Deiminated Epstein-Barr virus nuclear antigen 1 is a target of anti–citrullinated protein antibodies in rheumatoid arthritis




To test the hypothesis that deimination of viral sequences containing Arg–Gly repeats could generate epitopes recognized by anti–citrullinated protein antibodies (ACPAs) that are present in rheumatoid arthritis (RA) sera.


Multiple antigen peptides derived from Epstein-Barr virus (EBV)–encoded Epstein-Barr nuclear antigen 1 (EBNA-1) were synthesized, substituting the arginines with citrulline, and were used to screen RA sera. Anti–cyclic citrullinated peptide antibodies were purified by affinity chromatography and tested on a panel of in vitro deiminated proteins. Their ability to bind in vivo deiminated proteins was evaluated by immunoprecipitation, using EBV-infected cell lines.


Antibodies specific for a peptide corresponding to the EBNA-135–58 sequence containing citrulline in place of arginine (viral citrullinated peptide [VCP]) were detected in 50% of RA sera and in <5% of normal and disease control sera. In addition, affinity-purified anti-VCP antibodies from RA sera reacted with filaggrin-derived citrullinated peptides, with deiminated fibrinogen, and with deiminated recombinant EBNA-1. Moreover, anti-VCP antibodies immunoprecipitated, from the lysate of calcium ionophore–stimulated lymphoblastoid cell lines, an 80-kd band that was reactive with a monoclonal anti–EBNA-1 antibody and with anti–modified citrulline antibodies.


These data indicate that ACPAs react with a viral deiminated protein and suggest that EBV infection may play a role in the induction of these RA-specific antibodies.

Rheumatoid arthritis (RA) is one of the most common immune-mediated diseases, occurring in ∼1% of the adult population worldwide. Although the pathogenesis of RA is still poorly understood, involvement of both cellular and humoral autoimmune mechanisms has been clearly demonstrated.

A wide variety of autoantibodies have been detected in RA sera, but very few are specific enough to serve as reliable tools for diagnosis and treatment. A clear disease specificity was demonstrated for antiperinuclear factors (APFs) (1) and antikeratin antibodies (AKAs) (2), but for many years the antigen recognized by APFs and AKAs remained unknown. Finally, in 1993, Simon et al (3) showed that the target of AKA is filaggrin, and in 1995 Sebbag et al (4) demonstrated that there was a partial overlap between APFs and AKAs, and that filaggrin-binding IgG was reactive in both the AKA and APF tests.

Filaggrin, a protein involved in the aggregation of cytokeratin filaments, is synthesized in epithelial cells as a phosphorylated high molecular mass (>200 kd) precursor, profilaggrin. During cell differentiation, profilaggrin is dephosphorylated and cleaved; at this stage, the arginyl residues of filaggrin are converted into neutral citrullyl residues by peptidylarginine deiminase (PAD), generating more acidic isoforms (5). Antifilaggrin antibodies (AFAs) do not bind the precursor or the nondeiminated molecule; they react exclusively with in vivo and in vitro (6) deiminated filaggrin. In fact, AFAs and anticitrulline antibodies detect the same neutral/acidic isoforms of extracted filaggrin; conversely, treatment of recombinant filaggrin with PAD renders the molecules reactive with AFAs. PAD is a calcium-dependent enzyme that is inactive at normal intracellular Ca2+ concentrations (10−7M); events such as cell death, oxidative stress, or in vitro treatment with calcium ionophores can increase the calcium concentration up to the threshold level (10−5M), thus activating the enzyme and inducing protein deimination (7).

AFAs also react with synthetic peptides corresponding to the filaggrin sequences, in which arginine is substituted with citrulline. Schellekens at al (8) identified filaggrin sequences with a high antigenicity index and a large number of residues with a high probability of containing turns. Those containing several arginines were synthesized, substituting citrulline for arginine, and used as antigens on the solid phase to screen RA sera by enzyme-linked immunosorbent assay (ELISA). A high percentage of sera reacted with 1 or more of the modified filaggrin peptides, and in general the presence of 2 citrulline residues increased the antigenicity.

