statistical analysis The distribution of antibody activity values (OD values) of sera in each patient group (AIH-1, PBC) was compared to that of sera from the healthy donor group using the non-parametric Mann–Whitney test statistic. A p value less than 0.01 was considered statistically significant. 3. Results 3.1. Analysis of polyspecificity of purified anti-actin antibodies Six patients’ sera (three AIH-1 and three PBC) and three HD sera were used for the isolation of AAA by

affinity chromatography. All patients’ sera exhibited a strong anti-actin reactivity, whereas HD sera did not react with the immobilized actin by ELISA (data not shown). Affinity-purified N-AAA and D-AAA were further tested by ELISA against the eight panel antigens. Results are presented in Table 1. All AAA reacted with the denatured (A) as well as with the native monomeric (G) or polymeric (F) form of actin. However, each AAA exhibited its own reactivity profile against the panel antigens. As expected, N-AAA were polyspecific but interestingly, so were D-AAA from liver diseases, reacting with the homologous (actin) and at least one heterologous antigen. Almost all AAA (N-AAA and D-AAA) reacted with desmin. Apart from desmin, the main cross-reactivities of D-AAA were observed with myosin and tropomyosin. Only three AAA (one N-AAA and two D-AAA from PBC) reacted with thyroglobulin and none reacted with tubulin. 3.2. Reactivity of sera against proteolytic fragments of actin The nine sera used for AAA purification were tested by WB against the 26, 19 and 16 kDa proteolytic fragments of actin as well as with uncleaved actin (42 kDa). WB results are shown in Fig. 1. HD sera reacted neither with intact actin nor with any of the above fragments. AIH-1 sera reacted with uncleaved actin and the 16 kDa fragment but not with the 26 or 19 kDa fragments, suggesting that D-AAA from AIH-1 specifically bind to an epitope contained in the 16 kDa fragment. This fragment is also recognized by D-AAA from PBC. PBC sera also reacted with uncleaved actin and with the 26 kDa but not with the 19 kDa fragment; they recognize an epitope potentially being between residues 169 and 228 aa. All sera (normal and pathological) recognized an additional V8 proteolytic fragment

338

A. Zamanou et al. / Journal of Autoimmunity 20 (2003) 333–344

suggests that it lacks the 367–377 aa sequence (Fig. 2). The two bands corresponding to 35 kDa fragment were subsequently detected by silver staining of the gel (Fig. 3A), which is more sensitive than the usual staining of the gel by Coomassie blue or staining of the NC by Ponceau S (these two stains identify only the three main fragments, 26, 19 and 16 kDa). These bands as well as uncleaved actin were further digested by trypsin and the resulting tryptic peptides were analyzed by mass spectrometry. Fig. 3B shows the tryptic peptides isolated from each band (shaded sequences), which were identified by the mass spectrometer and were found to be part of actin sequence. Thus, it was confirmed that the 35 kDa bands are actin fragments (not contaminants) that lack its C-terminus (peptide 363–374 aa is identified only in uncleaved actin) (Fig. 3B).

3.3. Epitope mapping of the C-terminal end of actin (229–377 aa)

Fig. 1. Immunoblotting on the electrophoretically separated V8 proteolytic fragments of human skeletal actin, immobilized on NC. Each strip was separately incubated with one serum. Lanes 1–3: AIH-1 sera 1–3; lanes 4–6: PBC sera 1–3; lanes 7–9: HD sera 1–3; lane 10: uncleaved actin (42 kDa) and its proteolytic fragments (26, 19 and 16 kDa) as visualized after staining of NC with Ponceau S.

