Identification of citrullinated rheumatoid arthritis-specific epitopes in natural filaggrin relevant for antifilaggrin autoantibody detection by line immunoassay.код для вставкиСкачать
ARTHRITIS & RHEUMATISM Vol. 46, No. 5, May 2002, pp 1185–1195 DOI 10.1002/art.10229 © 2002, American College of Rheumatology Identification of Citrullinated Rheumatoid Arthritis–Specific Epitopes in Natural Filaggrin Relevant for Antifilaggrin Autoantibody Detection by Line Immunoassay Ann Union,1 Lydie Meheus,1 René Louis Humbel,2 Karsten Conrad,3 Guenter Steiner,4 Henri Moereels,1 Hans Pottel,1 Guy Serre,5 and Filip De Keyser6 Objective. To identify immunodominant epitopes in natural filaggrin that are reactive with antifilaggrin autoantibodies (AFA) in the sera of patients with rheumatoid arthritis (RA) and to explore their use in a diagnostic assay format. Methods. Based on the results of epitope mapping of human natural filaggrin as well as molecular modeling and computational chemistry, synthetic peptides together with recombinant citrullinated filaggrin were evaluated by a line immunoassay (LIA) for AFA detection. Diagnostic performance was assessed using 336 RA and 253 disease control sera and was compared with that of reference methods. Results. Several immunoreactive epitopes were identified in natural filaggrin, all of which contained at least 1 citrulline residue. Three antigenic substrates, including 2 synthetic peptides and recombinant citrullinated filaggrin showing maximal reactivity on LIA, were finally selected. Using the 3-antigen LIA3, overall sensitivity, specificity, and positive predictive value for RA were 65.2%, 98.0%, and 89.1%, respectively, compared with 61.9%, 98.8%, and 92.8% using the 2-antigen LIA2 (without recombinant protein). Thirty-seven per- cent of the rheumatoid factor (RF)–negative RA samples (30 of 81) were AFA-positive by LIA2, and 52 of 54 RF-positive control samples had no AFA detected on LIA2. Higher specificity and sensitivity were obtained by LIA2 versus anti-RA33 immunoblot, whereas good agreement was observed with antikeratin antibody testing. LIA performed significantly better than AFA immunoblotting using natural filaggrin, at a specificity level of 99% (P ⴝ 0.0047). Conclusion. Citrullinated residues are present in immunoreactive epitopes of natural human filaggrin. AFA can be readily detected by citrullinated peptides in an LIA-based test, resulting in high specificity and positive predictive value for RA. The LIA could serve as a user-friendly alternative to existing immunofluorescence tests and AFA immunoblot techniques. Given its complementarity to RF, this test can be a valuable tool in the differential diagnosis of arthritis. Rheumatoid arthritis (RA) is a major crippling systemic joint disease of unknown etiology that affects ⬃1% of the population. It has many features of an autoimmune disease, including the presence of a variety of autoantibodies in patients’ sera and, in animal models, the capacity to induce illness by transfer of pathogenic T cells (1,2). Inflammation of the joints—the hallmark of RA—is, however, not restricted to this disease but also occurs in osteoarthritis (OA), reactive arthritis, and gout. Because early diagnosis and prompt treatment of RA markedly improve outcome, using reliable diagnostic tools to rule out overlapping disorders is of major importance. However, clinical features characteristic of established RA are often lacking in the early stages of disease. Serologic support for the diagnosis, on the other hand, is limited and is based mainly on the presence of rheumatoid factor (RF). The rather Supported by grant 960267 from the Vlaams Instituut voor de Bevordering van het Wetenschappelijk-Technologisch Onderzoek in de Industrie. 1 Ann Union, PhD, Lydie Meheus, PhD, Henri Moereels, Hans Pottel, PhD: Innogenetics NV, Ghent, Belgium; 2René Louis Humbel: Centre Hospitalier de Luxembourg, Luxembourg; 3Karsten Conrad, MD: Institute of Immunology, Technical University Dresden, Dresden, Germany; 4Guenter Steiner, PhD: University of Vienna, Vienna, Austria; 5Guy Serre, MD, PhD: University of Toulouse, Toulouse, France; 6Filip De Keyser, MD, PhD: University Hospital Ghent, Ghent, Belgium. Address correspondence and reprint requests to Ann Union, PhD, Immune Diseases Group, Innogenetics NV, Industriepark Zwijnaarde 7/4, 9052 Ghent, Belgium. E-mail: firstname.lastname@example.org. Submitted for publication July 31, 2001; accepted in revised form December 19, 2001. 1185 1186 UNION ET AL Table 1. Overview of serum panels* Serum panel No. of RA samples A B C 107 127 336 D 50 No. of disease controls No. of healthy controls 61 SLE, 62 OA 84 SLE/SpA/OA 90 SLE/MCTD, 53 SSc, 50 SpA, 30 OA, 15 Sjögren’s syndrome, 15 myositis 25 non-RA inflammatory joint pain 51 71 Origin Application Ghent Ghent Luxembourg, Dresden, Vienna Epitope mapping of natural filaggrin Optimization of INNO-LIA RA Validation of INNO-LIA RA Ghent Comparison of AFA blot with LIA * RA ⫽ rheumatoid arthritis; SLE ⫽ systemic lupus erythematosus; OA ⫽ osteoarthritis; SpA ⫽ spondylarthropathy; INNO-LIA ⫽ Innogenetics line immunoassay; MCTD ⫽ mixed connective tissue disease; SSc ⫽ systemic scleroderma; AFA ⫽ antifilaggrin autoantibodies. low specificity of RF and the substantial number of RF-negative patients in whom chronic destructive disease subsequently develops (3–5) point to a clear need for RA-specific markers. During recent years, efforts to identify novel RA-specific autoantigens have led to the characterization of a group of antifilaggrin autoantibodies (AFA) that recognize human epidermal filaggrin, rat esophagus epithelial deiminated proteins, as well as buccal mucosa cell–derived profilaggrin (6–10). The high specificity of these antibodies for RA makes them valuable diagnostic markers. In this respect, the usefulness of natural filaggrin in Western blotting has been demonstrated (9), and its use in an enzyme-linked immunosorbent assay (ELISA) also yielded positive results (11). Because isolation of antigen from natural sources is inconvenient, alternative antigen forms are required. Substitution with the modified arginine residue citrulline (Cit) has recently been shown to be an indispensable peptide modification and one that is necessary for proper immune recognition of filaggrin by the autoantibodies present in human RA sera (10). In vitro deimination of recombinant human filaggrin by the peptidylarginine deiminase (PAD) enzyme