Involvement of Autoimmunity in the Pathogenesis of Aggressive Periodontitis

Introduction

Although the primary etiologic factor of periodontitis is the bacterial plaque biofilm, the factors that may determine the presence and rate of progression of the disease are complex and not fully understood. The potential involvement of autoimmunity in the pathogenesis of periodontitis has been considered intermittently for many years. Autoantibodies against a wide range of self-antigens, such as collagen type I (CI), host DNA, laminin, fibronectin, and desmosin, have been described previously (Anusaksathien and Dolby, 1991; Anusaksathien et al., 1992a,b; Govze and Herzberg, 1993). However, many of these reports have been inconsistent, some describing high levels of autoantibodies to CI (Ftis et al., 1986; Sugawara et al., 1992), while others report low levels of anti-CI (Hirsch et al., 1988). Recently, it has been suggested that autoimmune responses may be increased particularly in patients with AgP. De-Gennaro and co-workers (De-Gennaro et al., 2006) showed increased autoantibodies to fibronectin and laminin, but not to CI, in patients with AgP. In addition, Koutouzis and co-workers (Koutouzis et al., 2009), in a small pilot study, have shown increased tissue autoreactivity and anti-CI antibody titers specifically in patients with localized AgP.

The exact etiology of autoimmune reactivity in periodontitis is not known, but may be linked to the inflammatory process resulting from Porphyromonas gingivalis infection. Within the inflamed lesion, there is a high influx of immune cells that consume increased amounts of oxygen, thus leading to overproduction of reactive oxygen species (ROS) (Waddington et al., 2000). These include superoxide radical (O₂•−), hydroxyl radical (OH•), nitric oxide radical (NO•), peroxynitrate (ONOO–), and non-radical derivatives of oxygen, such as hydrogen peroxide (H₂O₂) and hypochlorous acid (HOCl) (Cheeseman and Slater, 1993). In addition, advanced glycation end-products (AGE) produced by sequential oxidative reactions (Cho et al., 2007) are known to be present in inflamed tissues (Bailey et al., 1995). The major protein components of the periodontal tissue are CI and CIII and thus are the major proteins that are susceptible to ROS attack. Post-translational modification of CI, CIII by ROS results in the formation of altered self-protein that may initiate an autoimmune response. Furthermore, breach of tolerance may also be initiated by enzymatic post-translational modification—for example, cleavage of extracellular proteins by matrix metalloproteases (Murphy and Nagase, 2008) or citrullination by peptidylarginine deiminase (PAD) (van Venrooij and Zendman, 2008). These two enzymatic post-translational modifications are of particular interest in periodontitis, since P. gingivalis produces both RgP-argingipain (Inomata et al., 2009) and peptidylarginine deiminase (PAD) (McGraw et al., 1999).

The aim of the present study was to test the hypothesis that patients with periodontal disease show autoimmune responses to native and post-translationally modified extracellular matrix proteins.

Materials & Methods

Patient Samples

Serum and stimulated saliva samples from AgP, CP, and G were collected from new patients attending the Periodontal Consultant Clinic, Institute of Dentistry, Barts & The London School of Medicine & Dentistry, according to clinical diagnostic criteria described in the Table, and as previously described (Hughes et al., 2006). Clinical diagnosis of Generalized AgP was performed according to World Workshop in Periodontology criteria (Armitage, 1999) as previously described (Hughes et al., 2006). Approval was granted by the City and East London Health Authority Local Research Ethics Committee (05/Q0601/80 & P/00/021). In addition, control sera from patients with rheumatoid arthritis (RA) (n = 12) from a previous study (Nissim et al., 2005) and with unknown periodontal disease status were tested. All participants provided written informed consent.

OPEN IN VIEWER

Table.

