Epitope-specific recognition of type II collagen by rheumatoid arthritis antibodies is shared with recognition by antibodies that are arthritogenic in collagen-induced arthritis in the mouse.код для вставкиСкачать
ARTHRITIS & RHEUMATISM Vol. 46, No. 9, September 2002, pp 2339–2348 DOI 10.1002/art.10472 © 2002, American College of Rheumatology Epitope-Specific Recognition of Type II Collagen by Rheumatoid Arthritis Antibodies Is Shared With Recognition by Antibodies That Are Arthritogenic in Collagen-Induced Arthritis in the Mouse Harald Burkhardt,1 Tobias Koller,1 Åke Engström,2 Kutty Selva Nandakumar,3 Javier Turnay,4 Hans G. Kraetsch,1 Joachim R. Kalden,1 and Rikard Holmdahl3 Objective. To analyze the fine specificity of IgG autoantibodies in sera from rheumatoid arthritis (RA) patients for type II collagen (CII) epitopes that are arthritogenic in collagen-induced arthritis (CIA), a relevant murine model of RA. Methods. For enzyme-linked immunosorbent assay (ELISA) analysis of conformation-dependent autoantibody binding, recombinant chimeric collagens that harbor the respective CII epitopes as an insertion within the frame of a constant type X collagen triple helix were constructed. In addition, synthetic peptides mimicking the native collagen structures were applied for the first time in the ELISA assessment of humoral CII autoimmunity. Results. The pathogenicity of IgG responses to certain CII determinants in CIA was demonstrated by arthritis development in BALB/c mice upon the combined transfer of 2 mouse monoclonal antibodies specific for precisely mapped conformational CII epitopes (amino acid residues 359–369 [C1III] and 551–564 [J1]), whereas antibodies to another epitope (F4) were not arthritogenic. To test whether human autoimmune responses are similarly directed to these conserved CII determinants, serum IgG was analyzed. The prevalence of sera with increased IgG binding to the C1III epitope was significantly higher in RA compared with sera from healthy donors or from patients with other rheumatic conditions, e.g., osteoarthritis (OA), systemic lupus erythematosus (SLE), or relapsing polychondritis (RP), whereas levels of antibodies specific for the nonarthritogenic F4 epitope were associated with OA rather than RA. Conclusion. Autoimmunity to CII, although detectable in different rheumatic conditions, differs in fine specificity between distinct disease entities. In RA, in contrast to degenerative joint disease, RP, and SLE, autoantibody responses are directed to an evolutionary conserved CII structure that is also targeted by pathogenic autoimmune responses in murine models of arthritis. Supported by grants from the Deutsche Forschungsgemeinschaft (SFB 263, project C3), the Bundesministerium für Bildung und Forschung (project C 2.1, BMBF 01GI9948), the King Gustaf V 80-Year Foundation, the Swedish Association Against Rheumatism, the Swedish Medical Research Council, and the European Commission (Bio4-98-0479). 1 Harald Burkhardt, MD, Tobias Koller, MD, Hans G. Kraetsch, MD, Joachim R. Kalden, MD: Friedrich-AlexanderUniversity of Erlangen–Nuremberg, Erlangen, Germany; 2Åke Engström, PhD: Uppsala University, Uppsala, Sweden; 3Kutty Selva Nandakumar, PhD, Rikard Holmdahl, MD, PhD: Lund University, Lund, Sweden; 4Javier Turnay, PhD: Universidad Complutense, Madrid, Spain. Address correspondence and reprint requests to Harald Burkhardt, MD, Department of Internal Medicine III, FriedrichAlexander-University of Erlangen–Nuremberg, Krankenhausstrasse 12, 91054 Erlangen, Germany. E-mail: firstname.lastname@example.org. uni-erlangen.de. Submitted for publication January 14, 2002; accepted in revised form May 3, 2002. Cartilage-specific molecules such as type II collagen (CII) have long been suspected to be targets of pathogenic autoimmune responses in the induction or perpetuation of joint inflammation in rheumatoid arthritis (RA). The hypothesis of CII-directed autoimmunity as a relevant pathogenic mechanism in RA is an appealing explanation for the chronicity of the arthritic process as well as for its fatal consequences for the integrity of joint cartilage. Several lines of evidence further support the concept of a CII-driven autoimmune process in RA. Direct evidence for the arthritogenic potential of CII autoimmunity is derived from experimental disease models of collagen-induced arthritis (CIA) in rodents as well as in nonhuman primates (1). In these models, the arthritis is initiated by complement-fixing autoantibodies 2339 2340 BURKHARDT ET AL that bind to CII in the cartilage matrix, and the formation of these antibodies upon immunization is major histocompatibility complex–restricted and T cell– dependent (2). In humans, anti–CII IgG in rheumatoid cartilage and synovium and circulating autoantibodies to native and denatured CII have been detected in the sera of RA patients (3–6). Moreover, anti–CII IgG– producing B cells have been detected in rheumatoid synovium and synovial fluid (7,8), suggesting an intraarticular antigen-driven immune process. In the present study, we focused on the question of whether the epitope specificity of anti-CII autoantibody formation in RA differs from that in other rheumatic diseases, such as osteoarthritis (OA), relapsing polychondritis (RP), and systemic lupus erythematosus (SLE). Recently, a systematic analysis of conformationdependent autoantibody binding in the murine CIA model in DBA/1 mice revealed a discrete recognition of sterically privileged sites on collagen fibrils, suggesting the possible importance of epitope accessibility in the cartilage matrix for the formation of arthritogenic immune complexes (9). Since recognition of the identified immunodominant domains is associated with arthritis development in CIA and since the CII domains are conserved in their amino acid sequences in mouse, rat, and human collagen, we have selected these epitopes for an analysis of the fine specificity of circulating human anti–CII IgG