A further enhancement of the diagnostic properties of these synthetic peptides was obtained by constraining the peptides to a beta-turn conformation; a cyclic peptide mimicking a beta-turn was obtained by substituting 2 serine residues with cysteine and subsequently oxidizing the molecule. Using RA sera, Schellekens and colleagues (9) demonstrated that an anti–cyclic citrullinated peptide (anti-CCP) ELISA was highly specific (98%) and “reasonably” sensitive (68%). However, filaggrin is selectively expressed only in epithelial cells, which do not represent a target of the RA autoimmune response and, on the contrary, are undetectable in synovial tissue. Thus, other molecules have been proposed as biologically relevant targets of AFAs. In 2001, Masson-Bessiere et al (10) showed that there is a broad overlap between AFAs and antibodies reacting with the deiminated form of the α- and β-chains of fibrinogen, and suggested that the deiminated form of fibrin deposited in synovia may be a major target of AFAs. Vimentin is another citrullinated protein that is detected in inflamed synovia and recognized by AFAs (11).

A comparative evaluation of the sequences recognized by AFAs showed that their most crucial feature is the presence of citrulline flanked by neutral amino acids such as glycine, serine, or threonine (6). Similar amino acid repeats are often found in nucleic acid–binding proteins; some of these are of viral origin (e.g., the transcription-regulating proteins in herpesvirus) (8). Epstein-Barr virus (EBV), a member of the Herpesviridae family, is known to infect human B lymphocytes and epithelial cells of the oropharynx, establishing a reservoir in both of these cellular compartments (12). A large number of other cells may be infected in a transient manner (13–15).

One of the nuclear proteins encoded by EBV, Epstein-Barr nuclear antigen 1 (EBNA-1), contains in its N-terminal region a sequence (amino acids 35–58) characterized by a 6-fold Gly–Arg repeat homologous to the C-terminal portion of SmD, the spliceosome protein recognized by autoantibodies in systemic lupus erythematosus sera (SLE) (16).

In the present study, we tested the hypothesis that deimination of a viral sequence containing Gly–Arg repeats could generate epitopes recognized by AFAs. To this end, we synthesized peptides corresponding to the N-terminus of EBNA-1, in which arginines were substituted by citrulline at various degrees, and were able to show that they are indeed specifically recognized by RA sera.



Sera were obtained from patients with systemic autoimmune diseases who were being followed up at the Rheumatology and Clinical Immunology Units of the University of Pisa.

Sera were obtained from 170 patients with RA (121 women and 49 men, mean age 60 years [range 20–88 years], mean disease duration 8 years [range 6 months to 41 years]) and from 238 control subjects, including 31 patients with mixed cryoglobulinemia, 30 patients with SLE, 45 patients with systemic sclerosis (SSc), 29 patients with psoriatic arthritis (PsA), 26 patients with ankylosing spondylitis (AS), and 77 normal healthy subjects (blood donors and healthy laboratory personnel, age- and sex-matched with the RA patients).

The diagnoses of RA (17), SLE (18), and SSc (19) were based on the American College of Rheumatology criteria. Mixed cryoglobulinemia was diagnosed in the presence of Meltzer's triad (purpura, weakness, and arthritis/arthralgia) and cryoglobulins in the serum. The diagnosis of AS was based on the revised New York criteria (20), while the diagnosis of PsA was based on the criteria described by Vasey and Espinoza (21).

Peptide synthesis.

Synthetic peptides were obtained by solid-phase synthesis using 9-fluorenylmethoxycarbonyl–protected amino acids, according to the method described by Merrifield (22) and modified by Atherton and Sheppard (23), in the form of linear peptides or as multiple antigen peptides (MAPs) bearing 4 identical sequences on a lysine scaffold (24). The sequences are shown in Table 1. The peptides were at least 90% pure, as deduced from their elution pattern on reverse-phase high-performance liquid chromatography and their relative absorption at 214 nm.

