Peptidylarginine deiminase from Porphyromonas gingivalis citrullinates human fibrinogen and α-enolase: Implications for autoimmunity in rheumatoid arthritis

Authors


Abstract

Objective

To investigate protein citrullination by the periodontal pathogen Porphyromonas gingivalis as a potential mechanism for breaking tolerance to citrullinated proteins in rheumatoid arthritis (RA).

Methods

The expression of endogenous citrullinated proteins was analyzed by immunoblotting of cell extracts from P gingivalis and 10 other oral bacteria. P gingivalis–knockout strains lacking the bacterial peptidylarginine deiminases (PADs) or gingipains were created to assess the role of these enzymes in citrullination. Citrullination of human fibrinogen and α-enolase by P gingivalis was studied by incubating live wild-type and knockout strains with the proteins and analyzing the products by immunoblotting and mass spectrometry.

Results

Endogenous protein citrullination was abundant in P gingivalis but lacking in the other oral bacteria. Deletion of the bacterial PAD gene resulted in complete abrogation of protein citrullination. Inactivation of arginine gingipains, but not lysine gingipains, led to decreased citrullination. Incubation of wild-type P gingivalis with fibrinogen or α-enolase caused degradation of the proteins and citrullination of the resulting peptides at carboxy-terminal arginine residues, which were identified by mass spectrometry.

Conclusion

Our findings demonstrate that among the oral bacterial pathogens tested, P gingivalis is unique in its ability to citrullinate proteins. We further show that P gingivalis rapidly generates citrullinated host peptides by proteolytic cleavage at Arg-X peptide bonds by arginine gingipains, followed by citrullination of carboxy-terminal arginines by bacterial PAD. Our results suggest a novel model where P gingivalis–mediated citrullination of bacterial and host proteins provides a molecular mechanism for generating antigens that drive the autoimmune response in RA.

Rheumatoid arthritis (RA) is characterized by disease-specific autoimmunity to citrullinated proteins. Citrullination is a posttranslational modification of arginine residues that is mediated by the family of peptidylarginine deiminases (PADs). Citrullinated fibrin(ogen) and α-enolase are 2 of the physiologic proteins that are targeted by anti–citrullinated protein antibodies in RA (1–5). Fibrinogen is the precursor of fibrin, and autoantibodies to citrullinated fibrin(ogen) are found in up to 66% of patients with RA (6). Alpha-enolase is an evolutionarily conserved, multifunctional protein (7) that is best known for its role in glucose metabolism and, more recently, as a plasminogen-binding protein on the surface of various mammalian and prokaryotic cell types (8, 9). Autoantibodies to citrullinated α-enolase can be found in 40–60% of patients with RA (4–6). The pathogenicity of these autoantibodies may be mediated by the formation of immune complexes with citrullinated host proteins in the joint and by activation of downstream inflammatory pathways via complement fixation and Fcγ receptor activation (10–13).

It is not yet known which factors trigger the breakdown of tolerance to citrullinated proteins. Protein citrullination is part of healthy physiology, with citrullinated filaggrin having been identified in healthy skin (14), and is part of the inflammatory response in general (15), whereas the formation of autoantibodies to citrullinated proteins is largely restricted to RA (16). Deposits of citrullinated fibrin have been found in a variety of inflammatory joint conditions without an accompanying autoantibody response (17, 18). Additional environmental and genetic risk factors are therefore likely to be required.

To date, tobacco exposure and the presence of certain alleles in the HLA–DRB1 locus with a common peptide-binding motif, collectively known as the shared epitope, have been identified as susceptibility factors for the development of autoantibodies to citrullinated proteins (19, 20), α-enolase and vimentin in particular (21), but these do not explain the total risk. Additional etiologic pathways require consideration, with the periodontal pathogen Porphyromonas gingivalis being a prime candidate for investigation.

Periodontitis, in which P gingivalis is a major causative agent, is a chronic inflammatory disease of the supporting tissues of the teeth, with an estimated prevalence of 4.2% in the US population (22). P gingivalis can be detected in 80–90% of periodontitis patients and in 10–30% of healthy subjects (23, 24). The bacterium has recently attracted interest based on epidemiologic links between RA and periodontitis (25) and the description of a novel bacterial PAD (26) (hereinafter called PPAD), suggesting a potential etiologic role of P gingivalis in RA through the generation of citrullinated antigens.

The pathophysiologic mechanisms of periodontitis are similar to those of RA. The condition is characterized by the resorption of the supporting bony structure around the teeth and is mediated by a variety of proinflammatory molecules, including tumor necrosis factor α, interleukin-1β, prostaglandin E2, and matrix metalloproteinases (27). A number of studies have indicated a positive association between the prevalence of periodontitis and RA (25, 28), even when adjusted for smoking, which is a major risk factor for both diseases. We have shown that RA-specific autoantibodies to citrullinated α-enolase peptide 1, the immunodominant B cell epitope of human α-enolase, cross-react with in vitro–citrullinated enolase from P gingivalis (5), raising the possibility of molecular mimicry between epitopes from citrullinated bacterial and human enolases. P gingivalis is the only prokaryote described to date that expresses a functional bacterial PAD, though its physiologic substrates are unknown, as are the molecular mechanisms of citrullination.