generates a protein that is reactive with human RA sera during immunoblotting, while the noncitrullinated protein remains unreactive. Furthermore, 2 synthetic peptides sharing a central tripeptide (Ser-Cit-His) showed reactivity to a small series of RA sera in an ELISA (10). Several other citrullinated peptides have also been evaluated by ELISA and showed different patterns of reactivity with RA sera (12,13). In order to compare these non-native antigenic substrates with naturally occurring forms, we investigated the presence of citrulline-containing epitopes in natural filaggrin and evaluated the diagnostic usefulness of naturally derived and synthetic citrullinated peptides in the line immunoassay (LIA) system. MATERIALS AND METHODS Serum samples. Different panels of human sera were used throughout this study (Table 1) and were collected in accordance with local ethics legislation. The epitope mapping studies used 107 serum samples from patients who fulfilled the American College of Rheumatology (ACR; formerly, the American Rheumatism Association) criteria for RA (14) and 174 control sera, derived from 51 healthy persons, 61 with systemic lupus erythematosus (SLE), and 62 with OA. Optimization of the INNO-LIA RA (Innogenetics, Ghent, Belgium) was used to test 127 sera from patients with established RA and 134 sera derived from 84 disease controls (patients with SLE, spondylarthropathy [SpA], or OA) and 71 healthy controls. In addition, 26 samples from patients with Crohn’s disease and 24 from patients with gout were tested. At a later stage, validation of the INNO-LIA RA was performed using 336 sera from patients with RA fulfilling the ACR criteria and 253 sera from disease controls (90 from patients with SLE or mixed connective tissue disease, 53 from scleroderma patients, 50 from SpA patients, 30 from OA patients, 15 from patients with Sjögren’s syndrome, and 15 from patients with myositis), all of which were obtained from the Centre Hopitalier de Luxembourg, the Division of Rheumatology in Vienna, Austria, and the University Hospital Carl Gustav Carus in Dresden, Germany. In addition, a group of 75 sera were tested for AFA in a blinded manner, using the standard immunoblotting reference method, at the Hopital Purpan, Toulouse, France. These sera included 25 samples from patients with RA fulfilling the ACR criteria in whom disease duration was ⬍1 year, 25 samples from RA patients with disease duration ⬎4 years, and 25 from patients who were examined for inflammatory joint pain but in whom RA could be excluded. Preparation of human filaggrin. Using the protocol described by Simon et al (6), natural filaggrin was purified from human skin that was obtained anonymously after abdominoplasty, in accordance with local ethics regulations. The epidermis was separated from the dermis by incubating the skin fragments at 56°C in phosphate buffered saline (PBS) containing 5 mM EDTA. The material was stored dry at ⫺20°C until used. The epidermis was cut into small pieces that were homogenized in 40 mM Tris HCl (pH 7.4), 150 mM NaCl, 5 mM EDTA, 0.5% Nonidet P40, 0.1% sodium azide, and 0.1 CITRULLINATED PEPTIDES FOR DETECTION OF AFA ON LIA mM phenylmethylsulfonyl fluoride (0.2 ml/cm2) and stirred overnight at 4°C. The homogenate was centrifuged at 15,000g for 15 minutes, and the extracted proteins in the supernatant were precipitated overnight at ⫺20°C by adding 5 volumes of absolute ethanol. After centrifugation for 15 minutes at 10,000g, the protein pellet was vacuum dried and subsequently resuspended in water. The amount of protein was determined by the Bradford protein assay as modified by Peterson (15), using bovine serum albumin (BSA) standard curves. For AFA detection, the crude filaggrin preparation was subjected to 10% tricine–sodium dodecyl sulfate– polyacrylamide gel electrophoresis (SDS-PAGE) using the Bio-Rad (Philadelphia, PA) mini-gel apparatus and then electrophoresed and blotted under standard conditions (8). Blot strips were probed with human sera overnight at a 1:100 dilution in PBS, 0.05% Tween 20, and 0.1% gelatin. As secondary antibody, the anti-human IgG–alkaline phosphatase conjugate (Sigma, St. Louis, MO) was added at a 1:1,000 dilution; visualization occurred upon addition of nitroblue tetrazolium (NBT)/BCIP chromogenic substrate. Two-dimensional (2-D) gel electrophoresis of filaggrin. The semipurified filaggrin material was separated by 2-D gel electrophoresis using Immobiline DryStrips (pH 3–10; Amersham Pharmacia Biotech, Uppsala, Sweden) in the first dimension and 10% tricine–SDS-PAGE in the second dimension according to standard procedures. Preparative gels were loaded with 300 g of protein and stained with Coomassie Brilliant Blue R-250 (Bio-Rad). The large, comma-shaped filaggrin spot was identified by immunoreaction with the antifilaggrin monoclonal antibody (Biomedical Technologies, Stoughton, MA). The pI of this heterogeneous protein ranged from 6.7 to 8.5, and molecular-weight forms of 35–68 kd were detected, with the more acidic isoforms representing the highest masses. Electroelution of filaggrin. The large filaggrin spot was cut out of the gel and separated into 3 parts: 1) acidic fraction (pI 6.7–7.5), 2) neutral fraction (pI 7.5–8.1), and 3) basic fraction (pI 8.1–8.5). Each fraction was electroeluted according to the method described by Hunkapiller et al (16), using 50 mM (NH4)HCO3, 0.1% SDS as elution buffer. Coomassie stain and SDS were removed from the eluted proteins by ion-pair extraction as described by Königsberg and Henderson (17). Vacuum-dried protein pellets were redissolved in the appropriate buffer for further analysis. Peptide mapping. Purified electroeluted filaggrin fractions of ⬃100 g were dissolved in 40 l of 100 mM (NH4)HCO3 (pH 8.0), 10% acetonitrile and digested with trypsin (enzyme:substrate ratio 1:40). After overnight incubation at 37°C, the digest was stored at ⫺20°C until used. Peptides were separated by reverse-phase high-performance liquid chromatography (RP-HPLC) on a 2.1 ⫻ 250–mm C4 Vydac column (Hesperia, CA) using a 140B solvent delivery system (Applied Biosystems, Foster City, CA) and eluted with linear gradient (8–70%) of 70% acetonitrile in 0.1% trifluoroacetic acid (TFA). Detection using a 1,000S diode array detector (Applied Biosystems) occurred at 214 nm, and peptides were manually recovered. Dot-spot analysis and microsequencing. Ninety percent of each peak fraction was vacuum dried and