autoantibodies. PATIENTS AND METHODS Arthritis transfer experiments. According to their previously characterized distinct specificities for conformational epitopes on CII (9), 3 monoclonal antibodies (mAb) were selected for antibody transfer experiments of experimental arthritis. Whereas mAb CII-C1 (␥2a) recognizes the triplehelical domain amino acids (aa) 359–369 of CII, mAb M2.139 (␥2b) binds to aa 551–564, and mAb CII-F4 (␥2a) binds to aa 926–936 of triple-helical CII. The mAb were purified from culture supernatants by affinity chromatography on protein G (GammaBind Plus; Pharmacia, Uppsala, Sweden). BALB/c mice (The Jackson Laboratory, Bar Harbor, ME) were bred and used in the Medical Inflammation Unit animal housing facility in a specific pathogen–free environment as defined at http://net.inflam.lu.se. Groups of mice (n ⫽ 7) were injected with a single mAb or with 2 (C1 ⫹ M2.139) or 3 (C1 ⫹ M2.139 ⫹ F4) mAb cocktails intravenously on day 0. Control mice received phosphate buffered saline (PBS) only. On day 5, all mice were injected intraperitoneally with lipopolysaccharide (50 g/mouse) from Escherichia coli serotype O55:B5 (Sigma, St. Louis, MO). Arthritis in each paw was scored daily on a scale of 0 (low) to 15 (high) for a minimum of 21 days after antibody transfer. Study populations. A pilot study on the specific recognition of 5 conformational CII epitopes, originally character- ized in murine CIA by human serum IgG, was performed using an enzyme-linked immunosorbent assay (ELISA) with recombinant chimeric collagen molecules (explained below). The study population consisted of 22 patients with an established diagnosis of RA (mean ⫾ SD age 56.3 ⫾ 10.5 years), as defined by the 1987 revised classification criteria of the American College of Rheumatology (ACR; formerly, the American Rheumatism Association) (10), and 22 patients with OA (mean ⫾ SD age 64 ⫾ 10.4 years) according to the ACR classification criteria for OA of the hip or knee (11,12). In an extended study, the specific autoantibody response to selected CII epitopes (C1III and J1) was subsequently investigated in patients with 4 distinct rheumatic diseases: RA (n ⫽ 48, mean ⫾ SD age 49.9 ⫾ 12.3 years, mean ⫾ SD disease duration 10.0 ⫾ 6.9 years, mean ⫾ SD erythrocyte sedimentation rate [ESR] 41.8 ⫾ 20.1 mm/hour), OA (n ⫽ 22, age 68.7 ⫾ 11.0 years, ESR 22.2 ⫾ 8.1 mm/hour), RP (n ⫽ 26, age 48.8 ⫾ 11.9 years, disease duration 7.2 ⫾ 5.0 years, ESR 45.2 ⫾ 44.9 mm/hour), and SLE (n ⫽ 38, age 42.3 ⫾ 13.3 years, disease duration 9.32 ⫾ 5.78 years, ESR 25.2 ⫾ 19.81 mm/ hour). A modified ELISA technique using synthetic triplehelical peptides as antigens (see below) was applied, and all OA and RA patients were newly recruited to exclude any bias by preselection of CII-reactive sera from the initial study. Inclusion into the new RA and OA cohorts also required the fulfillment of the above-mentioned ACR classification criteria. The diagnosis of RP was based on the criteria of Michet et al (13), and the diagnosis of SLE was based on the ACR 1982 revised criteria (14). In none of the cohorts was any preselection of patient sera made on the basis of a pretest for circulating anti–CII IgG or certain clinical characteristics, e.g., disease severity or duration. The study protocol was approved by the review board of the Friedrich-Alexander-University Erlangen–Nuremberg, and informed consent was obtained from all individuals before entering the study. Recombinant collagen constructs. The vector for expression of chimeric collagen molecules has been described previously and is based on a human type X collagen complementary DNA (cDNA) clone (15). At the position that corresponds to the amino-terminal end of the secreted protein, a sequence encoding a 6xHis-tag was introduced to facilitate purification of the recombinant molecules from supernatants of transfected cells. A human CII cDNA clone (kindly provided by Dr. S. W. Li, Thomas Jefferson University, Philadelphia, PA) was used as a template for polymerase chain reaction (PCR) amplification of the respective CII fragments. Specific restriction sites (sites Pst I/Bam HI), which allowed cloning of these fragments, were introduced via PCR. The CII fragments were ligated in-frame with type X collagen cDNA into the Pst I/Bam HI–restricted pUC 57 vector (Fermentas, St. Leon-Rot, Germany), resulting in the corresponding chimeric CX–CII constructs that were finally cloned in pRCMV (Promega, Madison, WI). Prior to the transfection procedure, the chimeric constructs were controlled by in vitro translation and DNA sequencing. The following CII fragments encoding frequently recognized B cell epitopes in murine CIA were inserted into the CX frame for expression of chimeric collagens (Figure 1): C1III (1623–1655 bp, aa 359–369), J1 (2199–2240 bp, aa 551–564), D3 (2607– 2642 bp, aa 687–698), E/F10II (2877–2897 bp, aa 777–783), F4 SHARED B CELL EPITOPES IN CIA AND RA Figure 1. Applied recombinant chimeric collagen constructs. The constant type X collagen frame is shown in the shaded area. The homotrimeric collagenous portion is flanked by noncollagenous (NC) globular domains at the amino-terminal (NC2) and at the carboxyterminal (NC1). The amino acid sequences of the short type II collagen (CII) cassettes representing the epitopes under investigation are depicted in single-letter code. Boldface letters highlight amino acid residues (aa) that have already been demonstrated to be critical for binding of murine monoclonal antibodies (9). *The italic letter E, for glutamic acid, in the control construct indicates the single amino acid mismatch by which it differs from the J1 construct. (3326–3356 bp, aa 926–936). In addition, a recombinant collagen molecule was constructed as a control for nonspecific IgG binding in ELISA by the introduction of a point mutation into the chimeric construct for the J1 epitope (Figure 1). The mutation led to a nonconservative amino acid exchange (R ⬎ E) in the CII epitope