PPAD displays no amino acid sequence similarity to the human PAD enzymes, and a previous study indicated that it might preferentially target carboxy-terminal arginine residues (26). This is in contrast to the human enzymes, which efficiently deiminate internal arginine residues (29). Citrullination of bacterial and host proteins and peptides by P gingivalis PAD could therefore create new epitopes and, given the infectious context providing endogenous and exogenous danger signals, trigger a latent antibody response to citrullinated bacterial and host proteins in susceptible individuals.

In the present study, we aimed to elucidate the molecular requirements for bacterial and human protein citrullination by P gingivalis PAD and thus advance our understanding of the potential underlying mechanisms for the generation of citrullinated antigens and the induction of autoimmunity in RA.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Porphyromonas gingivalis wild-type strain (W83), P gingivalis clinical isolates obtained from patients with severe periodontitis (MaRL, D243, JH16, and J430), and P gingivalis mutants (Δppad, ppad+, Δrgp, Δkgp, and Δrgp+kgp) were grown in Schaedler anaerobe broth (Oxoid), supplemented with 5% sheep blood, at 37°C in an anaerobic chamber (90% N2, 5% CO2, and 5% H2). Erythromycin or tetracycline was used at 5 μg/ml or 1 μg/ml, respectively, on solid media. The concentrations were doubled for selective growth in liquid culture.

Other anaerobic oral bacteria (Prevotella intermedia H13 [clinical isolate], Prevotella oralis [ATCC 33269], Capnocytophaga gingivalis [ATCC 33624], and Capnocytophaga ochracea [ATCC 27872]) were grown in Schaedler anaerobe broth, supplemented with 2.5 μg/liter of vitamin K, at 37°C in an anaerobic chamber (90% N2, 5% CO2, and 5% H2). Fusobacterium nucleatum (ATCC 10953) was grown in Schaedler anaerobe broth at 37°C in an anaerobic chamber (80% N2, 10% CO2, and 10% H2). Aggregatibacter actinomycetemcomitans (ATCC 43718) was grown in tryptic soy broth (Sigma), supplemented with 6% yeast extract and 8% glucose, at 37°C in an atmosphere consisting of 5% CO2. Aerobic bacteria (Streptococcus constellatus [ATCC 27823], Streptococcus gordonii [ATCC 10558], Streptococcus sanguinis [ATCC 10556], and Streptococcus salivarius [ATCC 7073]) were grown on Columbia agar plates, supplemented with 8% defibrinated sheep blood or brain/heart infusion broth.

Construction of P gingivalis mutant strain ppad.

A 1-kb region 3′ to the P gingivalis ppad gene (GenBank accession no. 2552184; locus tag PG1424) was amplified by polymerase chain reaction (PCR; primers 5′-GCTCTAGATGGAATCCGTGAGACAATG and 5′-TAAGCATGCGATATTTGTCGGAAGGACTC) for insertion into the Xba I and Sph I sites of the pUC19 plasmid (New England BioLabs). An erythromycin resistance cassette ermF/ermAM from plasmid pVA2198 was amplified and inserted into the Sma I and Xba I sites of the modified pUC19 plasmid. The resultant plasmid was modified further by incorporating an amplified 1-kb region 5′ to the ppad gene (primers 5′-AAGAGCTCAAGCACGTAATAAGGACAATGA and 5′-TTATCCCGGGTGTTCCTGAACATATGATAAGATCT) into the Sac I and Sma I sites to create the deletional inactivation plasmid construct (pΔppad) or by incorporating the entire ppad gene and a 1-kb region 5′ to the gene (primers 5′-AAGAGCTCAAGCACGTAATAAGGACAATGA and 5′-TTATCCCGGGTGTCTACCTGAGGAGTATTCT) into the Sac I and Sma I sites to create the control mutant construct (pppad+) to control for possible polar effects.

The correct placement and orientation of the DNA segments were confirmed by sequencing. The modified plasmid constructs were integrated into the P gingivalis W83 genome by a double-crossover recombination event by electroporation using standard protocols (30). Erythromycin-resistant clones were subcultured on selective plates, and genomic integration was confirmed by PCR using primers from outside of the cloned regions surrounding the ppad gene.

Construction of P gingivalis mutant strains rgp and kgp.