Table 1. Sequences of synthetic peptides*
  • *

    EBNA-1 = Epstein-Barr nuclear antigen 1; X = citrulline; MAPs = multiple antigen peptides; VCP = viral citrullinated peptide; GC = Gly–Cit repeat.

Linear peptides  
 EBNA-135–58 Cit P4GPAGPXGGGXGXGXGXGXGHNDGGArg41+43+45+47+49+53→Cit

PAD treatment.

Purified PAD from rabbit skeletal muscle was purchased from Sigma (St. Louis, MO). MAP EBNA-135–58, plasminogen-depleted human fibrinogen (95% pure; Calbiochem, La Jolla, CA), recombinant EBNA-1 (a generous gift from Dr. L. Frappier, University of Toronto), and bovine serum albumin (BSA; Sigma) were incubated at 2 mg/ml with 7.5 units/ml of PAD in 0.1M Tris HCl (pH 7.4), 10 mM CaCl2, and 5 mM dithiothreitol for 2 hours at 50°C. The same proteins, diluted in the deimination buffer and kept at 50°C for 2 hours, were used as the control. After PAD treatment, protein integrity was checked by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and Coomassie staining, and deimination by an anti–chemically modified citrulline kit (Anti-Citrulline [modified] Detection Kit; Upstate Biotechnology, Lake Placid, NY).

Purification of anti–MAP EBNA-I35–58Cit (anti–viral citrullinated peptide [anti-VCP]) antibodies.

MAP EBNA-135–58Cit (VCP) was conjugated to CNBr-activated Sepharose (Sigma) according to standard procedures. Total immunoglobulins from 11 sera containing anti-VCP antibodies were precipitated with 50% saturated ammonium sulfate; the precipitates were dissolved in phosphate buffer (pH 7.4) and dialyzed overnight against phosphate buffered saline (PBS). Enriched immunoglobulin preparations were applied to the column, and the flowthrough was collected for subsequent analysis. The column was extensively washed with 20 mM Na2HPO4, 150 mM NaCl (pH 7.2), and the antibodies bound to the column were eluted by 0.1M glycine buffer (pH 2.8) (0.5 ml/fraction), immediately neutralized with 50 μl Tris 1M (pH 8.0), and dialyzed overnight against PBS. The anti-VCP antibody content in the eluates and flowthrough was tested by ELISA.

Binding of anti–citrullinated peptides/proteins.

Antipeptide antibodies were detected in the sera or in the eluates and flowthroughs of the peptide column by ELISA, as previously described (25). Briefly, the different peptides (either linear or MAP) were added to polystyrene plates (Nunc MaxiSorp F96; Nunc, Roskilde, Denmark) at 2 μg/ml in PBS and incubated overnight. Saturation was carried out with PBS containing 3% BSA for 45 minutes at room temperature. Sera diluted 1:200 and purified antibodies at different concentrations (20 μg/ml to 0.3 μg/ml) in PBS containing 1% BSA and 0.05% Tween 20 (PBSBT) were incubated on the plates for 3 hours at room temperature. After washings with PBS–1% Tween and PBS, anti-human IgG alkaline phosphatase (Sigma) conjugated 1:3,000 in PBSBT was added to the wells, and the plates were incubated for 2 hours at room temperature. Alkaline phosphatase activity was revealed with p-nitrophenyl phosphate in 50 mM Na2CO3 (pH 9.6). Anti–deiminated fibrinogen and anti–deiminated EBNA-1 antibodies were determined by the same procedure, except that the proteins were coupled to the plates at 5 μg/ml in 50 mM Na2CO3 (pH 9.6). Anti-CCP antibodies were detected with a commercial kit (QUANTA Lite CCP, Inova Diagnostics, San Diego, CA), according to the manufacturer's instructions. Antifilaggrin-derived deiminated peptide antibodies were detected using a research version of a line immunoassay (INNO-LIA RA; Innogenetics, Alpharetta, GA) based on 2 citrullinated synthetic peptides derived from human filaggrin (26). Protein G–purified immunoglobulins from normal human sera were included as negative controls.