subsequently resuspended in a small volume of 10% acetonitrile, 50 mM NaCO3 buffer (pH 9.5). The peptides were dot-spotted onto 1187 Immunodyne ABC membranes (Pall Europe, Portsmouth, UK), which were probed with human sera in order to assign the immunoreactive epitopes. First, membranes were blocked with PBS, 0.5% caseine, and incubated overnight with sera diluted 1:50 in PBS, 0.5% caseine, 0.1% Triton X-705 (Sigma), 10 mM MgCl2.6H2O. After washing with PBS, 0.05% Tween 20, anti-human IgG was added, and membranes were developed in 100 mM NaCl, 100 mM Tris HCl (pH 9.8), 50 mM MgCl2.6H2O substrate buffer containing the chromogenic substrate NBT/BCIP. Reaction was stopped by addition of 0.2N H2SO4. The remaining 10% of the immunoreactive fractions was used for microsequencing. For this, fractions were directly analyzed on a model 477A pulsed liquid sequencer equipped with an on-line 120 phenylthiohydantoin analyzer (Applied Biosystems). Cloning, expression, purification, and citrullination of human filaggrin. Four candidate human filaggrin sequences were cloned by polymerase chain reaction (PCR) using genomic DNA isolated from human lymphocytes as template. The PCR sense primer was chosen to overlap the filaggrin linker sequence and was designed to introduce a functional translation initiation codon upstream from the linker sequence FLYQVST. The antisense primer was located just upstream from the next filaggrin linker sequence introducing the translation stop codon TAG. The amplified filaggrin sequence therefore consisted of the filaggrin linker followed by an integral filaggrin repeat and 3 additional amino acids resulting from the cloning strategy. The PCR-amplified fragments were cloned in a pBLSK vector (Stratagene, Amsterdam, The Netherlands) as an Eco RI–Xba I fragment, allowing sequencing of the amplified complementary DNA (cDNA). The cDNA inserts were recloned as Eco RV–Ecl 136II fragments of 1,030 bp in the Escherichia coli expression vector pI GRHISA (Innogenetics). The filaggrin proteins were expressed as recombinant filaggrin–His-6 fusion proteins in 3 different E coli strains, resulting in high Coomassie-stainable expression levels, and were further purified by metal-affinity chromatography under denaturing conditions. Enrichment of the full-size 50-kd protein was obtained by RP-HPLC. Enzymatic citrullination was performed as described by Senshu et al (18), using PAD enzyme (Sigma) at a 1:60–1:120 enzyme:substrate ratio, for 15 minutes to 6 hours at 37°C. The reaction was stopped by addition of 0.1% TFA. The degree of citrullination was verified by analyzing mobility shifts on SDS-PAGE or by isoelectrofocusing and amino acid analysis. Peptide synthesis, molecular modeling, and computational chemistry. Biotinylated peptides were synthesized according to standard 9-fluorenylmethoxycarbonyl chemistry as described previously (19) and purified by RP-HPLC using a linear gradient (8–70%) of 70% acetonitrile in 0.1% TFA. Molecular modeling and computational chemistry were performed using Insight II (version 97.2; Accelrys, San Diego, CA) and Discover software (Molecular Simulations, Orsay, France). LIA for antifilaggrin detection. Peptides and recombinant antigens were applied directly on a nylon membrane with a plastic backing, and strips were developed, with minor modifications, as previously described (20). In a final layout of the INNO-LIA RA, recombinant citrullinated filaggrin and 2 citrulline-containing peptides were retained. During each test run, a reference cutoff control strip was included, providing a 1188 UNION ET AL Figure 1. Epitope mapping of human natural filaggrin. A, Two-dimensional gel electrophoretic protein pattern of human epidermal extracts enriched for filaggrin. Separation was performed using Immobiline DryStrips (pH 3–10) and 10% Tricine–sodium dodecyl sulfate– polyacrylamide gel electrophoresis (SDS-PAGE). Preparative gels were loaded with 300 g of protein and stained with Coomassie Brilliant Blue R-250. The heterogeneous filaggrin protein (arrow) was divided into 3 fractions: acidic (ac), neutral (n), and basic filaggrin (ba). B–E, Tryptic digests of acidic and neutral filaggrin were separated by reverse-phase high-performance liquid chromatography (B and C), and peptides were dot-spotted on membranes (D and E) and probed with rheumatoid arthritis or control sera. ⴱ indicates positive controls (1 g natural filaggrin from 2 preparations). IEF ⫽ isoelectric focusing; AU ⫽ absorbance units; OD ⫽ optical density; nr ⫽ number. cutoff intensity for each antigen line. Air-dried strips were read visually or scanned using a Scanjet 5P scanner (HewlettPackard, McMinnville, OR) and an in-house LIA interpretation software program. A serum sample was considered to be positive when at least 1 of the 3 antigen lines scored positive. The performance of the INNO-LIA RA in terms of sensitivity and specificity was calculated using clinical diagnosis as the gold standard. Receiver operating characteristic (ROC) curves were determined for each antigen using an independent sample population of 127 sera from RA patients, 84 from disease controls, and 71 from healthy subjects, which enabled us to define optimal cutoff values and to generate a corre- sponding cutoff serum. Subsequently, validation of the INNOLIA RA was performed in a multicenter analysis, using 336 RA and 253 disease control samples, as indicated above. Reference serum tests. For serum panel D, AFA were identified by standard immunoblotting using human epidermal extracts enriched for the neutral-acidic variant of filaggrin (8). The intensity of color development was estimated according to a semiquantitative scale. The cutoff values for this assay (95% or 99% specificity) were based on a larger independent data set consisting of 190 RA and 480 non-RA control sera and were fixed at 0.25 and 2.25, respectively. For the multicenter validation, each center used either CITRULLINATED PEPTIDES FOR DETECTION OF AFA ON LIA 1189 Table 2. Identification of immunoreactive peptide fractions derived from epitope mapping of acidic and neutral human natural filaggrin* Synthetic peptide Immunoreactive peptide fraction† Acidic filaggrin R2HSASQDGQDTIRGHPG R2HSGIGHGQASSAVR R2DSGHRGYSGSQASDNEGH R2HSTSQEGQDTIHGHRGS R2QGSRHQQAR R2AGHGHSADSSR R2HGSHHQQSADSSR Neutral filaggrin R2HSGIGHGQASSAVR R2HSASQDGQDTIRGHPG R2HSTSQEGQDTIHGHRGS R2DSGHRGYSGSQASDNEGH R2GYSGSQASDNEGHSE R2NDEQSGDGSR Citrulline-containing IGP1155: IGP1156: IGP1157: IGP1158: IGP1249: HSASQDGQDTIRGHPGSS HSGIGHGQASSAVRDSGHRGYS DSGHRGYSGSQASDNEGH HSTSQEGQDTIHGHRGS GGQGSRHQQAR Nonmodified IGP1179: IGP1180: IGP1181: IGP1182: HSASQDGQDTIRGHPGSS HSGIGHGQASSAVRDSGHRGYS DSGHRGYSGSQASDNEGH HSTSQEGQDTIHGHRGS IGP1250: GGAGHGHSADSSR IGP1251: GGHGSHHQQSADSSR IGP1156 IGP1155 IGP1158 IGP1157 IGP1252: NDEQSGDGSR * R ⫽ citrulline; R2 indicates tryptic cleavage site. † Amino acid sequences were retrieved by microsequencing on a 477A sequencer equipped with an online 120 phenylthiohydantoin analyzer. nephelometric methods, an Imtec ELISA (Zepernick, Germany), or an in-house ELISA prepared at the laboratory of R. L. Humbel, Luxembourg) for RF detection. The presence of anti-RA33 antibodies was determined by immunoblotting (20) or ELISA (Imtec), and testing for antikeratin antibody (AKA) was carried out by indirect immunofluorescence on rat esophagus sections (9). Evaluation for antiperinuclear factor (APF) was performed using human buccal mucosa cells as previously described (9). Statistical methods. Results of the comparisons of the INNO-LIA RA with AFA immunoblotting and with testing for anti-RA33 or AKA were validated using McNemar’s test statistics and Cohen’s kappa. Discriminant and principal component analyses were performed to determine the individual peptide contribution to maximal variation and sensitivity. RESULTS Epitope mapping of human natural filaggrin. Human filaggrin isolated from epidermal sheets was electroeluted from preparative 2-D gels (Figure 1A). Because the protein consisted of a heterogeneous group of polypeptides, 3 different fractions were isolated according to their pI. Tryptic digests of the acidic, neutral, and basic filaggrin were separated by HPLC, and peptide fractions were subsequently analyzed in a dot-spot assay system (Figures 1B–E). Because of the low amount of material in the basic fraction, no further analysis was performed. Four APF-positive RA sera that were also reactive with human filaggrin on Western blot and 1 APF-negative control serum sample were used to determine immunoreactivity. For the acidic filaggrin, peptide fractions T32–41 showed a clear positive reaction (Figure 1D), in contrast to the APF-negative serum. A similar procedure was performed for neutral filaggrin, resulting in reactive fractions T28, T65, T81, and the series T29–39 (Figure 1E). Amino acid sequencing revealed that most of these fractions contained ⬎1 filaggrin-derived peptide, as could be expected based on the very heterogeneous nature of the protein (Table 2). During sequencing, citrulline was found in a number of peptides, as shown by 1) its specific retention time, which was different from that of arginine, 2) the lack of cleavage (which is to be expected if arginine were present), and 3) the absence of arginine residues during sequencing. Use of a citrulline standard confirmed these findings. Use of synthetic peptides for AFA detection on LIA. Based on the epitope mapping results, 12 peptides (IGP1155–1158, IGP1179–1182, and IGP1249–1252) were synthesized and evaluated in the LIA system (Table 2). For 4 citrullinated peptides, the nonmodified counterparts containing arginine instead of citrulline residues were synthesized in order to compare their respective reactivities. Using serum panel A, reactivity was noted with citrullinated peptides IGP1155, 1156, 1157, 1158, and 1249, with a combined sensitivity of 48% (Table 3). Upon taking into account only those sera that reacted with human natural filaggrin on Western blot, markedly higher sensitivities were reached for each synthetic peptide compared with the filaggrin-negative RA sera. Consequently, the combined sensitivity of 77.0% (95% CI confidence interval [95% CI] 63–87%) 1190 UNION ET AL Table 3. Immunoreactivity of 5 synthetic citrullinated peptides identified by epitope mapping of human natural filaggrin and of recombinant citrullinated filaggrin (Rec fg) on line immunoassay* Serum group No. of sera IGP1155 IGP1156 IGP1157 IGP1158 IGP1249 Combination of peptides Rec fg Peptides ⫹ Rec fg RA Filaggrin blot positive Filaggrin blot negative Control Healthy SLE Osteoarthritis 107 52 55 22 (20.6) 17 (32.7) 5 (9.1) 31 (29.0) 24 (46.2) 7 (12.7) 4 (3.7) 3 (5.8) 1 (1.8) 26 (24.3) 21 (40.4) 5 (9.1) 16 (15.0) 13 (25.0) 3 (5.5) 51 (47.7) 40 (77.0) 11 (20.0) 52 (48.6) 37 (71.2) 15 (27.2) 64 (59.8) 45 (86.5) 19 (34.5) 51 61 62 1 (2.0) 1 (1.6) 1 (1.6) 3 (5.9) – 1 (1.6) 1 (2.0) 1 (1.6) 1 (1.6) 1 (2.0) – – 1 (2.0) 2 (3.3) 1 (1.6) 3 (5.9) 4 (6.6) 2 (3.2) 1 (2.0) 7 (11.5) 3 (4.8) 4 (7.8) 9 (14.8) 5 (8.0) * Combined reactivity was defined by positivity toward at least 1 antigen. Values are the number (%) of reactive sera. RA ⫽ rheumatoid arthritis; Filaggrin blot ⫽ Western blotting using human natural filaggrin as antigenic substrate; SLE ⫽ systemic lupus erythematosus. in the filaggrin-positive group was higher than that of 20.0% (95% CI 10–33%) in the filaggrin-negative group. Control sera showed only minor reactivities with the peptides, resulting in a specificity of ⬎96%. In all groups, IGP1156 and 1158 reacted most strongly; immunoreaction toward IGP1157 was always very faint and did not increase diagnostic sensitivity. Interestingly, the positive LIA signals observed with these citrullinecontaining peptides were not seen with the nonmodified counterparts IGP1179, 1180, 1181, or 1182. The group of peptides (derived from the epitope mapping) that did not contain citrulline (IGP1250, 1251, and 1252) showed no substantial immunoreactivity. Diagnostic value of citrullinated recombinant filaggrin. Two of 4 available human filaggrin clones that differed in their amino acid sequence by 5% were expressed in E coli. Upon SDS gel analysis, in addition to the full-size protein, a 25-kd filaggrin band was present (Figure 2). Kinetic experiments were performed for the enzymatic deimination of filaggrin using PAD at a 1:60–1:120 enzyme:substrate ratio. A clear mobility shift on SDS gel was observed with respect to time of deimination: from 15 minutes onward, the protein had already migrated to a higher molecular-weight position, reaching a plateau after 2 hours of citrullination (Figure 2). The loss of positive charges, resulting from the modification of arginines to citrullines, was demonstrated by isoelectric focusing electrophoresis, which confirmed the change in pI associated with longer incubation times (results not shown). Amino acid analysis also showed an increased