The general procedure for construction of the Δrgp and Δkgp mutants has been described elsewhere (30). Homologous recombination of the ▵kgp plasmid into the P gingivalisrgp mutant genome resulted in a kgp-rgp–deficient mutant (Δrgp+kgp). The respective phenotypes were confirmed by enzymatic assays and Western blot analysis.

Preparation of bacterial whole-cell extracts.

Bacterial cultures were grown in liquid media until the early stationary phase. Twenty milliliters of the culture was centrifuged at 10,000g for 15 minutes at 4°C, and the resulting bacterial pellet was resuspended in phosphate buffered saline (PBS). The optical density (OD) at 600 nm was measured and adjusted to 1.0 with PBS, and the suspension was sonicated on ice. Sodium azide (final concentration 0.02% volume/volume) was added to all samples as a preservative.

Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE).

Protein samples were mixed with reducing 4× lithium dodecyl sulfate (LDS) sample buffer (Invitrogen), heated for 10 minutes at 70°C, and resolved on 12% NuPAGE Bis-Tris gels (Invitrogen) using MOPS running buffer. After electrophoresis, proteins were stained with the Coomassie-based stain InstantBlue (Triple Red) or transferred to nitrocellulose membranes for immunoblotting. For analysis of fibrinogen/enolase-derived peptides, 10–20% Tricine gels, 2× Tricine sample buffer, and Tricine SDS running buffer (all from Invitrogen) were used, and protein bands were visualized using a standard silver staining protocol.

Detection of citrullinated proteins by immunoblotting and dot-blotting.

Citrullinated proteins were detected using an anticitrulline (modified) detection kit (Upstate/Millipore) in accordance with the manufacturer's instructions. For dot-blotting, 10 μl of sample was spotted onto an equilibrated nitrocellulose membrane (0.1-μm Protran membrane; Whatman) and allowed to dry before proceeding with the standard Western blotting protocol. Controls were performed in which the modification step or the primary antibody was omitted to control for nonspecific binding by the primary antibody to structures other than modified citrulline side chains or for binding of the secondary antibody to proteins other than the primary antibody, respectively.

Analysis of fibrinogen and α-enolase citrullination by live P gingivalis.

P gingivalis was cultured as described above, and the OD600 nm was measured. Bacterial cells were pelleted, washed in ice-cold PBS, and resuspended in assay buffer (10 mM HEPES, 150 mM NaCl, 1 mM CaCl2, pH 7.5, and 10 mML-cysteine) to a final OD600 nm of 1.0. Purified fibrinogen was purchased from Sigma-Aldrich (catalog no. F3879), and recombinant human α-enolase was expressed in Escherichia coli as previously described (31). The proteins were diluted in HEPES buffer at a concentration of 0.5 mg/ml. Equal volumes of protein solution and bacterial cell suspension were mixed, and an aliquot was immediately withdrawn (corresponds to time point of 1 minute). The cultures were then incubated at 37°C on a shaking platform, and further aliquots were withdrawn after 1.5 hours, 3 hours, and 6 hours. Bacterial cells were immediately removed from all aliquots by centrifugation. The resulting supernatant was used for analysis on SDS-PAGE gels and immunoblotting as described above and for protein precipitation with 15% meta-phosphoric acid, leaving small peptides in solution, and subsequent analysis by high-performance liquid chromatography (HPLC). HPLC peak fractions were collected and subjected to peptide analysis by mass spectrometry.

HPLC analysis.

HPLC was performed with a Shimadzu VP series chromatograph equipped with a Supelcosil LC-318 reverse-phase column measuring 25 cm × 4.6 mm (Sigma-Aldrich). Samples (100 μl) were injected and eluted with a gradient of H2O/0.1% trifluoroacetic acid (TFA) (solution A) and 80% acetonitrile/0.08% TFA (solution B) and were monitored at 215 nm with a Shimadzu SPD-10A UV-Vis detector. Peaks were collected manually and freeze-dried prior to analysis by mass spectrometry.

In-gel digestion of proteins for mass spectrometry.

Protein bands were excised with a scalpel, and in-gel digestion was performed using a robotic system (Investigator ProGest; Genomic Solutions). The bands were washed in 100 mM ammonium bicarbonate buffer and dehydrated in 100% acetonitrile. Cysteine residues were reduced with 10 mM DTT, then carboamidomethylated with 55 mM iodoacetamide. Digestion was performed for 6 hours at 37°C by the addition of modified porcine trypsin (10 μl at 6.5 ng/μl in 25 mM ammonium bicarbonate), and peptides were recovered by sequential extraction with 25 mM ammonium bicarbonate buffer, 5% formic acid, and acetonitrile. Extracts were pooled, lyophilized, and redissolved in 0.1% formic acid prior to performing mass spectrometry.

Analysis by mass spectrometry.