Competition assays.

For the competition assays, the amount of anti-VCP antibodies that yielded 50% of maximum binding on the solid phase was determined. This amount of antibodies diluted in PBSBT was then preincubated with the synthetic peptides for 30 minutes at 37°C before being transferred to the antigen-coated plates. Thereafter, ELISAs were carried out in a manner similar to that for direct binding. MAP EBNA-135–58Arg was used as the negative control, and the same peptide as the one coupled to the plate was used as the positive control.

EBV transformation.

Lymphoblastoid cell lines were obtained by EBV transformation of peripheral blood mononuclear cells (PBMCs). B lymphocytes were purified from PBMCs using immunomagnetic selection with CD19 microbeads (Miltenyi Biotech, Bergisch Gladbach, Germany). All CD19+ preparations contained <3% of other PBMCs. Viral preparations of EBV strains from the B95.8 marmoset cell line were produced by the standard procedure (27). Cells were infected with EBV (105 transforming units) and cultured in complete medium in 75-cm2 tissue culture flasks.

Calcium ionophore treatment.

Lymphoblastoid cells (2 × 106) were suspended in 1 ml of Locke's solution (150 mM NaCl, 5 mM KCl, 10 mM HEPES HCl, pH 7.3, 2 mM CaCl2, 0.1% glucose) and incubated at 37°C for 15 minutes in the presence of 4 μM A23187, followed by a further incubation of 1 hour in serum-free RPMI at 37°C.


Calcium ionophore–treated cells were lysed in 25 mM Tris, 1 mM EDTA, pH 7.4, 150 mM NaCl, 1% Triton X-100, and protease inhibitors. The total cell lysate (2 × 106 cells/sample of immunoprecipitation) diluted in TETN250 buffer (25 mM Tris HCl, 5 mM EDTA, pH 7.4, 250 mM NaCl, 1% Triton X-100) was preabsorbed on heat-killed Staphylococcus aureus cells (Pansorbin; Calbiochem).

Protein G Sepharose beads (Sigma) were incubated with purified anti-VCP antibodies or control immunoglobulins for 4 hours at room temperature. After repeated washings, precleared lysate was added, and the tubes were incubated overnight at 4°C. The beads were washed twice with TETN250 and twice with 25 mM Tris HCl, 5 mM EDTA (pH 7.4). The immunoprecipitates were eluted in SDS-PAGE sample buffer, subjected to SDS-PAGE under nonreducing conditions, and blotted onto a polyvinylidene difluoride (PVDF) membrane (Hybond-P; Amersham, Little Chalfont, UK).

The membrane strips were saturated for 30 minutes at room temperature in Tris buffered saline (TBS) containing 3% BSA and incubated overnight at 4°C with a mouse monoclonal antibody specific for EBNA-1 (mouse monoclonal antibody anti–EBNA-1 p72/87; Advanced Biotechnologies, Columbia, MD) diluted 1:1,000 in TBS containing 3% BSA. After washings with TBS containing 0.1% Tween 20 (TBST), anti-mouse IgG–horseradish peroxidase diluted at 0.2 μg/ml in TBST was added, and the strips were incubated at room temperature for 2 hours. The Anti-Citrulline (modified) Detection Kit (Upstate Biotechnology) was used, following the manufacturer's instructions, to detect deiminated proteins. Peroxidase activity was visualized by means of enhanced chemiluminescence using SuperSignal West Dura Extended Duration Substrate (Pierce, Rockford, IL). Images were acquired and analyzed using the VersaDoc Imaging System and QuantityOne analysis software (Bio-Rad, Hercules, CA).


Binding of deiminated EBNA-135–58 peptides by RA sera.