(two bands of 35 kDa approximately), which contains most probably epitopes recognized by both N-AAA and D-AAA and according to our results, it does not seem to be specific for autoimmune liver diseases. Molecular weight (MW) of this fragment was initially calculated using a specific program for the analysis of gels and immunoblots (IMAGE MASTER, Pharmacia Uppsala Sweden), which is based on the comparison of the electrophoretic mobility of the unknown protein bands on the gel with the one of reference molecules of known molecular weight. In order to check the purity of this actin sample and exclude the possibility for the immunologically detected 35 kDa fragment to be a contamination, we performed WB using several monoclonal and polyclonal antibodies specific to cytoskeleton proteins. These proteins represent possible impurities of actin sample (Fig. 2). The 35 kDa fragment was recognized only by specific polyclonal AAA raised against native G-actin, suggesting that 35 kDa is a part of actin sequence and not a contaminant. However, this fragment was not recognized by monoclonal and polyclonal AAA specific for the C-terminal end of actin, which

Based on the specificity of D-AAA from AIH-1 and PBC against the C-terminal end of actin (16 kDa fragment), we further analyzed this sequence in order to identify a smaller fragment containing the specific epitope. For this purpose, we developed an ELISA using synthetic 12mer peptides: p23–p38, which cover the entire sequence from residue 229 to 377 aa. The 20mer peptide TgP41 from human thyroglobulin was used as control. Results are shown in Fig. 4A and B. AAA from AIH-1 preferentially reacted with P36 peptide and the same holds true for PBC patients. On the other hand, N-AAA exhibited a significantly lower reactivity against P36, as compared to D-AAA. This result is additionally supported by the binding curves of AAA to P36, shown in Fig. 5: the titer of the six D-AAA against P36 is significantly higher than that of N-AAA from the individual HD sera and of N-AAA from the pool of HD IgG. P36 corresponds to the sequence 351–362 aa of
-skeletal actin or 349–360 aa of cytoplasmic non-muscle  and -actins, found in liver cells.

3.4. Specificity of AAA to P36 peptide In order to further confirm the specificity of AAA from AIH-1 and PBC to P36, we developed a competitive ELISA using immobilized P36 and soluble P36, G-actin and BSA as inhibitors. D-AAA were preincubated with 12, 3, 0.75 and 0.1875 µmol/ml of P36, G-actin and control BSA. D-AAA from AIH-1 and PBC were equally inhibited by P36 (I50: 7.5–12 and 11–12 µmol/ml, respectively) (Table 2). D-AAA were also inhibited by the native antigen, G-actin (I50: 3–12 µmol/ml for AIH-1 and 8–12 µmol/ml for PBC) (Table 2). BSA did not inhibit any of the AAA. These

A. Zamanou et al. / Journal of Autoimmunity 20 (2003) 333–344

339

Fig. 2. Western blotting on the electrophoretically separated V8 proteolytic fragments of human skeletal actin, immobilized on NC, using control monoclonal and polyclonal anti-cytoskeleton antibodies. Lane 1: incubated with monoclonal AAA specific for the C-terminal end of actin, lane 2: incubated with polyclonal AAA specific for the C-terminal end of actin, lane 3: incubated with polyclonal AAA raised against native G-actin, lane 4: incubated with polyclonal anti-myosin antibody, lane 5: incubated with monoclonal anti-tropomyosin antibody, lane 6: incubated with monoclonal anti-troponin antibody, lane 7: incubated with monoclonal anti-
-actinin antibody, lane 8: incubated with monoclonal anti-desmin antibody, lane 9: uncleaved actin and its V8 proteolytic fragments as visualized after staining of NC with Ponceau S. Arrows on the left, show the fragments which are recognized by the antibodies.

data verify the specificity of D-AAA from AIH-1 and PBC against the 351–362 aa sequence of actin. 3.5. Screening of sera against P36 In order to further evaluate the significance of P36 as a specific target of autoantibodies in AIH-1 and PBC, we tested 39 sera from AIH-1, 24 from PBC and 21 from HD. The anti-P36 reactivity of individual sera is shown in Fig. 6. Statistical analysis of the results by Mann– Whitney, showed that AIH-1 sera exhibited a significantly increased IgG antibody reactivity to P36 as compared to healthy donors (p<0.01) and so did PBC sera.