Tandem electrospray mass spectra were recorded with a Q-Tof hybrid quadrupole/orthogonal acceleration time-of-flight spectrometer (Waters) interfaced to a CapLC chromatograph. Freeze-dried peptide samples were redissolved in 0.1% formic acid, and 6 μl was injected onto a PepMap C18 column (300 μm × 0.5 cm; LC Packings) and eluted with an acetonitrile/0.1% formic acid gradient at a flow rate of 1 μl/minute. The capillary voltage was set to 3,500V, and data-dependent product ion scans were performed on precursor ions with charge states of 2, 3, or 4 over a survey mass range of m/z 400 to 1,400. The raw spectra were smoothed, deisotoped, transformed onto a singly charged mass/charge (m/z) axis using a maximum entropy method as implemented in the peptide auto module of MassLynx (Waters), and then saved in the peaklist (pkl) format prior to database searching.

Proteins were identified by correlation of uninterpreted spectra to entries in the Swiss-Prot/TrEMBL database using the ProteinLynx Global Server (version 1.1; Waters) and a local installation of Mascot, version 2.2 (www.matrixscience.com). The database used was a FASTA format composite constructed in-house by merging Swiss-Prot, TrEMBL, and associated splice variants (release date May 26, 2009; 8,413,758 sequences). Searches were run in error-tolerant mode, and no mass or taxomic constraints were applied. The initial enzyme specificity was set to trypsin, but subsequent searches of the P gingivalis digestions of fibrinogen and enolase were repeated with no enzyme specificity in order to match peptides resulting from the combination of gingipain activity with other enzymes, such as aminopeptidases and carboxypeptidases. All spectra matching citrullinated peptides were reviewed manually by interpretation of sequence-specific fragment ions to confirm the presence and location of the citrulline residue and to exclude other modifications, such as deamidation of aspartic acid, which also result in a mass increase of 1 dalton.

RESULTS

Expression of endogenous citrullinated proteins is unique for P gingivalis.

In order to test whether P gingivalis citrullinates its own endogenous proteins, cultures of P gingivalis, comprising reference strain W83 and 4 clinical isolates from patients with periodontal disease, were grown to stationary phase, and whole-cell lysates were analyzed by immunoblotting using anti–modified citrulline (AMC) antibody. We observed strong, distinct bands, with a similar pattern of citrullinated proteins in all strains tested (Figure 1A). The gene encoding PPAD was detected in all P gingivalis strains tested, including the clinical isolates, as determined by PCR (data not shown). Subcellular fractionation of the wild-type strain further showed that the majority of citrullinated proteins were associated with the periplasm and the outer and inner membrane fractions (data not shown), which is a typical feature of bacterial virulence factors and antigens.

Figure 1.

Expression of endogenous citrullinated proteins is ubiquitous in Porphyromonas gingivalis, but not in 10 other oral bacteria. A, Protein citrullination in total cell extracts of the P gingivalis wild-type reference strain W83 (lane 3) and in 4 clinical isolates (lanes 4–7, corresponding to strains MaRL, D243, JH16, and J430) was analyzed by immunoblotting with anti–modified citrulline (AMC) antibody. Controls in which the modification step (lane 1) or the secondary antibody (lane 2) had been omitted were run in parallel. Molecular mass markers are shown on the left. B, Total cell extracts of 10 other prominent oral bacteria were tested for endogenous protein citrullination using the AMC antibody. Background signals were confirmed to stem from nonspecific binding of the primary antibody (control). Pg = Porphyromonas gingivalis; Fn = Fusobacterium nucleatum; Aa = Aggregatibacter actinomycetemcomitans; Pi = Prevotella intermedia; Po = Prevotella oralis; Cg = Capnocytophaga gingivalis; Co = Capnocytophaga ochracea; Sc = Streptococcus constellatus; Sg = Streptococcus gordonii; Sn = Streptococcus sanguinis; Sl = Streptococcus salivarius.

To examine whether endogenous citrullination is a unique ability of P gingivalis within the community of oral pathogens, we tested whole cell lysates of 10 other oral organisms for the presence of citrullinated proteins. None were detected except in P gingivalis (Figure 1B), suggesting that functional PAD enzymes are absent from the other strains tested. Weak bands were noticeable in a number of strains, but were the result of nonspecific antibody binding (see controls in Figure 1B).

Endogenous protein citrullination is dependent on the bacterial PAD enzyme in cooperation with protein cleavage by arginine gingipains.

To confirm that the observed endogenous protein citrullination in P gingivalis is due to the enzymatic activity of PPAD, and to rule out the possibility of a second, uncharacterized, bacterial PAD enzyme, we created a P gingivalis W83–knockout strain (Δppad) by replacement of the entire ppad-encoding DNA sequence with an antibiotic cassette. A strain in which the antibiotic cassette was inserted behind the ppad gene was used as a control against polar effects from genetic manipulations (ppad+). Immunoblotting for citrullinated proteins showed that Δppad entirely lacked endogenous citrullinated proteins, while the wild-type strain and the control strain (ppad+) showed a similar pattern and intensity of citrullinated proteins (Figure 2A). These data demonstrated that PPAD is essential for citrullination of endogenous proteins in P gingivalis and that it possesses only 1 peptidylarginine deiminase.