degree of citrullination over time: within 15 minutes, half the arginines were converted into citrullines, reaching near completion within 6 hours (Table 4). Immunoreactivity of recombinant proteins both with and without citrulline residues was further assessed on LIA. Reactivity with RA sera was noted within 15 minutes of citrullination and thereafter, while the nonmodified protein was mainly unreactive. The diagnostic value of the recombinant citrullinated filaggrin was compared with the reactivity of the 5 synthetic filaggrin peptides sprayed as individual lines, using the same battery of sera (Table 3). About half of the RA sera showed a positive reaction with the recombinant citrullinated filaggrin, which was similar to the combined reactivity with the peptides. Because of the partial complementary reactivity observed between recombinant protein and peptides, the sensitivity of the test for RA could be increased to nearly 60%, with a corresponding specificity of 95%. Figure 2. Citrullination of recombinant filaggrin. Escherichia coli– derived filaggrin was enzymatically deiminated using peptidyl-arginine deiminase at a 1:60 enzyme:substrate ratio, and the degree of citrullination at different time points was analyzed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE). In addition to a major 50-kd full-size protein, a 25-kd breakdown protein was observed. A clear mobility shift was visible with respect to time of deimination, indicating the conversion of arginine residues to citrullines. CITRULLINATED PEPTIDES FOR DETECTION OF AFA ON LIA 1191 Table 4. Amino acid composition of in vitro citrullinated recombinant human filaggrin* L-citrulline ⫹ L-ornithine Arginine Time 0 Time 1 Time 2 Time 3 Time 4 Time 5 0.6 (0.3) 25.5 (12.9) 10.6 (6.3) 10.4 (6.1) 13.3 (8.7) 5.4 (3.5) 15.4 (10.7) 2.9 (2.0) 15.0 (11.5) 1.5 (1.1) 15.9 (11.3) 1.0 (0.7) * Amino acid composition at different time points was determined by hydrolysis at 156°C in 6N HCl and analysis on a 477 amino acid analyzer. Ornithine is a well-known degradation product of citrulline formed during hydrolysis, and is taken into account when calculating the degree of citrullination. Time 1 ⫽ 15 minutes, Time 2 ⫽ 30 minutes, Time 3 ⫽ 1 hour, Time 4 ⫽ 2 hours, Time 5 ⫽ 6 hours. Values are the quantity in nanograms (%). Design of variant peptides by molecular modeling and computational chemistry. In order to further improve LIA sensitivity and shorten serum incubation times, novel peptides were designed so that citrulline residues would be spatially positioned for maximal accessibility. Following this modeling approach, numerous peptide conformations were considered, and peptides were subsequently synthesized and analyzed for immunoreactivity on LIA to investigate the influence of individual residue changes. Statistical processing of the results revealed that a limited number of peptides contributed to maximal variation and the highest combined sensitivity. ROC curve analysis was performed for each peptide in order to determine sensitivities at each specificity level. In contrast to the signals from the peptides derived from the epitope mapping studies, much stronger signals were obtained, and serum incubation time was reduced from overnight to just 1 hour. Finally, 2 peptide variants (peptides A and B) mapping in the filaggrin region of amino acid 51–100 (SWISS-PROT accession number P20930), together with the recombinant citrullinated filaggrin, were selected for further evaluation with serum panel B. A theoretical cutoff value was defined by ROC curve analysis, corresponding to a specificity of 99% for each antigen line (Figure 3). Figure 3. Receiver operating characteristic curves. Serum panel B (127 rheumatoid arthritis sera, 84 sera from disease controls, and 71 sera from healthy controls) was used to establish the curves. On the basis of these values, a cutoff control serum was prepared for use on a reference strip, thereby providing a cutoff intensity for each antigen line (Figure 4). Using this cutoff control, an overall specificity of 98.1% with a corresponding sensitivity of 67% was obtained (serum panel B). Evaluation of an additional 26 sera from patients with Crohn’s disease and 24 from patients with gout revealed no aspecific reactivity of these control samples. Assay validation. In a multicenter retrospective clinical study, performance of the INNO-LIA RA was assessed and compared with that of individual reference methods, using serum panel C. As seen in Table 5, a distinction was made between LIA3, containing the recombinant filaggrin and 2 citrullinated peptides, and LIA2, which took only the 2 peptides into consideration. Sensitivity for RA using LIA3 was slightly higher than that using LIA2 (65.2% and 61.9%, respectively), but specificity was lower (98.0% and 98.8%, respectively). The number of false positives with LIA2 was 3 of 253 and included 2 SSc samples and 1 Sjögren’s syndrome Figure 4. Immunoreactivity of recombinant filaggrin (Rec fg) and synthetic citrullinated peptides (pepA, pepB) on line immunoassay. Ten antifilaggrin autoantibody–positive rheumatoid arthritis (RA) sera were compared with a control (cut-off) strip that provided cutoff intensities for each corresponding antigen line. 1192 UNION ET AL Table 5. Performance of the INNO-LIA RA in a multicenter, retrospective study* Center LIA3 LIA2 No. of RA samples No. of control samples Sensitivity, % Specificity, % Sensitivity, % Specificity, % 125 112 99 336 25 60 168 253 80.0 53.5 59.6 65.2 100 98.4 97.6 98.0 78.4 49.1 55.6 61.9 100 98.4 98.8 98.8 1 2 3 Total * Strips were read against a cutoff control serum applied on a separate strip. LIA3 ⫽ recombinant citrullinated filaggrin ⫹ peptide A ⫹ peptide B; LIA2 ⫽ peptide A ⫹ peptide B (see Table 1 for other definitions). sample. Two of these sera were tested for RF and were also found to be positive. Because the prevalence of RA patients in normal rheumatologic practice is ⬃20%, and assuming that the disease controls are representative, the positive predictive values of LIA3 and LIA2 for RA were 89.1% and 92.8%, respectively. Therefore, only LIA2 was used for further comparison with other serologic tests. Because of the sample bias with respect to RF (use of RF-positive controls and the fact that presence of RF is one of the ACR criteria for RA), we could not statistically compare results of RF determination and LIA, but complementarity between these methods could be investigated (Table 6). A total of 30 (37%) of 81 RF-negative RA samples proved to be AFA-positive on LIA2. Furthermore, all RF-negative control samples were also AFA-negative, while 52 of 54 RF-positive control sera had no AFA detected on LIA2. In addition, LIA showed a higher sensitivity and specificity compared with the anti-RA33 assay (P ⬍ 0.0001), whereas good agreement was observed with AKA testing (kappa value 0.725; 95% CI 0.633–0.817). Comparison between LIA and AFA immunoblot. In order to compare the results of LIA with detection of AFA by the standard immunoblotting procedure using natural filaggrin, a limited serum panel (panel D) that was blinded to observers was analyzed on blot and scored semiquantitatively. Cutoff values for the LIA were adapted to obtain an overall specificity of either 95% or 99%, equal to the blot specificities. As shown in Table 7, LIA performed similarly (46% sensitivity) at both preset specificities (95% and 99%). The AFA blot resulted in higher sensitivity (56% at the 95% specificity level) but detected only 30% of the RA sera at the 99% specificity level. The 3 false-positive AFA blot control sera that were also APF-positive were diagnosed as OA, scapulohumeral periarthritis, and lumbar arthrosis. Comparison between the total number of correct or incorrect results showed that the 2 tests performed similarly at the 95% specificity level, while LIA proved statistically better than the AFA blot at the 99% level (P ⫽ 0.0047 by McNemar’s test). Among the 8 samples that were correctly identified by LIA only, 4 were from patients with early RA, and 4 were from patients with longstanding RA. Furthermore, 2 of 17 RF-negative RA samples were detected on LIA but not on Western blot. When this sample collection was evaluated using the prototype LIA2, LIA performance at the 99% specificity level was significantly better than that of the AFA blot (P ⫽ 0.007; Table 7). DISCUSSION In recent years, the presence of AFA in the sera of RA patients has been well-documented. However, the precise epitopes in filaggrin that are responsible for the Table 6. Comparison between antifilaggrin detection on LIA2 and reference methods for detection of RF, anti-RA33, and AKA* RF Anti-RA33 Reference test/LIA2 RA (n ⫽ 334) Disease control (n ⫽ 188) RA (n ⫽ 211) Disease control (n ⫽ 154) AKA: RA (n ⫽ 232) –/– –/⫹ ⫹/– ⫹/⫹ 51 30 76 177 134 0 52 2 77 83 24 27 138 2 14 0 72 20 10 130 * Disease controls were from serum panel C (see Table 1). Values are the number of positive (⫹) or negative (⫺) sera. LIA2 ⫽ line immunoassay with peptide A ⫹ peptide B; RF ⫽ rheumatoid factor; RA ⫽ rheumatoid arthritis; AKA ⫽ antikeratin antibodies. CITRULLINATED PEPTIDES FOR DETECTION OF AFA ON LIA 1193 Table 7. Sensitivity and specificity comparison of AFA detection by AFA blot using natural filaggrin versus LIA* Preset specificity 95%† RA samples (n ⫽ 50) Disease controls (n ⫽ 25) Preset specificity 99%‡ Cutoff control INNO-LIA RA§ AFA blot INNO-LIA RA AFA blot INNO-LIA RA LIA3 LIA2 28 (56) 3 23 (46) 1 15 (30) 1 23 (46) 1 24 (48) 2 24 (48) 1 * AFA ⫽ antifilaggrin autoantibodies; INNO-LIA ⫽ Innogenetics line immunoassay; RA ⫽ rheumatoid arthritis; LIA3 ⫽ recombinant citrullinated filaggrin ⫹ peptide A ⫹ peptide B; LIA2 ⫽ peptide A ⫹ peptide B. Values are the number (%) of reactive sera. † P ⫽ 0.37 by McNemar’s test. ‡ P ⫽ 0.0047 by McNemar’s test. § P ⫽ 0.007 by McNemar’s test. observed immunoreaction have not been determined until now, and their identification might help to improve diagnostic performance of assays relying on non-native antigenic substrates. The epitope mapping described in the present study indicates that the immunoreactive epitopes in natural filaggrin contain citrulline residues. Although the presence of citrulline in natural human filaggrin has been described previously (22,23), our results confirm the specific recognition by AFA of a citrullinated linear epitope in the natural antigen. These results are in accordance with previous findings that the presence of citrulline is necessary in order to obtain proper immunoreactivity of patient sera toward recombinant filaggrin or synthetic peptides (10,12). Based on the recognition of this essential secondary modification, it is now apparent why previous attempts to produce reactive recombinant antigen intended for AFA detection were not successful (Union A, et al: unpublished observations). Upon citrullination of recombinant E coli filaggrin by the enzyme PAD, however, a reactive antigen was generated (10). Other studies using citrullinated peptides have also proven useful for the development of a diagnostic ELISA (12,13). The peptides selected in this study partially overlapped those described by Schellekens et al (12) but did not contain the cfc1 sequence (13). This is most probably attributable to our use of trypsin for enzymatic cleavage of the protein, which completely destroys the RGRSRGRSGRSGS cfc1 sequence, hence generating fragments that are too small to be detected in the epitope mapping dot-spot analysis. In addition to our previous findings indicating that the citrulline residue can be situated in the particular protein context Ser-Cit-His (10), the current results also show that several other environmental amino acid residues are found in the natural protein (Table 2). Synthetic peptides from the earlier study, however, showed only minor reactivity in the LIA system (10). According to our calculations of molecular modeling and computational chemistry, a His residue appears to form a hydrogen bond with the adjacent citrulline, by which the accessibility of the citrulline is reduced. In general, an inverse correlation was found between the availability of citrulline for AFA and the size of the neighboring amino acids. In order to design novel peptides with higher immunoreactivity than those derived from the epitope mapping, the directed approach of molecular modeling was used instead of random synthesis or combinatorial chemistry. Evaluation by LIA revealed 2 selected peptides that were highly useful for AFA detection, with 62% sensitivity in established RA samples at a specificity of 98.4%. The additional use of recombinant deiminated filaggrin slightly increased sensitivity of the test but at the expense of specificity. We observed a partial complementarity with RF determination, because on LIA, 37% of the RF-negative RA samples were positive for the peptides. This percentage can vary according to the particular