concentration of total IgG. Results are shown in Fig. 7A and B. The incidence of IgG1 and IgG3 differed between AAA. D-AAA from AIH-1 contained a higher proportion of IgG1 as compared to N-AAA: 63–72% of IgG were IgG1 in AIH-1 D-AAA versus 29–46% in N-AAA. Furthermore, D-AAA from PBC contained a higher proportion of IgG3 as compared to N-AAA: 23–51% of IgG were IgG3 in PBC D-AAA versus 4.2–5.2% in N-AAA. The proportion of IgG2 and IgG4 varied among individual autoantibodies of the three groups and showed no significant difference between N-AAA and D-AAA (data not shown).

3.6. Quantitation of IgG subclasses of AAA

4. Discussion

The concentration of IgG subclasses: IgG1, IgG2, IgG3, IgG4 present in N-AAA and D-AAA was measured by quantitative ELISA. The proportion of IgG subclasses in individual AAA was calculated as follows: (cIgG1 or cIgG2 or cIgG3 or cIgG4/ cIgG)100, where cIgG1, cIgG2, cIgG3, cIgG4 are the concentrations of the different subclasses and cIgG is the

AAA are considered to be an important group of autoantibodies as they are frequently detected in both physiological and pathological situations, mainly in AIH-1 [17,19]. This study is the first attempt to investigate the mechanism of D-AAA production and their relationship with N-AAA. In the present study, the comparison of immunological properties of these

340

A. Zamanou et al. / Journal of Autoimmunity 20 (2003) 333–344

Fig. 3. Sequence analysis of the 35 kDa fragment by mass spectrometry. (A) The V8 proteolytic fragments of actin are separated in SDS-PAGE 12.5% and revealed by silver staining which detects two bands (showed by the respective arrows) corresponding at a molecular weight (MW) of 35 kDa. (i) Molecular weight markers of 54, 48 and 35 kDa and (ii) V8 proteolytic fragments after silver staining. (B) Band 1 corresponding to uncleaved actin and bands 2 and 3 corresponding to a MW of 35 kDa (as shown by SDS-PAGE) were further digested by trypsin. The resulting peptides were analyzed by mass spectrometry and their sequence was searched for homology with actin. The shaded sequences correspond to the tryptic peptides of each band which were identified by mass spectrometry. Bands 2 and 3 lack C-terminal sequences, identified only in uncleaved actin. Nevertheless, the exact termini of these fragments could not be verified with this approach, since some peptides resulting from tryptic digestion may not be extracted from the gel, separated by HPLC or ionized in the mass spectrometer.

two AAA populations, revealed both similarities and differences. Analysis of AAA specificity against various cytoskeletal antigens showed that polyspecificity is not only a property of N-AAA but also of D-AAA occurring in AIH-1 and PBC. Some other studies also recognize polyspecificity as a common feature between NAbs and DAbs, e.g. anti-DNA autoantibodies occurring in systemic lupus erythematosus and normal individuals, IgG anti-red blood cell antibodies occurring in patients with warm autoimmune hemolytic anemia and normal donors and polyspecific rheumatoid factor (RF) in rheumatoid arthritis and normal subjects [10,33,34]. Among these studies, only the last one proved the common origin of polyspecific NAbs and DAbs, as it showed that both natural and disease RFs are produced by the same

B-cell clones (CD5+). Thus in the present study, in order to check the common origin of N-AAA and D-AAA, we looked for some additional similarities between these two AAA groups. Indeed, fine specificity and subclass analysis revealed that both populations: (1) reacted simultaneously with native (F-) and denatured (A-) actin, suggesting that N-AAA and D-AAA bind to conformational and linear epitopes as well, (2) reacted with the 35 kDa V8 proteolytic fragment of actin, which covers a major part of the molecule sequence, lacking its C-terminus; this fragment probably contains a common epitope for N-AAA and D-AAA, and (3) belonged to all four IgG subclasses. The fragment of 35 kDa has been proven to be part of actin sequence and not a contamination of actin sample with another cytoskeleton protein, as it was recognized in WB only by the specific