Figure 2.

Citrullination in Porphyromonas gingivalis depends on the bacterial peptidylarginine deiminase (PPAD) and is influenced by arginine gingipain–mediated proteolytic cleavage of substrate proteins. A, A P gingivalis mutant strain lacking the PPAD gene (Δppad) was constructed, and total cell extracts were analyzed for the presence of citrullinated proteins by immunoblotting with anti–modified citrulline (AMC) antibody or by staining with the Coomassie-based stain InstantBlue. A control mutant containing the entire ppad gene and the antibiotic cassette (ppad+) was created to control for possible polar effects. The P gingivalis wild-type (WT) strain W83 was used as positive control. Molecular mass markers are shown on the left. B,P gingivalis mutant strains lacking arginine gingipain (Δrgp), lysine gingipain (Δkgp), or both (Δrgp+kgp) proteolytic activities were analyzed for the presence of citrullinated proteins by immunoblotting with AMC antibody or by staining with the Coomassie-based stain InstantBlue.

We then examined how citrullination depends on the activity of the major virulence factors in P gingivalis, which are called gingipains (32, 33). Gingipains are potent proteases and cleave various proteins/peptides after either arginine residues (i.e., arginine gingipain [Rgp]) or lysine residues (i.e., lysine gingipain [Kgp]), resulting in peptides with carboxy-terminal arginine or lysine residues. The reported preference of native PPAD for carboxy-terminal arginine residues in vitro (26) could be mediated through the activity of arginine gingipains, and as such, this would have important consequences for the type of citrullinated peptides that can be generated by P gingivalis. Hence, we studied P gingivalis mutants lacking functional arginine (Δrgp), lysine (Δkgp), or both (Δrgp+kgp) types of gingipains for endogenous citrullination. Immunoblotting of whole-cell lysates showed a significantly decreased level, but not complete abrogation, of citrullinated proteins in the Δrgp and Δrgp+kgp strains, but not in the Δkgp strain (Figure 2B), confirming that arginine gingipains play a role in protein citrullination, probably by generating proteins with carboxy-terminal arginine residues that are subsequently citrullinated by PPAD. The residual citrullinated proteins seen in the Δrgp and Δrgp+kgp strains might be due to the presence of proteins that naturally contain a carboxy-terminal arginine residue, which had not been proteolytically processed, and therefore appeared at different molecular weights as compared with the wild-type.

P gingivalis rapidly generates citrullinated fibrinogen and α-enolase peptides by proteolytic cleavage at Arg-X peptide bonds, followed by citrullination of carboxy-terminal arginines.

Having found that citrullination of endogenous P gingivalis proteins depends on the presence of PPAD and is influenced by arginine gingipains, the question arose whether the same principles apply to human proteins. We initially chose human fibrinogen for this study, since it is a major RA autoantigen in its citrullinated form (2, 3, 34) and a major part of the inflammatory response in general because of its function in the coagulation and platelet aggregation cascade. Fibrinogen is also involved in the pathogenesis of periodontitis, where it is abundantly found in the periodontal lesion, being an established target protein of gingipains (35). We investigated the potential of P gingivalis to citrullinate human fibrinogen using intact, live wild-type, Δppad-knockout, and Δrgp-knockout strains.

Fibrinogen was rapidly cleaved by wild-type P gingivalis (Figure 3A), which is consistent with previous reports (36). A similar degradation pattern was observed with the Δppad strain, indicating that citrullination of substrate proteins, including gingipains themselves, is not essential to the proteolytic potency of gingipains. As expected, fibrinogen samples incubated with the Δrgp strain showed considerably decreased proteolytic cleavage and a different pattern of cleaved peptides, with the residual proteolytic activity being mainly due to cleavage by lysine gingipains and other proteinases and peptidases from P gingivalis (37–40). Our analysis of the identity of the protein bands by mass spectrometry confirmed that the majority were derived from any of the 3 fibrinogen chains (Figure 3A). To exclude cleavage of protein by plasma-derived proteases, which may contaminate fibrinogen, we further performed control reactions in which protein alone was incubated in assay buffer, and no such cleavage was detected.

Figure 3.