We previously showed that antibodies reactive with EBNA-135–58 can be detected in normal subjects and in patients with EBV-related diseases or autoimmune disorders (28). In fact, using ELISA we showed that 30% of normal sera, 12% of sera from patients with Burkitt's lymphoma, 25% of sera from patients with RA, and 38% of sera from patients with SLE reacted with EBNA-135–58 synthesized as a MAP. When this peptide was subjected to in vitro deimination by PAD treatment and used as antigen on the solid phase, the pattern of sera reactivity was totally different. No reactivity with normal or SLE sera was detected, whereas 54% of RA sera bound deiminated MAP EBNA-135–58.

It is conceivable that the products of in vitro deimination are heterogeneous, i.e., that the peptides obtained differ in the number of arginine residues that have been transformed into citrulline. To better define the antibody specificity, we synthesized 4 linear peptides corresponding to the 35–58 sequence, in which a varying number of arginine residues were substituted by citrulline (Table 1). We discovered that the antibody binding was proportional to the number of citrulline residue substitutions; in fact, 28% of RA sera bound EBNA-135–58Cit P1, 31% bound EBNA-135–58Cit P2, 35% bound EBNA-135–58Cit P3, and 43% bound EBNA-135–58 Cit P4.

This VCP was then synthesized as a MAP and was used to screen sera from patients with different autoimmune disorders. As shown in Figure 1, VCP detected almost exclusively RA antibodies: 50% of RA sera bound VCP, compared with ∼3.5% of PsA, MC, and AS sera, 2% of SSc sera, and 0% of SLE sera.

Figure 1.

Binding of sera from patients with rheumatoid arthritis (RA) and other chronic arthritides. Sera diluted 1:200 were incubated on viral citrullinated peptide (VCP)–coated plates, and bound antibodies were detected with alkaline phosphatase–labeled anti-IgG antibodies. The upper limit of normal, set at the 97.5 percentile of normal healthy sera (NHS), was 27% (broken line). Anti-VCP antibodies were detected in 1 of 29 patients with psoriatic arthritis (PsA), 85 of 170 patients with RA, 1 of 31 patients with mixed cryoglobulinemia (MC), none of 30 patients with systemic lupus erythematosus (SLE), 1 of 26 patients with ankylosing spondylitis (AS), and 1 of 45 patients with systemic sclerosis (SSc). Results are expressed as the percentage of a reference serum.

Binding of Gly–Cit repeats by RA-specific antibodies.

To more precisely define the epitope recognized by the anti-VCP antibodies, we synthesized a peptide corresponding to the core sequence of the VCP containing 6 Gly–Cit repeats (MAP GC) (Table 1). VCP-specific antibodies were purified from 11 patient sera by affinity chromatography. All of the anti-VCP–purified antibodies bound MAP GC. When compared in liquid-phase inhibition assays, both MAP GC and VCP inhibited the binding of antibodies to solid-phase VCP (Figure 2A); 50% inhibition was obtained with <10 μg/ml peptide. These data from inhibition assays, together with the results of direct binding (performed in high detergent conditions), suggest a high affinity of anti-VCP antibodies.

Figure 2.

Inhibition of binding of anti–viral citrullinated peptide (anti-VCP) antibodies to VCP and to citrullinated peptide (CCP). Affinity-purified anti-VCP antibodies were preincubated with different amounts of multiple antigen peptide (MAP) Epstein-Barr nuclear antigen 1 (EBNA-1)35–58Arg, VCP, or MAP Gly–Cit repeat (GC) and then transferred to A, VCP-coated plates or B, CCP-coated plates. VCP and MAP GC inhibited the binding of anti-VCP antibodies to VCP (A) and CCP (B), while MAP EBNA-135–58Arg exerted no inhibition. Results are expressed as the percent binding (100% = binding in the absence of peptides). Values are the mean.

Binding of in vitro deiminated recombinant EBNA-I by affinity-purified anti-VCP antibodies.