A. Zamanou et al. / Journal of Autoimmunity 20 (2003) 333–344

341

Fig. 4. Reactivity profile of AAA against synthetic peptides spanning the carboxy-terminal end (residues 229–377) of human skeletal actin. Affinity-purified D-AAA isolated from three patients 1–3 with AIH-1 (A; j) and three patients 4–6 with PBC (B; ) were compared for reactivity with N-AAA from HD (A and B; ,). In all cases, AAA were assayed at 10 µg/ml. Antibody reactivity is expressed as OD405 nm values.

polyclonal AAA raised against native G-actin and mass spectrometry revealed sequence homology with actin. Reaction of both normal and patients’ sera with 35 kDa fragment was very strong, while normal sera did not react with the uncleaved molecule or with the fragments of 26, 19 and 16 kDa. This could be explained by the fact that changes in the primary structure of a protein, either inside or outside the antigenic regions (e.g. aminoacid deletions), can alter (augment or suppress) the binding activity in these sites [35]. In addition, it has been mentioned that there are epitopes which are not recognized on the unprocessed or native protein but are only exposed after its treatment with denaturing agents (e.g. SDS) [36]. It is possible then that fragmentation of actin at specific points induces such changes (e.g. chemical changes, like charge transfer) that contribute to the exposure of the epitope contained in the 35 kDa fragment.

Apart from similarities, some important differences also occurred between N-AAA and D-AAA, suggesting that D-AAA production is not exclusively the result of polyclonal B-cell activation. D-AAA from AIH-1 and PBC recognized, as compared to N-AAA, an additional epitope which seems to be disease-specific and resides in the C-terminal part of actin; they preferably bound to the 16 kDa fragment of actin (residues 229–377 aa) and especially to the small 12mer peptide P36: LSTFQQMWITKQ, corresponding to residues 351– 362 aa of human
-skeletal actin and 349–360 aa of non-muscle actin. Inhibition experiments showed that binding of D-AAA to P36 is specific. In addition, sera from AIH-1 and PBC specifically reacted with P36 as compared to normal sera. These data suggest that during the course of the disease at least a population of N-AAA is subjected to changes of its fine specificity, probably through antigen driven somatic mutations.

342

A. Zamanou et al. / Journal of Autoimmunity 20 (2003) 333–344

Fig. 5. Binding curves of affinity-purified AAA to immobilized P36. From top to bottom D-AAA.1-3 from AIH-1: (+); D-AAA.4-6 from PBC: (); N-AAA.1-3 from HD: (,) and N-AAA from the pool of HD IgG (j). Anti-P36 activity is expressed as OD405 nm values. Table 2 Competitive ELISA D-AAAa

D-AAA.1 D-AAA.2 D-AAA.3 D-AAA.4 D-AAA.5 D-AAA.6

Inhibitors P36

G-actin

BSA

7.5 11 12 11 10 12

5 3 12 11 8 12

NIb NI NI NI NI NI

Inhibition of binding of D-AAA to immobilized P36 by soluble P36, native G-actin and control BSA in decreasing concentrations, ranging from 12 to 0.1875 µmol/ml. Values correspond to the concentration (µmol/ml) of inhibitor required to inhibit the binding of AAA to P36 coated plates by 50%. (I50). a D-AAA, disease AAA; D-AAA.1–3 from AIH-1 sera 1–3 and D-AAA.4–6 from PBC sera 1–3. b NI, no inhibition.