Porphyromonas gingivalis rapidly cleaves human fibrinogen through arginine gingipain activity, and the resulting peptides are citrullinated at the carboxy-terminus by bacterial peptidylarginine deiminase (PAD). A, Fibrinogen fragments were incubated for 1 minute, 1.5 hours, 3 hours, or 6 hours with P gingivalis wild-type (WT) strain or with mutant strains lacking bacterial PAD (Δppad) or arginine gingipain (Δrgp), resolved by sodium dodecyl sulfate–polyacrylamide gel electrophoresis, and visualized using silver staining. A fibrinogen sample prior to incubation with P gingivalis (pre) served as a control. Protein bands were analyzed by mass spectrometry. Bands labeled α, β, and γ indicate the α, β, and γ chains of fibrinogen. B, All samples with visible protein bands from the experiments shown in A were tested for citrullination by immunoblotting with anti–modified citrulline (AMC). Controls for the preincubation sample were performed in which the modification step (pre ctrl1) or the secondary antibody (pre ctrl2) was omitted. C, All samples from the experiments shown in A were analyzed for the presence of citrullinated peptides by dot-blotting. Areas on the membrane where samples were not applied are marked as ×.

Next, we examined whether the cleaved fibrinogen fragments had been citrullinated by P gingivalis. Immunoblotting of all samples that contained protein bands (see Figure 3A) detected 2 citrullinated peptide bands at ∼8.5 kd, mapping to the amino-terminal region of the fibrinogen α-chain, in samples incubated with P gingivalis wild-type, but not Δppad or Δrgp (Figure 3B), confirming that Arg-X proteolytic cleavage of fibrinogen is a prerequisite for subsequent citrullination by PPAD. We further observed a weak positive signal in fibrinogen samples taken before incubation with P gingivalis and in the Δrgp sample taken at 1 minute, while the corresponding controls (the modification and the conjugate controls) were negative (Figure 3B), suggesting that purified human fibrinogen, as used in these experiments, is already endogenously citrullinated by human PADs.

The fact that the majority of fibrinogen had been degraded within minutes and that only 2 citrullinated proteins/peptides could be detected by immunoblotting suggested that the majority of generated peptides was smaller than the size limit of the peptide gels (∼3 kd) used for this analysis. Thus, we applied a dot-blot technique, which confirmed that the wild-type P gingivalis cells rapidly degraded and citrullinated fibrinogen into small citrullinated peptides (Figure 3C) and that both arginine gingipains and PPAD were required, since no positive signals were observed with the Δppad and Δrgp strains.

We then aimed to identify the amino acid sequence of the citrullinated fibrinogen peptides and determine the position of the citrulline residue. To this end, we fractionated the peptides derived from the wild-type and Δppad strains by HPLC and analyzed the eluted peak fractions by liquid chromatography tandem mass spectrometry. We identified a total of 30 peptides derived from fibrinogen (results not shown). The majority of the identified peptides were the product of proteolytic cleavage after either an arginine residue or a lysine residue, which is consistent with the results described above, but further proteolytic processing by other peptidases, particularly at the amino-terminus after glycine, alanine, and serine residues, was also evident. In samples incubated with P gingivalis wild-type, but not Δppad, we found 4 peptides that contained a carboxy-terminal citrulline residue (Figure 4): 1ADSGEGDFLAEGGGVCit16, 31PAPPPISGGGYCit42, 253GGSTSYGTGSETESPCit268, and 540ESSSHHPGIAEFPSCit554. Carboxy-terminal arginine-containing peptides were detected only in the Δppad, but not the wild-type, strain (Table 1), suggesting that citrullination of fibrinogen is tightly linked to cleavage by arginine gingipains. The combined data support the concept that target proteins such as human fibrinogen are cleaved by arginine gingipains, generating suitable peptide substrates for subsequent citrullination at the exposed carboxy-terminal arginine residue by P gingivalis PAD.

Figure 4.

Sequences of citrullinated peptides from human fibrinogen generated after incubation with Porphyromonas gingivalis. Amino acid sequences of human fibrinogen α-chain (Fib A; Swiss-Prot entry P02671) and fibrinogen β-chain (Fib B; Swiss-Prot entry P02675) are shown. Peptides were detected using liquid chromatography tandem mass spectrometry. Citrullinated peptides detected after incubation of fibrinogen with P gingivalis wild-type strain are underlined. Fibrinopeptides A and B and the thrombin cleavage sites are indicated. Cit = citrulline.

Table 1. Mass spectrometry of citrullinated peptides generated after incubation of human fibrinogen or α-enolase with Porphyromonas gingivalis wild-type strain and their respective arginated peptides generated with the Δppad strain*
Protein, sequenceAmino acid positionsP gingivalis strainm/z ratioMascot score
  • *

    Amino acids in the uncleaved proteins located carboxy–terminal and amino–terminal to the identified peptides are indicated to demonstrate the sites of proteolytic cleavage. Citrulline (Cit) residues that were identified are underlined. For the mass/charge (m/z) ratio, the numbers in parentheses are the peptide ion charge state.