J1 (MPGERGAAGIAGPK), rendering it a non–naturally occurring collagen sequence within the frame of wild-type CX. The mutation destroyed the binding of J1-specific murine mAb completely and was not recognized by any of the murine CII-specific mAb tested. For the expression of the collagen chimeras, HEK 293 cells were transfected with plasmid DNA constructs using the calcium phosphate precipitation method described previously (9). The transfected cells were grown in Dulbecco’s modified Eagle’s medium (DMEM)/Ham’s F-12, 5% fetal calf serum (FCS). After 48 hours, the selection was started by supplementation with 800 g/ml G-418 (Gibco, Paisley, UK). Medium was renewed every 2 days, and collection of supernatants was started when the G-418–resistant cells reached confluency. During harvesting, the transfected HEK 293 cells were kept in FCS-free DMEM/Ham’s F-12 supplemented with ascorbate. The 6xHis-tagged recombinant collagens in the culture super- 2341 natants were affinity purified on Ni-NTA Superflow agarose (Qiagen, Hilden, Germany). The purity of the preparations was controlled by sodium dodecyl sulfate–polyacrylamide gel electrophoresis using 10% gels. The proper folding of the recombinant collagen chimeras was tested in ELISA using mouse mAb that bind in a strict conformation-dependent manner to the respective CII inserts. ELISA. The collagen molecules required for the investigation of antibody specificities were extracted from cartilage preparations (entire CII, see ref. 9), purified as recombinant proteins from supernatants of transfected cells, or synthesized as triple-helical polypeptides (as described below). Microtiter plates (Nunc, Wiesbaden, Germany) were coated overnight with purified collagens at 4°C and blocked with 2% bovine serum albumin in PBS for 1 hour. Standardized coating efficiencies of purified CII and of the recombinant chimeric collagens were controlled by the immunoreactivity of mouse mAb specific for distinct conformational epitopes on CII (9). The following mAb were applied for the standardization procedures: CII-C1 (epitope specificity C1III), M2.139 (epitope specificity J1), LN2 173 (epitope specificity D3), CII-E10 (epitope specificity E/F10II), CII-F4 (epitope specificity F4), and the CX-specific mAb X53 (16). The CII specificities of the human antibodies were determined by adding the sera diluted in PBS to the collagencoated microtiter wells for 1 hour at room temperature. Antibody binding was detected using horseradish peroxidase– conjugated rabbit anti-human IgG (Dianova, Hamburg, Germany) and ABTS as substrate (Boehringer, Mannheim, Germany). A quantitative analysis of specific IgG binding to the CII epitopes embedded in the recombinant chimeric collagens required a normalization of the absolute values of absorbance for antibody interactions with the large CX domain of the different constructs. In parallel with the determination of IgG binding to recombinant chimeric collagens containing the different CII epitopes under investigation, absorbance was also assessed in microtiter wells coated with the mutated J1 control construct (Figure 1). Since the control construct shared the CX frame with all other constructs (accounting for ⬎90% of the entire sequence), it was used to measure IgG binding to the constant part, not specific for a particular CII insert in each sample. The ratio of absorbanceepitope/absorbancecontrol-construct was determined in triplicate, and the mean value (arbitrary units [AU]) was used as a parameter of epitope-specific binding for the statistical analysis. Synthesis of triple-helical collagen peptides. Triplehelical collagen peptides were synthesized according to the method described by Grab et al (17). The sequence of the 2 triple-helical peptides containing either the C1III or the J1 epitope is shown in Figure 6. A triple C-terminal anchor was obtained by assembling the peptide NH2-Lys(Dde)-Lys(Dde)Tyr(tBu)-Gly on Wang resin (p-benzyloxybenzyl alcohol resin). After removing the Dde protecting groups by treating the peptide resin 3 times with 2% hydrazine in N,Ndimethylformamide for 3 minutes, each of the 3 strands in the triple helix was assembled in parallel using the 3 free amino groups on the anchor peptide still on the synthesis resin. The peptide resin was treated with acid to release the 3-stranded peptides from the resin and to remove the side chain protecting groups. Following precipitation, the peptide was washed in 2342 BURKHARDT ET AL Figure 2. a, Arthritis induction in 4-month-old BALB/c mice by intravenous injection of monoclonal antibody (mAb) cocktail. Lipopolysaccharide was given intraperitoneally as a secondary immune stimulus. Typical clinical signs of joint inflammation in the mAb (M2.139 ⫹ CII-C1)–injected mouse (right) compared with the phosphate buffered saline–injected control (left) are shown. b, Inhibition of mAb-induced arthritis by the CII-F4 mAb. Different concentrations of 2 or 3 mAb cocktails were given, and arthritis was scored daily for 3 weeks. Each group consisted of 7 mice. Values are the mean area under the arthritis curve (AUC). diethyl ether. The 3-stranded peptides were used without further purification. Matrix-assisted laser desorption ionization–time-offlight (MALDI-TOF) mass spectrometry analysis was performed on tryptic peptide fragments. The expected mass-tocharge ratio of the fragments was found in the mass spectra. Peptide synthesis was performed in an ABI 430 peptide synthesizer (Applied Biosystems, Foster City, CA) using 9-fluorenylmethoxycarbonyl (FMOC) chemistry with activation of amino acid derivatives with HBTU. For MALDI-TOF analysis, a Kratos Kompakt IV mass spectrometer was used (Kratos Analytical, Manchester, UK). Amino acid derivatives were obtained from Alexis Biochemicals (Grünberg, Germany) or Novabiochem (Laufelfingen, Switzerland). FMOCGly-Wang resin and FMOC-Lys(-1[4,4-dimethyl-2,6dioxocyclohexylidene]-1-ethyl) were purchased from Novabiochem. Circular