Somatic mutations have been reported to accumulate during autoimmune disease in the CDR3 sequences of autoantibodies, which mainly contribute to antigen binding and fine specificity [14]. As far as differences are concerned between N-AAA and D-AAA, a significantly higher proportion of IgG1

Fig. 6. Distribution of anti-P36 reactivity values (OD405 nm) of sera from the three groups tested: AIH-1, PBC and HC. The cut-off point (calculated as the mean OD405 nm values of the healthy control sera+2.5 SD) is indicated by a horizontal line.

and IgG3 was detected in AIH-1 and PBC, respectively, as compared to N-AAA. It seems that in addition to fine specificity differentiation, D-AAA undergo subclass switching as well. It should be mentioned that autoantibodies of IgG1 and IgG3 subclass might have a distinct role in pathogenesis as these two

A. Zamanou et al. / Journal of Autoimmunity 20 (2003) 333–344

Fig. 7. Proportion of IgG subclasses present in the individual AAA: percentage of IgG1 subclass (A); percentage of IgG3 subclass (B). The percentage of IgG subclasses was calculated as follows: (cIgGy/ cIgG)100; where cIgGy is the concentration of IgG1 or IgG3 and cIgG, the concentration of total IgG. ,: N-AAA.1–3 from healthy controls, j: D-AAA.1-3 from AIH-1, : D-AAA.4–6 from PBC.

subclasses are known to be able to activate complement [37]. In a previous study, we demonstrated that AAA are not a serological marker of AIH-1 as they are also frequently detected in PBC patients’ sera [23]. This finding is confirmed in the present study, since we found that AAA from AIH-1 cannot be differentiated in terms of fine specificity from AAA from PBC. This is in disagreement with data from other investigations associating AAA exclusively with AIH-1 [18,19,21]. This disagreement could be attributed to the different methodologies used; the latter studies were based on IFL whereas in our study detection and characterization of AAA was based on ELISA. It should be mentioned that in PBC, most serological studies are focused on anti-mitochondrial autoantibodies (AMA) which are

343

markers of the disease and little is known about AAA also occurring in the disease [38]. As a consequence, the serological overlapping between AIH-1 and PBC as shown in the present and our previous study (>50%) [23], was more significant than the one mentioned in other investigations (10–20%) which was mainly based on conventional criteria [21,38,39]. At this point, we should underline that according to the conventional serological and histological criteria [38,39] our patients did not belong to the AIH-1/PBC overlap syndrome. The similar serological profile (ACTA and AAA) between the two diseases, might further imply the existence of common pathogenic mechanisms in AIH-1 and PBC. To summarize, AAA occurring in autoimmune liver diseases consist of at least two populations: one which shares common characteristics (e.g. 35 kDa peptide recognition) with N-AAA and another which is differentiated from N-AAA in terms of fine specificity and subclasses. Such differences have been previously investigated in other pathological situations as for example in autoimmune thyroiditis and lymphoproliferative disorders [13,15,16]. Anti-thyroglobulin antibodies from patients with autoimmune thyroiditis recognized additional epitopes as compared to their natural counterparts from healthy subjects, suggesting that autoimmune response to thyroglobulin is not the result of random polyclonal B-cell activation but is an antigen-driven response [15,16]. The same was concluded with antitubulin autoantibodies which become monospecific during disease, recognize different epitopes as compared to NAbs and undergo isotype and subclass switching, as well [13]. Based on the above data, we can further support that D-AAA from AIH-1 and PBC representing a qualitative rather than a quantitative change in autoantibody repertoire (they recognize a disease specific epitope), are most likely generated by a specific immune response to the antigen and may even have a role in the pathogenesis (G1/G3 subclass switch as compared to N-AAA). Future studies focused on the immunogenicity of P36 peptide in experimental animals or the cross-reactivity between anti-P36 AAA and liver-specific antigens, could allow us to get insight into their role in the disease and their relationship with a severe outcome of AIH-1. Acknowledgements We wish to thank Professor S. Avrameas and Dr C. Vorgias for their kind gifts of synthetic actin peptides and desmin, respectively. References [1] Guilbert B, Dighiero G, Avrameas S. Naturally occurring antibodies against nine common antigens in human sera I.