Fibrinogen α-chain    
 −.ADSGEGDFLAEGGGVCit.G1–16Wild type769.29 (+2)56
 −.ADSGEGDFLAEGGGVR.G1–16Δppad768.77 (+2)99
Fibrinogen α-chain    
 R.GGSTSYGTGSETESPCit.N253–268Wild type546.91 (+3)51
 R.GGSTSYGTGSETESPR.N253–268Δppad786.82 (+2)78
Fibrinogen α-chain    
 K.ESSSHHPGIAEFPSCit.G540–554Wild type819.83 (+2)98
 S.SHHPGIAEFPSR.G543–554Δppad445.54 (+3)39
Fibrinogen β-chain    
 R.PAPPPISGGGYCit.A31–42Wild type585.30 (+2)26
 R.PAPPPISGGGYR.A31–42Δppad584.77 (+2)65
α-enolase    
 S.TGIYEALELCit.D41–50Wild type583.30 (+2)24
 S.TGIYEALELR.D41–50Δppad582.79 (+2)56

To further test this concept, we performed analogous experiments using recombinant human α-enolase. Similar to the findings with fibrinogen, α-enolase was rapidly degraded by the wild-type and Δppad strains and less so by the Δrgp strain (Figure 5A). Using immunoblotting on peptide SDS-PAGE gels (Figure 5B) as well as dot-blotting (Figure 5C), no citrullination could be detected. Analysis of the samples derived from the wild-type and Δppad strains by mass spectrometry revealed only 1 citrullinated peptide in the wild-type (41TGIYEALELCit50) (Figure 5D and Table 1), among a total of 17 peptides detected. The arginine-containing counterpart of this citrullinated peptide was detected in the samples incubated with Δppad. Analogous to fibrinogen, no peptides with carboxy-terminal arginine were detected in the wild-type samples. The proportion of peptides cleaved at residues other than arginine and lysine was higher than that found in fibrinogen, suggesting extensive cleavage by non–arginine/lysine-specific peptidases. Combined with the higher relative number of lysine residues in α-enolase (8.8% versus 6.9% in fibrinogen), this might result in the generation of short peptides, some of which might be citrullinated but would be too short to be detected using these methods. We therefore incubated P gingivalis Δkgp with α-enolase (Figures 5A–C) and detected 5 citrullinated peptides by mass spectrometry (Figure 5D), confirming that PPAD is able to citrullinate α-enolase peptides. Using the AMC antibody dot-blot, which relies on long peptides that are hydrophobic enough to bind to the membrane, we observed weak positive signals with the Δkgp strain, which decreased with time (Figure 5C), again suggesting extensive proteolytic degradation by other proteinases.

Figure 5.

Human α-enolase is rapidly cleaved by Porphyromonas gingivalis gingipains, and citrullinated peptides are detectable by mass spectrometry. A, Analogous to the experiments with fibrinogen shown in Figure 3, human α-enolase was incubated for 1 minute, 1.5 hours, 3 hours, or 6 hours with P gingivalis wild-type (WT) strain or with mutant strains lacking bacterial peptidylarginine deiminase (Δppad), arginine gingipain (Δrgp), or lysine gingipain (Δkgp), resolved by sodium dodecyl sulfate–polyacrylamide gel electrophoresis, and visualized using silver staining. An α-enolase sample prior to incubation with P gingivalis (pre) served as a control. B, All samples with visible protein bands from the experiments shown in A were tested for citrullination by immunoblotting with anti–modified citrulline (AMC). C, All samples from the experiments shown in A were analyzed for the presence of citrullinated peptides by dot-blotting. Areas on the membrane where samples were not applied are marked as ×. D, Citrullinated peptides detected by mass spectrometry after incubation of enolase with P gingivalis wild-type and Δkgp are underlined with a continuous line and with a dashed line, respectively. The amino acid sequence of human α-enolase (Swiss-Prot entry P06733) is shown. Cit = citrulline.

DISCUSSION

In the present study, we found evidence that the periodontal pathogen Porphyromonas gingivalis is an alternative source in the human host for generating citrullinated proteins and peptides. The underlying mechanism—proteolytic cleavage and subsequent citrullination at carboxy-terminal arginine residues—differs from that of the human PAD enzymes, which citrullinate internal arginine residues in whole proteins the most efficiently. This finding suggests that protein citrullination by the bacterial PAD has the potential to generate epitopes to which immunologic tolerance does not exist, not only due to the presence of foreign citrullinated proteins from the bacterium, but also through a foreign mode of proteolytic processing and posttranslational modification of host antigens. It also indicates that citrullination of bacterial proteins at internal arginines, as a potential mechanism for triggering autoantibodies via molecular mimicry (5), is more likely to be due to the action of human PAD enzymes that are present at the site of inflammation.