dichroism measurements. Far-ultraviolet circular dichroism spectra were recorded between 180 and 250 nm in a J-715 spectropolarimeter (Jasco, Easton, MD) equipped with an RTE-111 thermostat (Thermo Neslab, Portsmouth, NH) using 0.01-cm optical pathlength thermostatized cuvettes and a peptide concentration of 0.07 mM in PBS. The spectra of triple-helical peptides dissolved in 0.1M acetic acid were almost identical to those obtained in PBS. Ellipticity is expressed as the mean residue weight molar ellipticity (MRW) in degrees ⫻ cm2 ⫻ dmole⫺1 units. Statistical analysis. For the evaluation of statistical differences in immunoreactivity and clinical parameters between the different cohorts (RA, OA, RP, SLE, and healthy subjects), analysis of variance (ANOVA) and the MannWhitney U-test were applied. RESULTS Transfer of arthritis by mouse mAb with specificity for conformational epitopes on CII. To assess the arthritogenicity of mAb that bind to 3 well-characterized conformational CII determinants frequently targeted by autoantibody responses in murine CIA, antibody transfer experiments were performed by systemic administration of mAb with the respective epitope specificities C1, J1, and F4. Whereas the intravenous injection of any single mAb at doses of up to 4.5 mg did not result in arthritis development in recipient BALB/c mice (results not shown) as in PBS-injected controls, the combined application of CII-C1 and CII-J1 antibodies, however, led to severe joint inflammation (Figure 2a). A pronounced amelioration of arthritis severity was detectable when CII-F4 antibodies were added to the cocktail of CII-specific mAb. The resulting reduction of the area under the curve of arthritis scores (Figure 2b), was maintained even at high doses of the antibody cocktail (18 mg) applied in the transfer experiments. Thus, a pure dilution of the arthritogenic CII-C1/J1 combination SHARED B CELL EPITOPES IN CIA AND RA in the triple mixture cannot account for the ameliorating effect of the CII-F4 mAb. Accordingly, the results suggest, in addition to the arthritis-promoting function of the CII-C1 and CII-J1 antibodies, a protective role of the CII-F4 antibody in the model of anti-CII antibody– induced arthritis. Analysis of the fine specificity of circulating anti–CII IgG autoantibodies with recombinant chimeric collagen molecules. To investigate whether circulating human anti-CII antibodies share specificities with murine IgG autoantibodies that arise early in experimental arthritis (9) and are capable of transferring disease to naive recipient mice (Figure 2), a systematic ELISA study was performed on patients’ sera. Recombinant chimeric collagen molecules that harbor the relevant CII domains C1III, J1, D3, E/F10II, and F4 in a native conformation were used as antigens for the ELISA analysis. In preliminary titration experiments with some selected patient sera, concentration-dependent specific IgG binding to the C1III epitope in comparison with the recombinant control construct could be demonstrated at antibody dilutions of up to 1:480 (data not shown). For a comparative study of CII epitope–specific IgG in OA and RA sera, all assays were performed at a standard dilution of 1:60, and the results were expressed as the ratio of absorbanceepitope/absorbancecontrol-construct for the statistical analysis as described. Two chimeric constructs containing the J1 and E/F10II epitopes were not recognized to any significant extent in any of the sera tested (Figure 3). Since IgG binding to the J1 and E/F10II constructs did not considerably differ from the respective controls, resulting in AU values that rarely exceeded 1.2, immunoreactivity to these constructs was regarded as negative and was used for the definition of a threshold (in AU) above which an ELISA result with one of the other epitopes (C1III, F4, D3) was considered positive. Accordingly, this threshold is defined as the mean ⫹ 2SD of all J1- and E/F10IIspecific ELISA results (in AU) and is equivalent to 1.2. It is indicated by the horizontal line in Figures 3a–c. The cutoff value for negative immunoreactivity was further validated by analysis of sera from 10 age-matched control patients without inflammatory or degenerative joint diseases (7 women, 3 men, mean ⫾ SD age 61 ⫾ 12 years) without circulating anti–CII IgG. The range of absorbance obtained with these 10 control sera for binding to either the control construct or the different epitopes (C1III, J1, E/F10II, F4, D3) was between 0.07 and 0.10 (not shown), resulting in AU values below the threshold of 1.2 (mean 1.0, range 0.7–1.2) in all epitopespecific ELISAs. Autoantibody binding to the D3 epitope was 2343 Figure 3. Differences in fine specificity of type II collagen (CII)– directed immune responses between rheumatoid arthritis (RA) and osteoarthritis (OA). The scatterplots show the results in sera from patients with OA and RA as assessed with an epitope-specific enzymelinked immunosorbent assay (ELISA) using recombinant chimeric collagens for the detection of CII autoantibodies. Arbitrary units (AU) are given as a quantitative measure of the ELISA results. a, IgG binding to the C1III epitope (amino acids 359–369) was increased in the RA compared with the OA population, indicating an RA-specific immunorecognition of this CII domain. b, Autoantibody formation with specificity for the F4 epitope was detectable at an elevated level in OA sera compared with RA (P ⬍ 0.02). c, Immunoreactivity to the D3 epitope (amino acids 687–698), the J1 epitope (amino acids 551–564), and the E10/F10 epitope (amino acids 777–783) does not differ between both cohorts. Horizontal lines show the mean ⫹2SD (in AU) of all J1- and E/F10-specific ELISA results. detectable in RA and OA sera. The results of parallel measurements using epitope-specific ELISAs revealed a statistically significant increase in circulating D3-specific IgG above the threshold for negative immunoreactivity according to the above definition (P ⬍ 0.01 for comparison of all data presented in Figure 3c). However, the levels of D3-specific IgG did not differ between the OA and RA sera, indicating a lack of relationship between circulating IgG with D3 specificity and the underlying disease condition in the joints. The C1III epitope was preferentially recognized by IgG autoantibodies from RA patients. In this respect, statistically significant quantitative differences became apparent in comparison with the ELISA results obtained with the OA-derived antibodies (P ⬍ 0.01) (Figure 3a). In contrast, an increased ELISA reactivity with the F4 epitope was detectable in OA sera (P ⬍ 0.02) (Figure 3b). Thus, the results demonstrate binding of human IgG autoantibodies to target structures on CII (epitopes 2344 C1III, D3, and F4) that are conserved across species barriers. Moreover, the results depicted in Figure 3 pinpoint differences in epitope specificity of circulating CII autoantibodies in the sera of patients, depending on the underlying disease, with a preferential recognition of the C1III epitope in RA. In this respect, it is noteworthy that all 22 OA and RA sera that were included in this initial investigation contained CII-specific IgG as measured under identical ELISA conditions in microtiter plates coated with the entire native CII (data not shown). This initial approach to analysis of the fine specificity of CII-directed autoantibody responses in RA and OA sera was subsequently extended, leading to the inclusion of SLE and RP patients as additional control cohorts in order to test the hypothesis on the RA specificity of C1III epitope–directed autoantibody responses. However, in the extended study, a novel technology was introduced for the assessment of CII epitope–specific IgG that circumvents some disadvantages of the recombinant collagen approach (see Discussion). Assessment of C1III-specific autoantibodies with triple-helical synthetic peptides. Triple-helical peptides containing the C1III or J1 epitope sequences were synthesized using a methodology for solid-phase assembly of branched triple-helical peptides as described previously (17). These epitopes were selected because they had been found to be arthritogenic in BALB/c mice. Three nascent peptide chains were carboxyterminal linked to 1 N␣ amino and 2 N amino groups of lysine residues and stabilized by 7 GPP* repeats. Triplehelical conformation of the synthetic peptides was assessed by circular dichroism spectroscopy, as shown for the C1III peptide in Figure 4a. At low temperatures, the ␣1(II) 359–369 peptide exhibited typical features of a collagen-like conformation, such as a large negative MRW at ⫽ 198 nm and a positive MRW at ⫽ 225 nm. Identical circular dichroism spectroscopic results were obtained for the ␣1(II) 551–564 peptide (J1) (data not shown). Further evidence for the native conformation of the synthetic peptides is provided in Figure 4b, showing their immunorecognition by collagen-specific antibodies in ELISA. The mouse mAb CII-C1 (C1III specific) and M2.139 (J1 specific) recognize the entire CII in a strict conformation-dependent manner, as demonstrated by a lack of binding to heat-denatured collagen. Both antibodies also bound to the specific synthetic triple-helical peptides that contained the sequence of their corresponding epitopes. The titration curves for recognition BURKHARDT ET AL Figure 4. a, Circular dichroism (CD) spectra of the synthetic collagen peptide containing the C1III sequence (amino acids 359–369). Spectra were recorded at 10°C and 85°C. At low temperatures, the ␣1(II) 359–369–containing peptide exhibits typical features of a collagen-like conformation, e.g., a large negative () molar ellipticity at ⫽ 198 nm and a positive at ⫽ 225 nm. The CD characteristics of a triple-helical molecule are lost upon heat denaturation. b, Mouse monoclonal antibodies (mAb) CII-C1 and M2.139 bind in a conformation-dependent manner to type II collagen (CII) and recognize the synthetic collagen mimetics in enzyme-linked immunosorbent assay (ELISA). Binding of the mAb CII-C1 and M2.139 to the branched synthetic collagen peptides ␣1(II) 359–369 and ␣1(II) 551– 564 are shown at different antibody dilutions in ELISA. Identical titration curves obtained with native CII or the respective synthetic peptides indicate that the collagen-mimetic molecules contain a perfect image of the conformational epitopes. In contrast, the mAb do not bind to heat-denatured CII. SHARED B CELL EPITOPES IN CIA AND RA Figure 5. Titration of serum IgG binding to the synthetic C1III collagen peptide (␣1[II] 359–369) in enzyme-linked immunosorbent assay (ELISA). The sera were from 6 rheumatoid arthritis patients (P1–P6). ELISA reactivity (absorbance at 405 nm) could be detected in samples P4–P6, even at higher serum dilutions in phosphate buffered saline, whereas P1–P3 remained negative. of the synthetic triple-helical peptides were identical to those obtained with the entire native CII after purification from cartilage for both mAb. These results provide clear experimental evidence for the adequate representation of native collagen structures by the synthetic collagen mimetic molecules. The results of an ELISA of 6 patient sera using the synthetic C1III epitope are shown in Figure 5. The titration curves of IgG binding show that the peptide is indeed recognized in 3 of the patients, whereas the remaining 3 were negative. Detection of IgG serum autoantibodies with specificity for triple-helical synthetic peptides. Sera from RA (n ⫽ 48), OA (n ⫽ 24), RP (n ⫽ 24), and SLE (n ⫽ 38) patients were assessed at a standard dilution of 1:100 in PBS for CII-directed autoantibodies with specificity for the J1 and C1III epitopes using ELISA plates coated with synthetic triple-helical peptides containing either the C1III epitope or the J1 epitope. Sera from patients with no clinical signs of a rheumatic disease were assessed in parallel with healthy donor control samples (n ⫽ 19, 14 