Clinical Data of the Three Patient Groups with Periodontal Disease

	N	Age (yrs; mean ± SD, range)	Males	Sites with PD < 5 mm	Plaque Score (mean ± SD)	Current Smoker
Gingivitis	18	38.3 ± 11.5 24 – 60	8 (43%)	0	64 ± 45%	2 (11%)
Chronic PD	14	52.3 ± 10.06 43 – 62	7 (50%)	21 ± 18.4	70.1 ± 28.1%	3 (21%)
Aggressive PD	25	32.6 ± 4.8 22 – 39	12 (48%)	27 ± 15.4	49 ± 20%	5 (20%)

Twenty-five aggressive periodontitis (AgP), 14 chronic periodontitis (CP), and 18 gingivitis (G) patients were tested. AgP samples were from patients under the age of 40, having a minimum of 6 teeth with pockets deeper than 5 mm and a clinical diagnosis of Generalized AgP according to World Workshop in Periodontology criteria (Armitage, 2000). Patients with CP were over the age of 40, with a minimum of 2 teeth affected by pockets deeper than 5 mm. Gingivitis patients exhibited gingival bleeding and no sites deeper than 5 mm pocket depth. AgP samples were taken from a subset of patients from a previously reported study (Hughes et al., 2006). CP and G samples were collected separately.

Results

SDS-PAGE Analysis and 3-D Fluorescence Profile Study of ROS-modified CI and CIII

On SDS-PAGE, native CI showed two major electrophoretic bands that migrated to the region below the 150-kDa protein marker, corresponding to the constituent α₁ and α₂ chains of CI. A higher-molecular-weight band above the 225-kDa protein marker corresponded to the β chain of CI (Fig. 1A, lane 1). Glycation by ribose caused a shift in the CI α- and β-chain bands to a position of slightly higher molecular weight, as well as loss of intact CI. Hypochlorous acid, hydroxyl radical, and peroxynitrite treatments resulted in marked loss of most of the intact CI polypeptide, together with fragmentation presented as a light smear. The 3-D fluorescence profile study demonstrated major changes in the fluorescence profile of glycated CI in comparison with native CI (Fig. 1B). Minor changes were observed for peroxynitrate-treated CI, but not to HOCl and OH• modification (Appendix Fig. 1).

OPEN IN VIEWER

Figure 1.

SDS-PAGE analysis and 3-D fluorescence spectra of chemically modified collagen type I (CI) and collagen type III (CIII). Samples containing CI and CIII were incubated overnight at 37°C with or without the various free-radical generators. Panels A and C are 7.5% SDS-PAGE analysis under reducing conditions. The position of the molecular-weight markers (in kDa) is shown on the left. (A) Lane 1 shows native CI with two major electrophoretic bands that migrated to the region below the 150-kDa marker, corresponding to the constituent α₁ and α₂ chains of CI. A higher-molecular-weight band above the 225-kDa protein marker corresponds to the β chain of CI. Glycation by ribose caused a shift in the CI α- and β-chain bands to a position of slightly higher molecular weight, as well as loss of intact CI (lane 2). Hypochlorous acid, hydroxyl radical, and peroxynitrite treatments (lanes 3-5) resulted in marked loss of most of the intact CI polypeptide, together with fragmentation presented as a light smear. (B) The observed 3-D fluorescence spectra of CI demonstrated major changes in the fluorescence profile of glycated CI in comparison with native CI, but minor changes for peroxynitrate-treated CI. (C) SDS-PAGE analysis of CIII. (Lane 1) Native CIII with a major band that migrated to the region below the 150-kDa protein marker, corresponding to the constituent α-chain of CIII. While the effect of modification on the SDS-PAGE profile is not as significant (lanes 2-5), the 3-D profile showed dramatic change after treatment with ROS, with major changes observed for both glycated and peroxynitrate-treated CIII (D).