Anti-VCP antibodies were tested by ELISA on undigested and in vitro deiminated recombinant EBNA-1. Antibodies purified from patients with RA bound exclusively deiminated EBNA-1, with the exception of antibodies from RA patient 5, which also bound native EBNA-1, albeit to a lower extent (Figure 3). It is likely that in this case a population of antibodies reactive with the nondeiminated sequences of VCP was co-purified together with antibodies to the Gly–Cit repeat. In contrast, control immunoglobulins showed the same reactivity with deiminated and native EBNA-1, suggesting that they recognize epitopes not containing citrulline.

Figure 3.

Binding of anti–viral citrullinated peptide (anti-VCP) antibodies to deiminated recombinant Epstein-Barr nuclear antigen 1 (EBNA-1). Affinity-purified anti-VCP antibodies and control immunoglobulins (10 μg/ml) were tested on plates coated with undigested or peptidylarginine deiminase–treated recombinant EBNA-1. Bound antibodies were detected with alkaline phosphatase–labeled anti-IgG antibodies. All of the anti-VCP antibodies except those from rheumatoid arthritis (RA) patient 5 bound exclusively deiminated EBNA-1. C = control.

Immunoprecipitation of in vivo deiminated EBNA-I by affinity-purified anti-VCP antibodies.

Lymphoblastoid cell lines (LCLs) were treated with the calcium ionophore A23187, lysed, and immunoprecipitated with either anti-VCP antibodies or control immunoglobulins. The immunoprecipitate was subjected to SDS-PAGE, transferred to PVDF, and probed with a monoclonal anti–EBNA-1 antibody and with polyclonal rabbit anti–modified citrulline antibodies. An 80-kd band was detected on the immunoprecipitate of anti-VCP antibodies by the anti–EBNA-1 antibody and by anti–modified citrulline antibodies, suggesting that the immunoprecipitated protein is indeed deiminated EBNA-1 (Figure 4). An additional band (∼65 kd) immunoprecipitated by anti-VCP antibodies was detected with anti–modified citrulline antibodies but not with anti–EBNA-1. Expression of this protein in other infected or noninfected cell lines and its characterization are undergoing testing.

Figure 4.

Immunoprecipitation of deiminated Epstein-Barr nuclear antigen 1 (EBNA-1) from calcium ionophore–treated lymphoblastoid cell lines (LCLs). LCLs were stimulated with the calcium ionophore A23187, lysed, and incubated with anti–viral citrullinated peptide (anti-VCP) antibodies or control immunoglobulins. The protein G–bound complexes were eluted in nonreducing sample buffer, loaded onto the gel, and transferred to a polyvinylidene difluoride membrane. Immunoprecipitated proteins were incubated with a monoclonal anti–EBNA-1 antibody or with anti–modified citrulline (anti-MC) antibody. An 80-kd protein was detected in the immunoprecipitate of anti-VCP antibodies, with both anti–EBNA-1 and anti-MC antibodies.

Binding of filaggrin-derived deiminated peptides and other deiminated proteins by affinity-purified anti-VCP antibodies.

The CCP assay is the test most widely used to detect antibodies specific for deiminated filaggrin. Anti-VCP antibodies were all positive in the CCP assay, and their binding was completely inhibited by preincubation with VCP or MAP GC (Figure 2B).

Finally, when tested on nitrocellulose strips containing filaggrin-derived deiminated peptides (26), anti-VCP antibodies all reacted with the 2 bands corresponding to the deiminated peptides (Table 2). Deiminated fibrinogen/fibrin has been proposed as the main target of the RA-specific immune response (10). We therefore tested the capacity of anti-VCP antibodies to bind in vitro deiminated fibrinogen. Anti-VCP antibodies all bound the deiminated fibrinogen, whereas no binding was observed with untreated fibrinogen or with deiminated BSA (Table 2).