women, 5 men, age mean ⫾ SD 63.1 ⫾ 11.1 years). The prevalence of positive ELISA results (data not shown) for the binding capacity of serum IgG to the 2345 entire native CII differed between the cohorts under investigation: 66.7% in RA (32 positive of 48 tested), 62.5% in OA (15 of 24), 96% in RP (23 of 24), 31.6% in SLE (12 of 38), and 5.3% in healthy subjects (1 of 19). As shown in Figure 6, enhanced binding of IgG autoantibodies to the ␣1(II) 359–369 peptide was detectable in the RA sera. This result on the preferential recognition of the C1III epitope by RA-derived autoantibodies was statistically highly significant compared with data obtained with serum samples from all other cohorts (P ⬍ 0.0001 by ANOVA). However, a slight increase in IgG binding to C1III epitope-coated microtiter wells above the level measured in the OA and SLE sera was also detectable in sera from RP patients. Assays for the J1 epitope showed slightly increased absorbance values with RA and RP sera (P ⬍ 0.001 by ANOVA), but the absolute measures and the quantitative differences in the respective OA and SLE controls remained rather low. A comparison of all ELISA results obtained with the 2 synthetic CII peptides in the different cohorts revealed a selective epitope recognition in association with a given disease entity exclusively for the dominance of C1III-specific IgG in the RA sera. Thus, the investigations with the synthetic collagen mimetics confirmed the initial results obtained with the recombinant chimeric collagens (see above) on fine specificity of antiCII autoantibodies. DISCUSSION In addition to cartilage-specific proteins, ubiquitously expressed proteins have been suggested as candidate autoantigens in the pathogenesis of RA. Thus, strong evidence for a potential role of glucose-6phosphate isomerase (G6PI) has recently been provided by the arthritogenic potential of anti-G6PI IgG in a murine arthritis model and by the detection of elevated anti-G6PI autoantibodies in RA sera and synovial fluids in up to 64% of the samples tested (18). However, RA, as defined by the commonly accepted classification criteria, is a rather heterogenous disease, and several different pathways may result in similar clinical manifestations, as reflected by the animal models that can be induced by distinct provocations. Several lines of evidence suggest a critical role of genetically linked autoimmunity to cartilage-specific CII in arthritic joint destruction. CII-specific B cells in the inflamed synovium and in synovial fluid have been shown to produce IgG autoantibodies that can bind to native collagen structures, indicating an intraarticular antigen-driven immune process in RA (7,8,19). The locally produced 2346 BURKHARDT ET AL Figure 6. Binding of serum IgG to enzyme-linked immunosorbent assay plates coated with synthetic collagen peptides (structure is shown with amino acid residues in one-letter code) that contain either the C1III epitope (␣1[II] 359–369) or the J1 epitope (␣1[II] 551–564). Sera were from patients with rheumatoid arthritis (RA), osteoarthritis (OA), relapsing polychondritis (RP), systemic lupus erythematosus (SLE), and normal healthy donors. Each box represents the 25th to 75th percentiles. Lines outside the boxes represent the 10th and the 90th percentiles. Lines inside the boxes represent the median. P values show significant differences in the absorbance at 405 nm, by Mann-Whitney U test. Absorbance values for binding of circulating IgG to the C1III epitope were significantly elevated in RA patients compared with all other disease entities. In contrast, the absorbance values measured for J1-specific IgG binding in serum samples from all cohorts evaluated remained at the detection limit. Significant epitope-specific differences in IgG binding were only detectable in the RA cohort with the preferential recognition of the C1III-containing peptide (ⴱ). Ahx ⫽ aminohexanoic acid, P* ⫽ hydroxyproline. autoantibodies may be largely consumed in the joint by complex formation with the cartilage matrix (20), thereby giving rise to activation of the complement cascade. Consistent with antibody escape into the circulation, CII-specific IgG is also detectable in the sera of RA patients, although at a lower frequency compared with the joint compartment (21). Therefore, the serum titers of CII-specific autoantibodies might reflect, at least to a certain extent, the vigor of the local immune response. Consequently, several studies have been performed to detect correlations between circulating antiCII antibodies and clinical parameters in RA patients (3–6). However, formation of autoantibody to CII also occurs in other rheumatic diseases, e.g., in RP (22), SLE (5), and degenerative joint disease. Although immunoreactivity to denatured collagen seems to prevail in OA sera, autoantibodies to triple-helical CII are also detect- able (23). It therefore appears rather difficult to decide whether the occurrence of anti-CII autoantibodies plays a role in the pathogenesis of RA or is merely a result of the exposure of destroyed cartilage. More direct experimental evidence of a pathogenic potential has been demonstrated by the ability of human and murine CII-specific autoantibodies to transfer arthritis to a normal mouse (21,24). However, the fine specificity of the arthritogenic anti-CII autoantibodies has remained obscure. In this respect, the arthritis transfer experiments and the ELISA analysis of CII epitope-specific human IgG in the present investigation provide the first experimental evidence for the pathogenicity of autoantibody formation directed to an evolutionary conserved conformational determinant that is located between aa 359–369 of the CII triple helix. Our results revealed that the CII region that is an immunodominant target of arthritogenic murine B cell SHARED B CELL EPITOPES IN CIA AND RA responses in CIA is also recognized by human IgG autoantibodies in RA sera. Moreover, the prevalence of IgG