Native CIII showed one major band that migrated to the region below the 150-kDa protein marker, corresponding to the constituent α-chain of CIII (Fig. 1C). The effect of modification on the SDS-PAGE profile was not as extensive as that seen for CI. The 3-D profile showed major changes observed for both glycated and peroxynitrate-treated CIII (Fig. 1D, Appendix Fig. 1).

Antibody Binding to Modified CI and CIII in Patients with AgP, CP, G, and RA

Binding to both native and ROS-modified CI and CIII was observed only in the AgP samples, while no binding was observed in either CP or G serum samples (AgP vs. CP or G, p < 0.01) (Figs. 2a-2d). Although there was a trend to greater binding to native collagens compared with ROS-modified collagens, there was no significant difference in the binding of AgP sera samples to native CI or CIII vs. ROS-CI and ROS-CIII (Bonferroni corrected pairwise comparisons, p > 0.05). In addition, no significant binding was seen to the control antigens CII and BSA (Figs. 2e, 2f). Control sera from patients with RA showed no significant binding to either CI or CIII (Figs. 2c, 2d), but bound strongly to ROS-CII (Fig. 2e) as previously described (Nissim et al., 2005). In AgP, autoantibody titers to native collagen significantly correlated with autoantibody titers to ROS-modified collagen, although the correlation coefficients were relatively modest (R² values of 0.44, 0.83, 0.55, and 0.63 for Ribose, HOCL, OH•, and OONO-, respectively) (Appendix Fig. 2).

OPEN IN VIEWER

Figure 2.

Binding of serum samples to native and reactive oxygen species-modified CI and CIII. Serum samples from patients with aggressive periodontitis (AP, n = 25), chronic periodontitis (CP, n = 14), gingivitis (G, n = 18), and rheumatoid arthritis (RA, n = 12) were tested. AgP samples bound to both CI and CIII whether native or ROS-modified (a and b, respectively). Binding to CI and CIII was significantly higher in AgP in comparison with CP and G (*p < 0.01, c and d). In addition, control sera from patients with RA showed no significant binding to either CI or CIII, but bound strongly to CII in comparison with AgP, CP, and G (*p < 0.001, e). No significant binding was seen to BSA (f).

To test the specificity of the binding of autoantibodies to native CI in AgP, we established a cut-off point of 0.1 OD ELISA units post hoc, based on the data. Based on this cut-off point, detection of autoantibodies in AgP compared with CP had an Odds Ratio of 20.6 (95% CI 3.6-118), with sensitivity of 0.87 and specificity of 0.76 (Appendix Fig. 3).

Binding to Citrullinated Filaggrin-derived Peptides

Although anti-CCP is used for the diagnosis of rheumatoid arthritis, P. gingivalis is the only bacterium known to express a PAD enzyme (Wegner et al., 2010), suggesting that periodontitis could be associated with the production of citrullinated proteins. Out of 25 AgP serum samples, 2 had high anti-CCP autoreactivity, and another 4 samples had a borderline titer (Fig. 3). No reactivity was detected in either C or G samples.

OPEN IN VIEWER

Figure 3.

Binding to citrullinated filaggrin-derived peptides (CCP) in ELISA. The peptide is generated by post-translational modification (citrullination) of protein-bound arginine by peptidylarginine deiminase (PAD). P. gingivalis express a PAD, an enzyme that has been reported to be significantly associated with RA. We observed binding to CCP only in AgP serum samples; 2 had high anti-CCP autoreactivity, and another 4 samples had a borderline titer. No reactivity was detected in either CP or G samples.