Table 2. Reactivity of anti-VCP antibodies and control immunoglobulins with deiminated peptides and proteins*
PatientMAP EBNA-135–58 ArgVCPCCPMAP GCFilaggrin-derived deiminated peptidesDeiminated fibrinogenUndeiminated fibrinogenDeiminated BSA
  • *

    Anti-VCP = anti–viral citrullinated peptide; MAP = multiple antigen peptide; EBNA-1 = Epstein-Barr nuclear antigen 1; CCP = cyclic citrullinated peptide; GC = Gly–Cit repeat; BSA = bovine serum albumin; C = control; RA = rheumatoid arthritis.



The data presented here indicate that sera from patients with RA react with a deiminated protein encoded by EBV. Antibodies that are present exclusively in RA sera bind the citrullinated peptide corresponding to sequence 35–58 of EBNA-1, recognize it in the context of the whole protein, and, as suggested by immunoprecipitation, bind in vivo deiminated EBNA-1. These results suggest a role for EBV infection in the induction of disease-specific antibodies in RA. Furthermore, the viral citrullinated peptide may offer a new diagnostic tool for RA.

The EBNA-1 protein is expressed in the nuclei of infected cells (29). It initiates replication by binding to the EBV DNA episome via its COOH-terminal domain and then crosslinking the episome to mitotic chromosomes as a protein anchor. It has an unusual structure in that it contains in its central portion a Gly–Ala repeat that constitutes one-third of the molecule (30). This repeat represents a dominant epitope in the anti–EBNA-1 immune response that follows EBV infection (31, 32), but antibodies against other portions of the molecule are present in immune sera. Healthy individuals, as well as those affected by autoimmune or EBV-related disorders, produce antibodies to the N-terminal EBNA-135–58 sequence that contains the 6-fold Gly–Arg repeat (28). Reactivity with this sequence does not differ between healthy persons and patients with RA. On the contrary, the substitution of arginine with citrulline transforms this sequence into a specific target of RA antibodies. Thus, 50% of RA sera but only 0–3% of sera from healthy individuals or from patients with other connective tissue disorders bind the citrullinated EBNA-135–58 peptide. Moreover, anti-VCP antibodies bind recombinant EBNA-1 after PAD treatment, thus demonstrating that the protein is susceptible to deimination and that the antibodies can recognize the 35–58 deiminated peptide in the context of the entire protein.

From a biologic standpoint, these findings are supported by the observation that anti-VCP antibodies bind in vivo deiminated EBNA-1 from calcium ionophore–stimulated LCLs. In fact, they are able to immunoprecipitate an 80-kd protein recognized by both monoclonal anti–EBNA-1 antibody and by anti–modified citrulline antibodies.

Anti-VCP antibodies also recognize filaggrin-derived deiminated peptides and in vitro deiminated fibrinogen, and both bindings are cross-inhibited by preincubation with VCP, indicating that the same antigen-binding site is involved in the recognition of these deiminated proteins.

It was previously reported that anti–deiminated filaggrin antibodies from patients with RA react with in vitro deiminated fibrinogen, and that purified anti–deiminated fibrinogen antibodies from the same patients bind all of the epithelial and synovial targets of AFAs (10). Taken together with the results of our study, these data provide biochemical evidence for the presence of common epitopes on different deiminated proteins. The recognition of such epitopes characterizes a family of RA-associated antibodies that share the same disease specificity and therefore have been labeled anti–citrullinated protein antibodies (ACPAs) (33).