344

[2]

[3]

[4] [5]

[6] [7]

[8]

[9]

[10]

[11] [12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

A. Zamanou et al. / Journal of Autoimmunity 20 (2003) 333–344 Detection, isolation and characterization. J Immunol 1982; 128:2779–87. Dighiero G, Lymberi P, Holmberg D, Lundquist I, Coutino A, Avrameas S. High frequency of natural autoantibodies in normal newborn mice. J Immunol 1985;134:765–71. Dighiero G, Lymberi P, Guilbert B, Ternynck T, Avrameas S. Natural autoantibodies constitute a substantial part of normal circulating immunoglobulins. Ann N Y Acad Sci 1986; 475:135–45. Avrameas S. Natural autoantibodies: from “horror autotoxicus” to “gnothi seauton”. Immunol Today 1991;12:154–9. Poletaev AB. The immunological homunculus (immunculus) in normal state and pathology. Biochemistry (Moscow) 2002; 67:600–8. Schwartz R. Anti-DNA antibodies and the problem of autoimmunity. Cell Immunol 1986;99:38–43. Mahana W, Guilbert B, Avrameas S. Suppresion of antiDNA antibody production in MRL mice by treatment with anti-idiotypic antibodies. Clin Exp Immunol 1987;70:538–45. Baccala R, Quang TV, Gilbert M, Ternynck T, Avrameas S. Two murine natural polyreactive autoantibodies are encoded by nonmutated germ-line genes. Proc Natl Acad Sci USA 1989; 86:4624–8. Miescher SM, Horn MP, Pachlopnik JM, Baldi L, Vogel M, Stadler BM. Natural anti-FcepsilonRIalpha autoantibodies isolated from healthy donors and chronic idiopathic urticaria patients reveal a restricted repertoire and autoreactivity on human basophils. Hum Antibodies 2001;10:119–26. Casali P, Notkins A. CD5+ B lymphocytes, polyreactive antibodies and the human B-cell repertoire. Immunol Today 1989; 10:364–8. Castell M. Cytoskeletal autoantibodies. In: Peter JB, Shoenfeld Y, editors. Autoantibodies. Amsterdam: Elsevier; 1996;217–26. Jasani B, Ternynck T, Lazarus JH, Phillips DI, Avrameas S. Natural antibody status in patients with Hashimoto’s thyroiditis. J Clin Lab Immunol 1999;51:9–20. Houdayer M, Bouvet JP, Wolff A, Magnac C, Guillemot JC, Borche L et al. Simultaneous presence, in one serum, of four monoclonal antibodies that might correspond to different steps in a clonal evolution from polyreactive to monoreactive antibodies. J Immunol 1993;150:311–9. Shlomchik M, Marshak-Rothstein A, Wolfwicz C, Rothstein T, Weigert M. The role of clonal selection and somatic mutation in autoimmunity. Nature 1987;328:805–11. Bresler H, Burek L, Hoffman W, Rose N. Autoantigenic determinants on human thyroglobulin II. Determinants recognized by autoantibodies from patients with chronic autoimmune thyroiditis compared to autoantibodies from healthy subjects. Clin Immunol Immunopathol 1990;54:76–86. Dietrich G, Piechaczyk M, Pau B, Kazatchkine MD. Evidence for a restricted idiotypic and epitopic specificity of anti-thyroglobulin autoantibodies in patients with autoimmune thyroiditis. Eur J Immunol 1991;21:811–4. Avrameas S, Guilbert B, Mahana W. Recognition of self and non-self constituents by polyspecific autoreceptors. Int Rev Immunol 1988;3:1–15. Abuaf N, Johanet C, Homberg JC. Autoantibodies in autoimmune active chronic hepatitis. In: Krawitt EL, Wiesner RH, editors. Autoimmunne liver diseases. New York, NY: Raven Press; 1991;93–109. Czaja A, Cassani F, Cataleta M, Valentini P, Bianchi F. Frequency and significance of antibodies to actin in type 1 autoimmune hepatitis. Hepatology 1996;24:1068–73.