autoantibodies binding to this particular conformational epitope in RA clearly varies from other rheumatic conditions, suggesting disease-related differences in fine specificity of CII-directed autoimmunity. The intriguing hypothesis that differences in B cell epitope specificities might be linked to distinct functional aspects of humoral autoimmune responses is further supported by the predominant IgG binding to the CII region aa 926–936 (F4 epitope) measured in OA compared with RA sera. Thus, a mouse mAb (CII-F4) with specificity for this region exhibited a remarkable disease-ameliorating effect in arthritis transfer experiments in combination with 2 arthritogenic CII-specific mouse mAb. Although the underlying mechanism of the protective effect of mAb CII-F4 is presently unknown, at least 2 hypothetical modes of action can be envisioned. One possibility is the induction of antiidiotypic immunoregulatory responses directed to cross-reactive idiotopes on anti-CII antibodies (25). Alternatively, a steric hindrance of proteolytic degradation of cartilage collagen in the N-telopeptide region close to the crosslinks in the fibril by stromelysin could account for the arthritis-ameliorating effect of the F4 mAb, because the F4 epitope colocalizes in the quaternary collagen structure with the stromelysin cleavage sites on adjacent molecules (9,26). For elucidation of disease-associated qualitative differences in fine specificities of CII-directed humoral autoimmunity, a novel technology was introduced in the present investigation. In this approach, recombinant chimeric collagens were replaced by small synthetic collagen-mimetic peptides that were synthesized according to a protocol described by Grab et al (17) for the assessment of conformation-dependent epitope recognition in ELISA. The assay conditions have the advantage that they do not require additional control for the exclusion of false-positive results that may arise from antibody recognition of the type X collagen frame surrounding the CII epitope inserts in the recombinant chimeric collagens, and it was also found to mimic the antibody recognition of native CII. The application of the new ELISA technology confirmed the role of the C1III epitope as a preferential target of CII-directed IgG autoantibody responses in RA sera, whereas J1-specific IgG remained at the limit of detectability. Remarkably, the levels of C1III-directed autoantibody binding in RA not only greatly exceeded the values determined in OA and SLE sera, but also the ELISA results obtained with the RP samples, although inflammatory cartilage destruction and CII autoimmunity are common features in 2347 both disease conditions. However, the primary target tissue of the destructive autoimmune process in RP is the extraarticular cartilage, e.g., in the ear, nose, and respiratory tract, whereas the joint is only occasionally affected by nonerosive inflammation. In RP, the involvement of extraarticular cartilage is accompanied by autoimmune responses to matrilin 1, a cartilage matrix component that is not expressed in joint cartilage, suggesting its potential importance as the major autoantigen (27). Accordingly, CII autoantibody formation in this condition might arise as a secondary phenomenon during epitope spreading, thus lacking the epitope specificity that is required for the induction of erosive joint inflammation. In rodents and, as shown by the results of our study, in humans as well, autoantibody responses with C1III fine specificity are associated with an inflammatory disease leading to the destruction of hyaline cartilage in the joints. The autoantibodies are directed to a collagen structure that is conserved in evolution; it is localized on the cyanogen bromide fragment 11 of CII that has previously been shown to possess all the structural requirements of an arthritogenic immunogen for the induction of CIA in DBA/1 mice (28,29). The systematic assessment of sera from patients with different rheumatic diseases for immunoreactivity to CII epitopes that were originally characterized in an experimental arthritis model in rodents revealed striking differences in fine specificities of the immune response that usually remain invisible with detection systems that use the entire antigen. Two of the epitopes under investigation (J1 and E/F10II) were not recognized by human autoantibodies despite the conserved amino acid sequence in human CII and were detected as major epitopes in the mouse CIA model. A possible explanation is that sera obtained from CIA mice are obtained during the priming response and during the early acute phase of arthritis, phases of disease that are not possible to address in humans. The focus of this pilot study on well-defined collagen structures relevant to the pathogenesis of murine experimental arthritis implies that the recognition of additional epitopes in human anti-CII autoimmunity during early events of the disease or the spreading to other CII determinants in the course of the disease cannot be excluded. However, the new technical approaches using recombinant collagen chimeras and synthetic triple-helical collagen mimetics also offer promising strategies for future prospective investigations of larger patient populations to answer these yetunresolved questions of CII-directed autoimmunity in 2348 BURKHARDT ET AL the pathogenesis of inflammatory joint disease. Therefore, this first comparative study on epitope specificity of circulating anti-CII autoantibodies in different rheumatic diseases demonstrates the potential of the approach to uncover hidden associations between clinical conditions and immunorecognition that may relate to the underlying pathogenic mechanisms and should encourage further work on the fine specificity of jointspecific autoimmunity. ACKNOWLEDGMENTS The authors are grateful to Eva Bauer and Carlos Palestro for excellent technical assistance. REFERENCES 1. Holmdahl R, Andersson M, Goldschmidt TJ, Gustafsson K, Jansson L, Mo JA. Type II collagen autoimmunity in animals and provocations leading to arthritis. Immunol Rev 1990;118:193–232. 