In order to more precisely map the epitope recognized by anti-VCP antibodies within the EBNA-1 peptide, we synthesized a peptide characterized by 6 Gly–Cit repeats (MAP GC) contained within the VCP. Most sera that are reactive with VCP recognize MAP GC (data not shown). Results of inhibition assays indicate that preincubation of anti-VCP antibodies with MAP GC (but not with MAP EBNA-135–58Arg) inhibits binding to solid-phase VCP. However, MAP GC is a less efficient competitor than VCP, suggesting that the amino acids flanking the repeat are important for antibody recognition—either forming part of the epitope or serving as a scaffold for its 3-dimensional structure. The analysis of clonal populations of antibodies may help to elucidate this point. Experiments are now in progress to accurately characterize the epitopes contained in the VCP sequence and the fine specificity of the antibodies that bind it. MAP GC also inhibits the binding of anti-VCP antibodies to CCP, thus showing that one of the epitopes recognized by ACPAs is indeed a stretch of citrullines flanked by uncharged amino acids with relatively small side chains.

The lack of monoclonal ACPAs and the general difficulty of inducing ACPAs in experimental animals have hampered the biochemical analysis of their specificity, and to date a comprehensive picture of the mechanisms leading to their production is lacking. ACPAs are high-affinity IgG antibodies and as such are produced with the aid of T cells, but the specificity of these helper T cells remains elusive. T cell epitopes have been identified on 2 deiminated self proteins that are targets of ACPAs: a deiminated vimentin peptide has been shown to induce a strong T cell proliferative response in mice transgenic for HLA–DRB1 (34), and T cells specific for the glycosylated epitopes on type II collagen may help in the production of antibodies directed to deiminated type II collagen sequences (35).

Several self proteins (filaggrin, fibrin, vimentin, and possibly type II collagen) are in vivo deiminated and can elicit or maintain the immune response to deiminated proteins in patients with RA. In contrast, the involvement of exogenous antigens in the production of ACPAs has not been previously suggested.

The reactivity of RA sera with a deiminated protein encoded by EBV raises the issue of the role of the virus in inducing ACPAs and opens up new perspectives that may help us decipher the mechanisms leading to the production of these antibodies in RA. EBV is considered to be one of the environmental agents that contributes to the pathogenesis of RA. EBV infection is widespread, and 95% of all adults display serologic signs of a previous infection. It is known that patients with RA have elevated levels of antibodies to latent and replicative EBV proteins (36) and in particular to EBNA-1 (37, 38).

It has been convincingly demonstrated that peripheral T lymphocytes from patients with RA are not efficient in the killing of autologous EBV-infected lymphoblastoid B lines (39), and that the frequency of EBV-infected peripheral B lymphocytes is higher in patients with RA than in controls (40).

More recently, it was shown that the EBV DNA content in PBMCs from patients with RA is 10-fold higher than that in PBMCs from healthy individuals or from patients with other inflammatory arthritides (41). Another study detected EBV infection in synovial tissue from patients with RA but not in patients with osteoarthritis (42, 43), although other investigators failed to confirm these findings (44). Although it is not yet clear whether this viral overload is specific for EBV, it certainly could provide a chronic antigenic stimulus leading to the processing of viral proteins and the presentation of EBV-derived peptides.

EBNA-1–specific CD4+ T cells (45, 46), mainly of the Th1 phenotype (47), have recently been detected in healthy EBV carriers and may provide help for the production of antibodies directed to the different epitopes of this protein. In the presence of a higher EBV load, events such as apoptosis, infection, damage, or any stimulus involving calcium influx may activate PAD, which is highly expressed in RA synovia (48, 49), and induce the deimination of different proteins, including viral proteins. As a result, EBNA-1 may contain posttranslationally acquired deiminated sequences. Naive B cells specific for deiminated antigen and normally present in the B cell repertoire (50) could undergo affinity maturation with the help of EBNA-1–specific T cells, giving rise to the production of ACPAs that takes place in patients with RA.

Based on this model, EBNA-1–specific T cells could play a role in ACPA production, and anti-VCP antibodies would then be the very first ACPAs produced. Subsequently, the immune response at both the T cell level and the B cell level could spread toward the deiminated self proteins. Experiments are now under way to test these hypotheses.


We are indebted to Dr. Lori Frappier (University of Toronto) for the generous gift of recombinant EBNA-1 and to Ms Lisa Chien for revising the manuscript.