[20] George J, Shoenfeld Y. Actin autoantibodies. In: Peter JB, Shoenfeld Y, editors. Autoantibodies. Amsterdam: Elsevier; 1996;10–12. [21] Whittingham S, Mackay MD. Smooth muscle autoantibodies. In: Peter JB, Shoenfeld Y, editors. Autoantibodies. Amsterdam: Elsevier; 1996;761–73. [22] Dighiero G, Lymberi P, Monot C, Abuaf N. Sera with high levels of antismooth muscle and antimitochondrial antibodies frequently bind to cytoskeleton proteins. Clin Exp Immunol 1990; 82:52–6. [23] Zamanou A, Tsirogianni A, Terzoglou C, Balafas A, Economidou I, Lymberi P. Anti-smooth muscle (ASMA) and anti-cytoskeleton antibodies (ACTA) in liver diseases: a comparison of classical indirect immunofluorescence with ELISA. J Clin Lab Anal 2002;16:194–201. [24] Takaya M, Kawahara S, Namba T, Grob D. Antibodies against myofibrillar proteins in myasthenia gravis patients. Tokai J Exp Clin Med 1992;17:35–9. [25] Ternynck T, Bleux C, Gregoire J, Avrameas S, Kanellopoulos-Langevin C. Comparison between autoantibodies arising during Trypanosoma cruzi infection in mice and natural autoantibodies. J Immunol 1990;144:1504–11. [26] Pateraki E, Portocala R, Labrousse H, Guesdon JL. Antiactin and antitubulin antibodies in carine viscernal leishmaniasis. Infect Immun 1983;42:496–500. [27] Vandekerckove J, Weber K. Chordate muscle actins differ distingly from invertabrate muscle actins. The evolution of the different vertebrate muscle actins. Mol Biol 1984;179:391–413. [28] Spudish JA, Watt S. The regulation of rabbit skeletal muscle contraction. J Biol Chem 1971;246:4866–71. [29] Whalen RG, Butler-Browne S, Gros F. Identification of a novel form of myosin light chain present in embryonic muscle tissue and cultured muscle cells. J Mol Biol 1978;126:415–31. [30] Schardt CW, McLachlan SM, Matheson J, Rees Smith B. An enzyme-linked immunoassay for thyroid microsomal antibodies. J Immunol Methods 1982;55:155–68. [31] Roustan C, Benyamin M, Bertrand E, Audemard E, Jauregui-Adell J. Conformational changes induced by Mg2+ on actin monomers. FEBS Lett 1985;181:119–23. [32] Thrasyvoulides A, Sakarellos-Daitsiotis M, Souvatzoglou A, Sakarellos C, Lymberi P. B-cell autoepitopes on the acetylcholinesterase-homologous region of human thyroglobulin: association with Graves’ disease and thyroid eye disease. Eur J Endocrinol 2001;145:119–27. [33] Stahl D, Lacroix-Desmazes S, Kaveri SV, Kazatchkine MD. Polyreactivity of disease-associated anti-RBC IgG autoantibodies of patients with warm autoimmune haemolytic anaemia and natural anti-RBC IgG autoantibodies of healthy individuals. Scand J Immunol 2000;52:190–9. [34] Hurez V, Dietrich G, Kaveri SV, Kazatchkine MD. Polyreactivity is a property of natural and disease-associated human autoantibodies. Scand J Immunol 1993;38:190–6. [35] Atassi MZ. Antigenic structures of proteins. Eur J Biochem 1984; 145:1–20. [36] Sara EM. Epitope mapping Immunochemical protocols. In: Manson M, editor. Methods in molecular biology. Totowa, NJ: The Humana Press; 1992;vol. 10:105–16. [37] Schur P. IgG subclasses—a review. Ann Allergy 1987;58:89–96. [38] Mackay IR. Autoimmune diseases of the liver: chronic active hepatitis and primary biliary cirrhosis. The autoimmune diseases. London: Academic Press, 1985 p. 291–337. [39] Czaja AJ. Frequency and nature of the variant syndromes of autoimmune liver disease. Hepatology 1998;28:360–5.