Discussion

In the present pilot study, we tested the immune response in serum from three different groups of patients with periodontal diseases to two major components of the periodontal attachment structures, CI and CIII. In addition, we exposed CI and CIII in vitro to ROS that are involved in acute and chronic inflammation (Nissim et al., 2005). Although IL-1α levels were similar in all three groups, reflecting the ongoing inflammation, binding to both native and ROS-modified CI and CIII was observed only in the AgP samples, while no binding was observed in either CP or G serum samples (p < 0.01). Out of 25 AgP samples tested, 19 showed a positive reaction, and only 6 were negative to CI. In contrast, only 2 samples with CP and 1 with G reacted against CI. Similarly, only samples from AgP, but not CP and G samples, responded to CIII. As control target proteins we used CII, which is a known autoantigen in rheumatoid arthritis, as well as BSA, both of which did not show any reactivity with periodontal patients’ samples. The fact that no reactivity was seen against collagen type II, which is the major component of articular cartilage and a known autoantigen in rheumatoid arthritis, suggests that the autoreactivity toward CI and CIII in the AgP samples is diseased-tissue-specific. Hence, sera from patients with RA showed significant autoantibody titers to CII, but not to CI or CIII.

Analysis of autoantibody binding to ROS-modified collagens by Western blots suggests that some of the autoantibodies present bound specifically to ROS-modified epitopes (not shown). However, we did not see increased reactivity to ROS-modified CI or CIII. This is very different from our previous observation in RA samples, where reactivity to ROS-CII was higher than that to native CII (Nissim et al., 2005). Our findings appear particularly novel in three important respects: first, the demonstration that collagen autoantibody production is specific to the diseased tissue, since no reactivity was observed to CII, which is the major component of the articular cartilage; second, the autoreactivity was seen specifically in patients with generalized aggressive periodontitis; and third, the demonstration of autoantibody production to ROS-modified collagens. Our observation is supported by two recent studies, although their findings differ from the data presented here in both clinical detail and types of autoantibodies described (De-Gennaro et al., 2006; Koutouzis et al., 2009). Overall, this may possibly explain the previous inconsistent results (e.g., Ftis et al., 1986; Hirsch et al., 1988; Sugawara et al., 1992).

To address the possibility of additional aberrant post-translational modification of self-proteins, we measured the reactivity against citrullinated peptides. Citrulline is a non-standard amino acid generated by post-translational modification of arginine residues by peptidylarginine deiminase (PAD). The presence of antibodies directed toward citrullinated proteins in the serum of patients with RA has been described in detail. Anti-CCP can be detected in up to 80% of RA patients with 98% specificity (Zendman et al., 2004). We reasoned that citrullination might also be involved in breaching of tolerance in periodontitis, due to similar inflammatory processes in RA and periodontitis and also due to the fact that P. gingivalis is the only bacteria that is known to express PAD (Wegner et al., 2010). Analysis of our data demonstrated that only sera from AgP patients have autoreactivity against citrullinated peptide, while no reactivity was demonstrated in either C or G samples. Therefore, the breach of tolerance in AgP patients may also result from aberrant enzymatic activity.

Patients with AgP are reported to have a “hyper-inflammatory” phenotype (Armitage, 1999), and analysis of our data raises the question of whether the production of autoantibodies may be a primary pathogenic event in this condition, or whether the production of autoantibodies is secondary to other inflammatory events. However, it was notable that there were no significant differences among patient groups in salivary IL-1α concentrations, all of which were similarly elevated and which decreased post-initial periodontal treatment, confirming once more that IL-1α expression is associated with inflammation. In some respects, the homogeneity of the responses seen in AgP was surprising, given that AgP may be considered a clinical syndrome with different etiological factors playing a distinct role in different patients, such as different genetic and microbial factors (Armitage, 2000; Parkhill et al., 2000; Kamma et al., 2004; Haubek et al., 2008). The finding that the presence of autoantibodies may have specific utility as a diagnostic marker of AgP, based on a level of ELISA units set post hoc with this dataset, would benefit from further study with an independent dataset.

Overall, the results suggest that autoantibody production is seen specifically in patients with AgP and suggest a potentially very significant mechanism in the pathogenesis of this condition. In addition, the observation has potential utility for the diagnostic testing of patients with AgP. In view of the importance of this observation, and its variance from other reported studies, further investigation in an independent patient cohort is a high priority.