Of the 11 oral bacterial species tested, endogenous citrullinated proteins were detected exclusively in P gingivalis, indicating that a bacterial PAD gene is expressed or is active only in this bacterium among those tested. To substantiate this finding, we performed similarity searches using BLAST and PSI-BLAST (41). This revealed numerous ortholog distantly related to PPAD among prokaryotes, including several of the oral organisms we tested. Most share the predicted conserved catalytic residues of PPAD and other members of the guanidino group–modifying enzyme superfamily, although they most likely possess agmatine iminohydrolase or arginine deiminase, rather than peptidylarginine deiminase activity (42).

Using fibrinogen as a model antigen, we showed that P gingivalis rapidly generated small fibrinogen peptides with carboxy-terminal citrulline residues. Fibrinopeptide A, which normally results from thrombin cleavage of the fibrinogen α-chain after arginine-16, was also detected in its citrullinated form (1ADSGEGDFLAEGGGVcit16) in samples incubated with P gingivalis wild-type, but only in the native, arginine-containing form in the Δppad samples. It is known that P gingivalis gingipain–mediated degradation of human fibrinogen inhibits fibrinogen polymerization and results in the localized bleeding tendency that is typical of chronic periodontitis (35). A recent study showed that in intact fibrinogen, internal citrullination at arginine-16 by mammalian PAD impairs thrombin-catalyzed cleavage and fibrin polymerization (43), indicating at least 2 possible pathogenic roles of citrullinated fibrinogen in RA: serving as an autoantigen and disturbing the coagulation cascade and linked pathways.

The pathophysiologic role of fibrinogen peptides with carboxy-terminal citrulline residues, which are generated by the concerted action of gingipain and PPAD, is as yet unknown, thus opening up a novel area for future investigations. Similarly, it is known that P gingivalis arginine gingipains cleave a number of other human proteins, releasing biologically active peptides with important roles in immunity and inflammation, such as C5a (44) and bradykinin (45, 46), and simultaneous citrullination of these peptides by P gingivalis PAD might have a previously unappreciated role in human disease.

The lower levels of detectable citrullination of α-enolase peptides with the P gingivalis wild-type strain are likely to be the result of a combination of physiologic and technical factors. Enolase has a lower percentage of arginine residues (3.9%) as compared with fibrinogen (5.2%) and a higher percentage of lysine residues (enolase 8.8% versus fibrinogen 6.9%). It also appears to be more extensively cleaved by non-Arg/Lys peptidases. In combination, this would result in fewer suitable PPAD substrates, and very small peptides overall, which would not be detected with the methods used in the present study. Thus, lower levels of detectable citrullination may simply be due to a relative paucity of the substrate and technical shortcomings with the detection of short peptides.

Herein, we have demonstrated that P gingivalis efficiently citrullinates its own proteins and peptides from host fibrinogen and, to a lesser extent, α-enolase. The 2 major findings of this study—that proteolytic processing is required for citrullination by P gingivalis and that host peptides with exclusively carboxy-terminal citrulline residues are generated—provides a strong basis for future in vivo studies aimed at identifying citrullinated peptides at the site of gingival inflammation and exploring their potency for triggering a T cell and/or B cell response. Citrullinated host peptides generated by P gingivalis are likely to expose epitopes previously hidden from immune surveillance, which in the context of bacterial infection in a genetically susceptible host, may trigger an immune response. The slightly increased prevalence of anticitrullinated protein antibodies reported in patients with periodontitis as compared with healthy controls (47, 48) supports this concept, but the lower frequency and titer than are found in RA patients suggest that P gingivalis infection is not sufficient on its own for the mature autoimmune response. However, once tolerance is breached, we predict that exposure to host proteins in the inflamed joint, which have been citrullinated by human PADs (31), leads to intra- and intermolecular epitope spreading to additional peptides from the initiating proteins and other autoantigens.

We therefore propose a “two-hit” model of RA, based first on the breakdown of tolerance to specific citrullinated peptides generated by P gingivalis at the site of gingival inflammation and followed by epitope-spreading to other host citrullinated proteins in the inflamed joint. This self-sustaining immune response would then result in the chronic and destructive inflammation that typifies RA. The unique nature of the bacterial deiminase, along with its location on the cell surface of the bacterium (26), provides a target for treatment designed to prevent this otherwise incurable disease.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Venables had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Wegner, Wait, Sroka, Lundberg, Potempa, Venables.

Acquisition of data. Wegner, Wait, Sroka, Eick, Nguyen, Kinloch, Culshaw, Venables.

Analysis and interpretation of data. Wegner, Wait, Sroka, Potempa, Venables.

Acknowledgements

We are thankful to Drs. Andrzej Kozik, Maria Rapala-Kozik, and Anna Golda (Jagiellonian University, Krakow, Poland) for their help with the HPLC analysis and to Mr. Anto Jose (Glasgow Dental School, Glasgow, UK) for preparation of bacterial strains.

Ancillary