study and detection method but seems largely dependent on the study population: 30% or 42% in patients with established RA who fulfill the ACR criteria (9,13) compared with 14–18% in early arthritis cohorts (11,24,25). This difference may be attributable to the fact that RF levels fluctuate during the course of the disease, while AFA titers are more stable over time (26,27). Nevertheless, both autoantibodies are markers of the underlying immunopathologic process in RA, because specific plasma cells can be identified from rheumatoid synovium (28,29). There is a pressing need for new serologic RA tests that provide diagnostic information that is more specific than that provided by the currently available RF determination. The LIA results of the present study (98.8% specificity) as well as those obtained with the anti–citrullinated cyclic peptide ELISA (98% specificity) (13) meet this challenge. Furthermore, it has been 1194 shown that appearance of AFA occurs early in the course of RA and may precede telltale clinical signs (30–32), while there is also evidence that RF-negative RA patients who are AFA-positive have a severe disease course and poor prognosis (26,33). Recently, it was demonstrated that positivity for both RF and AFA is associated with an even worse prognosis (24). In view of current, expensive biologic treatments, detection of both AFA and RF could help to define which patients are eligible for such treatment. In this study, it was noteworthy that RG sequences in filaggrin were the preferential citrullination positions and were also observed to be the favorite dimethylation site in the SmD protein acting as a specific autoantigen in SLE (19). The glycine residue, being the smallest, probably favors the amount of available space around the arginine, which may facilitate arginine modification. Furthermore, citrullinated myelin basic protein also occurs in patients with multiple sclerosis (34). Apparently, a number of posttranslational modification phenomena play a key role in the breakdown of immune tolerance and triggering of autoimmune reactivities. In addition to citrullinated and dimethylated autoantigens, epitopes generated by caspase cleavage and phosphorylation are all targeted by autoantibodies (35; for review see ref. 36). The biologic significance of deimination and its involvement in autoimmune responses remain to be elucidated. Intracellular and highly specific anticitrulline staining was recently observed in synovial immunohistochemistry studies (37). Filaggrin itself was not detectable (37,38), but biochemical characterization of deiminated synovial tissue proteins led to the identification of ␣ and ␤ chains of fibrin, which could correspond to the genuine target of AFA (38). A defect in the apoptotic process may be responsible for the uncontrollable proliferation of synovial fibroblasts. Although extensive DNA fragmentation has been observed in rheumatoid synovium, only a few cells actually complete the apoptotic process (39). Because citrullination of vimentin has been linked to apoptosis (40), such deiminated proteins could be released at the inflammatory site, hence creating additional neoepitopes for both B cell and T cell immune recognition. In conclusion, our results indicate that citrulline is present in the immunoreactive epitopes of natural filaggrin, and that AFA can be readily detected on a membrane solid phase using citrullinated peptides. Due to its partial complementarity to RF and its high prognostic value, this LIA provides a valuable tool for the diagnosis of RA and may be of use in the choice of UNION ET AL particular therapies. The specificity of AFA undoubtedly makes them intruiging candidates for further study. ACKNOWLEDGMENTS We are very thankful to An Vos, Liesbet Amerijckx, Martine Dauwe, and Patrick Schmit for their skillful technical assistance; to Stéphanie Dincq, Katrien De Bosschere, and Frank Hulstaert (Medical Affairs), as well as to the Department of Protein Expression, Purification/Peptide Synthesis of Innogenetics for valuable contributions; and to Fred Shapiro for editorial support. We also wish to thank Dr. P. Verlende (Plastische Heelkunde, Volkskliniek, Ghent) for providing human skin from abdominoplasty, Prof. Dr. G. Verbruggen and Dr. I. Peene (Rheumatology, UZ Ghent) for collection of osteoarthritis samples, and Dr. K. Lüthke (Rheumatology, Dresden) for serum sample collection and clinical data. We are especially indebted to Jos Raymackers (Innogenetics) who so unexpectedly and regrettably died during the course of this study. REFERENCES 1. Trentham DE, Dynesius RA, David JR. Passive transfer by cells of type II collagen-induced arthritis in rats. J Clin Invest 1978;62: 359–66. 2. Sakata A, Sakata K, Ping H, Ohmura T, Tsukano M, Kakimoto K. Successful induction of severe arthritis by the transfer of in vitro-activated synovial fluid T cells from patients with rheumatoid arthritis (RA) in severe combined immunodeficient (SCID) mice. Clin Exp Immunol 1996;104:247–54. 3. Masi AT, Feigenbaum SL. Seronegative rheumatoid arthritis: fact or fiction? Arch Intern Med 1983;43:2167–72. 4. Lichtenstein MJ, Pincus T. Rheumatoid arthritis identified in population based cross sectional studies: low prevalence of rheumatoid factor. J Rheumatol 1991;18:989–93. 5. Pincus T, Callahan LF. How many types of patients meet classification criteria for rheumatoid arthritis? J Rheumatol 1994;21: 1385–9. 6. Simon M, Girbal E, Sebbag M, Gomès-Daudrix V, Vincent C, Salama G, et al. The cytokeratin filament-aggregating protein filaggrin is the target of the so-called “antikeratin antibodies,” autoantibodies specific for rheumatoid arthritis. J Clin Invest 1993;92:1387–93. 7. Sebbag M, Simon M, Vincent C, Masson-Bessière C, Girbal E, Durieux J-J, et al. The antiperinuclear factor and the so-called antikeratin antibodies are the same rheumatoid arthritis-specific autoantibodies. J Clin Invest 1995;95:2672–9. 8. Vincent C, Simon M, Sebbag M, Girbal-Neuhauser E, Durieux JJ, Cantagrel A, et al. Immunoblotting detection of autoantibodies to human epidermis filaggrin: a new diagnostic test for rheumatoid arthritis. J Rheumatol 1998;25:838–46. 9. Vincent C, de Keyser F, Masson-Bessière C, Sebbag M, Veys E, Serre G. Anti-perinuclear factor compared with the so called “antikeratin” antibodies and antibodies to human epidermis filaggrin, in the diagnosis of arthritides. Ann Rheum Dis 1999;58:42–8. 