2. Holmdahl R, Vingsbo C, Mo JA, Michaelsson E, Malmstrüm V, Jansson L, et al. Chronicity of tissue-specific experimental autoimmune disease: a role for B cells? Immunol Rev 1995;144: 109–35. 3. Terato K, Shimozuru Y, Katayama K, Takemitsu Y, Yamashita I, Miyatsu M, et al. Specificity of antibodies to type II collagen in rheumatoid arthritis. Arthritis Rheum 1990;33:1493–500. 4. Morgan K, Clague RB, Collins I, Ayad S, Phinn SD, Holt PJ. Incidence of antibodies to native and denatured cartilage collagens (types II, IX, and XI) and to type I collagen in rheumatoid arthritis. Ann Rheum Dis 1987;46:902–7. 5. Rowley MJ, Mackay IR, Brand CA, Bateman JF, Chan D. Epitope specificity of antibodies to type II collagen in rheumatoid arthritis and systemic lupus erythematosus. Rheumatol Int 1992;12:65–9. 6. Terato K, DeArmey DA, Ye XJ, Griffiths MM, Cremer MA. The mechanism of autoantibody formation to cartilage in rheumatoid arthritis: possible cross-reaction of antibodies to dietary collagens with autologous type II collagen. Clin Immunol Immunopathol 1996;79:142–54. 7. Tarkowski A, Klareskog L, Carlsten H, Herberts P, Koopman WJ. Secretion of antibodies to types I and II collagen by synovial tissue cells in patients with rheumatoid arthritis. Arthritis Rheum 1989; 32:1087–92. 8. Rudolphi U, Rzepka R, Batsford S, Kaufmann SH, von der Mark K, Peter HH, et al. The B cell repertoire of patients with rheumatoid arthritis. II. Increased frequencies of IgG⫹ and IgA⫹ B cells specific for mycobacterial heat-shock protein 60 or human type II collagen in synovial fluid and tissue. Arthritis Rheum 1997;40:1409–19. 9. Schulte S, Unger C, Mo JA, Wendler O, Bauer E, Frischholz S, et al. Arthritis-related B cell epitopes in collagen II are conformation-dependent and sterically privileged in accessible sites of cartilage collagen fibrils. J Biol Chem 1998;273:1551–61. 10. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, Cooper NS, et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 1988;31:315–24. 11. Altman R, Alarcon G, Appelrouth D, Bloch D, Borenstein D, Brandt K, et al. The American College of Rheumatology criteria for the classification and reporting of osteoarthritis of the hip. Arthritis Rheum 1991;34:505–14. 12. Altman R, Asch E, Bloch D, Bole G, Borenstein D, Brandt K, et al. Development of criteria for the classification and reporting of osteoarthritis: classification of osteoarthritis of the knee. Arthritis Rheum 1986;29:1039–49. 13. Michet CJ Jr, McKenna CH, Luthra HS, O’Fallon WM. Relapsing polychondritis: survival and predictive role of early disease manifestations. Ann Intern Med 1986;104:74–8. 14. Hochberg MC. Updating the American College of Rheumatology revised criteria for the classification of systemic lupus erythematosus [letter]. Arthritis Rheum 1997;40:1725. 15. Reichenberger E, Beier F, LuValle P, Olsen BR, von der Mark K, Bertling WM. Genomic organization and full-length cDNA sequence of human collagen X. FEBS Lett 1992;311:305–10. 16. Girkontaite I, Frischholz S, Lammi P, Wagner K, Swoboda B, Aigner T, et al. Immunolocalization of type X collagen in normal fetal and adult osteoarthritic cartilage with monoclonal antibodies. Matrix Biol 1996;15:231–8. 17. Grab B, Miles AJ, Furcht LT, Fields GB. Promotion of fibroblast adhesion by triple-helical peptide models of type I collagenderived sequences. J Biol Chem 1996;271:12234–40. 18. Schaller M, Burton DR, Ditzel HJ. Autoantibodies to GPI in rheumatoid arthritis: linkage between an animal model and human disease. Nature Immunol 2001;2:746–53. 19. Rönnelid J, Lysholm J, Engstrüm-Laurent A, Klareskog L, Heyman B. Local anti–type II collagen antibody production in rheumatoid arthritis synovial fluid: evidence for an HLA–DR4–restricted IgG response. Arthritis Rheum 1994;37: 1023–9. 20. Jasin HE. Autoantibody specificities of immune complexes sequestered in articular cartilage of patients with rheumatoid arthritis and osteoarthritis. Arthritis Rheum 1985;28:241–8. 21. Wooley PH, Luthra HS, Singh SK, Huse AR, Stuart JM, David CS. Passive transfer of arthritis to mice by injection of human anti-type II collagen antibody. Mayo Clin Proc 1984;59:737–43. 22. Foidart JM, Abe S, Martin GR, Zizic TM, Barnett EV, Lawley TJ, et al. Antibodies to type II collagen in relapsing polychondritis. N Engl J Med 1978;299:1203–7. 23. Charriere G, Hartmann DJ, Vignon E, Ronziere MC, Herbage D, Ville G. Antibodies to types I, II, IX, and XI collagen in the serum of patients with rheumatic diseases. Arthritis Rheum 1988;31: 325–32. 24. Terato K, Hasty KA, Reife RA, Cremer MA, Kang AH, Stuart JM. Induction of arthritis with monoclonal antibodies to collagen. J Immunol 1992;148:2103–8. 25. Nordling C, Kleinau S, Klareskog L. Down-regulation of a spontaneous arthritis in male DBA/1 mice after administration of monoclonal anti-idiotypic antibodies to a cross-reactive idiotope on anti-collagen antibodies. Immunology 1992;77:144–6. 26. Wu JJ, Lark MW, Chun LE, Eyre DR. Sites of stromelysin cleavage in collagen types II, IX, X, and XI of cartilage. J Biol Chem 1991;266:5625–8. 27. Hansson AS, Heinegård, D, Piette JC, Burkhardt H, Holmdahl R. The occurrence of autoantibodies to matrilin 1 reflects a tissuespecific response to cartilage of the respiratory tract in patients with relapsing polychondritis. Arthritis Rheum 2001;44:2402–12. 28. Terato K, Hasty KA, Cremer MA, Stuart JM, Townes AS, Kang AH. Collagen-induced arthritis in mice: localization of an arthritogenic determinant to a fragment of the type II collagen molecule. J Exp Med 1985;162:637–46. 29. Malmstrüm V, Michaelsson E, Burkhardt H, Mattsson R, Vuorio E, Holmdahl R. Systemic versus cartilage-specific expression of a type II collagen-specific T-cell epitope determines the level of tolerance and susceptibility to arthritis. Proc Natl Acad Sci U S A 1996;93:4480–5.