10. Girbal-Neuhauser E, Durieux J-J, Arnaud M, Dalbon P, Sebbag M, Vincent C, et al. The epitopes targeted by the rheumatoid arthritis-associated antifilaggrin autoantibodies are posttranslationally generated on various sites of (pro)filaggrin by deimination of arginine residues. J Immunol 1999;162:585–94. 11. Aho K, Palosuo T, Lukka M, Kurki P, Isomäki H, Kautiainen H, CITRULLINATED PEPTIDES FOR DETECTION OF AFA ON LIA 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. et al. Antifilaggrin antibodies in recent-onset arthritis. Scand J Rheumatol 1999;28:113–6. Schellekens GA, de Jong BA, van den Hoogen FH, van de Putte LB, van Venrooij WJ. Citrulline is an essential constituent of antigenic determinants recognized by rheumatoid arthritis-specific autoantibodies. J Clin Invest 1998;101:273–81. Schellekens GA, Visser H, de Jong BA, van den Hoogen FH, Hazes JM, Breedveld FC, et al. The diagnostic properties of rheumatoid arthritis antibodies recognizing a cyclic citrullinated peptide. Arthritis Rheum 2000;43:155–63. 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. Peterson GL. Determination of total protein. Methods Enzymol 1983;91:95–119. Hunkapiller MW, Lujan E, Ostrander F, Hood LE. Isolation of microgram quantities of proteins from polyacrylamide gels for amino acid sequence analysis. Methods Enzymol 1983;91:227–36. Königsberg WH, Henderson L. Removal of sodium dodecyl sulfate from proteins by ion-pair extraction. Methods Enzymol 1983;91:254–9. Senshu T, Akiyama K, Kan S, Asaga H, Ishigami A, Manabe M. Detection of deiminated proteins in rat skin: probing with a monospecific antibody after modification of citrulline residues. J Invest Dermatol 1995;105:163–9. Brahms H, Raymackers J, Union A, de Keyser F, Meheus L, Lührmann R. The C-terminal RG dipeptide repeats of the spliceosomal Sm proteins D1 and D3 contain symmetrical dimethylarginines, which form a major B-cell epitope for anti-Sm autoantibodies. J Biol Chem 2000;275:17122–9. Meheus L, van Venrooij WJ, Wiik A, Charles PJ, Tzioufas AG, Meyer O, et al. Multicenter validation of recombinant, natural and synthetic antigens used in a single multiparameter assay for the detection of specific anti-nuclear autoantibodies in connective tissue disorders. Clin Exp Rheumatol 1999;17:205–14. Hassfeld W, Steiner G, Studnicka-Benke A, Skriner K, Graninger W, Fischer I, et al. Autoimmune response to the spliceosome: an immunologic link between rheumatoid arthritis, mixed connective tissue disease, and systemic lupus erythematosus. Arthritis Rheum 1995;38:777–85. Lynley AM, Dale BA. The characterization of human epidermal filaggrin: a histidine-rich, keratin filament-aggregating protein. Biochim Biophys Acta 1983;744:28–35. Harding CR, Scott IR. Histidine-rich proteins (filaggrins): structural and functional heterogeneity during epidermal differentiation. J Mol Biol 1983;70:651–73. Van Jaarsveld CH, ter Borg EJ, Jacobs JW, Schellekens GA, Gmelig-Meyling FH, van Booma-Frankfort C, et al. The prognostic value of the antiperinuclear factor, anti-citrullinated peptide antibodies and rheumatoid factor in early rheumatoid arthritis. Clin Exp Rheumatol 1999;17:689–97. Goldbach-Mansky R, Lee J, McCoy A, Hoxworth J, Yarboro C, Smolen JS, et al. Rheumatoid arthritis associated autoantibodies in patients with synovitis of recent onset. Arthritis Res 2000;2: 236–43. 1195 26. Janssens X, Veys EM, Verbruggen G, Declercq L. The diagnostic significance of the antiperinuclear factor for rheumatoid arthritis. J Rheumatol 1988;15:1346–50. 27. Paimela L, Palosuo T, Aho K, Lukka M, Kurki P, Leirisalo-Repo M, et al. Association of autoantibodies to filaggrin with an active disease in early rheumatoid arthritis. Ann Rheum Dis 2001;60: 32–5. 28. Munthe E, Natvig JB. Immunoglobulin classes, subclasses and complexes of IgG rheumatoid factor in rheumatoid plasma cells. Clin Exp Immunol 1972;12:55–70. 29. Masson-Bessière C, Sebbag M, Durieux JJ, Nogueira L, Vincent C, Girbal-Neuhauser E, et al. In the rheumatoid pannus, antifilaggrin autoantibodies are produced by local plasma cells and constitute a higher proportion of IgG than in synovial fluid and serum. Clin Exp Immunol 2000;119:544–52. 30. Cordonnier C, Meyer O, Palazzo E, de Bandt M, Elias A, Nicaise P, et al. Diagnostic value of anti-RA33 antibody, antikeratin antibody, antiperinuclear factor and antinuclear antibody in early rheumatoid arthritis: comparison with rheumatoid factor. Br J Rheumatol 1996;35:620–4. 31. Berthelot JM, Maugars Y, Castagné A, Audrain M, Prost A. Antiperinuclear factors are present in polyarthritis before ACR criteria for rheumatoid arthritis are fulfilled. Ann Rheum Dis 1997;56:123–5. 32. Aho K, von Essen R, Kurki P, Palosuo T, Heliövaara M. Antikeratin antibody and antiperinuclear factor as markers for subclinical rheumatoid disease process. J Rheumatol 1993;34:1278–81. 33. Westgeest AA, Boerbooms AM, Jongmans M, Vandenbroucke JP, Vierwinden G, van de Putte LB. Antiperinuclear factor: indicator of more severe disease in seronegative rheumatoid arthritis. J Rheumatol 1987;14:893–7. 34. Wood DD, Moscarello MA. The isolation, characterization, and lipid-aggregating properties of a citrulline containing myelin basic protein. J Biol Chem 1989;264:5121–7. 35. Utz PJ, Hottelet M, Schur PH, Anderson P. Proteins phosphorylated during stress-induced apoptosis are common targets for autoantibody production in patients with systemic lupus erythematosus. J Exp Med 1997;185:843–54. 36. Utz PJ, Anderson P. Posttranslational protein modifications, apoptosis, and the bypass of tolerance to autoantigens. Arthritis Rheum 1998;41:1152–60. 37. Baeten D, Peene I, Union A, Meheus L, Sebbag M, Serre G, et al. Specific presence of intracellular citrullinated proteins in rheumatoid arthritis synovium: relevance to antifilaggrin autoantibodies. Arthritis Rheum 2001;44:2255–62. 38. Masson-Bessière C, Sebbag M, Girbal-Neuhauser E, Nogueira L, Vincent C, Senshu T, et al. The major synovial targets of the rheumatoid arthritis-specific antifilaggrin autoantibodies are deiminated forms of the alpha- and beta-chains of fibrin. J Immunol 2001;166:4177–84. 39. Firestein GS, Nguyen K, Aupperle KR, Yeo M, Boyle DL, Zvaifler NJ. Apoptosis in rheumatoid arthritis: p53 overexpression in rheumatoid arthritis synovium. Am J Pathol 1996;149:2143–51. 40. Asaga H, Yamada M, Senshu T. Selective deimination of vimentin in calcium ionophore-induced apoptosis of mouse peritoneal macrophages. Biochem Biophys Res Commun 1998;243:641–6.